For patients with relapsed acute promyelocytic leukemia (APL), all-trans retinoic acid-based salvage regimens can achieve second complete remission (CR2), but the optimal post-remission strategy for APL patients after CR2 remains unclear. Hematopoietic stem cell transplantation (HSCT) during CR2 might be effective, but data on the role of HSCT for APL patients after CR2 are limited in Japan. We retrospectively analyzed outcomes for 57 relapsed APL patients who achieved CR2 in the JALSG APL97 study. Of those, six received autologous (auto)-HSCT, 21 received allogeneic (allo)-HSCT, and 30 received various regimens other than HSCT. The 5-year event-free survival (EFS) rate, overall survival (OS) rate and cumulative incidence of relapse (CIR) were 50.7%, 77.4% and 51.0% in the non-HSCT group, 41.7%, 83.3% and 58.3% in the auto-HSCT group and 71.1%, 76.2% and 9.8% in the allo-HSCT group, respectively. Both the EFS rate and CIR were significantly better in the allo-HSCT group than in other groups. Allo-HSCT appears effective in APL patients in CR2, with a low relapse rate beyond a relatively early transplantation-related mortality (19%). Among older patients (age ≥40 years), the 5-year OS was significantly better in the non-HSCT group than in the HSCT group (78.0% vs 40.5%; P = 0.04). Further prospective studies with larger patient numbers are required to confirm the impact of HSCT alone and in combination with arsenic trioxide on outcomes for patients with APL in CR2.
The introduction of all-trans retinoic acid (ATRA) has brought about marked progress in the treatment of acute promyelocytic leukemia (APL), but relapse still occurs in approximately 15–25% of patients.[1-3] Most of the relapsed patients were able to achieve second complete remission (CR2) using ATRA-based salvage regimens[4-6] or recent arsenic trioxide (ATO)-based salvage regimens.[7, 8] After achieving CR2, most patients need to receive post-remission treatments to reduce minimal residual disease (MRD). A variety of post-remission strategies have been used, including further consolidation chemotherapy, hematopoietic stem cell transplantation (HSCT),[6, 9-11] continued treatment with ATO[7, 8, 12] or a combination of such therapies; however, the optimal post-remission therapy remains controversial. Previous studies have reported that ATO-based post-remission therapy for patients with APL in CR2 resulted in superior survival compared with chemotherapy alone or HSCT alone. Likewise, HSCT strategies for patients with APL in CR2 resulted in better outcomes than chemotherapy alone, despite being associated with high transplantation-related mortality (TRM).[9-11] Moreover, autologous HSCT (auto-HSCT) was much better than allogeneic HSCT (allo-HSCT) for patients in CR2 who achieved molecular remission.[6, 9]
Recently, in a phase 2 prospective study, our Japan Adult Leukemia Study Group (JALSG) reported the efficacy of sequential treatment using ATO followed by auto-HSCT for 25 patients with relapsed APL. However, evidence has been lacking in terms of the role of auto-HSCT alone on the cumulative relapse rate or efficacy for patients with APL in CR2 who were ineligible for the phase 2 study regimens. Moreover, in situations where no guidelines regarding the optimal choice of auto- or allo-HSCT in CR2 have been determined, the role of HSCT alone in post-remission therapies for patients with APL in CR2 is yet to be evaluated. Therefore, the present study aimed to evaluate in detail the efficacies of HSCT alone for APL patients in CR2 by comparing outcomes, including cumulative relapse rate, both for APL patients who underwent auto-HSCT or allo-HSCT during CR2 and for those who did not receive HSCT during long-term follow up.
Materials and Methods
Information on patients with APL in CR2 and the salvage treatment applied were obtained from the JALSG APL97 study. Between May 1997 and June 2002, a total of 302 adult patients with previously untreated de novo APL were registered in this study. The main eligibility criteria included diagnosis of APL with t(15;17) and/or the PML-RARA fusion gene and age between 15 and 70 years. For remission induction therapy, patients received ATRA either alone or with chemotherapy, followed by three courses of consolidation therapy consisting of cytarabine and anthracyclines. After completing consolidation therapy, patients negative for the PML-RARA fusion gene were randomly allocated to undergo either six courses of intensified maintenance chemotherapy or observation alone. More detailed eligibility criteria and the treatment schedule have been described previously. Of the 283 assessable patients with t(15;17) and/or PML-RARA, 267 (94.3%) achieved complete remission (CR). Of the 267 patients who achieved CR, 67 (26.1%) experienced a first relapse during the median follow-up duration of 100 months (range, 11–155 months) from first achieving CR.
