The role of maintenance therapy in acute promyelocytic leukemia in the first complete remission

  • Review
  • Intervention

Authors


Abstract

Background

Acute promyelocytic leukemia (APL) is the most curable type of leukemia. A consensus exists regarding the need for administration of both induction and consolidation treatments, albeit using different approaches. However, there is conflicting evidence for the role of maintenance treatment in APL patients.

Objectives

To examine the efficacy and safety of maintenance therapy in APL patients and to establish the optimal regimen for maintenance.

Search methods

We searched the Cochrane Central Register of Controlled Trials (CENTRAL) (The Cochrane Library 2012, Issue 6), MEDLINE (January 1966 to July 2012), LILACS (1982 to July 2012), relevant conference proceedings (2000 to 2012) and databases of ongoing and unpublished trials.

Selection criteria

Randomized controlled trials assessing maintenance treatment in patients with newly diagnosed APL in first complete remission (CR) following induction or induction and consolidation therapy.

Data collection and analysis

Two review authors assessed the quality of trials and extracted data. We estimated and pooled hazard ratios (HR) and risk ratios (RR) with 95% confidence intervals (CI) using the fixed-effect model. If significant heterogeneity was present we explored potential causes for such heterogeneity and if not found we used also the random-effects model.

Main results

We included 10 randomized controlled trials enrolling 2072 patients in the systematic review, and conducted meta-analysis on nine of them. There was no statistically significant effect on overall survival (OS) in the three main comparisons (HR for any maintenance treatment versus observation 0.79, 95% CI 0.49 to 1.27; HR for all-trans retinoic acid (ATRA)-based maintenance versus non-ATRA based maintenance 1.21, 95% CI 0.73 to 1.98; HR for ATRA alone maintenance versus ATRA and chemotherapy 0.99, 95% CI 0.69 to 1.43). However, disease free survival (DFS) was improved with any maintenance therapy compared to observation (HR 0.59, 95% CI 0.48 to 0.74; 5 trials, 1209 patients) and with ATRA and chemotherapy compared to ATRA alone maintenance (HR for ATRA alone compared to ATRA and chemotherapy 1.38, 95% CI 1.09 to 1.76; 4 trials, 1028 patients). DFS was probably improved with ATRA-based regimens compared to non-ATRA based regimens, but the effect did not reach statistical significance (HR 0.72, 95% CI 0.51 to 1.01; 4 trials, 670 patients). Analysis of clinically relevant adverse events could not be conducted due to paucity of data. Yet, increased reports of grade 3/4 adverse events were noted for any maintenance versus observation and for combined ATRA and chemotherapy versus ATRA alone treatment. The major limitation of this review lies in the variability between the included trials in both maintenance and pre-maintenance parameters. We tried to address this variability and to reduce its potential biases by conducting three separate main comparisons, as outlined above, leaving less statistical power to the presented results.

Authors' conclusions

Maintenance therapy compared to observation in APL patients improved DFS but not OS. Similarly, ATRA and chemotherapy compared to ATRA and probably ATRA based regimens compared to non-ATRA based regimens improved DFS but not OS. The significance of these findings is limited due to clinical heterogeneity between studies.

Résumé scientifique

Le rôle du traitement d'entretien dans la leucémie aiguë promyélocytaire dans la première rémission complète

Contexte

La leucémie aiguë promyélocytaire (LAP) est le type de leucémie le plus curable. Il existe un consensus au sujet de la nécessité d'administrer les deux traitements d'induction et de consolidation, bien qu'utilisant des approches différentes. Toutefois, les données concernant le rôle du traitement d'entretien chez des patients atteints de LAP sont contradictoires.

Objectifs

Examiner l'efficacité et la sécurité d'emploi du traitement d'entretien chez des patients atteints de LAP et établir le schéma posologique d'entretien optimal.

Stratégie de recherche documentaire

Nous avons effectué une recherche dans le registre Cochrane des essais contrôlés (CENTRAL) (The Cochrane Library 2012, numéro 6), MEDLINE (de janvier 1966 à juillet 2012), LILACS (de 1982 à juillet 2012), des actes de conférences pertinents (2000 à 2012) et les bases de données d'essais en cours et non publiés.

Critères de sélection

Les essais contrôlés randomisés évaluant le traitement d'entretien chez les patients atteints de LAP récemment diagnostiquée dans la première rémission complète (RC) suite au traitement d'induction ou d'induction et de consolidation.

Recueil et analyse des données

Deux auteurs de la revue ont évalué la qualité des essais et extrait des données. Nous avons estimé et regroupé les hazard ratios (HR) et les rapports de risque (RR) avec des intervalles de confiance (IC) à 95 % au moyen du modèle à effets fixes. Lorsqu'il y avait une hétérogénéité significative, nous avons exploré les causes potentielles de cette hétérogénéité et si elles n'étaient pas détectées, nous avons aussi utilisé le modèle à effets aléatoires.

Résultats principaux

Nous avons inclus 10 essais contrôlés randomisés totalisant 2 072 patients dans la revue systématique, et réalisé une méta-analyse sur neuf d'entre eux. Il n'y avait pas d'effet statistiquement significatif sur la survie globale (SG) dans les trois principales comparaisons (HR pour un traitement d'entretien comparé à l'observation 0,79, IC à 95 % 0,49 à 1,27 ; HR pour un traitement d'entretien à base d'acide all-trans-rétinoïque (AATR) comparé à un traitement d'entretien ne comprenant pas d'AATR 1,21, IC à 95 % 0,73 à 1,98 ; HR pour un traitement d'entretien à base d'AATR en monothérapie comparé à l'AATR et la chimiothérapie 0,99, IC à 95 % 0,69 à 1,43). Toutefois, la survie sans maladie (SSM) a été améliorée avec un traitement d'entretien comparé à l'observation (HR 0,59, IC à 95 % 0,48 à 0,74 ; 5 essais, 1 209 patients) et avec l'AATR et la chimiothérapie comparés à un traitement d'entretien à base d'AATR en monothérapie (HR pour l'AATR en monothérapie comparé à l'AATR et la chimiothérapie 1,38, IC à 95 % 1,09 à 1,76 ; 4 essais, 1 028 patients). La SSM a probablement été améliorée avec les schémas posologiques à base d'AATR comparés aux schémas posologiques ne comprenant pas d'AATR, mais l'effet n'a pas atteint une signification statistique (HR 0,72, IC à 95 % 0,51 à 1,01 ; 4 essais, 670 patients). L'analyse des événements indésirables cliniquement pertinents n'a pu être réalisée en raison de la pénurie de données. Cependant, des rapports plus nombreux d'événements indésirables de grade 3/4 ont été consignés pour un traitement d'entretien comparé à l'observation et pour l'association d'AATR et de chimiothérapie comparée à un traitement à base d'AATR en monothérapie. La limitation majeure de cette revue réside dans la variabilité entre les essais inclus dans tous les paramètres d'entretien et de pré-entretien. Nous avons essayé de résoudre cette variabilité et de réduire ses biais potentiels en effectuant trois principales comparaisons séparées, comme cela a été exposé ci-dessus, en accordant moins de puissance statistique aux résultats présentés.

Conclusions des auteurs

Un traitement d'entretien comparé à l'observation chez des patients atteints de LAP a amélioré la SSM mais pas la SG. De même, l'AATR et la chimiothérapie comparés à l'AATR en monothérapie et probablement les schémas posologiques à base d'AATR comparés aux schémas posologiques ne comprenant pas d'AATR ont amélioré la SSM mais pas la SG. La signification de ces résultats est limitée en raison de l’hétérogénéité clinique entre les études.

Plain language summary

Maintenance therapy for acute promyelocytic leukemia

Acute promyelocytic leukemia (APL) is a subtype of acute myelogenous leukemia (AML), comprising about 10% of AML cases. APL has a distinct therapeutic approach and is the most curable type of AML. It is characterized by a specific chromosome translocation, the t(15;17). The integrated effect of induction and consolidation in APL currently results in high long-term survival rates, reaching as high as 90% in some series. Contrary to the established role of initial treatment phases (termed induction and consolidation phases), in APL there is conflicting evidence regarding the place of a continued low-toxicity treatment (termed maintenance treatment) in these patients. This debate has become more intense after the introduction of all-trans retinoic acid (ATRA), which targets the PML-RARα fusion gene, and even more so with the recent addition of arsenic trioxide to the therapeutic options of APL.

We aimed to evaluate the effects of maintenance therapy on survival in APL patients as well as to assess its influence on other parameters, such as disease recurrence rate and adverse events. We also tried to establish the best maintenance regimen for these patients.

We therefore conducted a systematic review and meta-analysis of 10 randomized controlled trials including 2072 patients. Our main results showed that any maintenance treatment compared to observation prolongs the freedom from disease duration but not overall survival. Similarly, ATRA and chemotherapy compared to ATRA alone improves freedom from disease duration but not overall survival. Moreover, ATRA-based regimens compared to non-ATRA based regimens probably improves freedom from disease duration but not overall survival.

Finally, we showed that any maintenance treatment compared to observation as well as maintenance combining ATRA and chemotherapy compared to ATRA alone are more toxic, potentially limiting patient adherence to treatment. Our results imply that in patients with newly diagnosed APL, ATRA-based maintenance therapy may be added to the standard therapy in order to improve freedom from disease. Yet, one should bear in mind that this treatment does not improve overall survival and adds toxicity.

Our results are limited mainly by the diversity of trials in terms of maintenance regimens and treatments antedating maintenance administration. Additionally, quality of life (QOL) parameters were not reported, and therefore are worth evaluating in future trials since maintenance therapy has a direct impact on QOL.

Résumé simplifié

Traitement d'entretien de la leucémie aiguë promyélocytaire

La leucémie aiguë promyélocytaire (LAP) est un sous-type de leucémie myéloïde aiguë (LMA), représentant environ 10 % des cas de LMA. La LAP a une approche thérapeutique distincte et est le type de LMA le plus curable. Elle se caractérise par une translocation de chromosome spécifique, le t(15;17). L'effet intégré de l'induction et de la consolidation dans la LAP donne actuellement des taux de survie à long terme élevés, atteignant 90 % dans certaines séries de cas. Contrairement au rôle établi des phases de traitement initial (appelées phases d'induction et de consolidation) dans la LAP, il y a des preuves contradictoires concernant la place d'un traitement continu à faible toxicité (appelé traitement d'entretien) chez ces patients. Ce débat s'est intensifié après l'introduction de l'acide all-trans-rétinoïque (AATR), qui cible le gène de fusion PML-RARα, et d'autant plus encore avec la récente addition de trioxyde d'arsenic aux options thérapeutiques de la LAP.

Nous avons cherché à évaluer les effets du traitement d'entretien sur la survie chez des patients atteints de LAP ainsi qu'à évaluer son influence sur d'autres paramètres, tels que le taux de récidive de la maladie et les événements indésirables. Nous avons aussi essayé d'établir le meilleur schéma posologique d'entretien pour ces patients.

Nous avons par conséquent réalisé une revue systématique et une méta-analyse de 10 essais contrôlés randomisés incluant 2 072 patients. Nos principaux résultats ont montré qu'un traitement d'entretien comparé à l'observation prolonge la durée sans maladie mais pas la survie globale. De même, l'administration de l'AATR et d'une chimiothérapie comparativement à l'AATR en monothérapie améliore la durée sans maladie mais pas la survie globale. De plus, les schémas posologiques à base d'AATR comparés aux schémas posologiques ne comprenant pas d'AATR ont probablement amérlioé la durée sans maladie mais pas la survie globale.

En dernier lieu, nous avons démontré qu'un traitement d'entretien comparativement à l'observation ainsi qu'un traitement d'entretien associant l'AATR et la chimiothérapie comparativement à l'AATR en monothérapie sont plus toxiques, limitant potentiellement l'observance du traitement par le patient. Nos résultats impliquent que chez les patients atteints de LAP récemment diagnostiquée, le traitement d'entretien à base d'AATR peut être rajouté au traitement standard en vue d'améliorer la durée sans maladie. Cependant, on doit garder à l'esprit que ce traitement n'améliore pas la survie globale et augmente la toxicité.

Nos résultats sont limités principalement par la diversité des essais en termes de schémas posologiques d'entretien et de traitements anticipant l'administration d’entretien. En outre, les paramètres de qualité de vie (QdV) n'ont pas été rapportés, et méritent par conséquent d'être évalués dans de futurs essais puisque le traitement d'entretien a un impact direct sur la QdV.

Notes de traduction

Traduit par: French Cochrane Centre 22nd March, 2013
Traduction financée par: Instituts de Recherche en Sant� du Canada, Minist�re de la Sant� et des Services Sociaux du Qu�bec, Fonds de recherche du Qu�bec-Sant� et Institut National d'Excellence en Sant� et en Services Sociaux pour la France: Minist�re en charge de la Sant�

Summary of findings(Explanation)

Summary of findings for the main comparison. 
Any maintenance compared with observation for patients with APL in first complete remission

Patient or population: patients with APL in first complete remission

Settings: Patients diagnosed with APL following induction and consolidation treatments

Intervention: any maintenance

Comparison: observation

OutcomesIllustrative comparative risks* (95% CI)Relative effect
(95% CI)
No of Participants
(studies)
Quality of the evidence
(GRADE)
Comments
Assumed riskCorresponding risk
Any maintenanceObservation

Overall survival

[range of median follow-up: 2 years to 10.1 years]

Study population HR [1.07] ([0.23] to [4.86])[892]
([3])

⊕⊕⊝⊝

low

Moderate quality of evidence due to:

  • Relatively few events producing a wide confidence interval around the effect estimate

  • Heterogeneity among trials was significantly high

  

Disease-free survival

[range of median follow-up: 2 years to 10.1 years]

Study population HR [0.63] ([0.41] to [0.97])[1209]
([5])

⊕⊕⊕⊕

high

High quality of evidence due to:

  • Heterogeneity among trials was significantly high

  • However, effect was large (HR < 0.5) in 3 out of 6 trials

  

Adverse events requiring discontinuation

[range of median follow-up: 6.2 years to 7.8 years]

Study population RR [20.72] ([2.83] to [151.66])[906]
([3])
⊕⊕⊕⊝
moderate

Moderate quality of evidence due to:

  • Heterogeneity among trials was significantly high

  • Relatively few events producing a wide confidence interval around the effect estimate

  • However, effect was very large (RR > 5) in 2 trials.

[28] per 1000 [0] per 1000

Relapse rate

[range of median follow-up: 6.2 years to 10.1 years]

study population RR [0.74] ([0.50] to [1.11])[896]
([4])
⊕⊕⊕⊝
moderate

Moderate quality of evidence due to:

  • Heterogeneity among trials was significantly high

294 per 1000 488 per 1000
*The basis for the assumed risk (e.g. the median control group risk across studies) is provided in footnotes. The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI).
CI: Confidence interval; RR: Risk Ratio;HR: Hazard Ratio
GRADE Working Group grades of evidence
High quality: Further research is very unlikely to change our confidence in the estimate of effect.
Moderate quality: Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate.
Low quality: Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate.
Very low quality: We are very uncertain about the estimate.

Summary of findings 2

Summary of findings 2. 
ATRA-based maintenance compared with non-ATRA based maintenance for patients with APL in first complete remission

Patient or population: patients with APL in first complete remission

Settings: patients diagnosed with APL following induction and consolidation treatments

Intervention: ATRA-based maintenance

Comparison: non-ATRA-based maintenance

OutcomesIllustrative comparative risks* (95% CI)Relative effect
(95% CI)
No of Participants
(studies)
Quality of the evidence
(GRADE)
Comments
Assumed riskCorresponding risk
ATRA-based maintenancenon-ATRA-based maintenance

Overall survival

[range of median follow-up duration: 1.5 years to 10.1 years]

study population HR [1.21] ([0.73] to [1.98])[632]
([3])
⊕⊕⊕⊝
moderate

Moderate quality of evidence due to:

  • Relatively few events producing a wide confidence interval around the effect estimate

  

Disease-free survival

[range of median follow-up duration: 1.5 years to 10.1 years ]

study population HR [0.72] ([0.51] to [1.01])[670]
([4])
⊕⊕⊕⊕
High
 
  

Relapse rate

[range of median follow-up duration: 1.5 years to 10.1 years]

study population RR [0.88] ([0.59] to [1.30])[425]
([3])
⊕⊕⊕⊝
moderate

Moderate quality of evidence due to:

  • Relatively few events producing a wide confidence interval around the effect estimate

184 per 1000 189 per 1000
*The basis for the assumed risk (e.g. the median control group risk across studies) is provided in footnotes. The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI).
CI: Confidence interval; RR: Risk Ratio;;HR: Hazard Ratio
GRADE Working Group grades of evidence
High quality: Further research is very unlikely to change our confidence in the estimate of effect.
Moderate quality: Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate.
Low quality: Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate.
Very low quality: We are very uncertain about the estimate.

Summary of findings 3

Summary of findings 3. 
ATRA alone maintenance compared with ATRA and chemotherapy for patients with APL in first complete remission

Patient or population: patients with APL in first complete remission

Settings: Patients diagnosed with APL following induction and consolidation treatments

Intervention: ATRA alone maintenance

Comparison: ATRA and chemotherapy maintenance

OutcomesIllustrative comparative risks* (95% CI)Relative effect
(95% CI)
No of Participants
(studies)
Quality of the evidence
(GRADE)
Comments
Assumed riskCorresponding risk
ATRA alone maintenanceATRA and chemotherapy maintenance

Overall survival

[range of median follow-up: 3.9 years to 10.1 years]]

Study population HR [1.13] ([0.62] to [2.07])[1033]
[4]
⊕⊕⊕⊝
moderate

Moderate quality of evidence due to:

  • Relatively few events producing a wide confidence interval around the effect estimate

  

Disease-free survival

[range of median follow-up: 3.9 years to 10.1 years]

Study population HR [1.47] ([1.03] to [2.10])[1028]
[4]
⊕⊕⊕⊕
high
 
  

Subgroup analysis in pediatric patients: Overall survival

[range of median follow-up: 3.8 to 7.8 years]

Study population HR [2.05] ([0.47] to [8.82])[133]
[2]

⊕⊕⊝⊝

low

Low quality of evidence due to:

  • Relatively few patients and few events producing a wide confidence interval around the effect estimate

  • Heterogeneity among trials was moderately increased.

