The complete remission (CR) rate of adult patients with acute lymphoblastic leukaemia (ALL) has improved to about 90% with modern intensive chemotherapy. However, many patients eventually relapse, and the long-term leukaemia-free survival rate is only 30–40% (Litzow, 2009). Many relapsed patients receive various salvage therapies after the first relapse, and several studies have reported that 38–56% of relapsed patients can achieve second remission (CR2; Giona et al, 1997; Thomas et al, 1999; Camera et al, 2004; Tavernier et al, 2007; Cornelissen et al, 2009). While allogeneic haematopoietic stem cell transplantation (Allo-HSCT) in CR2 is considered to be the only curative strategy, early relapse and/or organ dysfunction after salvage chemotherapy and the lack of a suitable donor often prevent Allo-HSCT at this stage. Therefore, the prognosis of adult patients with relapsed ALL is extremely poor (Fielding et al, 2007; Tavernier et al, 2007; Oriol et al, 2010; Gokbuget et al, 2012). We collected clinical data after the first relapse in adult patients with Philadelphia chromosome (Ph)-negative ALL who were treated in institutions all over Japan, and performed a retrospective analysis to clarify the prognosis and prognostic factors for the outcome in relapsed patients. Patients with Ph-positive ALL were not included in our analysis because the outcome of treatment in these patients has improved dramatically since tyrosine kinase inhibitors became available (Ottmann & Pfeifer, 2009).
To analyse the outcome of adult patients who developed a first relapse of acute lymphoblastic leukaemia (ALL), we collected the clinical data of 332 patients with Philadelphia-chromosome (Ph) negative ALL, aged 16–65 years, who relapsed after first complete remission (CR1) between 1998 and 2008 in 69 institutions all over Japan, including 58 patients who relapsed after allogeneic haematopoietic stem cell transplantation (Allo-HSCT) in CR1. The overall survival (OS) was 43·4% at 1 year, and 16·3% at 5 years from relapse in patients who received chemotherapy alone in CR1. Among patients who relapsed after chemotherapy alone in CR1, 123 (52·5%) achieved a second remission (CR2) following salvage chemotherapy, of whom 62 subsequently underwent Allo-HSCT during CR2. Allo-HSCT in CR2 was significantly associated with better OS. Moreover, the type of salvage chemotherapy influenced OS from relapse. A doxorubicin, vincristine, and predonisone-based (AdVP-type) regimen was related to better OS in patients with longer CR1 (more than 1 year), but was related to worse OS in patients with shorter CR1. In conclusion, the prognosis of patients with relapsed Ph-negative ALL is poor. Allo-HSCT after a first relapse could improve the prognosis. Selection of the optimal salvage chemotherapy might depend on the duration of CR1.
We retrospectively collected clinical data of patients with Ph-negative ALL, aged 16–65 years, who relapsed after achieving first remission (CR1) between 1998 and 2008. Patients with lymphoblastic lymphoma were not included. Patients had received either chemotherapy alone, autologous stem cell transplantation (Auto-HSCT), or Allo-HSCT in CR1. Patients who did not achieve remission with chemotherapy but eventually achieved remission after HSCT were excluded. CR was defined based on haematological findings, and not on molecular findings. The data collected included clinical information about the disease at both diagnosis and relapse, the content of therapies both at diagnosis and after relapse, and the clinical course. This study was approved by the Institutional Review Board of Saitama Medical Centre, Jichi Medical University.
Differences between groups were examined using the Fisher's exact test for categorical variables and the Mann-Whitney U-test for continuous variables. Overall survival (OS) was calculated using the Kaplan-Meier method. To evaluate the influence of factors for OS, the log-rank test and proportional-hazards modelling were used for univariate and multivariate analyses, respectively. In a univariate analysis, the impact of Allo-HSCT in CR2 was assessed by a landmark analysis that was limited to patients who achieved CR2 and survived without relapse for at least 68 d, which was the median duration between the achievement of CR2 and the performance of Allo-HSCT in CR2. A post-hoc multicomparison test using the Holm method was performed for comparisons among three groups.