Salvage treatment in first relapse
All 67 relapses occurred between 1998 and 2005, during which time ATRA was mainly used as the salvage treatment for relapsed patients because ATO was not commercially available in Japan. Among the relapsed patients, two were unable to complete the follow-up survey and 65 received salvage treatment with ATRA alone (n = 17), ATRA plus chemotherapy (n = 33), tamibarotene (Am80) alone (n = 7), chemotherapy alone (n = 6), allo-HSCT alone (n = 1) or unknown (n = 1). Of those patients who received salvage treatments, 58 (89%) achieved CR2.
Of the 58 patients who achieved CR2, 27 had received HSCT (auto-HSCT, n = 6; allo-HSCT, n = 21) during CR2, 30 had not and one was unassessable. Therefore, the present study included 57 patients. We defined 27 patients in CR2 who received HSCT (six auto-HSCT and 21 allo-HSCT) as the HSCT group and 30 patients in CR2 who received regimens other than HSCT as the non-HSCT group. Clinical characteristics of the 57 APL patients in CR2 are summarized in Table 1.
Table 1. Clinical characteristics of the 57 APL patients in CR2 according to treatment after CR2
(n = 6)
(n = 21)
(n = 30)
(n = 57)
No. (%) or median (range)
No. (%) or median (range)
No. (%) or median (range)
No. (%) or median (range)
Allo-HSCT, allogeneic HSCT; APL, acute promyelocytic leukemia; ATRA, all-trans retinoic acid; Auto-HSCT, autologous HSCT; CR, complete remission; CR2, second complete remission; HLA, human leukocyte antigen; HSCT, hematopoietic stem cell transplantation; Non-HSCT, patients who received regimens other than HSCT; WBC, white blood cell.
WBC counts (x109/L)
10.0 or higher
At first relapse
First CR duration (months)
ATRA plus chemotherapy
In CR2 achievement
Age at CR2 (years)
Time to HSCT after CR2 (months)
Hematopoietic stem cell transplantation group
Stem cells for auto-HSCT were harvested in CR2 from peripheral blood in all six patients. Peripheral blood stem cell (PBSC) collection was made after mobilization using granulocyte colony-stimulating factor (G-CSF) following chemotherapy. All patients who underwent auto-HSCT achieved molecular CR of PML-RARA in bone marrow according to nested reverse transcriptase–polymerase chain reaction (RT-PCR) (n = 3), real-time quantitative PCR (RQ-PCR) (n = 2) or RT-PCR (n = 1) just before PBSC collection. For allo-HSCT, bone marrow cells were used in 15 patients, G-CSF-mobilized PBSC in four patients and cord blood cells in two patients. Donors were unrelated in 13 patients (bone marrow, 11 patients; cord blood, two patients). Seven of 15 patients who were examined for PML-RARA in the marrow before allo-HSCT were positive for MRD.
Patients were administered various conditioning regimens for HSCT. All six autografted patients received a myeloablative regimen using total body irradiation (TBI)/cyclophosphamide (CY) (n = 1), busulfan (BU)/CY (n = 3), BU/melphalan (n = 1) or BU/etoposide/cytarabine (n = 1). Allografted patients received a myeloablative regimen using TBI/CY (n = 11), TBI/BU/CY (n = 2), BU/CY (n = 6) or a non-myeloablative fludarabine-based regimen (n = 2).
Patients in this group received various consolidation and/or maintenance regimens with chemotherapy and/or ATRA (Table 1). Reasons for not undergoing HSCT in CR2 included age >60 years (n = 6), relatively poor condition (n = 4), patient refusal (n = 5), lack of an appropriate donor (n = 1), medical decision (n = 13) and unknown reasons (n = 1). Among patients in the non-HSCT group, 10 received allo-HSCT (n = 8) or auto-HSCT (n = 2) during the third CR (CR3) or more. After achievement of CR2, patients were treated with a variety of consolidation regimens, including chemotherapy, ATO, gemtuzumab ozogamycin and observation alone (Table S1).
Hematological CR was defined as the presence of all of the following: <5% blasts in bone marrow; no leukemic blasts in peripheral blood or extramedullary sites; and recovery of peripheral blood counts. Relapse was defined as the presence of at least one of the following: two consecutive positive RT-PCR obtained 1 month apart after achieving molecular remission; recurrence of >10% leukemic cells in bone marrow; recurrence of any leukemic cells in peripheral blood; or development of extramedullary disease.