  

Subgroup analysis in pediatric patients: Disease-free survival

[range of median follow-up: 3.8 to 7.8 years]

Study population HR [2.17] ([1.09] to [4.31])[127]
[2]

⊕⊕⊝⊝

low

Low quality of evidence due to:

  • Relatively few patients and few events producing a wide confidence interval around the effect estimate

  • Heterogeneity among trials was moderately increased.

  
*The basis for the assumed risk (e.g. the median control group risk across studies) is provided in footnotes. The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI).
CI: Confidence interval; RR: Risk Ratio;HR: Hazard Ratio
GRADE Working Group grades of evidence
High quality: Further research is very unlikely to change our confidence in the estimate of effect.
Moderate quality: Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate.
Low quality: Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate.
Very low quality: We are very uncertain about the estimate.

Background

Description of the condition

Acute promyelocytic leukemia (APL) is a variant of acute myeloid leukemia (AML), which accounts for 5% to 10% of AML cases. As such, the disease is relatively uncommon, with an estimated annual incidence of 900 to 1000 new cases per year in the United States with equal male-to-female distribution. A higher incidence is reported among those of Latin American origin (Douer 2003). Further epidemiological data show that the incidence of APL increases during childhood until the age of 20 years and then plateaus. This observation, which stands in contrast to the increased incidence of AML with age, might be explained by the relatively few genetic mutations needed to induce this type of malignancy (Douer 2003). Indeed, one of the distinctive hallmarks of APL is a translocation involving the retinoic acid receptor-alpha (RAR-α) locus on chromosome 17 and the PML gene on chromosome 15 (Tallman 2009). This translocation, crucial but not sufficient to induce a leukemic phenotype, dysregulates differentiation and self-renewal of myeloid progenitor cells and confers resistance to apoptosis (de Thé 2010). An important clinical utility of this translocation is confirmation of the diagnosis of APL by cytogenetics (either karyotype or fluorescence in situ hybridization (FISH) or molecular reverse transcription-polymerase chain reaction (RT-PCR) methods demonstrating this translocation). Furthermore, this unique translocation enables the physician to confirm whether the patient has achieved remission following treatment. It may also enable the monitoring of high-risk patients for relapse during the follow-up period, after completion of all treatment phases.

Accordingly, the treatment scheme for APL is distinctively different from the standard treatment for AML in several aspects. One of the more significant differences is the use of two agents, all-trans retinoic acid (ATRA) and arsenic trioxide (ATO), as part of the treatment plan at different phases. These two agents target the PML-RARα fusion gene and have revolutionized the treatment and prognosis of this type of leukemia, which is considered today the most curable type of AML.

Two treatment phases are well validated in APL, the:

  • remission induction phase which aims to achieve complete remission (CR);

  • consolidation (post-remission) phase directed at achieving a durable molecular remission and eventually cure. 

The vast majority of patients with newly diagnosed APL are given induction therapy that incorporates ATRA with anthracycline-based chemotherapy. This approach results in hematological complete remission rates of over 90% at the end of the induction phase. Despite the high complete remission rate at the end of induction, patients have a high chance of relapse if consolidation is not administered. Thus, with the integration of the induction remission and consolidation phases, the long-term survival rates for APL are now exceeding 70% in most clinical trials (Sanz 2005), reaching as high as 90% in some series (Hu 2009; Liu 2010).

A risk-assigned approach has been developed by some authors, stratifying the risk of relapse in accordance with clinical and biological features at presentation (Sanz 2003). The PETHEMA group suggests that a white cell count at presentation higher than 10 X 109/L is the strongest predictor for both hematological and molecular relapse (Sanz 2000).

Description of the intervention

Maintenance therapy, a treatment phase that follows the consolidation period and aims to prevent disease recurrence, is not an integral part of the standard treatment for AML. This fact might explain the scarce data regarding maintenance treatment in APL before the introduction of ATRA. In the pre-ATRA era, only two small retrospective studies (Kantarjian 1987; Marty 1984) addressed the issue of maintenance therapy in APL, showing a superior disease free survival (DFS) for patients with APL receiving low-dose 6-mercaptopurine (6-MP) and methotrexate (MTX) compared to no maintenance.

However, with the advent of ATRA as part of the treatment for APL, and the subsequently improved long-term survival, the issue of maintenance therapy in APL has been raised. This subject is particularly relevant for patients achieving complete molecular remission at the end of intensive consolidation chemotherapy with reverse transcription-PCR negativity for the PML-RARα transcript (Sanz 2009), who on the one hand might benefit from maintenance therapy but on the other hand might not need it since they are less prone to relapse (Grimwade 2002).

Maintenance therapy may include any combination of ATRA, chemotherapy and ATO for different time periods. Several randomized controlled trials have addressed the role of maintenance therapy in APL. These trials evaluated maintenance therapy (either ATRA-based, chemotherapy-based or a combination of both) compared to observation. While some trials have shown superiority of maintenance therapy, either ATRA or chemotherapy-based (Adès 2010; Tallman 2002), in terms of either DFS or overall survival (OS), others could not shown a statistically significant effect (Avvisati 2002; Avvisati 2011) or even demonstrated a negative impact of maintenance (Asou 2007) on OS.

The type of anthracycline used during the induction and consolidation phases might also influence the need for maintenance. For example, trials applying daunorubicin as the anthracycline showed superiority for the incorporation of maintenance, while those using idarubicin could not prove it. These differences in the need for maintenance based on the anthracycline used contributed to the confusion regarding the role of maintenance treatment in APL (Park 2011).

How the intervention might work

On a molecular basis, ATRA exerts its anti-leukemic effect through differentiation of the leukemic cells upon exposure to this agent (Tallman 2009). Moreover, ATRA is potentially responsible for induction of caspase-like activity that degrades the PML-RARα fusion transcript (Douer 2000). Similarly, ATO exerts its action on the leukemic cells by direct degradation of the PML-RARα fusion transcript, inducing predominantly apoptosis but also differentiation of the leukemic cells (Tallman 2009). Therefore, the anti-leukemic effect of each of these agents if administrated as maintenance in remission, and given separately or in combination with conventional chemotherapy, may prevent re-expansion of the quiescent APL leukemic clone and as a consequence prevent a clinical relapse. A trial which followed 33 out of 100 APL patients throughout the maintenance phase demonstrated a PML-RARα transcript reduction starting during the induction phase, continuing throughout the consolidation phase and becoming undetectable during the maintenance phase (Yao 2008), suggesting a biological  effect of maintenance. 

On clinical grounds, as maintenance therapy might preserve clinical (and even molecular) remission, the consequences of omitting maintenance therapy could eventually be translated into disease relapse, necessitating salvage therapy with high-dose chemotherapy with or without autologous or allogeneic hematopoietic cell transplantation. Thus, omitting maintenance therapy might eventually result in a negative effect on OS for patients with APL.

Why it is important to do this review

With the advent of differentiating agents (that is ATRA, ATO) APL has become the most curable subtype of AML. A consensus exists regarding the administration of both induction and consolidation treatments, albeit using different approaches. However, there is conflicting evidence for the effect of maintenance treatment on survival and on DFS in APL patients, making it a controversial treatment. We have decided to conduct a systematic review and meta-analysis to establish the role of maintenance therapy in APL patients achieving complete remission. Furthermore, as the disease is highly curable and disease relapse risk depends on patients' individual risk score at presentation, it seems necessary to define which subgroups of patients might benefit most from maintenance therapy.

Objectives

Primary objective:

to examine the efficacy and safety of maintenance therapy in terms of overall survival (OS), disease-free survival (DFS) and adverse effects for patients with APL.

Secondary objective:

to compare different protocols for maintenance therapy in APL and to establish the optimal protocol in terms of efficacy and safety.

Methods

Criteria for considering studies for this review

Types of studies

Randomized controlled trials, irrespective of publication status, year of publication and language. We excluded from the review quasi-randomized trials and cross-over trials.

Types of participants

Patients with newly diagnosed APL in complete remission (hematological or molecular) following either induction or induction and consolidation therapy. Randomization to maintenance could be performed before induction or after the achievement of complete remission, either after induction or induction and consolidation treatments. Both pediatric and adult patients were included.

Types of interventions

We evaluated three main comparisons, as follows.

Comparison I
  • Investigational intervention arm: any maintenance (ATRA, chemotherapy, ATO or combination of any of the above)

  • Control arm: observation, i.e. no maintenance

We evaluated different maintenance strategies as follows.

Comparison II
  • Investigational intervention arm: ATRA-based maintenance

  • Control arm: non-ATRA based maintenance

Comparison III
  • Investigational intervention arm: ATRA alone maintenance

  • Control arm: ATRA in combination with any chemotherapy maintenance

Types of outcome measures

We based definitions of measured outcomes on the revised recommendations of the National Cancer Institute (NCI) for diagnosis, response criteria and treatment outcome in AML (Cheson 2003).

Primary outcomes
  • Overall survival (OS), defined as the time from enrollment to maintenance till death from any cause. If a trial reported on OS from study entry or from any time point after the diagnosis of APL, and not from randomization to maintenance, we did not exclude it from the meta-analysis.

  • Disease-free survival (DFS), defined as the time from enrollment to maintenance until relapse or death.

Secondary outcomes
  • Treatment-related mortality.

  • Hematological relapse.

  • Molecular relapse.

  • Central nervous system (CNS) relapse.

  • Rate of patients undergoing hematopoietic cell transplantation.

  • Quality of life.

  • Adverse events (a comparison was done only between two active treatment arms):

    • adverse events requiring discontinuation of therapy,

    • grade 3/4 adverse events,

    • grade 3/4 hematologic toxicity,

    • APL differentiation syndrome (formerly ATRA syndrome).

Search methods for identification of studies

Electronic searches

We searched the following databases in accordance with the search strategy outlined in the Cochrane Handbook for Systematic Reviews of Interventions (Lefebvre 2011):

  • Cochrane Central Register of Controlled Trials (CENTRAL) (The Cochrane Library 2012, Issue 6) (see Appendix 1);                       

  • MEDLINE (1966 to 8 July 2012) (through PubMed) (see Appendix 2);                      

  • LILACS (1982 to 8 July 2012) (see Appendix 3).

We crossed the term 'acute promyelocytic leukemia' and similar OR 'APL' and similar with the term 'maintenance' and similar (see Appendix 1; Appendix 2; Appendix 3 for the detailed search strategies).

We combined the search terms with the highly sensitive search strategy for identifying reports of randomized controlled trials (Robinson 2002) in the MEDLINE search.

Searching other resources

We searched the conference proceedings of the American Society of Hematology (2000 to 2011), American Society of Clinical Oncology Annual Meeting (2000 to 2011) and the European Hematology Association (2000 to 2011) for relevant abstracts.

We searched databases of ongoing and unpublished trials: http://www.controlled-trials.com; http://www.clinicaltrials.gov/ct; http://clinicaltrials.nci.nih.gov.

We contacted the first or corresponding author of each included study and the researchers active in the field for information regarding unpublished trials or complementary information on their own trial.

We checked the citations of included trials and major reviews for additional studies.

Data collection and analysis

Selection of studies

We performed selection of studies in accordance with the guidelines in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011b). Two review authors (LV, EM) inspected the title and, when available, the abstract of each reference identified in the search and applied the inclusion criteria. Where relevant articles were identified, we obtained the full article and the two review authors inspected it independently.

We included trials regardless of publication status, date of publication and language.

We used a PRISMA flow diagram to report the numbers of identified records, excluded records and articles and included studies (Moher 2009).

Data extraction and management

We conducted data extraction according to the guidelines proposed in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011c), with two review authors (LV, EM) independently extracting the data from included trials. We discussed any disagreement between the two review authors. If the disagreement remained unresolved a third review author (AG) independently extracted the data. We discussed the data extraction, documented disagreements and their resolution and, where necessary, contacted the authors of the studies for clarification. If this was unsuccessful, we reported disagreements.

We documented the justification for excluding studies from the review. We collected all data on an intention-to-treat basis, where possible.

We extracted, checked and recorded the following data.

1.   Characteristics of trials
  • Author, title, source, publication date, country, language, duplicate publications

  • Publication status: published, published as abstract, unpublished

  • Year (defined as recruitment initiation year) and country or countries of study

  • Trial sponsor (academic, industrial)

  • Intention-to-treat analysis: performed, possible to extract, efficacy analysis

  • Design and possible sources of bias (method of allocation generation and concealment, blinding (participants personnel, outcome assessors), incomplete outcome data, selective outcome reporting, other sources of bias)

  • Duration of study follow-up

  • Response definition, event definitions

  • Case definitions used (inclusion and exclusion criteria)

  • Assessment of mortality (primary outcome, secondary outcome, safety)

  • Study endpoints as reported in the methods section of each trial

2.   Characteristics of patients
  • Number of participants in each group

  • Age (mean and standard deviation)

  • Gender

  • Ethnicity

  • Participants lost to follow-up

  • Histological subtype, classical M3, microgranular variant (vM3)

  • Additional diagnoses

  • Type of remission confirmation (molecular, morphological)

  • Number and type of previous induction and consolidation cycles

  • Risk stratification group according to white blood cell count at presentation: below or equal to 10 X 109/L, above 10 X 109/L

  • Performance status

3. Characteristics of interventions
  • Experimental intervention:

    • dose, number of administered doses and total duration of therapy,

    • additional drugs (combination).

  • Treatment of control group:

    • observation.

  • Maintenance: regimen, dose, number of administered doses and total duration of therapy.

4. Characteristics of outcome measures (extracted for each group and total events)
  • Overall survival (OS):

    • number of patients available for survival analysis at the end of follow-up,

    • hazard ratio (HR) of OS and its standard error (SE), confidence interval (CI) or P value,

    • crude mortality at the end of follow-up (used for dichotomous all-cause morality analysis if over 50% of the included trials lacked sufficient time-to-event data),

    • Kaplan-Meier curve (yes or no).

  • Disease-free survival (DFS).

  • Number of patients with relapsed disease.

  • HR of progression-free survival and its SE, CI or P value.

  • Number of patients achieving complete response.

  • Adverse effects (grade 3 and 4, requiring discontinuation of treatment, infection-related).

  • Number of patients excluded from outcome assessment after randomization and the reasons for their exclusion.

  • Quality of life (scale and score).

The following parameters for OS, DFS and grade 3 or 4 adverse events are described in the 'Summary of findings for the main comparison'.

  • Point estimate and 95% CI.

  • Number of participants and trials.

  • Risk of event in control and experimental groups.

  • Quality of evidence (GRADE), reasons for downgrading and upgrading.

Assessment of risk of bias in included studies

Two review authors (LV, EM) independently and individually assessed the trials for methodological quality. Allocation concealment, sequence generation, blinding, incomplete outcome data and selective outcome reporting according to The Cochrane Collaboration's tool for assessing bias (Higgins 2011) were described and assessed. We resolved any disagreement by discussion. If disagreement persisted, a third review author (AG) extracted the data independently. We discussed data extraction, document disagreements and their resolution and, where necessary, we contacted the authors of the studies for clarification.

Measures of treatment effect

We measured risk ratio (RR) for dichotomous data and hazard ratio (HR) for time-to-event outcomes. We estimated standardized mean difference (SMD) for quality of life assessment.

Unit of analysis issues

For dichotomous outcomes, the unit of analysis was the number of participants assigned to the investigational arm and the number of participants assigned to the control arm. For continuous outcomes, the mean, standard deviation and the number of participants in the investigational and control arms were the unit of analysis.

For time-to-event outcomes we used log HR and the standard error of log HR as the unit of analysis (Deeks 2011).

Dealing with missing data

We used methods to deal with missing data as outlined in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011a). Briefly, we contacted authors of the included trials in order to obtain additional information not reported in the publications of the trials.

If not all randomized patients were analyzed we made assumptions on the missing data as to whether they were randomly or non-randomly missing or could not be categorized. We made assumptions based on the reports of the reasons for missing data and the distribution between allocated groups. If such data were not provided we categorized the missing data as unclear. We performed sensitivity analysis of the primary outcome based on these assumptions to detect the impact of assumption modifications on the results.

For dichotomous parameters we imputed missing data for patients who were lost to follow-up after randomization, assuming a poor outcome (worst-case scenario) for missing individuals. We performed sensitivity analysis of the primary outcome, without trials with above 20% of participants lost to follow-up.

In addition, we addressed the effect of missing data on outcomes in the Discussion section of the review.

Assessment of heterogeneity

We assessed the heterogeneity of the trials' results with the Chi2 test of heterogeneity and the I2 statistic of inconsistency. Statistically significant heterogeneity was defined as a P value of less than 0.1 or an I2 statistic greater than 40%. We explored potential sources of heterogeneity through subgroup stratification by age, type of induction treatment, disease-risk groups, allocation concealment, blinding and size of studies (Deeks 2011).

Assessment of reporting biases

We aimed to inspect the funnel plot of the treatment effect against the precision of trials (plot of the log of the RR for efficacy against the standard error) in order to estimate potential asymmetry that may indicate selection bias (the selective publication of trials with positive findings) or methodological flaws in the small studies (Sterne 2011). However, due to low numbers of trials included in each comparison, this methodology evaluation was not performed as the test is underpowered in these circumstances.

Data synthesis

We entered data into the Cochrane statistical package Review Manager 5 (RevMan 5) (RevMan 2011). For dichotomous data, we estimated risk ratios (RR) and their CIs using the Mantel-Haenszel method. We pooled log HR for time-to-event outcomes using an inverse variance method. If the HR and its SE (or CI) were not reported, we estimated ‘O – E’ and ‘V’ statistics indirectly using the methods described by Parmar 1998 and Tierney 2007.

In order to better deal with the heterogeneity between the included trials in both pre-maintenance and maintenance therapies, we created three main comparisons, as stated above, to group trials as similarly as possible.

In all comparisons, we used the fixed-effect model. However, in the case of significant heterogeneity (I2 > 40%) we tried to explore heterogeneity by subgroup and sensitivity analyses as outlined in the subsequent paragraphs. If we could not explain the reasons for heterogeneity we also used the random-effects model and reported results accordingly.

Subgroup analysis and investigation of heterogeneity

We explored potential sources of heterogeneity through stratifying the patient subgroups given below; allocation concealment; blinding and size of studies.