For the multivariate analysis, we included all covariates at first, and used the backward stepwise selection of covariates. Finally, P values of less than 0·05 were considered statistically significant. Potential confounding factors at diagnosis that were considered in the analysis were sex, phenotype of ALL, white blood cell (WBC) count, karyotype (patients with t(4;11) or complex karyotype (5 or more chromosomal abnormalities) were considered high-risk, and were compared to those with other karyotypes), presence of central nervous system (CNS) invasion of ALL, intensity of chemotherapy, and duration between the diagnosis and the achievement of CR1. Moderate-intensity regimens at diagnosis were defined as those that included both high-dose cytarabine and high-dose methotrexate, and the other regimens were regarded as conventional-intensity. In addition, potential confounding factors at relapse that were considered in the analysis were relapse year, age, duration of CR1, sites of relapse, WBC count, and additional chromosomal abnormalities. Allo-HSCT in CR2 was also included as a time-dependent covariate in the multivariate analysis.
We specifically evaluated the influence of salvage chemotherapy after relapse on OS. Salvage chemotherapy regimens were categorized into three types. The first was combination chemotherapy based on conventional doses of doxorubicin, vincristine, and predonisone (AdVP-type). The second type was fractionated cyclophosphamide, vincristine, doxorubicin, and dexamethasone alternating with high doses of methotrexate and cytarabine (hyperCVAD/MA; Kantarjian et al, 2000), and the third type included high-dose cytarabine (Kantarjian et al, 1986), the combination of mitoxantrone, etoposide, and cytarabine (MEC; Amadori et al, 1991), and the combination of high-dose cytarabine and mitoxantrone (HAM; Lejeune et al, 1990). Potential confounding and selection biases for these three kinds of salvage regimens were considered through the Fisher's exact test.
All statistical analyses were performed with EZR (Saitama Medical Centre, Jichi Medical University; Kanda, 2012), which is a graphical user interface for R (The R Foundation for Statistical Computing, version 2.13.0).
A total of 332 patients, treated in 69 institutions all over Japan, were included in this study. The characteristics of the patients are summarized in Table 1. Their median age at relapse was 35 years, and 165 patients were male. The median duration of CR1 was 290 d (range 15–7162 d), and the median follow-up time after relapse was 1344 d (range 12–3689 d). Two-hundred and seventy patients relapsed after chemotherapy alone in CR1, and 58 and 4 patients relapsed after Allo- and Auto-HSCT in CR1, respectively. When the patients who relapsed after chemotherapy alone in CR1 were compared to those who relapsed after Allo-HSCT in CR1, the distribution of the phenotype and karyotypes of leukaemic cells were significantly different. In addition, older age at relapse, shorter duration of CR1, and lower rate of the appearance of additional chromosomal abnormalities at relapse were observed in patients who relapsed after chemotherapy alone in CR1.
|All patients (n = 332)||Relapse after CTx alone (n = 270)||Relapse after Allo-HSCT (n = 58)||P valuea|
|Patient characteristics at diagnosis|
|Age, years; median (range)||34 (15–65)||37 (15–65)||28 (16–58)||0·006|
|Male||165 (50%)||132 (49%)||30 (52%)||0·773|
|Female||167 (50%)||138 (51%)||28 (48%|
|T||65 (20%)||47 (17%)||17 (29%)||0·038|
|B||253 (76%)||213 (79%)||37 (64%)|
|Other||6 (2%)||4 (1%)||2 (3%)|
|WBC, ×109/l; median (range)||11·1 (0·6–759·7)||11·0 (0·6–759·7)||14·4 (0·6–537·9)||0·264|
|normal||162 (49%)||138 (51%)||20 (34%)||0·025|
|t(4;11)||10 (3%)||10 (4%)||0 (0%)|
