GIMEMA AIDA 0493 amended protocol for elderly patients with acute promyelocytic leukaemia. Long-term results and prognostic factors

Authors


Roberto Latagliata, MD, Dipartimento di Biotecnologie Cellulari ed Ematologia, Università‘La Sapienza’, Via Benevento 6 – 00161 Rome, Italy. E-mail: rob.lati@libero.it

Summary

To reduce toxicity in elderly patients with acute promyelocytic leukaemia, in 1997 the Gruppo Italiano Malattie Ematologiche Dell’Adulto (GIMEMA) started an amended protocol for patients aged >60 years, with the same induction [all-trans retinoic acid (ATRA) + idarubicin] as in younger patients, followed by a single consolidation course (idarubicin + cytarabine) and maintenance with intermittent ATRA. Among 60 enrolled patients, 54 (90%) achieved haematological remission and six died during induction. Four additional patients died in complete remission (CR) from haemorrhage (2) and infection (2) prior or during consolidation therapy. Eleven patients relapsed at a median time of 17·5 months from CR. The 5-year overall survival (OS), disease-free survival (DFS) and cumulative incidence of relapse (CIR) rates were 76·1%, 64·6% and 27·4%, respectively. Univariate analysis identified a performance score (PS) = 2 as the only significant adverse prognostic factor for both OS (P = 0·017) and DFS (P = 0·0003). Male sex had an unfavourable impact on DFS (P = 0·021) and on CIR (P = 0·019), but not on OS (P = 0·234). In multivariate analysis for DFS, only PS = 2 retained prognostic significance (HR = 4·5, P = 0·0083). In conclusion, the amended GIMEMA protocol is effective, with similar relapse rate and inferior toxicity compared to the original AIDA 0493. However, considering the recent availability of effective new agents, a less aggressive approach should be tested in this setting.

The introduction of all-trans retinoic acid (ATRA) as tailored treatment for acute promyelocytic leukaemia (APL) has markedly improved the overall results in this disease (Sanz et al, 2009). Moreover, unlike that observed for elderly patients with other acute myeloid leukaemias (AML), such improvement has been also reported in elderly APL patients, with complete remission (CR) and disease-free survival (DFS) rates similar to those achieved in younger patients (Mandelli et al, 1997; Sanz et al, 2004; Ades et al, 2005).

In APL patients aged ≥60 years, the Gruppo Italiano Malattie Ematologiche Dell’Adulto (GIMEMA) previously reported a CR rate of 86% and a 5-year overall survival (OS) rate exceeding 55% with the original AIDA 0493 (ATRA plus idarubicin) protocol containing three consolidation cycles (Mandelli et al, 2003); however, severe toxicity during consolidation and excessive rates of death in CR were found using this regimen and similarly reported in other studies, leading the GIMEMA group to reduce treatment intensity in this patient category. An age-adapted protocol was thus derived from the original AIDA 0493 protocol, using the same induction with ATRA and idarubicin but limiting consolidation to one single chemotherapy course. Preliminary results of this study showed a significant reduction of therapy-related toxicity and similar anti-leukaemic efficacy of the amended protocol as compared to the AIDA 0493 (Mandelli et al, 2003). We now report the long-term outcome results and prognostic factors in elderly APL patients treated with this amended AIDA protocol.

Materials and methods

Diagnostic criteria

APL diagnosis was initially established according to the French-American-British (FAB) criteria (Bennett et al, 1985) and confirmed in all patients by molecular identification of the PML/RARA hybrid gene by reverse transcription polymerase chain reaction (RT-PCR) and/or karyotypic detection of the t(15;17) in leukaemic cells.

Therapy

Since March 1997, an age-adapted AIDA protocol with reduced consolidation was adopted by GIMEMA centres for newly diagnosed APL patients aged ≥60 years. The induction phase was identical to the original AIDA protocol and consisted of oral ATRA (45 mg/m2/d) until haematological CR and intravenous Idarubicin (12 mg/m2/d) given on days 2,4,6,8. As consolidation therapy, patients in CR received only the first consolidation course (Idarubicin 5 mg/m2 + cytarabine 1 g/m2 from day 1 to day 4) instead of the three courses scheduled in the original AIDA scheme (Mandelli et al, 2003). Following this single consolidation cycle, patients in molecular CR (as assessed by RT-PCR of PML/RARA, see below) received 2 years of maintenance treatment with intermittent ATRA alone (45 mg/m2/d for 15 d every 3 months).

Prophylaxis and treatment of differentiation syndrome (DS)

All patients received low dose steroids (0·5 mg/kg/d) during induction therapy. Management of DS consisted of prompt discontinuation of ATRA and administration of i.v. dexamethasone (10 mg every 12 h for a minimum of 4 d or until complete resolution of DS manifestations.