For the primary analysis, we performed subgroup analysis according to:

  • age (younger adults, age < 60 years; older adults, age ≥ 60 years);

  • pediatric patients (age < 15 years) versus adult patients (age ≥ 15 years);

  • risk stratification group according to white blood cell count at presentation (low risk, below or equal to 10 X 109/L; high risk, above 10 X 109/L);

  • type of prior induction and consolidation therapy (ATRA-containing treatment, daunorubicin versus idarubicin-containing treatment, cytarabine-containing treatment);

  • type of maintenance (ATRA alone, ATRA-containing regimen, ATO-containing regimen) compared with no maintenance.

We formally assessed differences between subgroups using the Chi2 test for differences between subgroups (Deeks 2001).

Sensitivity analysis

We performed sensitivity analyses using the method of allocation concealment (Schulz 1995); blinding (patients, caregivers and assessors); allocation generation; incomplete outcome data (adequately, inadequately addressed); selective reporting (Higgins 2011); time definition for OS and DFS measures (randomization to maintenance, other time point) and the type of publication (full paper, abstract, unpublished).

Results

Description of studies

Results of the search

A total of 1273 potentially relevant titles and abstracts were screened. At initial screening, 1193 references were excluded. Eighty full text articles were retrieved for a detailed evaluation. We retrieved an additional six abstracts from conference proceedings. Three ongoing trials were identified; none have currently reported results (Figure 1).

Figure 1.

Flow diagram.

Included studies

Ten trials (42 publications) enrolling 2072 patients, recruited between the years 1989 to 2005, were eligible for this systematic review. Meta-analysis was conducted on nine of them. Nine trials were published as full text and one trial was published as abstract only (Feusner 2010).

Each trial randomized between 60 and 586 patients. All but two trials (Lin 2007; Shen 2004) were multicenter trials. The median follow-up period ranged from to 1.5 to 10 years (as specified in Characteristics of included studies).

Type of patients

All trials evaluated patients with newly diagnosed APL. Two trials excluded children (Asou 2007; Powell 2010). Median age at study entry ranged between 33 and 48 years (range 1 to 81 years). In all trials APL was confirmed by the demonstration of t(15;17) or by the presence of the PML-RARA transcript, or both. Major organ dysfunction (that is heart, lung, kidney, liver) was an exclusion criterion in four trials (Asou 2007; Avvisati 2002; Avvisati 2011; Tallman 2002).

Study design

The treatment scheme in all included trials consisted of induction, consolidation and maintenance phases according to the various protocols (see in 'Additional tables': Table 1 for maintenance protocols; Table 2 for pre-maintenance treatment scheme). Detailed protocols are given in the Characteristics of included studies section). In six trials (Asou 2007; Avvisati 2002; Avvisati 2011; Feusner 2010; Lin 2007; Parovichnikova 2004) randomization to maintenance treatment of patients in complete remission was done after completion of the induction and consolidation phases. In three trials (Adès 2010; Powell 2010; Tallman 2002) two randomizations were carried out, the first one before the induction or consolidation phases and the second one before maintenance. In one trial (Shen 2004) randomization was carried out before induction initiation and patients were randomized to three different induction and maintenance treatments hitherto receiving the same consolidation treatment.

Table 1. Maintenance protocols of the included trials
Trial Intervention Comparator Duration Notes
Adès 2010

One of the following arms:

# ATRA (76 patients): 45mg/m2 per day for 15 days every 3 months

# MTX+6-MP (117 patients): 6-MP 90mg/m2 per day, MTX 15 mg/m2 per week

# ATRA+MTX+6-MP (129 patients):combination of the above

# observation (79 patients)

Each of the 4 arms was compared with the other arms2 years 
Asou 2007

Intensive chemotherapy (89 patients)

consisting of 6 cycles:

# cycle 1:behenoyl Ara-C (BHAC) (170 mg/m2, days 1 through 5), daunorubicin (30 mg/m2, days 1 and 4) and 6-mercaptopurine ( 70 mg/m2, days 1 through 7).

# cycle 2:BHAC and mitoxantrone (5 mg/m2, days 1 and 2).

# cycle 3: BHAC, etoposide (80 mg/m2, days 1, 3, and 5), and vindesine (2 mg/m2, days 1 and 8)

# cycle 4:BHAC, aclarubicin (14 mg/m2, days 1 through 4), and 6MP

# cycle 5:same as cycle no. 1

# cycle6:same as cycle no. 3

Observation (86 patients)6 cycles, span of 6 weeks between cycles 
Avvisati 2002 MTX+6-MP (58 patients): 6-MP 1mg/kg daily, MTX 0.25mg/kg weekly Observation (58 patients)2 years 
Avvisati 2011

One of the following:

# ATRA (83 patients before protocol amendment + 137 patients following protocol amendment ): 45mg/m2 per day for 15 days every 3 months

# MTX+6-MP (78 patients): 6-MP 90mg/m2 per day MTX 15 mg/m2 per week

# ATRA+MTX+6-MP (81 patients before protocol amendment + 131 patients following protocol amendment):alternating ATRA for 15 days with MTX+6-MP for 3 months.

# observation (76 patients)

Each of the 4 arms was compared with the other arms2 years

in the mid-course of the trial an amendment was made in the maintenance protocol and patients were randomized into the 2 following maintenance arms with no observation arm:

# ATRA

# ATRA+MTX+6-MP

Feusner 2010 ATRA (32 patients): 45mg/m2 for 7 days every other week ATRA+6-MP+MTX (38 patients): ATRA 45mg/m2 for 7 days every other week; 6-MP 60mg/m2 a day; MTX 20 mg/m2 a week1 yearsame protocol as Powell 2010
Lin 2007 ATRA+As4S4+6-MP+MTX (30 patients): ATRA 25mg/m2/d for one month; As4S4 5 tablets tid for 1 month; 6-MP 50 mg 2-3 daily for one month and MTX 30-40 mg once a week for 4 weeks ATRA+6-MP+MTX (30 patients): same dosing as in intervention2-3 years or more 
Parovichnikova 2004

Chemotherapy alone arm (34 patients):

Rotation of 5-days of cytarabine (100 mg/m2 bid) with one of the following: daunorubicin (45 mg/m2 for 2 days, up to a total dose of 650 mg/m2) or 6-MP (50mg/m2 for 5 days) or cyclophosphamide (800 mg/m2 IV 1 day) every 4 weeks

ATRA and chemotherapy (34 patients): Alternate same chemotherapy as in intervention arm (but half dose) with ATRA 45 mg/m2 for 5 days every 4 weeks2 years 
Powell 2010 ATRA (166 patients): 45mg/m2 for 7 days every other week ATRA+6-MP+MTX (165 patients): ATRA 45mg/m2 per day for 7 days every other week; 6-MP 60mg/m2 per day; MTX 20 mg/m2 per week1 year 
Shen 2004

One of the following:

# arm 1: ATRA+6-MP/MTX (19 patients): ATRA, 25 mg/m2 per day for 30 days; then 6-MP100 mg day for 30 days or 15 mg of MTX once a week for 4 weeks

# arm 2: ATO+6-MP/MTX (18 patients):As2O3, 0.16mg kg per day for 30 days; then 6-MP100 mg day for 30 days or 15 mg of MTX once a week for 4 weeks

# arm 3: ATRA+ATO+6-MP/MTX (20 patients): ATRA 25 mg/m2 per day for 30 days; then As2O3 0.16 mg kg per day for 30 days; then 6-MP100 mg day for 30 days or 15 mg of MTX once a week for 4 weeks

Each of the 3 arms was compared with the other 2 arms5 cycles (10 months for arm 1 or 2 ;15 months for arm 3).randomization was used for both induction and maintenance
Tallman 2002 ATRA (99 patients): 45mg/m2 daily Observation (105 patients)1 year 
Table 2. Pre-maintenance treatment scheme of the included trials
Trial Induction treatment consolidation treatment No. of consolidation cycles Comments
Adès 2010ATRA plus cytarabine and daunorubicin

Cytarabine plus daunorubicin

 

2Induction treatment was stratified by age and initial white cell count
Asou 2007

Stratified according to initial WBC count:

# Group A: ATRA alone

# Group B: ATRA+idarubicin+cytarabine

# Group C: ATRA+intensified idarubicin+cytarabine

# First cycle: mitoxantrone and cytarabine.

# Second cycle: cytarabine plus etoposide and daunorubicin.

# Third cycle: cytarabine.

3 
Avvisati 2002Idarubicin plus cytarabine

# First cycle: idarubicin and cytarabine

# Second cycle: mitoxantrone plus VP-16

# Third cycle: cytarabine and 6-TG

3 
Avvisati 2011Same as in Avvisati 2002Same as in Avvisati 20023 
Feusner 2010

ATRA plus cytarabine and daunorubicin

 

ATRA and daunorubicin

 

3 
Lin 2007ATRA

# 2 courses of daunorubicin and cytarabine

# 2 courses of homoharringtonine and cytarabine

# 2 courses of aclarubicin and cytarabine

6 
Parovichnikova 2004ATRA plus cytarabine and daunorubicin

daunorubicin plus cytarabine

 

2 
Powell 2010ATRA plus cytarabine and daunorubicin

ATRA and daunorubicin

 

3Following induction patients were randomized to receive or not receive 2 cycles of As2O3
Shen 2004

Three groups according to randomization:

# ATRA alone

# As2O3 alone

# Combination of ATRA and As2O3

 

# First cycle: daunorubicin and cytarabine

# Second cycle: intensive cytarabine

# Third cycle: homoharringtonine and cytarabine

3 
Tallman 2002Patients were randomized to ATRA until CR or 1 to 2 cycles of daunorubicine plus cytarabineDaunorbicine and cytarabine2 

In six trials (Adès 2010; Avvisati 2002; Feusner 2010; Powell 2010; Shen 2004; Tallman 2002) assessment of remission achievement was done morphologically, while in four trials (Asou 2007; Avvisati 2011; Lin 2007; Parovichnikova 2004) assessment of remission was done either molecularly or by morphology.

Intervention
Maintenance protocols

Seven trials compared two arms of maintenance (Asou 2007; Avvisati 2002; Feusner 2010; Lin 2007; Parovichnikova 2004; Powell 2010; Tallman 2002), one trial compared three arms (Shen 2004) and two trials compared four arms (Adès 2010; Avvisati 2011) (Table 1).

Types of maintenance protocols varied among trials and the following comparisons were included:

Type of anthracycline used during induction and consolidation

All included trials employed anthracycline at the induction or consolidation, or both, phases of treatment. Daunorubicin (Adès 2010; Feusner 2010; Lin 2007; Parovichnikova 2004; Powell 2010; Shen 2004; Tallman 2002) and idarubicin (Asou 2007; Avvisati 2002; Avvisati 2011) were usually administered. Mitoxantrone was given in three trials (Asou 2007; Avvisati 2002; Avvisati 2011) and aclarubicin was used in one trial (Lin 2007). For the purpose of comparing the anthracycline dose intensity prior to maintenance treatment, we converted the cumulative dose of anthracyclines in each trial into doxorubicin dose equivalents. The type of anthracycline used during the induction and consolidation phases and the cumulative dose in each trial, expressed in daunorubicin equivalents, are described in Table 3. We used anthracycline equivalents as described previously (Mandelli 2009).

Table 3. Anthracycline dosing at induction and consolidation in the included trials
Trial Anthracycline use Cumulative dose at induction Cumulative dose at consolidation Total dose in both induction and consolidation Total dose in daunorubicin equivalents
Adès 2010Daunorubicin180 mg/m2315 mg/m2495 mg/m2495 mg/m2
Asou 2007Idarubicin\Mitoxantrone\DaunorubicinIdarubicin: 36 mg/m2Idarubicin: 36 mg/m2 Mitoxantrone 21 mg/m2 Daunorubicin 150mg/m2Idarubicin: 72 mg/m2 Mitoxantrone 21 mg/m2 Daunorubicin 150mg/m597.5 mg/m2
Avvisati 2002Idarubicin\MitoxantroneIdarubicin: 48mg/m2

Idarubicin: 25 mg/m2

Mitoxantrone: 50 mg/m2

Idarubicin: 73 mg/m2

Mitoxantrone: 50 mg/m2

573.3 mg/m2
Avvisati 2011Idarubicin\MitoxantroneIdarubicin: 48 mg/m2

Idarubicin: 25 mg/m2

Mitoxantrone: 50 mg/m2

Idarubicin: 73 mg/m2

Mitoxantrone: 50 mg/m2

573.3 mg/m2
Feusner 2010Daunorubicin200 mg/m2200 mg/m2400 mg/m2400 mg/m2
Lin 2007Daunorubicin + AclarubicinnoneDaunorubicin: 135 mg/m2 Aclarubicin: 105 mg/m2Daunorubicin: 135 mg/m2 Aclarubicin: 105 mg/m2N/A
Parovichnikova 2004Daunorubicin180 mg/m2360 mg/m2540 mg/m2540 mg/m2
Powell 2010Daunorubicin200 mg/m2300 mg/m2500 mg/m2500 mg/m2
Shen 2004Daunorubicinnone135 mg/m2135 mg/m2135 mg/m2
Tallman 2002Daunorubicinby randomization to either Daunorubicin 135 mg/m2 or no anthracycline use225 mg/m2

225 mg/m2 or 360 mg/m2

(by randomization)

225 mg/m2 or 360 mg/m2

(by randomization)

Duration of intervention

The duration of the maintenance period varied between trials. In three trials (Feusner 2010; Powell 2010; Tallman 2002) maintenance was given for one year. In four trials (Adès 2010; Avvisati 2002; Avvisati 2011; Parovichnikova 2004) maintenance was given for two years. In one trial (Asou 2007) the duration of maintenance was reported as six cycles of treatment, with six weeks between each cycle. In Shen 2004 maintenance was measured as five cycles, each cycle lasting around two months. In Lin 2007 maintenance was reported as continued for two to three years or more.

Dosing of intervention
  • ATRA: the daily dose of ATRA was 45 mg/m2 in most trials, except for two trials (Lin 2007; Shen 2004) where the dose was 25 mg/m2 daily. In one trial (Tallman 2002) ATRA was given on a daily basis for one year. In all other trials ATRA was administered intermittently according to various schedules (Table 1).

  • Chemotherapy: dosing and schedules of chemotherapy varied between trials (Table 1).

Duration of follow-up

Follow-up in trials varied. It ranged between 1.5 to 10 years (see Characteristics of included studies).

Table 1 summarizes the maintenance protocols of the included trials.

Outcome assessment
Primary outcome measure

Overall survival data were available from six trials (Adès 2010; Asou 2007; Avvisati 2011; Feusner 2010; Parovichnikova 2004; Powell 2010). For three trials (Adès 2010; Asou 2007; Avvisati 2011) OS measurement was from randomization to maintenance. For two trials (Feusner 2010; Powell 2010) OS data were from enrollment to induction treatment, and in one trial (Parovichnikova 2004) data were missing for the starting point of OS measurement. DFS was defined as the time from attainment of complete remission (CR) or randomization to maintenance till relapse or death. DFS data were available from nine trials (Adès 2010; Asou 2007; Avvisati 2002; Avvisati 2011; Feusner 2010; Parovichnikova 2004; Powell 2010; Tallman 2002) and were calculated for all but one trial (Shen 2004) following CR achievement after the last consolidation cycle. The outcomes and their definitions for each trial included are described in the table 'Characteristics of included studies'.

Excluded studies

Forty-six articles were excluded after evaluation of full text publications for the following reasons (see table 'Characteristics of excluded studies'):

Risk of bias in included studies

Figure 2 summarizes the risk of bias of the included trials.

Figure 2.

Risk of bias summary: review authors' judgements about each risk of bias item for each included study.

Other potential sources of bias  

Allocation

Allocation was adequately concealed in seven out of 10 included trials (Adès 2010; Asou 2007; Avvisati 2002; Avvisati 2011; Feusner 2010; Powell 2010; Tallman 2002) and was not reported in three trials (Lin 2007; Parovichnikova 2004; Shen 2004). The sequence was adequately generated in six trials (Adès 2010; Asou 2007; Avvisati 2002; Avvisati 2011; Feusner 2010; Powell 2010) and was centrally performed. In four trials the method of sequence generation was not reported (Lin 2007; Parovichnikova 2004; Shen 2004; Tallman 2002). We judged the quality of these trials as unclear risk of bias.

Blinding

In all trials, blinding of patients and caregivers was not applied since prolonged blinded maintenance treatment was difficult to carry out, especially since most comparators included active treatment. We reported the quality of these trials as unclear risk of bias. Blinding of outcome assessors was not reported in all trials. We judged the quality of these trials as unclear risk of bias.

Incomplete outcome data

For the primary outcome analysis, three trials reported missing outcome data (Asou 2007; Avvisati 2002; Powell 2010). In all three trials the percentage of drop-outs was less than 10% of randomized patients. Reasons for missing data and treatment allocation were given in two trials (Asou 2007; Avvisati 2002). We reported the quality of these trials as low risk of bias. In another six trials there was no report on missing data (Adès 2010; Avvisati 2011; Feusner 2010; Lin 2007; Shen 2004; Tallman 2002) and all randomized patients were included in the analysis. We judged the quality of these trials as low risk of bias. One trial (Parovichnikova 2004) did not report missing data and we were not able to determine whether all randomized patients were included in the analysis. We judged this trial's quality as unclear risk of bias.

Selective reporting

The protocols of four trials (Adès 2010; Avvisati 2011; Feusner 2010; Powell 2010) were available for assessment. In these trials, all pre-planned outcomes were addressed in the trial publications. We reported those trials as low risk for reporting bias. For six trials the study's protocol was not available for assessment. However, in four of these trials (Asou 2007; Avvisati 2002; Lin 2007; Tallman 2002) survival and toxicity outcomes were reported. Therefore, we reported these trials as low risk for reporting bias. In two trials (Parovichnikova 2004; Shen 2004) the outcome reporting was incomplete (toxicity was missing in Parovichnikova 2004; overall survival reporting was missing in Shen 2004) and we reported these trials as high risk for reporting bias.

Other potential sources of bias

We focused on four other potential sources of bias in each of the included trials, as detailed below.