Overall survival (OS) was calculated from the date of CR2 to the date of death or last follow up. Event-free survival (EFS) was calculated from the date of CR2 to an event (relapse or death) or to the date of last follow up. Cumulative incidence of relapse (CIR) was calculated from the date of CR2 to the date of second relapse or last follow up for patients alive in CR2. Results were analyzed as of 31 March 2010, allowing for median follow ups of 84 months (range, 16–120 months) and 87 months (range, 2–136 months) from the date of CR2 for the HSCT and non-HSCT groups, respectively. Differences in categorical factors between the HSCT and non-HSCT groups were compared using the χ2 test. Age at CR2 was dichotomized using a cut-off point of 40 years to create a younger group (<40 years) and an older group (≥40 years) by taking the transplantation risk of age in the risk score of the European Group for Blood and Marrow Transplantation into consideration. Continuous data were compared using the Mann–Whitney test. The OS and EFS were estimated using the Kaplan–Meier method and compared using the log-rank test. To adjust for effects of the timing of HSCT in the survival analysis, HSCT was treated as a time-dependent covariate in the Kaplan–Meier estimates of OS and EFS. The CIR was estimated using the cumulative incidence method, where death in CR2 was considered as a competing risk and compared using Gray's test. All tests were two tailed and a value of P < 0.05 was considered statistically significant. All analyses were performed using statistica version 6.0 software (Statsoft Inc., Tulsa, OK, USA) and stata 11 software (STATA Corp LP, College Station, TX, USA).
The characteristics and prognosis of patients with APL who achieved CR2 by salvage treatment with HSCT (n = 27) or non-HSCT (n = 30) are summarized in Table 2.
Table 2. Clinical consequences of APL patients in CR2 according to treatment with HSCT or non-HSCT after CR2
(n = 27)
(n = 30)
Analyses were performed using a time-dependent covariate approach.
APL, acute promyelocytic leukemia; CI, confidence interval; CIR, cumulative incidence of relapse; CR, complete remission; CR2, second complete remission; EFS, event-free survival; HSCT, hematopoietic stem cell transplantation; N/A, not available; OS, overall survival; TRM, transplantation-related mortality; WBC, white blood cell.
In the 27 patients (six auto-HSCT and 21 allo-HSCT) with a median duration of first CR at 22 months (range, 6–81 months), six patients relapsed and seven patients died, including four patients with TRM (Table 2).
Among the six patients who received auto-HSCT, the median duration of first CR was 22 months (range, 10–81 months), the median time from achievement of CR2 to HSCT was 6 months (range, 4–20 months), no TRM was seen and four patients (67%) relapsed at 9, 29, 46 and 84 months after auto-HSCT. Among those who relapsed, one died from APL progression 12 months after auto-HSCT. Another three patients achieved CR3 through treatment with ATO, Am80 or high-dose cytarabine and remained in CR3 at 7, 22, and 54 months after CR3, respectively. Of those who received auto-HSCT in CR2, four relapsed and one died, and the remaining two patients were alive in CR2.
In the 21 patients who received allo-HSCT in CR2, the median duration of first CR was 22 months (range, 6–63 months) and the median time from achievement of CR2 to HSCT was 6 months (range, 1–13 months). Of the 21 patients, four patients (19%) died of TRM (two patients died due to graft-versus-host disease [GVHD] and two patients died due to multiple organ failure) and two patients (9.5%) relapsed at 4 and 34 months after salvage HSCT and died. No significant difference in 5-year OS, EFS rates and CIR in seven patients with MRD before allo-HSCT was observed compared with eight patients negative for MRD (data not shown). Among those who received allo-HSCT in CR2, four died of TRM, two relapsed and died and the remaining 15 patients were alive in CR2.
Clinical consequences for the non-HSCT group
In the 30 patients in CR2 who did not receive any HSCT as post-remission therapy, the median duration of first CR was 18 months (range, 6–91 months) (Table 2). In CR2, these patients received consolidation treatment with various chemotherapy regimens, sometimes followed by maintenance treatment with ATRA. Of the 30 patients, 14 (47%) remained in CR2 after a median of 69 months (range, 2–133 months), but 16 (53%) experienced a second relapse after a median of 14 months (range, 1–113 months). One of the 14 patients who remained in CR2 died from secondary acute lymphoblastic leukemia. Among the 16 patients who experienced a second relapse, eight received allo-HSCT (three in CR3, one in CR4, two in the second relapse and two in the third relapse) and two received auto-HSCT in CR3. Of these eight patients who received allo-HSCT, four died from TRM (GVHD in two patients, pneumonia in one patient and multiple organ failure in one patient), two died from APL progression with further relapse after HSCT and two survived in a disease-free state. Of the two patients who received auto-HSCT, both remained in CR3. Of the six patients who experienced a second relapse and did not receive HSCT, one failed to obtain CR and died from APL progression and five patients achieved CR3 (two died of APL progression after the third relapse, one died of myocardial infarction and two remained in CR3 as of 22 and 23 months). Of those who received no HSCT in CR2, 13 patients were alive in CR2 and one patient died in CR2. Of the remaining 16 patients who relapsed, 10 patients died and six were alive in CR3 or more.