Evaluation of response

Haematological CR was defined as the presence of normal bone marrow (BM) cellularity with absence of leukaemic promyelocytes, with peripheral blood polymorphonuclear (PMN) and platelet counts >1 × 109/l and >100 × 109/l, respectively. Resistant disease was defined as persistence in the BM of leukaemic promyelocytes after 90 d of ATRA treatment. Molecular remission was defined as reported elsewhere (Lo Coco et al, 1999).

RT-PCR analysis of the PML/RARA hybrid gene

BM aspirates were obtained at diagnosis, at haematological CR, at the end of consolidation therapy, every 6 months during the maintenance phase and after the end of treatment. Following isolation of the BM mononuclear fraction by centrifugation on a Ficoll-Hypaque gradient, cells were washed twice in phosphat-buffered saline, suspended in a 4 mol/l guanidium thiocyanate solution, stored at −20°C and shipped in dry ice to the reference laboratory (University ‘La Sapienza’ of Rome, Italy) for centralized molecular studies. Total RNA was extracted by the method of Chomczynsky and Sacchi (1987). Prior to RT-PCR analysis, RNA integrity was assessed by running samples on a formaldehyde minigel. The protocol and the oligonucleotide primers used for RT-PCR of the PML-RARA hybrid gene have been reported elsewhere (Diverio et al, 1996). Amplification of PML-RARA was carried out, in all cases, simultaneously with the amplification of the ABL1 gene as an internal control.

Statistical analysis

Differences in the distribution of individual parameters in subsets were assessed by nonparametric tests, at a significance level of P < 0·05. The analysis of long-term outcomes was done only on responders. As a consequence, OS was calculated from both the diagnosis and the date of achievement of CR to death due to any cause, or to the date of last follow-up for patients alive (censored). DFS was calculated from achievement of CR to relapse or death in CR, or to the date of last follow-up for patients alive in first CR (censored). Cumulative incidence of relapse (CIR) was calculated from CR to relapse or to the date of last follow-up for patients alive in first CR (censored), using the cumulative incidence method, where death in CR was considered as a competing risk. Cumulative incidence of non-relapse mortality (CINRM) was calculated from CR to death in CR or to the date of last follow-up for patients alive in first CR (censored), using the cumulative incidence method where relapse was considered as a competing risk. The outcomes for OS, DFS, CIR and CINRM were truncated after a 5-year period of follow-up, in order to avoid their underestimation due to under-reporting from participating centres. The probabilities of OS and DFS were estimated using the Kaplan–Meier method (Kaplan & Meier, 1958) and the Cox model (Cox, 1972), adjusting for other variables; the probability of CIR was estimated using the appropriate non-parametric method (Gooley et al, 1999). The log-rank test (Mantel, 1966) was used to compare risk factor categories for the Kaplan–Meier curves and the Gray test (Gray, 1988) for the incidence curves. All statistical comparisons were based on two-tailed tests. The statistical analysis was carried out using SAS statistical software (sas Release 8.02; SAS Institute Inc, Cary, NC, USA).

Results

Sixty patients entered the amended protocol between March 1997 and May 2004. Of these, six (10%) died during induction from haemorrhage (three patients) or infection (three patients) and 54 (90%) achieved haematological CR and were evaluable for post-remission follow-up. Our analysis focussed on the 5-year follow-up of these 54 patients (24 males and 30 females, median age 66·1 years, range 60·1–73·4), as the amendment was planned to change essentially the post-remission phase. Their clinical characteristics are shown in Table I. According to Sanz’s relapse risk score (Sanz et al, 2000), 18 patients (34·6%) were classified as low-risk, 27 (51·9%) as intermediate-risk and seven (13·5%) as high-risk. A previous primary tumour treated with chemotherapy, radiotherapy or both, was recorded in 9/54 patients (16·7%). Five patients (9·3%) developed retinoic acid syndrome during induction, which resolved after treatment with dexamethasone.

Table I.   Clinical characteristics of patients in haematological CR after induction.
CharacteristicN
Male/female24/30
Median age (range)66·1 years (60·1–73·4)
Performance status
 022 (40·7%)
 122 (40·7%)
 210 (18·6%)
WBC count: median (range)1·5 × 109/l (0·4–53·0)
Patients with WBC count >10 × 109/l7 (13·0%)
Platelet count: median (range)29·5 × 109/l (3·0–185·0)
Patients with platelet count <40 × 109/l32 (61·5%)
Risk category
 Low18 (34·6%)
 Intermediate27 (51·9%)
 High7 (13·5%)

Immediately after the achievement of morphological CR, two patients died (one from infection and one from cerebral haemorrhage), while 52 patients entered the planned consolidation course. During the aplastic phase that followed consolidation, two additional patients died from infection and haemorrhage respectively, while the remaining 50 patients (92·6%) achieved molecular remission after completing consolidation. Severe toxicities (World Health Organization grades 3–4) during the consolidation phase were recorded in 4/52 patients (7·7%) with two cases of haemorrhage (one fatal), one fatal case of infection and one case of cardiac arrhytmia. Twenty-eight patients (53·8%) had a fever of unknown origin (FUO). After consolidation, five patients were excluded from further treatment due to excessive toxicity (three patients) or refusal (two patients) and six patients were lost to follow-up while in molecular CR. The remaining 39 patients started the maintenance program.