# Funding: while seven trials reported grants awarded from non-pharmaceutical organizations, either academic or governmental, three trials (Avvisati 2002; Lin 2007; Parovichnikova 2004) did not report on financial support.

# Intention-to-treat (ITT) analysis: all included trials except for one (Parovichnikova 2004) reported analysis based on ITT.

# Sample size calculation: sample size calculations were reported in four out of 10 included trials (Adès 2010; Avvisati 2002; Avvisati 2011; Powell 2010). In all these trials the designed sample size was reached.

# Early termination of trial: one trial (Shen 2004) was stopped early due to superior outcome in the combined ATRA-ATO-6-MP-MTX group. As this trial's termination was not pre-defined in its protocol we reported this trial as high risk of bias.

Effects of interventions

See: Summary of findings for the main comparison; Summary of findings 2; Summary of findings 3

We defined three main comparisons for the effects of interventions in our protocol, as outlined in the 'Methods' section above (Types of interventions). Definitions of the primary outcomes were given in the 'Included studies' section above and specifically for each trial in the 'Characteristics of included studies' section. Briefly, OS was measured from randomization to maintenance (most trials) or from enrollment to induction. DFS was measured from attainment of CR or randomization to maintenance until relapse or death.

Any maintenance compared to observation

Five trials were included in this comparison (Adès 2010; Asou 2007; Avvisati 2002; Avvisati 2011; Tallman 2002). Three trials reported on the OS outcome (Adès 2010; Asou 2007; Avvisati 2011), and all five trials reported on the DFS outcome.

Primary outcome measures
Overall survival (OS)

Three trials (892 patients) reported on this outcome (Adès 2010; Asou 2007; Avvisati 2011). We found no effect on OS for any maintenance treatment compared to observation (HR 0.79, 95% CI 0.49 to 1.27, fixed-effect model; I2 for heterogeneity = 89%; Figure 3). Exploration for the significant heterogeneity was unsuccessful, probably reflecting the small number of participating trials in this comparison as well as the differences in both the maintenance and pre-maintenance treatments. Using the random-effects model for this comparison the HR for maintenance treatment compared to observation was 1.07 (95% CI 0.23 to 4.86).

Figure 3.

Forest plot of comparison: 1 any maintenance compared to observation, outcome: 1.1 Overall survival.

Disease-free survival (DFS)

Five trials (1209 patients) were included in this comparison (Adès 2010; Asou 2007; Avvisati 2002; Avvisati 2011; Tallman 2002). Any maintenance given to APL patients in the first complete remission was associated with improved DFS compared to observation (HR 0.59, 95% CI 0.48 to 0.74, fixed-effect model; I2 for heterogeneity = 72%; Figure 4). Heterogeneity in this comparison was mainly explained by the trial that used intensive chemotherapy in the maintenance phase (Asou 2007). As appears in the subgroup analysis below (low intensity maintenance versus observation), by excluding this trial heterogeneity was reduced to 31% with negligible impact on the point estimate and CI for this comparison.

Figure 4.

Forest plot of comparison: 1 any maintenance compared to observation, outcome: 1.2 Disease-free survival.

Secondary outcome measures
Treatment-related mortality

Treatment-related mortality was reported in two trials. In one trial (Asou 2007) there was no treatment-related mortality in both arms. In the other trial (Avvisati 2011) treatment-related mortality was similar between the active maintenance arm and the observation arm (1.6% in any maintenance versus 1.3% in observation arm). Due to the paucity of data, a meta-analysis could not be conducted.

Relapse rate

The risk of relapse was reduced in the maintenance arm compared to observation (RR 0.69, 95% CI 0.58 to 0.81, fixed-effect model; 4 trials, 896 patients; I2 for heterogeneity = 82%) (Adès 2010; Asou 2007; Avvisati 2002; Tallman 2002). The reasons for heterogeneity in this comparison probably stem from the different treatments in the comparator arms, but we failed to demonstrate reduced heterogeneity by subgroup analysis. Using the random-effects model, the relapse rate was found to be similar between any maintenance and observation (HR 0.74, 95% CI 0.50 to 1.11).

Adverse events
Adverse events requiring discontinuation of therapy

Adverse events requiring discontinuation of therapy were reported in three trials. In two trials the rate was 8.6% and 14.1% in the active treatment arm (Avvisati 2002; Tallman 2002), respectively. In Avvisati 2011 a discontinuation of treatment was not registered.

Grade 3 and 4 adverse events

Only one trial (Tallman 2002) reported on grade 3 to 4 adverse events for ATRA maintenance as compared to observation. An increased risk for this outcome in the treatment arm (ATRA alone) was noted (34% of patients versus 2.9% of patients in the control arm).

Grade 3 and 4 hematologic toxicity

None of the trials reported on grade 3 or 4 hematologic toxicity.

Infection-related adverse events

Only two studies reported on this adverse effect. In one trial (Adès 2010) the rate of infections related to the 'any maintenance arm' was 3.7% (12 patients, of whom six died of sepsis) compared to none in the observation arm. Similarly, in Tallman 2002 seven patients experienced infectious complications during the maintenance phase (ATRA alone) compared to none in the observation arm. Due to the paucity of trials providing data on this outcome, a meta-analysis was not feasible for this outcome.       

APL differentiation syndrome (formerly ATRA syndrome)

None of the trials reported on APL differentiation syndrome during maintenance treatment.

Subgroup analysis for overall survival (OS) and disease-free survival (DFS)
Low intensity maintenance versus observation

As the maintenance regimen employed in Asou 2007 was intensive chemotherapy-based and very different from the other trials included in this comparison (Adès 2010; Avvisati 2002; Avvisati 2011; Tallman 2002), which used either ATRA, low dose chemotherapy or both, we performed a subgroup analysis with the exclusion of the above exceptional trial.

The effect on OS for any maintenance treatment compared to observation remained similar, although borderline significance in favor of any maintenance was achieved (HR 0.60, 95% CI 0.36 to 1.01, fixed-effect model; I2 for heterogeneity = 91%) (Adès 2010; Avvisati 2011).

For the DFS outcome, by the exclusion of Asou 2007 any maintenance given to APL patients in the first complete remission remained associated with statistically significant improved DFS compared to observation as in the main analysis (HR 0.52, 95% CI 0.41 to 0.66, fixed-effect model) but with reduced statistical heterogeneity (I2 for heterogeneity = 31%). Finally, the relapse rate in this subgroup analysis was reduced with any maintenance compared to observation (HR 0.39, 95% CI 0.28 to 0.55, fixed-effect model) but with reduced heterogeneity compared to the main comparison, albeit remaining high (I2 for heterogeneity = 64% versus 82% in the main comparison).

Elderly patients (age > 60 years)

No data were available for subgroup analysis of the primary outcomes regarding elderly patients.

Pediatric patients (age < 15 years)

Only one trial (Tallman 2002), enrolling 36 pediatric patients, reported on the primary outcomes among children. This trial demonstrated that maintenance with ATRA alone did not improve OS compared to observation alone (5-year OS 83% in the treatment arm versus 55% in the observation arm, P = 0.13). However, maintenance with ATRA alone had a superior effect on DFS compared to observation alone (5-year DFS 61% in the treatment arm versus 15% in the observation arm, P = 0.0002).

Risk group stratification

There were not enough data to analyze the primary outcomes according to risk group stratification.

Type of remission confirmation

Maintenance therapy compared to observation did not have a statistically significant effect on OS in trials that defined remission only by morphology as well as in those that defined it morphologically or molecularly, or both. Conversely, DFS was improved by maintenance as compared to observation in trials where remission was defined morphologically (HR 0.52, 95% CI 0.41 to 0.67, fixed-effect model; 3 trials, 721 patients; I2 for heterogeneity = 54%) (Adès 2010; Avvisati 2002; Tallman 2002) but not in those where it was defined according to molecular criteria (HR 0.90, 95% CI 0.57 to 1.44, random effect model; 2 trials, 488 patients; I2 for heterogeneity = 82%) (Asou 2007; Avvisati 2011). The difference in this outcome between these two subgroups was statistically significant (P = 0.04).

Type of prior induction and consolidation therapy
Type of prior anthracycline use:

Analysis of trials that used idarubicin for induction or consolidation demonstrated a marginally statistically significant negative effect of maintenance therapy on OS compared to observation (HR 1.86, 95% CI 0.99 to 3.48, fixed-effect model; 2 trials, 491 patients; I2 for heterogeneity = 48%) (Asou 2007; Avvisati 2011). Only one trial that used daunorubicin for induction or consolidation reported on OS and showed superiority of any maintenance compared to observation (HR 0.25, 95% CI 0.12 to 0.52) (Adès 2010). The difference between these two subgroups was statistically significant (P < 0.0001).

Analysis of DFS in trials that used idarubicin for induction or consolidation did not demonstrate a statistically significant effect of maintenance therapy compared to observation (HR 0.85, 95% CI 0.61 to 1.18, fixed-effect model; 3 trials, 604 patients; I2 for heterogeneity = 65%) (Asou 2007; Avvisati 2002; Avvisati 2011). Conversely, when daunorubicin was used for induction or consolidation maintenance was associated with improved DFS compared to no maintenance (HR 0.44, 95% CI 0.33 to 0.60, fixed-effect model; 2 trials, 605 patients; I2 for heterogeneity = 0%) (Adès 2010; Tallman 2002). The difference between these two subgroups was statistically significant (P = 0.004).

Subtypes of maintenance comparisons
  • Chemotherapy alone versus observation: four trials were eligible for this comparison (Adès 2010; Asou 2007; Avvisati 2002; Avvisati 2011). Data on OS could be retrieved from two trials. Chemotherapy alone compared to observation for maintenance did not improve OS (HR 0.60, 95% CI 0.29 to 1.23, fixed-effect model; 2 trials, 369 patients; I2 for heterogeneity = 92%) (Adès 2010; Asou 2007). Similarly, DFS was not improved with chemotherapy alone compared to observation (HR 0.94, 95% CI 0.68 to 1.30, fixed-effect model; 3 trials, 369 patients; I2 for heterogeneity = 34%) (Asou 2007; Avvisati 2002; Avvisati 2011).

  • ATRA alone versus observation: three trials were eligible for this subgroup analysis (Adès 2010; Avvisati 2011; Tallman 2002). Only one trial with 155 patients (Adès 2010) reported on OS for this comparison, demonstrating a similar effect for ATRA alone compared to observation (HR 0.53, 95% CI 0.24 to 1.15). However, maintenance with ATRA alone improved DFS compared to observation (HR 0.47, 95% CI 0.33 to 0.66; 2 trials, 204 patients; I2 for heterogeneity = 0%) (Avvisati 2011; Tallman 2002).

  • ATRA-containing regimens versus observation: two trials were eligible for  this subgroup analysis (Avvisati 2011; Tallman 2002). Only one trial with 240 patients (Avvisati 2011) reported on OS for this comparison, demonstrating a similar effect for ATRA alone compared to observation (HR 1.40, 95% CI 0.67 to 2.96). However, ATRA-containing regimens (that is ATRA alone or ATRA combined with chemotherapy) compared to observation achieved a statistically significant improved DFS (HR 0.48, 95% CI 0.35 to 0.66; 2 trials, 444 patients; I2 for heterogeneity = 0%) (Avvisati 2011; Tallman 2002).

ATRA-based maintenance compared to non-ATRA based maintenance

Four trials were included in this comparison (Adès 2010; Avvisati 2011; Parovichnikova 2004; Shen 2004). Three trials reported on OS (Adès 2010; Avvisati 2011; Parovichnikova 2004) and all four trials reported on DFS.

Primary outcome measures
Overall survival (OS)

There was no statistically significant effect on OS for patients treated with ATRA-based maintenance compared to non-ATRA based maintenance (HR 0.98, 95% CI 0.64 to 1.51, fixed-effect model; three trials, 632 patients; I2 for heterogeneity = 0%) (Adès 2010; Avvisati 2011; Parovichnikova 2004) (Figure 5).

Figure 5.

Forest plot of comparison: 2 ATRA-based maintenance compared to non-ATRA based maintenance, outcome: 2.1 Overall survival.

Disease-free survival (DFS)

ATRA-based maintenance was probably associated with improved DFS compared to non-ATRA based maintenance, but result did not reach statistical significance (HR 0.72, 95% CI 0.51 to 1.01, fixed-effect model; 4 trials, 670 patients; I2 for heterogeneity = 0%) (Adès 2010; Avvisati 2011; Parovichnikova 2004; Shen 2004) (Figure 6).

Figure 6.

Forest plot of comparison: 2 ATRA-based maintenance compared to non-ATRA based maintenance, outcome: 2.2 Disease-free survival.

Secondary outcome measures
Treatment-related mortality

Only one study assessed this outcome (Avvisati 2011). Treatment-related mortality was not affected by ATRA-based maintenance compared to non-ATRA based maintenance (3.7% in ATRA based arm versus 2.6% in non-ATRA based arm).

Relapse rate

Risk of relapse was similar between patients who received ATRA-based maintenance and those who received non-ATRA based maintenance (RR 0.88, 95% CI 0.59 to 1.30, fixed-effect model; 3 trials, 425 patients; ; I2 for heterogeneity = 19%) (Adès 2010; Parovichnikova 2004; Shen 2004).

Adverse events
Adverse events requiring discontinuation of therapy

None of the trials reported on adverse events requiring discontinuation of therapy.

Grade 3 and 4 adverse events

None of the trials reported on grade 3 and 4 adverse events.

Grade 3 and 4 hematologic toxicity

None of the trials reported on grade 3 and 4 hematologic adverse events.

Infection-related adverse events

None of the trials reported on infection-related adverse events.

APL differentiation syndrome (formerly ATRA syndrome)

None of the trials reported on APL differentiation syndrome during maintenance.

Subgroup analysis for overall survival (OS) and disease-free survival (DFS)
Elderly patients (age > 60 years)

None of the trials in this comparison reported on elderly patients.

Pediatric patients (age < 15 years)

None of the trials in this comparison reported on pediatric patients.

Risk group stratification

There were not enough data to analyze primary outcomes according to risk group stratification.

Type of remission confirmation

No statistically significant effect on OS nor DFS was noted between patients allocated to ATRA-based maintenance compared to non-ATRA based maintenance by stratifying trials according to the type of remission confirmation, that is molecular versus hematological.

Type of prior induction and consolidation therapy
Type of prior anthracycline

OS was not affected by the allocated treatments in patients with APL who received daunorubicin during induction or consolidation, or both (Adès 2010; Parovichnikova 2004) (HR 1.31, 95% CI 0.64 to 2.70, fixed-effect model; 2 trials, 390 patients; I2 for heterogeneity = 0%). Only one trial used idarubicin, with no effect on OS with ATRA-based maintenance compared to non-ATRA based maintenance (HR 1.12, 95% CI 0.56 to 2.22) (Avvisati 2011). The difference between the subgroups was not statistically significant (P = 0.64).

DFS was not improved with the use of ATRA-based maintenance compared to non-ATRA based maintenance for APL patients receiving daunorubicin prior to maintenance treatment (HR 0.81, 95% CI 0.53 to 1.22, fixed-effect model; 3 trials, 428 patients; I2 for heterogeneity = 2%) (Adès 2010; Parovichnikova 2004; Shen 2004). Only one trial used idarubicin (Avvisati 2011), with no advantage for ATRA-based maintenance compared to non-ATRA based maintenance with regards to DFS (HR 0.56, 95% CI 0.30 to 1.03). The difference between the subgroups was not statistically significant (P = 0.33).

ATRA alone maintenance compared to ATRA and chemotherapy maintenance

Four trials were included in this comparison, all of them reported on OS and DFS (Adès 2010; Avvisati 2011; Feusner 2010; Powell 2010).

Primary outcome measures
Overall survival (OS)

There was no difference in the HR for OS between ATRA alone or ATRA and chemotherapy maintenance (HR 0.99, 95% CI 0.69 to 1.43, fixed-effect model; 4 trials, 1033 patients; I2 for heterogeneity = 51%) (Adès 2010; Avvisati 2011; Feusner 2010; Powell 2010) (Figure 7). Heterogeneity could be explained by the use of different anthracyclines in the pre-maintenance treatments, as described in the subgroup analysis below.

Figure 7.

Forest plot of comparison: 3 Maintenance with ATRA alone compared to maintenance with ATRA and chemotherapy, outcome: 3.1 Overall survival.

Disease-free survival (DFS)

ATRA alone maintenance was statistically significantly inferior to combined ATRA and chemotherapy maintenance in terms of DFS (HR 1.38, 95% CI 1.09 to 1.76, fixed-effect model; 4 trials, 1028 patients; I2 for heterogeneity = 46%) (Adès 2010; Avvisati 2011; Feusner 2010; Powell 2010) (Figure 8). Again, heterogeneity could be explained by the different anthracyclines used prior to maintenance treatment and also by the different types of remission confirmation (see subgroup analysis below).

Figure 8.

Forest plot of comparison: 3 Maintenance with ATRA alone compared to maintenance with ATRA and chemotherapy, outcome: 3.2 Disease-free survival.

Secondary outcome measures
Treatment-related mortality

Only one trial (Avvisati 2011) enrolling 432 patients reported on this outcome. In this trial, no difference in treatment-related mortality was demonstrated between ATRA alone and ATRA and chemotherapy maintenance (0.9% in ATRA alone arm versus 1.9% in the combination arm).

Relapse rate

Only one trial (Adès 2010) enrolling 205 patients reported the relapse rate for ATRA alone compared to ATRA and chemotherapy, demonstrating an increased relapse rate for ATRA alone (34% in ATRA alone arm versus 14% in the combination arm).

Adverse events
Adverse events requiring discontinuation of therapy

No data were available for this analysis.

Grade 3 and 4 adverse events

Only one trial (Powell 2010) including 331 patients reported grade 3 to 4 adverse events. This trial reported a reduced risk for grade 3 to 4 adverse events in the ATRA alone maintenance arm compared to the ATRA and chemotherapy maintenance arm (25% in the ATRA alone maintenance arm versus 36% in the combination arm, P = 0.033).

Grade 3 and 4 hematologic toxicity

Only one trial (Powell 2010) including 331 patients reported grade 3 to 4 hematological adverse events. This trial reported a reduced risk for grade 3 to 4 hematological adverse events in the ATRA alone maintenance arm as compared to the ATRA and chemotherapy maintenance arm (4% in the ATRA alone maintenance versus 18% in the combination arm, P < 0.0001).