|complex||50 (15%)||36 (13%)||14 (24%)|
|others||84 (25%)||66 (25%)||18 (31%)|
|−||312 (94%)||255 (94%)||53 (91%)||0·238|
|+||12 (4%)||8 (3%)||4 (7%)|
|Conventional||184 (55%)||149 (55%)||32 (55%)||1·0|
|Moderate||142 (43%)||117 (43%)||25 (43%|
|Days from diagnosis to CR1; median (range)||38 (14–337)||37 (14–337)||42 (21–143)||0·081|
|Patient characteristics at relapse|
|Upto end 2003||152 (46%)||127 (47%)||21 (36%)||0·147|
|2004 onwards||180 (54%)||143 (53%)||37 (64%)|
|Follow-up after relapse, days; median (range)||1344 (12–3689)||1381 (12–3689)||1001 (726–2799)||0·540|
|Duration of CR1|
|Median (range) (d)||290 (15–7162)||246 (15–7162)||465 (33–2185)||<0·001|
|<1 year||193 (58%)||171 (63%)||22 (38%)|
|≥1 year||135 (41%)||96 (36%)||35 (60%)|
|Sites of relapse|
|BM included||288 (87%)||237 (88%)||47 (81%)||<0·001|
|CNS alone||29 (9%)||28 (10%)||1 (2%)|
|others||13 (4%)||4 (1%)||9 (16%)|
|Age, years; median (range)||35 (16–65)||37 (16–65)||29 (18–59)||0·011|
|WBC, × 109/l; median (range)||4·5 (0·6–331·6)||4·4 (0·6–331·6)||5·1 (1·2–137·3)||0·168|
|Additional chromosomal abnormality|
|−||231 (70%)||198 (73%)||29 (50%)||0·01|
|+||52 (16%)||37 (14%)||15 (26%)|
|Patients that achieved CR2, n||163 (53·6%)||136 (54·0%)||26 (54·2%)||0·753|
Overall survival (OS) from relapse in the 270 patients who received chemotherapy alone in CR1 was 43·4% at 1 year and 16·3% at 5 years, while OS from relapse in the 58 patients who received Allo-HSCT in CR1 was 50·0% at 1 year and 10·6% at 5 years. The OS rate was not significantly different between these two groups (hazard ratio (HR): 0·97, 95% confidence interval (CI): 0·64–1·29, P = 0·868; Fig 1).
Outcome of relapsed patients after chemotherapy alone in CR1
Among patients who relapsed after chemotherapy alone in CR1, 43% had received moderate-intensity regimens at diagnosis, such as hyperCVAD/MA (Kantarjian et al, 2000) and regimens used in the Japan Adult Leukaemia Study Group (JALSG) ALL201 studies (C000000063 for older patients and C000000064 for younger patients; Sakura et al, 2012). The remaining patients had received conventional-intensity regimens, which were mainly used in JALSG ALL93 (Takeuchi et al, 2002) and ALL97 (Jinnai et al, 2010) studies.
Among the 270 patients who relapsed after chemotherapy alone in CR1, 234 received salvage chemotherapy, and 123 of these 234 (52·5%) achieved a second complete remission (CR2; Fig 2). Sixty-two of these 123 patients underwent Allo-HSCT in CR2, 16 patients from a human leucocyte antigen (HLA)-matched related donor, 6 from an HLA-mismatched related donor, and 30 from an unrelated donor. The remaining 10 received unrelated cord blood. Fifty-five of them received myeloablative conditioning. The median duration between the achievement of CR2 with salvage chemotherapy and Allo-HSCT in CR2 was 68 d (range 10–276 d). OS from CR2 was significantly better in patients who underwent Allo-HSCT in CR2 than in those who did not (73·7% vs. 50·0% at 1 year and 43·9% vs. 11·0% at 5 years, respectively; HR: 0·43, 95% CI: 0·26–0·72, P = 0·001) by a landmark analysis that was limited to patients who were alive without disease at 68 d after they had achieved CR2 (Fig 3). The results of a Mantel-Byar analysis that evaluated the influence of Allo-HSCT in CR2 were similar to that of a landmark analysis using the log-rank test (HR: 0·49, 95% CI: 0·31–0·77, P = 0·002). In a multivariate analysis using the backward stepwise selection of covariates, younger age at relapse (younger than 36 years), lower WBC count at relapse (less than 10 × 109/l), and Allo-HSCT in CR2 treated as a time-dependent covariate were associated with better OS among patients who achieved CR2 following salvage chemotherapy (Table 2).