Disease relapse occurred in 11 patients after a median time from morphological CR of 17·5 months (range 7·9–32·9), with a 27·4% CIR rate at 5 years. As to the type of relapse, five were haematological, three were molecular and three haematological with extramedullary involvement. At the 5-year follow-up, 39 patients were in first complete molecular remission. Death in CR occurred in four patients, with a 8·1% (95% CI: 7·8–8·4) CINRM rate at 5 years.

The 5-year OS from diagnosis for the whole population of 60 patients was 68·5% (95% CI: 59·98–78·14). The 5-year OS from CR and DFS was 76·1% and 64·6%, respectively (Fig 1). At univariate analysis, performance score (PS) = 2 was the only significant adverse prognostic factor for both OS (P = 0·017) and DFS (P = 0·0003) (Fig 2).

Figure 1.

 Disease-free survival (DFS) and overall survival (OS) from achievement of CR.

Figure 2.

 Disease-free survival (DFS) and overall survival (OS) according to performance score.

Male sex had an unfavourable prognostic impact on DFS [42·4% (95% CI: 33·1–54·4) vs. 80·3% (95% CI: 68·6–94), P = 0·021] and on CIR [48·3% (95% CI: 44–52·0) vs. 12·8% (95% CI: 11·8–13·8), P = 0·019] at 5 years, but not on OS (P = 0·234). White blood cell (WBC) count as a continuous variable was also significant for DFS (P = 0·037) but not for OS (P = 0·190). Age (both as continuous and as categorical variable), platelet count, WBC count with a cut-off of 10 × 109/l and relapse risk score did not show any prognostic significance either for OS or DFS (not shown).

Only statistically significant variables were inserted into multivariate models. The results of multivariate analysis for DFS are shown in Table II; only PS maintained its role as adverse prognostic factor (Hazard Ratio = 4·5, 95% CI: 1·472–13·778, P = 0·0083).

Table II.   Multivariate analysis for DFS.
VariablePr >ChiSqHazard ratio95% Confidence limits
PS: 2 vs. 0, 10·00834·5041·47213·778
Sex: male versus female0·08102·9180·8769·713
Risk score: high versus low/intermediate0·50761·4850·4614·787

Discussion

After the introduction of ATRA as a mainstay of APL treatment, elderly patients have been generally treated in the same manner as younger ones, with excellent results if compared with treatment outcomes in the pre-ATRA era. However, the need for a dose-reduction of post-remission therapy has been clearly suggested by some studies specifically dealing with elderly APL patients, which showed high toxicities and unacceptable rates of deaths in CR (Mandelli et al, 2003; Sanz et al, 2004; Ades et al, 2005).

The Spanish PETHEMA (Programa de Estudio y Tratamiento de las Hemopatías Malignas) group amended the original AIDA protocol for APL patients of all ages, by removing non-anthracycline drugs from the post-remission phase but leaving a total of three courses for consolidation together with maintenance treatment with ATRA and low-dose chemotherapy. The results of this approach were excellent also in the elderly population (aged >60 years), with a 6-year DFS of 79% and a cumulative incidence of relapse 8·5%; however, up to 8% of patients in this age group died in remission during consolidation or maintenance (Sanz et al, 2004).

In our amended age-adapted GIMEMA protocol, we decided to reduce the number of standard AIDA consolidation courses from 3 to only 1. Furthermore, the maintenance phase was also reduced by using intermittent ATRA alone and omitting low-dose chemotherapy. In terms of efficacy, the CIR at 5 years (27·4%) was only slightly increased when compared to that of our previous protocol (23·2%), while the rate of deaths in CR was reduced to 7·7%. As a consequence, the 5-year OS of the amended protocol compared favourably with the 5-year OS of our previous protocol (76·1% vs. 56%).

However, we recognize that the treatment-related toxicity for the amended protocol is still too high, considering the whole treatment plan (induction + post-remission phase); thus, the long-term DFS of this amended protocol is somewhat lower than expected. In fact, up to 35% of patients in this series were lost due to treatment-related complications, including death in CR in four cases.

The question remains as to how these results could be improved: in our opinion, there are at least two future directions. First of all, prognostic factors should help us to recognize those patients at lower risk, preventing them suffering excessive toxicity. In our study, relapse risk score, which is widely used in younger patients, had no influence while PS revealed a prognostic impact on both CR and DFS; it is conceivable that PS could represent a suitable tool for identifying the risk for individual elderly patients.

In addition, new effective agents that show high efficacy, such as arsenic trioxide and Mylotarg (Lo-Coco et al, 2004; Tsimberidou et al, 2006), are now available; their role in elderly patients should be tested based on a very promising toxicity profile.

In conclusion, this reduced-intensity approach is a step forward in the treatment of elderly APL patients, but further improvements are still needed in order to minimize toxicity and improve overall results in this patient category.

Ancillary