Infection-related adverse events

No data were available for this analysis.

APL differentiation syndrome (formerly ATRA syndrome)

No data were available for this analysis.

Subgroup analysis for overall survival (OS) and disease-free survival (DFS)
Adults patients (excluding pediatric patients)

No data were available for this analysis.

Elderly patients (age > 60 years)

No data were available for this analysis.

Pediatric patients (age < 15 years)

Two trials reported on a pediatric population (Avvisati 2011; Feusner 2010). Overall, ATRA alone maintenance had an effect on OS similar to ATRA and chemotherapy maintenance (HR 2.05, 95% CI 0.47 to 8.82, fixed-effect model; 2 trials, 133 patients; I2 for heterogeneity = 0%).

For DFS, the overall effect of ATRA maintenance alone was statistically significantly inferior to ATRA and chemotherapy maintenance (HR 2.17, 95% CI 1.09 to 4.31, fixed-effect model; 2 trials, 127 patients; I2 for heterogeneity = 40%).

Risk group stratification

There were not enough data to analyze the primary outcomes according to risk group stratification.

Type of remission confirmation

ATRA alone maintenance had no statistically significant effect on OS compared to ATRA and chemotherapy maintenance in trials which applied hematological remission confirmation prior to maintenance initiation (HR 1.55, 95% CI 0.92 to 2.63, fixed-effect model; three trials, 606 patients; I2 for heterogeneity = 0%) (Adès 2010; Feusner 2010; Powell 2010). Only one trial using molecular remission confirmation was available for this comparison (Avvisati 2011). This trial did not demonstrate a difference in OS between the two allocated maintenance arms (HR 0.65, 95% CI 0.39 to 1.08). However, the difference between the subgroups was statistically significant (P = 0.02) due to different points of estimate.

Trials which used hematological remission confirmation demonstrated a negative effect on DFS for patients allocated to ATRA alone maintenance compared to ATRA and chemotherapy maintenance (HR 1.72, 95% CI 1.23 to 2.41, fixed effect model; 3 trials, 596 patients; I2 for heterogeneity = 3%) (Adès 2010; Feusner 2010; Powell 2010). Only one trial using molecular remission confirmation was available for this comparison (Avvisati 2011), showing no effect on DFS with ATRA alone maintenance compared to ATRA and chemotherapy maintenance (HR 1.08, 95% CI 0.76 to 1.54). The difference between the subgroups was not statistically significant (P = 0.06).

Type of prior induction and consolidation therapy
Type of prior anthracycline used

Trials which used daunorubicin demonstrated a similar effect on OS for ATRA alone maintenance compared to ATRA and chemotherapy maintenance (HR 1.55, 95% CI 0.92 to 2.63, fixed-effect model; 3 trials, 606 patients; I2 for heterogeneity = 0%) (Adès 2010; Feusner 2010; Powell 2010). Only one trial reporting OS in this comparison used idarubicin (Avvisati 2011), with a similar OS for patients treated with ATRA alone for maintenance and those treated with ATRA and chemotherapy as maintenance (HR 0.65, 95% CI 0.39 to 1.08). The difference between the subgroups was statistically significant (P = 0.02) due to different points of estimate.

In terms of DFS, there was a negative impact for ATRA alone maintenance compared to ATRA and chemotherapy maintenance when daunorubicin was used, namely patients previously treated with daunorubicin faired less well when maintenance consisted of ATRA alone compared to ATRA and chemotherapy (HR 1.72, 95% CI 1.23 to 2.41, fixed-effect model; 3 trials, 596 patients; I2 for heterogeneity = 3%) (Adès 2010; Feusner 2010; Powell 2010). Only one trial used idarubicin in this comparison (Avvisati 2011), with similar DFS rates for patients treated with ATRA alone and those treated with ATRA and chemotherapy maintenance (HR 1.08, 95% CI 0.76 to 1.54). The difference between the subgroups was not statistically significant (P = 0.06).

Sensitivity analysis
Sensitivity analysis by allocation concealment and allocation generation

Three trials reported neither allocation generation nor the allocation concealment method (Lin 2007; Parovichnikova 2004; Shen 2004). One trial (Lin 2007) did not include an outcome report that could be used for primary outcome meta-analysis. The remaining two trials were included under the same comparison; therefore this type of sensitivity analysis was only available for the comparison of ATRA-based maintenance and non-ATRA based maintenance. No difference in OS was noted between the two trials reporting adequate methods of allocation concealment and generation (Adès 2010; Avvisati 2011) and the single trial (Parovichnikova 2004) which did not report it. As for DFS, trials with adequate methods of allocation concealment and generation did not demonstrate improved DFS with ATRA-based maintenance compared to non-ATRA based maintenance (HR 0.73, 95% CI 0.51 to 1.03, fixed-effect model; 2 trials, 564 patients; I2 for heterogeneity = 5%) (Adès 2010; Avvisati 2011). The two trials with unclear methods of allocation concealment and generation did not differ in terms of DFS regarding ATRA-based maintenance and non-ATRA based maintenance as well. The difference between the two subgroups was not statistically significant (P = 0.83).

Sensitivity analysis by duration of follow-up

Six trials had a follow-up duration of over five years (Adès 2010; Asou 2007; Avvisati 2002; Avvisati 2011; Powell 2010; Tallman 2002), whereas three trials had a follow-up duration of five years or less (Feusner 2010; Parovichnikova 2004; Shen 2004). Since all five trials included in the comparison of any maintenance compared to observation had longer than five years follow-up, sensitivity analysis was not applicable for this comparison. Stratifying follow-up duration into two subgroups according to this cutoff did not demonstrate improved OS in the other two main comparisons. However, among these two comparisons, DFS was improved in trials with follow-up longer than five years in all three main comparisons. In contrast, no effect on DFS was noted in trials where the follow-up duration was five years or less.   

Sensitivity analysis by measure of OS

Among the six trials reporting OS data, three trials reported OS data from the time of randomization to maintenance till death from any cause (Adès 2010; Asou 2007; Avvisati 2011), whereas three trials reported either OS data from study entry (Feusner 2010; Powell 2010) or from an undetermined time point (Parovichnikova 2004). Sensitivity analysis by stratifying trials on the basis of the survival duration measurement method was applicable for two main comparisons (comparison 2: ATRA-based maintenance versus non-ATRA based maintenance; and comparison 3: ATRA alone maintenance versus ATRA and chemotherapy maintenance) and did not result in a different OS effect between subgroups.

Sensitivity analysis by measure of DFS

One trial (Shen 2004) had only one randomization for both the induction and maintenance treatments. This randomization was conducted before induction initiation. Moreover, DFS in this trial was measured differently (after achieving CR following induction and prior to consolidation), in contrast to all other included trials where randomization was carried out after the consolidation phase. Therefore, we performed sensitivity analysis, excluding the DFS data from this trial. The results of this sensitivity analysis were similar to the main comparison, reflecting the negligible effect of this small-sized trial.

Discussion

Summary of main results

Ten randomized controlled trials enrolling a total of 2072 patients were compatible with the inclusion criteria and were therefore included in our systematic review. Nine of them (2012 patients enrolled) were included in the meta-analysis.

There was no evidence that any maintenance therapy improved OS compared to observation. DFS was improved with any maintenance compared to observation. When comparing the two active arms, ATRA and chemotherapy showed improved DFS but not OS compared to ATRA alone. For ATRA-based regimens compared to non-ATRA based regimens, DFS was probably improved, but OS was not. There was a major statistical and clinical heterogeneity between the trials included in our meta-analysis, reflecting the differences in both pre-maintenance treatments, namely induction and consolidation regimens, as well as in the maintenance protocols and disease monitoring.

Subgroup analysis suggested that DFS benefit may depend on pre-maintenance treatment parameters including the type of anthracycline used and the intensity of treatment as well as the definitions of the type of remission.

There was increased toxicity with any maintenance compared to observation and with combined maintenance protocols compared to monotherapy. This might negatively balance the improved disease control achieved with the use of maintenance.

Overall completeness and applicability of evidence

The objectives of this review were to examine the efficacy and safety of maintenance therapy for patients with APL, and to establish the optimal maintenance regimen. Our results show that according to the three main comparisons, maintenance therapy does not improve OS but improves DFS. This was shown according two main comparisons, that is any maintenance compared to observation, and ATRA alone compared to ATRA and chemotherapy (with statistically marginal proof of benefit in the second comparison of ATRA-based maintenance compared to non-ATRA based maintenance). It is perplexing that according to our review, DFS was prolonged in the maintenance arm while OS, treatment-related mortality and relapse rates were not different. As CR rates by the end of consolidation were between 70% and 90% and the relapse rates ranged between 20% and 30% in most trials (Table 4), it might be difficult to demonstrate an OS benefit for maintenance therapy as most patients might have been already cured prior to the maintenance phase. Therefore, the possibility of an underpowered meta-analysis for the OS measure should be considered.

Table 4. Summary of outcome measures of the included trials from induction
Trial Median FU CR Cumulative incidence of relapse Overall survival from induction
Adès 201010.1 years533/576=92.5%142/533=26.6%Estimated 10-year survival 77%
Asou 20075.3 years267/283=94.3%60/267=22.5%Predicted 6-year OS rate: 83.9%
Avvisati 20027.5 years (minimum)187/257=72.8%70/187=37.4%Estimated 8-year OS rate was 45% (95% CI, 37% to 54%) for patients randomized to arm A of induction and 36% (95% CI, 27% to 44%) for those randomized to arm B of induction
Avvisati 20117.8 years761/807=94.3%N\AEstimated 12 years OS 76.5% (95% CI, 74.1% to 78.9%)
Feusner 20103.9 years84%N\A3 years OS 87%
Lin 20073 years (unclear if median)100%14/60=23.3%OS rate in 3 years was 100.0% in As4S4 group and 83.3% in non-As4S4 group
Parovichnikova 20042.5 years (maximum)90%N\A2.5 years OS 77%
Powell 20106.2 years432/481=89.8%N\ASurvival at 3 years was 86% in the As2O3 arm compared to 81% in the standard arm
Shen 20041.5 years57/61=93.4%7/57=12.3%55/61=90.2% at median 18 month
Tallman 20026.2 years272/380=71.6%113/380=29.7%OS at 5 years was 45% (95% CI, 36 to 52) for patients randomized to induction chemotherapy and 69% (95% CI, 62 to 76) for patients randomized to ATRA induction

The improved DFS with no effect on OS may also stem from the toxicity of maintenance therapy that consequently negatively balanced the improved disease control, from the extended time to relapse with maintenance therapy which ultimately occurs and from the availability of a very effective salvage therapy upon relapse. Yet, even in the ATO salvage therapy era, survival following relapse in APL is reduced (50% to 81% at 2 years) (Lengfelder 2012). Moreover, most trials included in our review were performed in the pre-ATO era with a 2-year survival of 50% or less following relapse (Lo coco 1999). Therefore, the last explanation is less plausible for the discrepancy between DFS and OS.

Yet, improved DFS with no OS benefit, albeit all the limitations mentioned above, may be important for both treating physicians and patients as it may imply better disease control and save additional treatments upon relapse. Thus, although no clear survival advantage is demonstrated for maintenance therapy, this type of treatment may indirectly impact the quality of life of patients by reducing treatment burden with a better disease control. Still, as we do not have quality of life data from the included trials and as maintenance therapy by itself has a direct impact on patients' quality of life, this hypothesis should be appropriately evaluated in future trials.

Another interesting point is that although there was no difference in the relapse rate according to the random-effects model, DFS was better with maintenance. This finding may be due to the fact that relapse rate was not reported by all trials. Moreover, due to differences in the follow-up periods between the included trials, there might have been an impact on the time to relapse influencing DFS, which is a time to event variable. Furthermore, the relapse rate becomes statistically significant between maintenance and observation when the intensive chemotherapeutic maintenance versus observation trial published by Asou et al is excluded (Asou 2007). This observation highlights both the clinical and statistical complexity of this review and might imply that in APL patients achieving CR, a more intensive maintenance regimen does not translate necessarily into improved disease control.   

We found a positive interaction between certain maintenance regimens and DFS advantage when daunorubicin rather than idarubicin was used as the anthracycline in the induction and consolidation phases. Moreover, most trials used similar cumulative doses of anthracyclines for the induction or consolidation phases, or both (Table 3). This could be used as a surrogate to eliminate dose response differences between trials. Still, this interaction may be the consequence of a bystander effect, especially in light of the low number of included trials.Therefore, until a direct comparison is made between the various types of anthracyclines, there is no advantage for one anthracycline over the other with respect to the issue of maintenance.

The method of remission confirmation varied between trials. Some trials confirmed remission by molecular methods (Asou 2007; Avvisati 2011; Parovichnikova 2004) whereas others used morphological assessment (Adès 2010; Avvisati 2002; Feusner 2010; Lin 2007; Powell 2010; Shen 2004; Tallman 2002). Detection of molecular relapse is highly predictive for hematological relapse (Diverio 1998). There are reports of improved outcome if pre-emptive treatment is given at the time of molecular relapse rather than waiting for hematological relapse (Lo coco 1999). In trials where remission confirmation was based on morphology and not on molecular grounds, 2% to 8% of patients might be positive for the PML-RARa transcript if screened by molecular methods and are therefore at a higher risk for relapse while on maintenance treatment (Grimwade 2002). Yet, one has to bear in mind that 30% of patients who test negative on RT-PCR might still relapse. Although PCR-based remission confirmation and disease monitoring are superior to morphological confirmation and monitoring only, the true impact of this difference in remission assessment on the maintenance outcome is still uncertain.

We were not able to perform analysis by risk group stratification as only one trial (Powell 2010) reported survival data for maintenance according to the risk groups used in the PETHEMA trial (Sanz 2000). This issue needs to be addressed in future trials.

Due to a lack of detailed information concerning the elderly group, we could not conduct a meta-analysis for this group. One of the two trials (Adès 2010; Avvisati 2011) describing the outcome of maintenance among elderly patients reported a poorer survival in this group, not because of an increased relapse rate but due to death in CR during the maintenance period (Adès 2010). This data, albeit restricted to one trial, demonstrates the complexity of APL treatment among the elderly, which might discourage the use of maintenance treatment in this age group.

With regard to pediatric patients, four trials addressed the issue of maintenance in children (Adès 2010; Avvisati 2011; Feusner 2010; Tallman 2002). However, due to the range of comparisons and the low number of pediatric patients (reflecting the rarity of the disease among children), a meta-analysis (of two trials only) could be conducted only for the comparison of ATRA alone versus ATRA and chemotherapy. Thus, the role of maintenance in pediatric patients needs further investigation.

Extramedullary relapse occurs in nearly 3% to 5% of APL patients, manifested usually in the central nervous system and even more so since the introduction of ATRA (Raanani 2007). This type of relapse is associated with high white blood cell (WBC) count at presentation and carries a poor prognosis (de Botton 2006). We did not have enough data to evaluate the influence of maintenance treatment on the occurrence or prevention of extramedullary relapse.

Quality of the evidence

All trials that were included were randomized controlled trials. Allocation generation and concealment were adequate in seven out of 10 trials. None of the trials reported blinding of participants and personnel, and probably also outcome assessors. This might introduce both selection bias (mainly for trials with an observation control arm) and detection bias for all trials.

All trials but one reported analysis on the basis of intention to treat, and the drop-out rates were low. The risk of attrition bias is therefore low. The primary outcomes measures in this review are given as time-to-event measures. This type of outcome measure minimizes risk of bias when compared to outcome measures given by specified time points (that is 5-year survival).

Although included trials are of adequate methodological quality and their results are valid, it is difficult to draw robust conclusions due to the clinical and statistical heterogeneity. Using the GRADE scoring system, quality of evidence was graded for most measured primary outcomes as moderate or low quality. The main reasons for quality faults in the included trials were lack of blinding, low numbers of patients and events producing a wide confidence interval (i.e., imprecision) and high heterogeneity between trials. However, few outcome measures were upgraded with a large effect. This emphasizes the major limitations, as described above, and stresses the need for further randomized trials.

Potential biases in the review process

The major limitation of the present review is the clinical variability among the trials included stemming from the variability in the maintenance protocols, in the treatments prior to maintenance and in the definitions of outcomes. This variability led us to conduct three main comparisons to evaluate the effect of maintenance therapy, thus leaving less power for each comparison compared to conducting only one main comparison. Comparing any maintenance to observation demonstrates best this clinical heterogeneity, as it is well reflected in the significant statistical heterogeneity of the results of this comparison. However, analysis of subgroups of patients by the type of treatment arm (that is ATRA alone, ATRA-containing regimen, chemotherapy alone) in order to explore the reasons for this heterogeneity demonstrated improved DFS with no effect on OS only for ATRA-containing regimens (either ATRA alone or ATRA plus chemotherapy) compared to observation, without statistical heterogeneity between trials. However, the limitation of this subgroup analysis is underlined by the low number of trials available for each subgroup.

We conducted two subgroup analyses for each comparison to explore heterogeneity and to assess the effect of different factors on outcome measures. However, only very few trials were included in these subgroup analyses and thus a reliable conclusion cannot be drawn. Furthermore, with respect to the two subgroup analyses conducted, namely the type of anthracycline used in the remission induction and consolidation phase and the type of remission confirmation, it was difficult to separate the effect of anthracycline type from that of remission type since remission was confirmed morphologically in trials that used daunorubicin for induction or consolidation, while it was established molecularly in those using idarubicin as the anthracycline.

We also compared several outcome measures of trials which randomized patients before induction (Table 4). Complete response rate was similar between these trials and reached 90% or more, except for two (Avvisati 2002; Tallman 2002) reporting on a complete remission rate of 70%. As response endpoints achieved before the initiation of maintenance were similar among trials, it can be assumed that no attrition bias was introduced by prior induction and consolidation treatments.

Differences in duration of follow-up between trials might affect survival outcomes. The median duration of follow-up varied between 1.5 and 10 years among the included trials. Of note, in six trials the duration of follow-up was longer than five years. We therefore conducted a sensitivity analysis on the primary outcomes according to follow-up duration (over five years versus five years or less) with no significant influence on the results.