Comparisons between the HSCT and non-HSCT groups
Median age at CR2 was significantly younger in the HSCT group than in the non-HSCT group (P = 0.006) (Table 2). No significant differences were observed between these two groups in the frequency of male sex, white blood cell count at diagnosis or duration of first CR. The frequency of relapse after CR2 was significantly higher in the non-HSCT group (22% vs 53%; P = 0.016) (Table 2). However, the frequency of death did not differ between the two groups.
Although no significant differences in the 5-year OS rate (Table 2, Fig. 1a) or 5-year EFS rate (Table 2, Fig. 1b) were evident between the two groups, the CIR was significantly lower in the HSCT group than in the non-HSCT group (5-year CIR, 19.7% vs 51.0%; P = 0.018) (Table 2, Fig. 1c).
When we analyzed the data by dividing each group into two age subgroups of younger patients (age <40 years) and older patients (age ≥40 years), younger patients showed no significant difference in 5-year OS rate between the HSCT group (100%) and non-HSCT group (82.5%; P = 0.10), but did show a tendency in favor of allo-HSCT (Fig. 2a). Conversely, among the older patients, the OS rate was significantly higher in the non-HSCT group than in the HSCT group (5-year OS, 78.0% vs 40.5%; P = 0.04) (Fig. 2b). In the HSCT group, OS rate was significantly better in younger patients (age <40 years, n = 15; 5-year OS, 100%) than in older patients (age ≥40 years, n = 12; 5-year OS, 50.0%; P = 0.006) (Fig. 2c).
Comparisons among auto-HSCT, allo-HSCT and non-HSCT groups
We compared several outcomes among auto-HSCT, allo-HSCT and non-HSCT groups. No significant differences were seen in the 5-year EFS rate (auto-HSCT, 41.7%; allo-HSCT, 71.1%; non-HSCT, 45.4%) (Fig. 3a) or 5-year OS rate (auto-HSCT, 83.3%; allo-HSCT, 76.2%; non-HSCT, 75.3%) (Fig. 3b). However, 5-year CIR differed significantly between patients who underwent auto-HSCT (58.3%) and allo-HSCT (9.8%; P = 0.007) and between patients who underwent non-HSCT (51.0%) and allo-HSCT (9.8%; P = 0.009), while no significant difference was evident between the auto-HSCT and non-HSCT groups (P = 0.603) (Fig. 3c).
The main results of the present study indicate that the 5-year CIR was significantly better in patients who underwent allo-HSCT than in those who did not and the 5-year OS rate was significantly better in the non-HSCT group than in the HSCT group among older patients (age ≥40 years).
Several studies have demonstrated that auto-HSCT for APL in CR2 yields favorable results with a relatively low relapse rate.[6, 9] In an Italian study, it was reported that of 15 patients receiving auto-HSCT for APL in CR2 only two of eight patients who were negative for PML-RARA transcript by RT-PCR in bone marrow before auto-HSCT relapsed, whereas all seven patients with positive findings from the RT-PCR relapsed. In a study from the European Acute Promyelocytic Leukemia Group reported, among 28 auto-grafted patients who were in molecular remission at the time of stem cell harvest, only three relapsed (7-year EFS rate, 76.5%). A recent prospective study of our JALSG also observed a relatively low relapse rate, in which there were only three relapses among 23 auto-grafted patients with molecular remission at the time of stem cell harvest (5-year EFS rate, 65%). These studies show a prognostic importance of MRD negativity using molecular analysis before HSCT on the outcome. However, the results of the present study differ from previous report in that the MRD negativity is well associated with the low relapse rates in auto-HSCT. Contrary to our expectation, both the 5-year EFS rate (41.7%) and the 5-year CIR (58.3%) were worse for the auto-HSCT group than for the allo-HSCT group (Fig. 3a,c), even though all six patients were confirmed to have achieved molecular CR in bone marrow by nested RT-PCR or RT-PCR just before peripheral hematopoietic stem cell collection. Therefore, auto-HSCT was less effective for relapse in APL in CR2 and pre-transplant MRD had no predictive significance with respect to relapse in the present study. This might be due primarily to the small number of patients (n = 6) who received auto-HSCT in our analyses, which was the major limitation in the present study. Another possible explanation is the difference in sensitivity for the detection of MRD. In the APL97 study, although all patients who received auto-HSCT were MRD negative before transplantation, the detection limit of the PML-RARA fusion transcript was 10−4, whereas in the report by de Botton et al., nested RT-PCR for PML-RARA amplification was used with a sensitivity of 10−5 to 10−6. Although the 5-year EFS and 5-year CIR were worse in the auto-HSCT group in the present study, the 5-year OS rate (83.3%) was not inferior to that in the allo-HSCT and non-HSCT groups. No TRM was seen in the six patients who underwent auto-HSCT and all but one patient achieved CR3 by means of a range of post-relapse salvage treatments.