As for the availability of data from the included trials, most trials have reported satisfactory data. Whenever important data were missing we contacted the primary investigators for more detailed information, usually successfully.

Agreements and disagreements with other studies or reviews

To our knowledge this is the first systematic review and meta-analysis assessing the role of maintenance therapy in APL. Three meta-analyses recently conducted addressed the role of ATO in APL (Wang 2011; Xu 2009; Xu 2009a). These reviews did not assess its role for maintenance and included newly diagnosed patients as well as relapsed patients.

Maintenance therapy, despite being controversial, is part of many APL protocols worldwide. The Spanish PETHEMA group protocols LPA 99 and LPA 96 incorporated maintenance treatment consisting of ATRA and low dose methotrexate and mercaptopurine for two years in APL patients (Sanz 2004a). The German AML Cooperative Group (AMLCG) incorporated monthly maintenance chemotherapy for three years as part of their protocol for APL (Lengfelder 2009).

Clinical guidelines addressing APL treatment are few and are usually given as part of AML clinical guidelines (Milligan 2006; NCCN 2001). The National Comprehensive Cancer Network (NCCN) updated guidelines (NCCN 2001) on AML favor the use of maintenance therapy in APL but underline the unclear effect of maintenance therapy, particularly among low risk patients who achieve molecular remission at the end of consolidation. The British guidelines (Milligan 2006) are even less conclusive on the role of maintenance treatment in APL. Additionally, an expert panel on behalf of the European LeukemiaNet (Sanz 2009) stressed the relative benefit of maintenance in accordance with the prior induction and consolidation therapy. Therefore, the panel suggested using maintenance in conjunction with protocols in which a benefit for maintenance was demonstrated.

The Australian Leukaemia and Lymphoma Group (ALLG) has recently published the results of a prospective study conducted between the years 1997 and 2002 (Iland 2012) where APL patients were given two induction cycles based on the AIDA 0493 protocol (Avvisati 2011) and consolidation consisted of intermittent doses of ATRA only followed by an observational period. However, after an interim analysis, the protocol was amended and all the following patients received maintenance treatment consisting of a combination of ATRA and chemotherapy (6-MP+MTX) for two years. The investigators reported improved DFS, remission duration and failure-free survival with maintenance compared to observation, but like us they could not show advantage in terms of OS.

Authors' conclusions

Implications for practice

Maintenance treatment in APL does not affect OS according to existing data. However, mainly if ATRA-based, it prolongs DFS. With the growing use of ATO in APL for induction and consolidation as well as for relapse, the role of maintenance treatment might become even more obscure. Therefore, the current approach should be individually based, taking into consideration patient preferences as well as disease and patient biological parameters such as age, prognostic risk group, pre-maintenance treatment and type of remission status.

Implications for research

As survival data are incomplete and as there is considerable variability among included trials for both pre-maintenance and maintenance treatments, we recommend an individualized patient data meta-analysis as a measure that can validate the results of this review and answer some of the unsolved issues of maintenance therapy in APL.

As appears from the current review, the major gaps in the field of maintenance in APL that should be the focus of future research are as follows.

  • The characterization of the patients who might benefit most from maintenance and those who do not need it, by using validated risk group stratification.

  • The significance of the intensity of the pre-maintenance treatment with regard to the role of maintenance treatment.

  • The effect of different anthracyclines used in the pre-maintenance treatment on the role of maintenance.

  • The significance of type of remission and relapse definition (hematological versus molecular) and the effect on maintenance efficacy.

  • The optimal duration of maintenance.

  • The best maintenance regimen.

  • The effect of maintenance treatment among certain groups of patients, such as the elderly and pediatric patients.   

Acknowledgements

We would like to acknowledge the following researchers for providing us valuable and updated data from their trials: Prof Martin Tallman; Prof Giuseppe Avvisati; Dr Francesca Paola Paoloni; Prof Lionel Adès; Prof Pierre Fenaux; Prof Sylvie Chevret; Prof Bayard Powell; Prof James H Feusner; Prof Norio Asou; Dr Masako Iwanaga; and Dr Marco Vignetti.

We would like to thank also Dr Nicole Skoetz and Ms Bettina Schmidtke from the Cochrane Haematological Malignancies Group for reviewing this protocol and Ms Ina Monsef for her help in constructing the search strategy.

Data and analyses

Download statistical data

Comparison 1. Any maintenance compared to observation
Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size
1 Overall survival3892Hazard Ratio (Fixed, 95% CI)0.79 [0.49, 1.27]
2 Disease-free survival51209Hazard Ratio (Fixed, 95% CI)0.59 [0.48, 0.74]
3 Relapse rate4896Risk Ratio (M-H, Fixed, 95% CI)0.69 [0.58, 0.81]
4 Low intensity maintenance vs. observation: Overall survival2719Odds Ratio (Fixed, 95% CI)0.60 [0.36, 1.01]
5 Low intensity maintenance vs. observation: Disease-free survival41039Odds Ratio (Fixed, 95% CI)0.52 [0.41, 0.66]
6 Low intensity maintenance vs. observation: Relapse rate3721Odds Ratio (M-H, Fixed, 95% CI)0.39 [0.28, 0.55]
Analysis 1.1.

Comparison 1 Any maintenance compared to observation, Outcome 1 Overall survival.

Analysis 1.2.

Comparison 1 Any maintenance compared to observation, Outcome 2 Disease-free survival.

Analysis 1.3.

Comparison 1 Any maintenance compared to observation, Outcome 3 Relapse rate.

Analysis 1.4.

Comparison 1 Any maintenance compared to observation, Outcome 4 Low intensity maintenance vs. observation: Overall survival.

Analysis 1.5.

Comparison 1 Any maintenance compared to observation, Outcome 5 Low intensity maintenance vs. observation: Disease-free survival.

Analysis 1.6.

Comparison 1 Any maintenance compared to observation, Outcome 6 Low intensity maintenance vs. observation: Relapse rate.

Comparison 2. ATRA-based maintenance compared to non-ATRA based maintenance
Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size
1 Overall survival3632Hazard Ratio (Fixed, 95% CI)1.21 [0.73, 1.98]
2 Disease-free survival4670Hazard Ratio (Fixed, 95% CI)0.72 [0.51, 1.01]
3 Relapse rate3425Risk Ratio (M-H, Fixed, 95% CI)0.88 [0.59, 1.30]
Analysis 2.1.

Comparison 2 ATRA-based maintenance compared to non-ATRA based maintenance, Outcome 1 Overall survival.

Analysis 2.2.

Comparison 2 ATRA-based maintenance compared to non-ATRA based maintenance, Outcome 2 Disease-free survival.

Analysis 2.3.

Comparison 2 ATRA-based maintenance compared to non-ATRA based maintenance, Outcome 3 Relapse rate.

Comparison 3. Maintenance with ATRA alone compared to maintenance with ATRA and chemotherapy
Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size
1 Overall survival41033Hazard Ratio (Fixed, 95% CI)0.99 [0.69, 1.43]
2 Disease-free survival41028Hazard Ratio (Fixed, 95% CI)1.38 [1.09, 1.76]
Analysis 3.1.

Comparison 3 Maintenance with ATRA alone compared to maintenance with ATRA and chemotherapy, Outcome 1 Overall survival.

Analysis 3.2.

Comparison 3 Maintenance with ATRA alone compared to maintenance with ATRA and chemotherapy, Outcome 2 Disease-free survival.

Comparison 4. ATRA alone maintenance compared to ATRA and chemotherapy maintenance in pediatric patients
Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size
1 Overall survival2133Hazard Ratio (Fixed, 95% CI)2.05 [0.47, 8.82]
2 Disease-free survival2127Hazard Ratio (Fixed, 95% CI)2.17 [1.09, 4.31]
Analysis 4.1.

Comparison 4 ATRA alone maintenance compared to ATRA and chemotherapy maintenance in pediatric patients, Outcome 1 Overall survival.

Analysis 4.2.

Comparison 4 ATRA alone maintenance compared to ATRA and chemotherapy maintenance in pediatric patients, Outcome 2 Disease-free survival.

Comparison 5. Subgroup analysis: overall survival by type of remission confirmation
Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size
1 Any maintenance compare to observation3892Hazard Ratio (Fixed, 95% CI)0.79 [0.49, 1.27]
1.1 Molecular confirmation2491Hazard Ratio (Fixed, 95% CI)1.86 [0.99, 3.48]
1.2 Hemtological confirmation1401Hazard Ratio (Fixed, 95% CI)0.25 [0.12, 0.52]
2 ATRA based maintenance compare to non-ATRA based maintenance3632Hazard Ratio (Fixed, 95% CI)1.21 [0.73, 1.98]
2.1 Molecular confirmation2310Hazard Ratio (Fixed, 95% CI)1.24 [0.67, 2.29]
2.2 Hemtological confirmation1322Hazard Ratio (Fixed, 95% CI)1.15 [0.50, 2.67]
3 ATRA alone maintenance compare to ATRA and chemotherapy maintenance41033Hazard Ratio (Fixed, 95% CI)0.99 [0.69, 1.43]
3.1 Molecular confirmation1427Hazard Ratio (Fixed, 95% CI)0.65 [0.39, 1.08]
3.2 Hemtological confirmation3606Hazard Ratio (Fixed, 95% CI)1.55 [0.92, 2.63]
Analysis 5.1.

Comparison 5 Subgroup analysis: overall survival by type of remission confirmation, Outcome 1 Any maintenance compare to observation.

Analysis 5.2.

Comparison 5 Subgroup analysis: overall survival by type of remission confirmation, Outcome 2 ATRA based maintenance compare to non-ATRA based maintenance.

Analysis 5.3.

Comparison 5 Subgroup analysis: overall survival by type of remission confirmation, Outcome 3 ATRA alone maintenance compare to ATRA and chemotherapy maintenance.

Comparison 6. Subgroup analysis: disease-free survival by type of remission confirmation
Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size
1 Any maintenance compare to observation51209Hazard Ratio (Fixed, 95% CI)0.59 [0.48, 0.74]
1.1 Molecular confirmation2488Hazard Ratio (Fixed, 95% CI)0.90 [0.57, 1.44]
1.2 Hemtological confirmation3721Hazard Ratio (Fixed, 95% CI)0.52 [0.41, 0.67]
2 ATRA based maintenance compare to non-ATRA based maintenance4670Hazard Ratio (Fixed, 95% CI)0.72 [0.51, 1.01]
2.1 Molecular confirmation2310Hazard Ratio (Fixed, 95% CI)0.61 [0.35, 1.07]
2.2 Hematological confirmation2360Hazard Ratio (Fixed, 95% CI)0.79 [0.52, 1.21]
3 ATRA alone maintenance compare to ATRA and chemotherapy maintenance41028Hazard Ratio (Fixed, 95% CI)1.38 [1.09, 1.76]
3.1 Molecular confirmation1432Hazard Ratio (Fixed, 95% CI)1.08 [0.76, 1.54]
3.2 Hemtological confirmation3596Hazard Ratio (Fixed, 95% CI)1.72 [1.23, 2.41]
Analysis 6.1.

Comparison 6 Subgroup analysis: disease-free survival by type of remission confirmation, Outcome 1 Any maintenance compare to observation.

Analysis 6.2.

Comparison 6 Subgroup analysis: disease-free survival by type of remission confirmation, Outcome 2 ATRA based maintenance compare to non-ATRA based maintenance.

Analysis 6.3.

Comparison 6 Subgroup analysis: disease-free survival by type of remission confirmation, Outcome 3 ATRA alone maintenance compare to ATRA and chemotherapy maintenance.

Comparison 7. Subgroup analysis: overall survival by type of prior anthracycline use
Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size
1 Any maintenance compare to observation3892Hazard Ratio (Fixed, 95% CI)0.79 [0.49, 1.27]
1.1 Daunorubicin use1401Hazard Ratio (Fixed, 95% CI)0.25 [0.12, 0.52]
1.2 Idarubicin use2491Hazard Ratio (Fixed, 95% CI)1.86 [0.99, 3.48]
2 ATRA based maintenance compare to non-ATRA based maintenance3554Hazard Ratio (Fixed, 95% CI)1.21 [0.73, 1.98]
2.1 Daunorubicin use2390Hazard Ratio (Fixed, 95% CI)1.31 [0.64, 2.70]
2.2 Idarubicin use1164Hazard Ratio (Fixed, 95% CI)1.12 [0.56, 2.22]
3 ATRA alone maintenance compare to ATRA and chemotherapy maintenance41033Hazard Ratio (Fixed, 95% CI)0.99 [0.69, 1.43]
3.1 Daunorubicin use3606Hazard Ratio (Fixed, 95% CI)1.55 [0.92, 2.63]
3.2 Idarubicin use1427Hazard Ratio (Fixed, 95% CI)0.65 [0.39, 1.08]
Analysis 7.1.

Comparison 7 Subgroup analysis: overall survival by type of prior anthracycline use, Outcome 1 Any maintenance compare to observation.

Analysis 7.2.

Comparison 7 Subgroup analysis: overall survival by type of prior anthracycline use, Outcome 2 ATRA based maintenance compare to non-ATRA based maintenance.

Analysis 7.3.

Comparison 7 Subgroup analysis: overall survival by type of prior anthracycline use, Outcome 3 ATRA alone maintenance compare to ATRA and chemotherapy maintenance.

Comparison 8. Subgroup analysis: disease-free survival by type of prior anthracycline use
Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size
1 Any maintenance compare to observation51209Hazard Ratio (Fixed, 95% CI)0.59 [0.48, 0.74]
1.1 Daunorubicin use2605Hazard Ratio (Fixed, 95% CI)0.44 [0.33, 0.60]
1.2 Idarubicin use3604Hazard Ratio (Fixed, 95% CI)0.85 [0.61, 1.18]
2 ATRA based maintenance compare to non-ATRA based maintenance4592Hazard Ratio (Fixed, 95% CI)0.72 [0.51, 1.01]
2.1 Daunorubicin use3428Hazard Ratio (Fixed, 95% CI)0.81 [0.53, 1.22]
2.2 Idarubicin use1164Hazard Ratio (Fixed, 95% CI)0.56 [0.30, 1.03]
3 ATRA alone maintenance compare to ATRA and chemotherapy maintenance41028Hazard Ratio (Fixed, 95% CI)1.38 [1.09, 1.76]
3.1 Daunorubicn use3596Hazard Ratio (Fixed, 95% CI)1.72 [1.23, 2.41]
3.2 Idarubicin use1432Hazard Ratio (Fixed, 95% CI)1.08 [0.76, 1.54]
Analysis 8.1.

Comparison 8 Subgroup analysis: disease-free survival by type of prior anthracycline use, Outcome 1 Any maintenance compare to observation.

Analysis 8.2.

Comparison 8 Subgroup analysis: disease-free survival by type of prior anthracycline use, Outcome 2 ATRA based maintenance compare to non-ATRA based maintenance.

Analysis 8.3.

Comparison 8 Subgroup analysis: disease-free survival by type of prior anthracycline use, Outcome 3 ATRA alone maintenance compare to ATRA and chemotherapy maintenance.

Comparison 9. Sensitivity analysis: overall survival by method of allocation concealment, allocation generation
Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size
1 ATRA based maintenance compare to non-ATRA based maintenance3632Hazard Ratio (Fixed, 95% CI)1.21 [0.73, 1.98]
1.1 adequate allocation generation and concealment2564Hazard Ratio (Fixed, 95% CI)1.13 [0.66, 1.92]
1.2 unclear168Hazard Ratio (Fixed, 95% CI)1.88 [0.47, 7.55]
Analysis 9.1.

Comparison 9 Sensitivity analysis: overall survival by method of allocation concealment, allocation generation, Outcome 1 ATRA based maintenance compare to non-ATRA based maintenance.

Comparison 10. Sensitivity analysis: disease-free survival by method of allocation concealment, allocation generation
Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size
1 ATRA based maintenance compare to non-ATRA based maintenance4670Hazard Ratio (Fixed, 95% CI)0.72 [0.51, 1.01]
1.1 adequate allocation generation and concealment2564Hazard Ratio (Fixed, 95% CI)0.73 [0.51, 1.03]
1.2 unclear2106Hazard Ratio (Fixed, 95% CI)0.62 [0.16, 2.41]
Analysis 10.1.

Comparison 10 Sensitivity analysis: disease-free survival by method of allocation concealment, allocation generation, Outcome 1 ATRA based maintenance compare to non-ATRA based maintenance.

Comparison 11. Sensitivity analysis: overall survival by duration of follow-up
Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size
1 Any maintenance compare to observation3892Hazard Ratio (Fixed, 95% CI)0.79 [0.49, 1.27]
1.1 over 5 years3892Hazard Ratio (Fixed, 95% CI)0.79 [0.49, 1.27]
1.2 5 years or less00Hazard Ratio (Fixed, 95% CI)0.0 [0.0, 0.0]
2 ATRA based maintenance compare to non-ATRA based maintenance3711Hazard Ratio (Fixed, 95% CI)0.98 [0.64, 1.51]
2.1 over 5 years2643Hazard Ratio (Fixed, 95% CI)0.92 [0.58, 1.45]
2.2 5 years or less168Hazard Ratio (Fixed, 95% CI)1.88 [0.47, 7.55]
3 ATRA alone maintenance compare to ATRA and chemotherapy maintenance41033Hazard Ratio (Fixed, 95% CI)0.99 [0.69, 1.43]
3.1 over 5 years3963Hazard Ratio (Fixed, 95% CI)0.99 [0.68, 1.43]
3.2 5 years or less170Hazard Ratio (Fixed, 95% CI)1.14 [0.16, 8.08]
Analysis 11.1.

Comparison 11 Sensitivity analysis: overall survival by duration of follow-up, Outcome 1 Any maintenance compare to observation.

Analysis 11.2.

Comparison 11 Sensitivity analysis: overall survival by duration of follow-up, Outcome 2 ATRA based maintenance compare to non-ATRA based maintenance.

Analysis 11.3.

Comparison 11 Sensitivity analysis: overall survival by duration of follow-up, Outcome 3 ATRA alone maintenance compare to ATRA and chemotherapy maintenance.

Comparison 12. Sensitivity analysis: disease-free survival by duration of follow-up
Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size
1 Any maintenance compare to observation51209Hazard Ratio (Fixed, 95% CI)0.59 [0.48, 0.74]
1.1 over 5 years51209Hazard Ratio (Fixed, 95% CI)0.59 [0.48, 0.74]
1.2 5 years or less00Hazard Ratio (Fixed, 95% CI)0.0 [0.0, 0.0]
2 ATRA based maintenance compare to non-ATRA based maintenance4749Hazard Ratio (Fixed, 95% CI)0.61 [0.46, 0.83]
2.1 over 5 years2643Hazard Ratio (Fixed, 95% CI)0.61 [0.45, 0.83]
2.2 5 years or less2106Hazard Ratio (Fixed, 95% CI)0.62 [0.16, 2.41]
3 ATRA alone maintenance compare to ATRA and chemotherapy maintenance41028Hazard Ratio (Fixed, 95% CI)1.38 [1.09, 1.76]
3.1 over 5 years3964Hazard Ratio (Fixed, 95% CI)1.37 [1.07, 1.77]
3.2 5 years or less164Hazard Ratio (Fixed, 95% CI)1.54 [0.65, 3.64]
Analysis 12.1.

Comparison 12 Sensitivity analysis: disease-free survival by duration of follow-up, Outcome 1 Any maintenance compare to observation.

Analysis 12.2.

Comparison 12 Sensitivity analysis: disease-free survival by duration of follow-up, Outcome 2 ATRA based maintenance compare to non-ATRA based maintenance.

Analysis 12.3.

Comparison 12 Sensitivity analysis: disease-free survival by duration of follow-up, Outcome 3 ATRA alone maintenance compare to ATRA and chemotherapy maintenance.

Comparison 13. Sensitivity analysis by overall survival (OS) measure method
Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size
1 ATRA based maintenance compare to non-ATRA based maintenance3632Hazard Ratio (Fixed, 95% CI)1.21 [0.73, 1.98]
1.1 OS measure from time of randomization to maintenance2564Hazard Ratio (Fixed, 95% CI)1.13 [0.66, 1.92]
1.2 Unknown time measure for OS data168Hazard Ratio (Fixed, 95% CI)1.88 [0.47, 7.55]
2 ATRA alone maintenance compare to ATRA and chemotherapy maintenance41033Hazard Ratio (Fixed, 95% CI)0.99 [0.69, 1.43]
2.1 OS measure from time of randomization to maintenance2632Hazard Ratio (Fixed, 95% CI)0.84 [0.53, 1.32]
2.2 OS measure from study entry2401Hazard Ratio (Fixed, 95% CI)1.35 [0.73, 2.50]
Analysis 13.1.

Comparison 13 Sensitivity analysis by overall survival (OS) measure method, Outcome 1 ATRA based maintenance compare to non-ATRA based maintenance.

Analysis 13.2.

Comparison 13 Sensitivity analysis by overall survival (OS) measure method, Outcome 2 ATRA alone maintenance compare to ATRA and chemotherapy maintenance.

Comparison 14. Sensitivity analysis by disease-free survival (DFS) measure method
Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size
1 ATRA based maintenance compare to non-ATRA based maintenance4670Hazard Ratio (Fixed, 95% CI)0.72 [0.51, 1.01]
1.1 DFS measure from randomization to maintenance3632Hazard Ratio (Fixed, 95% CI)0.74 [0.52, 1.04]
1.2 DFS measure prior to maintenance initiation138Hazard Ratio (Fixed, 95% CI)0.12 [0.01, 1.83]
Analysis 14.1.

Comparison 14 Sensitivity analysis by disease-free survival (DFS) measure method, Outcome 1 ATRA based maintenance compare to non-ATRA based maintenance.

Comparison 15. Subgroup analysis: survival by subtype of maintenance compared to observation
Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size
1 Overall survival3679Hazard Ratio (Fixed, 95% CI)0.55 [0.36, 0.86]
1.1 Chemotherapy alone maintenance compared to observation2369Hazard Ratio (Fixed, 95% CI)0.60 [0.29, 1.23]
1.2 ATRA alone maintenance compared to observation1155Hazard Ratio (Fixed, 95% CI)0.53 [0.24, 1.15]
1.3 ATRA containing regimen compared to observation1155Hazard Ratio (Fixed, 95% CI)0.53 [0.24, 1.15]
2 Disease-free survival41088Hazard Ratio (Fixed, 95% CI)0.60 [0.50, 0.72]
2.1 Chemotherapy alone maintenance compared to observation3440Hazard Ratio (Fixed, 95% CI)0.94 [0.68, 1.30]
2.2 ATRA alone maintenance compared to observation2204Hazard Ratio (Fixed, 95% CI)0.47 [0.33, 0.66]
2.3 ATRA containing regimen maintenance compared to observation2444Hazard Ratio (Fixed, 95% CI)0.48 [0.35, 0.66]
Analysis 15.1.

Comparison 15 Subgroup analysis: survival by subtype of maintenance compared to observation, Outcome 1 Overall survival.

Analysis 15.2.

Comparison 15 Subgroup analysis: survival by subtype of maintenance compared to observation, Outcome 2 Disease-free survival.

Appendices

Appendix 1. Search strategy (CENTRAL)

Appendix 2. Search strategy (Ovid MEDLINE)

# Searches
1exp LEUKEMIA, MYELOID, ACUTE/
2LEUKEMIA, MYELOID/
3ACUTE DISEASE/
42 and 3
5(acut$ or akut$ or agud$ or aigu$).tw,kf,ot.
6((myelo$ or mielo$ or nonlympho$ or granulocytic$) and (leuk?em$ or leuc$)).tw,kf,ot.
75 and 6
8aml.tw,kf,ot.
9LEUKEMIA, PROMYELOCYTIC, ACUTE/
10(acut$ or akut$ or agud$ or aigu$).tw,kf,ot.
11((promyelocyt$ or promielocitic$ or promyelozyt$ or progranulocyt$) and (leuk?em$ or leuc$)).tw,kf,ot.
1210 and 11
13apl.tw,kf,ot.
141 or 4 or 7 or 8 or 9 or 12 or 13
15exp MAINTENANCE/
16maintenanc$.tw,kf,ot.
17((post-remission$ or postremission$) adj2 therap$).tw,kf,ot.
18or/15-17
1914 and 18
20randomized controlled trial.pt.
21controlled clinical trial.pt.
22randomized.ab.
23placebo.ab.
24drug therapy.fs.
25randomly.ab.
26trial.ab.
27groups.ab.
28or/20-27
29humans.sh.
3028 and 29
3119 and 30

Appendix 3. Search strategy (LILACS)

We searched the following terms:

acute promyelocytic leukemia

APL

What's new

DateEventDescription
10 June 2013AmendedMinor changes due to external feedback.

Contributions of authors

Eli Muchtar: conception and design, provision of study material, protocol development, search for trials, data extraction, analysis and data interpretation, writing of the review and final approval of the review.

Liat Vidal: conception and design, provision of study material, protocol development, search for trials, data extraction, analysis and data interpretation, final approval of the review.

Ron Ram: conception, protocol development, analysis and data interpretation, final approval of the review.

Anat Gafter-Gvili: protocol development, search for trials, analysis and data interpretation, final approval of the review.

Ofer Shpilberg: protocol development and design, analysis and data interpretation, final approval of the review, clinical and scientific advice.

Pia Raanani: conception and design, provision of study material, protocol development, analysis and data interpretation, final approval of the review, clinical and scientific advice.

Declarations of interest

None known to declare.

Sources of support

Internal sources

  • Sackler School of Medicine, Tel-Aviv University, Tel-Aviv, Israel.

External sources

  • No sources of support supplied

Differences between protocol and review

# We used the most updated available results for data analysis. In the case where more updated data were provided by the authors, different from the published data, we used the updated data.

# For both dichotomous data and time-to-event outcomes we used the fixed-effect model. Where significant heterogeneity was noted we explored reasons for heterogeneity and if not found we also used the random-effects model for the pooled data analysis.

# For sensitivity analysis we also applied the median follow-up period, stratifying for trials with a follow-up period over five years and for trials with a median follow-up of five years or less.

Characteristics of studies

Characteristics of included studies [ordered by study ID]

Adès 2010

Methods

Location: Europe

Period:1993 to 1998

Median follow up: 10.1 years

Participants

401 randomized, 401 evaluable

Main eligibility criteria: patients with confirmed APL by PML-RARa hybrid gene

Age restrictions: under 75 years

Treatment given before maintenance:

# induction: one cycle of ATRA plus cytarabine and daunorubicin. Induction treatment was stratified by age and initial white blood cell (WBC) count: Patients 65 years or younger with WBC count lower than 5X109/L were randomized to receive ATRA 45 mg/m2 per day followed by chemotherapy (CT) - daunorubicin (DNR) 60 mg/m2 per day for 3 days, and AraC 200 mg/m2 per day as continuous infusion for 7 days or ATRA plus CT, where CT was started on day 3 of ATRA treatment.
Patients with WBC count higher than 5X109/L at presentation (irrespective of their age) and patients aged 66 to 75 years with WBC count of 5X109/L or lower were not randomized but received ATRA with the addition of CT on day 1 of ATRA treatment (high WBC count group) and the same schedule as in the ATRA followed by CT group (elderly group), respectively.

# consolidation: After achieving complete remission (CR) patients received 2 cycles containing cytarabine plus daunorubicin (the first identical to the induction course, the second with daunorubicin 45 mg/m2 per day for 3 days and cytarabine 1 g/m2 per 12 hours for 4 days).

Median age: 46 years (range 28 to 72) (data from study entry)

Interventions

Investigational interventions:

# ATRA (76 patients)

# ATRA+MTX+6-MP (129 patients)

# MTX+6-MP (117 patients)

Comparator intervention: Observation (79 patients)

intervention period: 2 years

Outcomes

Primary outcome for maintenance treatment: time to relapse

was calculated 3-4 weeks following attainment of CR after the second consolidation.

Secondary outcomes for maintenance treatment: overall survival, event-free survival

were calculated from the date of randomization to maintenance following CR attainment.

Adverse events

Notes

Published as journal article

EFS from randomization to maintenance represents disease-free survival

Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Low risk"random generation (computer-based) using permutation blocks" by author's reply
Allocation concealment (selection bias)Low risk"Randomizations for induction and maintenance were performed through a centralized telephone assignment procedure"
Blinding of participants and personnel (performance bias)
All outcomes
Unclear riskThe study was open-label
Blinding of outcome assessment (detection bias)
All outcomes
Unclear riskNot reported
Incomplete outcome data (attrition bias)
All outcomes
Low riskAll patients were included in the analysis of overall survival and disease-free survival
Selective reporting (reporting bias)Low riskThe outcomes of the maintenance phase outlined in the trial's protocol were addressed
Other biasLow risk

1. Funding: the trial was granted by non-pharmaceutical organization (Programme Hospitalier de Recherche Clinique and the Centre Hospitalier Universitaire (CHU)
of Lille, France)

2. Intention-to-treat (ITT) analysis: yes

3. Sample size calculation: performed and carried-out

4. Early termination of trial: no

Asou 2007

Methods

Location: Japan

Period:1997 to 2002

Median follow-up: 5.3 years

Participants

175 randomised, 175 evaluable

Main eligibility criteria: diagnosis of APL based on t(15;17) and/or the PML-RARα fusion gene; ECOG performance status 0 to 3; sufficient function of major organs (heart, lungs, liver and kidneys)

Age restrictions: 15 to 70 years

Median age: 48 years (from study entry)

Treatment given before maintenance:

# induction: patients received ATRA. If patients had leukocyte counts below 3.0X109/L and APL cells below 109/L at the start of therapy, they were treated with ATRA alone (group A). ATRA combined with idarubicin plus cytarabine was given to patients with initial leukocyte counts between 3.0X109/L and 10.0X109/L, and those with leukocyte counts below 3.0X109/L and APL cells above 109/L (group B). Patients with initial leukocyte counts of 10.0X109/L or more received intensified chemotherapy (idarubicin plus cytarabine) in addition to ATRA (group C).

# consolidation: 3 cycles of consolidation. The first consolidation consisted of mitoxantrone and cytarabine. The second consolidation contained cytarabine plus etoposide and daunorubicin. The third consolidation consisted of cytarabine.

Interventions

Investigational interventions: intensified chemotherapy of 6 courses every 6 weeks. The first course consisted of behenoyl cytarabine (BHAC), daunorubicin and mercaptopurine (6MP). The second consisted of BHAC and mitoxantrone. The third consisted of BHAC, etoposide, and vindesine. The fourth consisted of BHAC, aclarubicin and 6MP. The fifth and sixth courses were the same as the first and third, respectively (89 patients).

Comparator intervention: Observation (86 patients).

intervention period: 6 cycles of 6 weeks each

Outcomes

Primary outcomes: OS; DFS

Overall survival and disease-free survival in patients randomized to either observation or maintenance chemotherapy were measured following achievement of molecular CR.

Adverse events

NotesPublished as journal article
Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Low riskAuthor's reply - "We used a computer-generated randomization method, using permuted blocks, and stratified according to age and initial leukocyte count"
Allocation concealment (selection bias)Low riskAuthor's reply - "We used a centralized computer system via the Internet"
Blinding of participants and personnel (performance bias)
All outcomes
Unclear riskThe study was open-label
Blinding of outcome assessment (detection bias)
All outcomes
Unclear riskNot reported
Incomplete outcome data (attrition bias)
All outcomes
Low riskMissing data were reported by trial's authors and reasons were given
Selective reporting (reporting bias)Low riskTrial's protocol was not available, however according to our judgement all necessary outcomes were reported
Other biasLow risk

1. Funding: the trial was granted by a governmental non-pharmaceutical organization (Japanese Ministry of Education, Culture, Sport, Science, and Technology and Japanese Ministry of Health, Labor, and Welfare).

2. Intention-to-treat (ITT) analysis: yes

3. Sample size calculation: not reported

4. Early termination of trial: no

Avvisati 2002

Methods

Location: Italy

Period:1989 to 1993

Median follow up: minimum 7.5 years

Participants

116 randomized, 116 evaluable

Main eligibility criteria: diagnosis of hypergranular APL (microgranular APL was excluded from the study) morphologically defined according to the FAB classification

Age restrictions: range of age 12 to 62 years

Median age: median 36.9 (from study entry)

Treatment given before maintenance:

# induction: one cycle of idarubicin plus cytarabine

# consolidation: 3 cycles of consolidation. The first consolidation consisted of idarubicin and cytarabine. The second consolidation contained mitoxantrone plus VP-16. The third consolidation consisted of idarubicin plus cytarabine and 6-TG.

Interventions

Investigational intervention: 6-MP plus MTX (58 patients).

Comparator intervention: observation (58 patients).

Intervention period: two years

Outcomes

Primary outcomes: DFS, OS

Disease-free survival was defined as the time from the achievement of CR to relapse or death in CR.

OS was defined as time from randomization to induction till death from any cause.

Adverse events

NotesPublished as journal article
Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Low risk"Investigators were masked to the potential treatment assignment by central randomization performed according to permuted blocks"
Allocation concealment (selection bias)Low riskCentral randomization
Blinding of participants and personnel (performance bias)
All outcomes
Unclear riskThe study was open-label
Blinding of outcome assessment (detection bias)
All outcomes
Unclear riskNot reported
Incomplete outcome data (attrition bias)
All outcomes
Low risk58 patients were randomized to maintenance. However, five patients in the maintenance arm did not complete the planned treatment because of toxicity, 1 underwent allogeneic bone marrow transplantation, and 1 refused to continue maintenance
Selective reporting (reporting bias)Low riskTrial's protocol was not available, however our judgement is that all necessary outcomes were reported
Other biasLow risk

1. Funding: not reported

2. Intention-to-treat (ITT) analysis: yes

3. Sample size calculation: performed and carried-out

4. Early termination of trial: no

Avvisati 2011

Methods

Location: Italy

Period:1993 to 2000

Median follow up: 7.8 years

Participants

586 randomized, 586 evaluable

Main eligibility criteria: patients with confirmed APL by t(15;17) or PML-RARa hybrid gene

Age restrictions: age under 75 years

Treatment given before maintenance:

# induction: one cycle of idarubicin and ATRA

# consolidation: 3 cycles. The first cycle consisted of idarubicin plus cytarabine; the second cycle consisted of mitoxantrone plus VP16; and the third cycle consisted of idarubicin plus cytarabine and 6-TG

Median age: 32.6, 41.6, 36.3 35.8, 33.4, 33.7 years for the different maintenance comparators

Interventions

Investigational intervention:

# ATRA (83 patients).

# ATRA+MTX+6-MP (81 patients).

# MTX+6-MP (78 patients).

Comparator intervention: Observation (76 patients).

In the mid-course of the study an amendment was made in the maintenance protocol. The reason for this not pre-defined amendment was observation at the first period after randomization that there was a worse DFS form observation arm compare to the other 3 original arms of maintenance. Therfore, arms of chemotherapy alone and observation were canceled and patients were randomized into the 2 following maintenance arms:

# ATRA (137 patients).

# ATRA+MTX+6-MP (131 patients).

Intervention period: two years, for both maintenance protocols.

Outcomes

Primary outcomes: molecular DFS

Secondary outocmes: OS; adverse events

OS was calculated from the date of diagnosis until death from any cause; molecular DFS was calculated from molecular CR following the last cycle of consolidation until death in hematological CR and molecular or hematologic relapse, depending on which one occurred first.

NotesPublished as journal article
Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Low riskPermuted blocks
Allocation concealment (selection bias)Low risk"Central randomization was performed stratifying the patients according to the participating Institutions"
Blinding of participants and personnel (performance bias)
All outcomes
Unclear riskThe study was open-label study
Blinding of outcome assessment (detection bias)
All outcomes
Unclear riskNot reported
Incomplete outcome data (attrition bias)
All outcomes
Low riskAll patients were included in the analysis of overall survival and disease-free survival.
Selective reporting (reporting bias)Low riskThe outcomes of maintenance phase outlined in study protocol were addressed.
Other biasLow risk

1. Funding: the trial was granted by non-pharmaceutical organization (Associazione Italiana contro le Leucemie and Associazione Italiana contro il Cancro).