In the present study, none of the young patients (age <40 years) died within 5 years from the date of CR2. Taken together with our results that the 5-year OS rate tended to be better in the allo-HSCT group than in the non-HSCT group among younger patients (Fig. 2a), the 5-year EFS rate was better in allo-HSCT than in the auto-HSCT and non-HSCT groups (Fig. 3a) and the 5-year CIR was lower in allo-HSCT than in the auto-HSCT and non-HSCT groups (Fig. 3c), we suggest that conventional allo-HSCT represents an effective option for young APL patients who achieve CR2. The reason for this is that allo-HSCT is originally aimed at producing a graft-versus-leukemia effect in addition to direct antitumor effects of conditioning and is also regarded as an acceptable method of treatment in patients who are positive for pre-transplant MRD.
It is to be expected that patients in the non-HSCT group (those who did not undergo transplantation during CR2) would be older and those with complications, so the outcomes would be poorer than those of the transplant groups. However, counter to our expectations, survival outcomes in the non-HSCT group were relatively high (5-year EFS rate, 45.4%; 5-year OS rate, 75.3%) and not inferior to those in the HSCT groups. A similar result was reported by the European Acute Promyelocytic Leukemia Group, in which a consistent proportion of relapsed APL patients in CR2 who did not undergo transplantation were almost completely cured (EFS rate, 30.4%; OS rate, 39.5%). Outcomes for the non-HSCT group were less favorable in the present study, but 13 of the 30 patients (43%) remained in CR2. The European APL study group also reported that 39% remained in CR2 in the non-HSCT group. Such findings suggest that HSCT might not always be necessary for all patients in CR2 to prevent further relapse, given the potential for unnecessary TRM.
More recently, ATO has been used worldwide for the treatment of relapsed APL patients,[7, 13, 18] and has been included in the design of several front-line studies, with the aim of reducing therapy-related toxicities and obtaining more profound molecular remission. However, the efficacy of ATO alone in relapsed APL patients remains contentious. A study from France that treated relapsed APL reported that OS in patients with an ATO-based regimen was superior to that in patients with conventional combination chemotherapy or allo-HSCT, but others have reported that an ATO-based regimen offered a high response rate but also a high relapse rate.[8, 19] Moreover, a recent study from India that treated relapsed APL patients who had achieved molecular CR with ATO reported that the EFS rate was significantly inferior in patients who underwent continuous administration of ATO+ATRA without auto-HSCT (34%, n = 19) compared with that in patients treated with auto-HSCT after CR2 (83%, n = 14; P = 0.001). The reason for such discrepancies in the effects of ATO-based regimen among different studies might be attributed to the small numbers of patients, selection bias and differences in economic constraints. Nevertheless, approximately 40% and 60% of patients receiving an ATO-based regimen relapsed in the French and Indian studies, respectively. In the present study, none of the relapsed patients were treated with ATO, because all relapsed before ATO gained approval for use in Japan. Only quite recently, our study group has reported better efficacy of a regimen of ATO followed by auto-HSCT for relapsed APL in the phase 2 study (n = 23; 5-year EFS, 65%).
In conclusion, the present study suggests that allo-HSCT is favorably recommended for younger APL patients during CR2, but for older APL patients, safer and less toxic treatments such as non-myeloablative transplantation might be preferable. Nevertheless, given the small number of patients in the present study and the retrospective nature of the analysis, clear conclusions are difficult to reach. Further prospective studies with larger numbers of patients are required to confirm the role of HSCT both alone and in combination with ATO on the outcomes for patients with APL in CR2.
This work was supported in part by Grants-in-Aid for Scientific Research from the Japanese Ministry of Education, Culture, Sport, Science, and Technology; the National Cancer Center Research and Development Fund (23-A-23); and Grants-in-Aid from the Cancer Research from the Japanese Ministry of Health, Labor and Welfare (Clinical Cancer Research 23-004). We thank the clinicians and leaders of the 92 institutions who entered their patients into the JALSG APL97 and provided the data that made the present study possible.