2. Intention-to-treat (ITT) analysis: yes

3. Sample size calculation: performed and carried out

4. Early termination of trial: no

Feusner 2010

Methods

Location: USA

Period:1999 to 2005

Median follow up: 3.9 years

Participants

70 randomized, 70 evaluable

Main eligibility criteria: patients with confirmed APL by PML-RARa hybrid gene

Age restrictions: age less than 15 years

Treatment given before maintenance:

# Induction: one cycle of ATRA plus cytarabine plus daunorubicin

# consolidation: 3 cycles, each contain ATRA and daunorubicin

Median age: not available

Interventions

Investigational intervention: ATRA (32 patients).

Comparator intervention: ATRA+6-MP+MTX (38 patients).

intervention period: one year

Outcomes

Primary outcome: DFS

DFS was defined as the time from attainment of CR to relapse or death.

Secondary outcomes: OS

OS was defined as the time from study entry to death.

Adverse events

Notes

Published as abstract.

Same trial design as Powell 2010. However, the trial reports exclusively results in children aged<15 years. By study design, children 15 years of age were assigned to the standard non-As2O3 arm during consolidation.

Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Low riskRandom numbers, based on the report in Powell 2010 as both publications report on the same trial
Allocation concealment (selection bias)Low riskCentral randomization, based on the report in Powell 2010 as both publications report on the same trial
Blinding of participants and personnel (performance bias)
All outcomes
Unclear riskOpen-label study
Blinding of outcome assessment (detection bias)
All outcomes
Unclear riskNot reported
Incomplete outcome data (attrition bias)
All outcomes
Low riskTwo lost to follow-up patients were reported, allocation unclear. As it represents only 3% of the study population the effect of this missing data is small
Selective reporting (reporting bias)Low riskThe outcomes of maintenance phase outlined in the study protocol are addressed in the trial publication, except for adverse events (toxicity outcome)
Other biasLow risk

1. Funding: the trial was granted by non-pharmaceutical organization (National Cancer Institute).

2. Intention-to-treat (ITT) analysis: yes

3. Sample size calculation: performed

4. Early termination of trial: no

Lin 2007

Methods

Location: China

Period: not mentioned

Median follow up: 3 years (not mentioned whether median)

Participants

60 randomized, 60 evaluable

Main eligibility criteria: patients with diagnosis of APL based on bone marrow morphology and t(15;17) translocation

Age restrictions: none

Treatment given before maintenance:

# induction: ATRA until remission achievement

# consolidation: 6 cycles of chemotherapy divided into: 2 courses of daunorubicin and cytarabine, 2 courses of homoharringtonine and cytarabine, and 2 courses of aclarubicin and cytarabine.

Median age: 34 and 32 years for each arm

Interventions

Investigational intervention:

ATRA + As4S4 + 6-MP+MTX (30 patients).

Comparator intervention: ATRA + 6-MP+MTX (30 patients).

Intervention period: 2 to 3 years and more

Outcomes

Overall survival, CCR (complete cytogenetic response) - not defined in the article

adverse effects

NotesPublished as journal article in Chinese, Mandarin
Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Unclear riskNot reported
Allocation concealment (selection bias)Unclear riskNot reported
Blinding of participants and personnel (performance bias)
All outcomes
Unclear riskThe study was open-label
Blinding of outcome assessment (detection bias)
All outcomes
Unclear riskNot reported
Incomplete outcome data (attrition bias)
All outcomes
Low riskAll patients were included in the analysis of overall survival and disease-free survival
Selective reporting (reporting bias)Low riskTrial's protocol was not available for assessment, however our judgement is that all necessary outcome were reported
Other biasUnclear risk

1. Funding: unclear

2. Intention-to-treat (ITT) analysis: yes

3. Sample size calculation: unclear

4. Early termination of trial: no

Parovichnikova 2004

Methods

Location: Russia

Period: 2001-2002

Median follow up: maximum 2.5 years

Participants

68 randomized, 68 evaluable

Main eligibility criteria:

Age restrictions: not mentioned

Median age: 34 years in the maintenance chemotherapy arm and 32 years in the chemotherapy and ATRA maintenance arm (range 16-72)

Treatment given before maintenance:

# induction: one cycle of daunorubicin+cytarabine (3+7) and ATRA for 30 days.

# consolidation: 2 cycles of daunorubicin+cytarabine in doses and timing identical to the induction phase.

Interventions

Investigational intervention: rotation of 5-days cytarabine (100 mg/m2 bid) in combination with one of the following drugs: daunorubicin (45 mg/m2 for 2 days, up to a total dose of 650 mg/m2) or 6-MP (50 mg/m2 for 5 days) or cyclophosphamide (800 mg/m2 i.v. 1 day). Interval between courses was 4 weeks (34 patients).

Comparator intervention: alternate same chemotherapy (but half dose) with ATRA 45 mg/m2 for 5 days. Interval between courses was 4 weeks (34 patients).

intervention period: 2 years

Outcomes

DFS

OS

NotesPublished as journal article in Russian; updated abstract was published in 2007
Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Unclear riskNot reported
Allocation concealment (selection bias)Unclear riskNot reported
Blinding of participants and personnel (performance bias)
All outcomes
Unclear riskOpen-label study
Blinding of outcome assessment (detection bias)
All outcomes
Unclear riskNot reported
Incomplete outcome data (attrition bias)
All outcomes
Unclear riskNot reported
Selective reporting (reporting bias)High riskTrial's protocol was not available for assessment
Other biasUnclear risk

1. Funding: not reported

2. Intention-to-treat (ITT) analysis: not reported

3. Sample size calculation: not reported

4. Early termination of trial: no

Powell 2010

Methods

Location: USA

Period:1999 to 2005

Median follow up: 6.2 years

Participants

331 randomized, 331 evaluable

Main eligibility criteria: patients with confirmed APL by PML-RARa hybrid gene

Age restrictions: age above 15

Treatment given before maintenance:

At study entry, patients were randomized to standard induction and consolidation therapy with or without two 25-day courses of As2O3 consolidation given after the standard induction and before the consolidation treatments.

# induction: one cycle of ATRA plus cytarabine and daunorubicin. Then patients were randomized to receive or not receive 2 cycles of As2O3.

# consolidation: 3 cycles containing ATRA and daunorubicin.

Median age: 43 years in the ATRA arm, 40 years in the ATRA and chemotherapy arm

Interventions

Investigational intervention: ATRA (166 patients).

Comparator intervention: ATRA+6-MP+MTX (165 patients).

Intervention period: one year

Outcomes

Primary outcome: DFS

DFS was defined as the time from attainment of CR to relapse or death.

Secondary outcomes: OS

Overall survival was defined as the time from study entry to death.

Adverse events

NotesPublished as journal article
Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Low riskRandom numbers, based on author's reply
Allocation concealment (selection bias)Low riskCentral randomization, based on author's reply
Blinding of participants and personnel (performance bias)
All outcomes
Unclear riskOpen-label study
Blinding of outcome assessment (detection bias)
All outcomes
Unclear riskNot reported
Incomplete outcome data (attrition bias)
All outcomes
Low riskMissing data in 7 patients: 5 were lost to follow-up after maintenance randomization and completion of the protocol treatment. 2 patients refused treatment after randomization. the allocation was not given. As the missing data concern a small proportion of the study population (7 out of 331 patients, 2.1%), we judged this study as low risk of bias for incomplete outcome data
Selective reporting (reporting bias)Low riskThe outcomes of maintenance phase outlined in the study protocol were addressed
Other biasLow risk

1. Funding: the trial was granted by an academic non-pharmaceutical body (National Cancer Institute).

2. Intention-to-treat (ITT) analysis: yes

3. Sample size calculation: performed and carried out

4. Early termination of trial: no

Shen 2004

Methods

Location: China

Period: 2001 to 2003

Median follow up: median 1.5 years

Participants

61 randomized, 57 evaluable

Main eligibility criteria: newly diagnosed APL without prior exposure to any anti-leukemic therapy

Age restrictions: not mentioned

Median age: 30.5, 39.5, 34 years (from study entry)

Treatment given before maintenance:

# induction: 3 groups according to randomization: ATRA alone, As2O3 alone, combination of ATRA and As2O3.

# consolidation: 3 cycles of consolidation. the first cycle contained daunorubicin and cytarabine the second cycle contained intensive cytarabine, the third cycle contained homoharringtonine and cytarabine.

Interventions

Investigational intervention:

# ATRA for 30 days; then 6-MP for 30 days or MTX for 4 weeks (19 patients).

# As2O3 for 30 days; then 6-MP for 30 days or MTX for 4 weeks (18 patients).

# ATRA for 30 days; then As2O3 for 30 days; then 6-MP for 30 days or MTX for 4 weeks (20 patients).

intervention period: 5 cycles

Outcomes

Primary coutcome: DFS

DFS was defined as the time from CR to relapse, death from any causes, or censoring of the data of patients. CR was calculated following induction and before initiation of consolidation and maintenance treatments.

NotesPublished as journal article
Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Unclear riskNot reported
Allocation concealment (selection bias)Unclear riskNot reported
Blinding of participants and personnel (performance bias)
All outcomes
Unclear riskNot reported
Blinding of outcome assessment (detection bias)
All outcomes
Unclear riskNot reported
Incomplete outcome data (attrition bias)
All outcomes
Low riskAll patients were included in the analysis of overall survival and disease-free survival
Selective reporting (reporting bias)High riskTrial's protocol was not available for assessment, no OS data were reported
Other biasHigh risk

1. Funding: partially by non-pharmaceutical organizations (Chinese National Basic Research Program, National Natural Science Foundation of China, Chinese National High Tech Program, Shanghai Municipal Commission for Science and Technology, Shanghai Leading Academic Discipline, and the Poˆle Sino-Français en Sciences du Vivant et en Génomique).

2. Intention-to-treat (ITT) analysis: yes

3. Sample size calculation: not reported

4. Early termination of trial: yes (reported in Hu 2009)

Tallman 2002

Methods

Location: USA

Period:1992 to 1995

Median follow up: 6.2 years

Participants

204 randomized, 204 evaluable

Main eligibility criteria: patients with diagnosis of APL based on bone marrow morphology and t(15;17) translocation

Age restrictions: none

Treatment given before maintenance:

# induction: patients were randomized to ATRA until CR or 1 to 2 cycles of daunorubicin plus cytarabine.

# consolidation: 2 cycles. The first cycle was identical to the chemotherapy administered in induction. The second cycle consisted of high-dose cytarabine plus daunorubicin.

Median age: available from induction phase (37 years in the daunorubicin group and 38 years in the ATRA group)

Interventions

Investigational intervention: ATRA (99 patients).

Comparator intervention: observation (105 patients).

Intervention period: one year

Outcomes

Primary outcome: DFS

DFS was defined as the time from attainment of complete remission to relapse, death from any cause, or censoring of the data of the patient.

Secondary outcomes: OS

defined as the time from registration to death.

Adverse events

NotesPublished as journal article
Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Unclear riskNot reported
Allocation concealment (selection bias)Low riskCentral randomization, based on author's reply
Blinding of participants and personnel (performance bias)
All outcomes
Unclear riskOpen-label study
Blinding of outcome assessment (detection bias)
All outcomes
Unclear riskNot reported
Incomplete outcome data (attrition bias)
All outcomes
Low riskAll patients were included in primary outcome analysis
Selective reporting (reporting bias)Low riskTrial's protocol was not available, however our judgement is that all necessary outcome were reported
Other biasLow risk

1. Funding: in part by non-pharmaceutical organizations (National Cancer Institute, National Institutes of Health and the Department of Health and Human Services).

2. Intention-to-treat (ITT) analysis: yes

3. Sample size calculation: not reported

4. Early termination of trial: no

Characteristics of excluded studies [ordered by study ID]

StudyReason for exclusion
Adès 2010ano maintenance evaluation
Asou 1998maintenance was not randomized
Asou 2001maintenance was not randomized (duplicate of Asou 1998)
Bapna 1998non-randomized study
Burnett 1999no maintenance evaluation
Burnett 2007maintenance was not randomized
Castaigne 1990non-randomized study
Dai 2009non-randomized study
Estey 2001non-randomized study
Estey 2005no maintenance evaluation
Fenaux 1991maintenance was not randomized
Fenaux 1992maintenance was not randomized
Fenaux 2000no maintenance evaluation
Frankel 1994non-randomized study
Gore 2010maintenance was not randomized (risk-based maintenance)
Head 1995non-randomized study
Hu 2000non-randomized study
Iland 2012non-randomized study
Jurcic 2000maintenance was not randomized
Kantarjian 1987non-randomized study
Kobayashi 2010non-randomized study (case report)
Koh 2001maintenance was not randomized
Lengfelder 2000maintenance was not randomized
Lengfelder 2009maintenance was not randomized (duplicate of Lengfelder 2000)
Lu 2002non-randomized study (review article)
Martinelli 1998maintenance was not randomized
Marty 1984non-randomized study
Mathews 2009same maintenance was randomized between 2 time periods
McMullin 2005non-randomized study (case report)
Ortega 2005non-randomized study
Santamaria 2007maintenance was not randomized
Sanz 1999maintenance was not randomized
Sanz 2004non-randomized study
Sanz 2004amaintenance was not randomized
Seiter 1995non-randomized study (case series)
Sun 1993non-randomized study
Warrell 1994non-randomized study
Willemze 1994maintenance was not randomized
Xu 2007non-randomized study
Zhang 2000non-randomized study
Zhang 2011maintenance was not randomized
Zheng 2010non-randomized study
Zubizarreta 2000non-randomized study

Characteristics of ongoing studies [ordered by study ID]

0903X-101128

Trial name or titleComparison of Two Treatments in Intermediate and High-risk Acute Promyelocytic Leukemia (APL) Patients to Assess Efficacy in 1st Hematological Complete Remission and Molecular Remission
MethodsRCT
Participants

Patients with untreated newly diagnosed and genetically proven APL with intermediate and high risk disease.

Male and female patients age > 18 years and < 61 years.

Interventions

Experimental: 2 doses of gemtuzumab ozogamicn administered at monthly intervals

Active Comparator: 2 years maintenance therapy with intermittent ATRA plus 6-mercaptopurine (6-MP) and methotrexate (MTX)

Outcomes

Primary outcome: efficacy in achieving first hematological complete remission and molecular remission

Secondary outcome: short and long-term toxicity of treatment, patient quality of life, overall survival

Starting dateMay 2002
Contact informationWyeth, wholly owned subsidiary of Pfizer
NotesStudy completion date: December 2007

CDR0000064499

Trial name or titleCombination Chemotherapy With or Without Bone Marrow Transplantation in Treating Patients With Acute Promyelocytic Leukemia
MethodsRCT, phase III, multi-center
Participants

Newly diagnosed acute promyelocytic leukemia. Must havePML-RARa transcript at disease presentation. Age: 16 to 74.

Patients who are PML-RARa PCR-negative after recovery from third consolidation or are PML-RARa-positive and ineligible for BMT after recovery from third consolidation are designed to maintenance treatment as outlined below (group A).

Patients proceed to allogeneic bone marrow transplantation (BMT) in group B if they are PML-RARa-positive, achieve CR, are under age 55, and have an HLA-A, -B, and -DR identical family donor after recovery from third consolidation.

Patients proceed to autologous BMT in group B if they are PML-RARa-positive, achieve CR, and have no identical family donor or are age 55 years and over after recovery from third consolidation.

Interventions

For maintenance:

Patients are randomized to 1 of 4 treatment arms.

Arm I: mercaptopurine (MP) daily and oral methotrexate (MTX) weekly.

Arm II: Beginning 3 months after recovery from third consolidation, patients receive oral ATRA on days 1-15.

Treatment on arms I and II continues every 3 months for 2 years in the absence of disease progression or unacceptable toxicity.

Arm III: Patients receive 1 course of arm I treatment, alternated by 1 course of arm II treatment. Alternating treatment continues every 3 months for 2 years in the absence of disease progression or unacceptable toxicity.

Arm IV: Patients undergo observation only.

 

For BMT:  

Eligible patients receive conditioning comprising cyclophosphamide (CTX) IV for 2 days followed by total body irradiation or oral busulfan on days -9 to -6 and CTX on days -5 to -2. Autologous or allogeneic bone marrow is infused on day 0 (within 4 months after initiation of third consolidation).

Outcomes

OBJECTIVES:

  • Determine the complete remission (CR) rate in patients with acute promyelocytic leukemia treated with induction comprising ATRA and idarubicin.

  • Determine the presence of the PML-RARa transcript using polymerase chain reaction (PCR) in patients with CR after 3 sequential consolidation regimens comprising cytarabine plus idarubicin, followed by mitoxantrone plus etoposide, and then idarubicin, cytarabine and thioguanine.

  • Determine the percentage of patients who complete the protocol, including PML-RARa-positive patients treated with post-consolidation bone marrow transplantation (BMT) and PML-RARa-negative patients treated with maintenance comprising mercaptopurine (MP) plus methotrexate (MTX) versus ATRA only versus MP plus MTX alternating with ATRA versus observation only.

  • Compare the disease-free survival (DFS) and overall survival of these patients treated with these regimens.

  • Determine the rate and type of grade 4 toxicity, treatment-related mortality, and time to granulocyte and platelet recovery associated with each phase of treatment in these patients.

  • Determine the DFS and overall survival of PML-RARa-positive patients who are ineligible for BMT and are treated with maintenance comprising MP plus MTX alternating with ATRA.

  • Compare the quality of life of patients treated with these regimens.

Starting dateOctober 1995
Contact informationContact investigator: Franco Mandelli, MD email: gimema@gimema.it
Notes 

CDR0000553210

Trial name or titleCombination Chemotherapy With or Without Gemtuzumab Followed By Tretinoin, Mercaptopurine, and Methotrexate or Observation in Treating Patients With Acute Promyelocytic Leukemia
MethodsRCT, multi-center study
ParticipantsPatients with previously untreated low- or intermediate-risk acute promyelocytic leukemia
InterventionsInduction and consolidation combination chemotherapy with or without gemtuzumab ozogamicin followed by maintenance therapy comprising tretinoin, mercaptopurine, and methotrexate versus observation
OutcomesDFS as primary outcome
Starting dateJune 2007
Contact information Sponsors and collaborators: Southwest Oncology Group. Contact investigator: Steven E Coutre, MD email: coutre@stanford.edu
NotesRandomization and observation arm closed as of May 27th, 2010

Ancillary