|Hazard ratio (95% CI)||P value|
|Allo-HSCT in CR2a|
|Age at relapse|
|>35 years||1·83 (1·18–2·82)||0·007|
|WBC count at relapse|
|≥10 × 109/l||2·87 (1·67–4·95)||<0·001|
|<10 × 109/l|
We further analysed the effect of each salvage regimen that patients received after the first relapse. Patients who received one of the three types of salvage regimen below were included in this analysis. Those who had isolated CNS relapse were excluded. Seventy-five patients received the AdVP-type salvage regimen, and 58 received the hyperCVAD/MA regimen (Kantarjian et al, 2000). The other 46 patients received a high-dose cytarabine (Kantarjian et al, 1986; n = 22), MEC (Amadori et al, 1991; n = 14), or HAM (Lejeune et al, 1990; n = 10) regimen. The type of salvage regimen did not significantly influence OS from relapse (Fig 4A). In the univariate analyses, a longer duration of CR1 (more than 1 year; P = 0·003), the intensity of chemotherapy at diagnosis (P < 0·001), and the relapse year (before 31 December 2003 or after 1 January 2004; P = 0·008) were significantly associated with the selected salvage regimen. Therefore, the patients included in this analysis were stratified into two groups according to the duration of CR1, intensity of chemotherapy at diagnosis, and relapse year, respectively. The duration of CR1 was less than 1 year in 96 patients and longer than 1 year in 71 patients. In patients who had a shorter duration of CR1, the type of salvage regimen influenced OS (P = 0·002; Fig 4B). In the post-hoc test, patients who received the AdVP-type regimen had worse OS, compared to those who received the hyperCVAD/MA regimen (P = 0·040) or other regimens (P < 0·001). The type of salvage regimen also influenced OS in patients who had a longer duration of CR1 (P = 0·048; Fig 4C). In contrast, the post-hoc test indicated that patients who received the AdVP-type regimen were associated with better OS with borderline significance, compared to those who received the hyperCVAD/MA regimen (P = 0·066). When patients were stratified into two groups according to the intensity of chemotherapy at diagnosis or the relapse year, the type of salvage regimen did not influence OS (P = 0·733 and 0·843, in patients who received conventional- and moderate-intensity chemotherapy at diagnosis, respectively, and P = 0·131 and 0·892, in patients who relapsed before 31 December 2003 and after 1 January 2004, respectively).
Outcome of Allo-HSCT in non-remission after relapse
In relapsed patients who had received chemotherapy alone in CR1, 18 patients directly underwent Allo-HSCT after relapse without salvage chemotherapy and 46 underwent Allo-HSCT in non-remission (NR) following failed salvage chemotherapy. OS from Allo-HSCT in patients who underwent Allo-HSCT in NR after salvage chemotherapy was 39·1% at 1 year and 20·7% at 5 years. OS from Allo-HSCT in patients who underwent Allo-HSCT directly after relapse was 38·9% at 1 year and 11·1% at 5 years. There was no difference in OS from Allo-HSCT between these two groups (HR: 0·81, 95% CI: 0·17–1·45, P = 0·51; Fig 5).
Several large clinical studies have demonstrated the outcomes of a relatively large number of patients with ALL who relapsed after uniform chemotherapy regimens (Fielding et al, 2007; Tavernier et al, 2007; Oriol et al, 2010; Gokbuget et al, 2012), and the results of uniform salvage chemotherapy regimens have been reported as clinical trials, each of which included a small number of relapsed patients with ALL (Giona et al, 1997; Koller et al, 1997; Montillo et al, 1997; Weiss et al, 2002; Camera et al, 2004; Specchia et al, 2005). The present study investigated the prognosis of relapsed patients with Ph-negative ALL, based on the clinical data of 332 patients from 69 institutions all over Japan. These patients had received various kinds of treatment strategies before and after relapse according to their respective institution, and therefore, these patients should reflect the more general population of relapsed patients with ALL.
Overall survival (OS) at 5 years from relapse was 16·3% in patients who received chemotherapy alone in CR1 and 10·6% in patients who received Allo-HSCT in CR1, and Allo-HSCT in CR1 did not influence the outcome after relapse in our study. These outcomes were comparable to those in three previous reports of large clinical studies (Fielding et al, 2007; Tavernier et al, 2007; Oriol et al, 2010; Table 3). Recently, Gokbuget et al (2012) reported the outcome of 547 relapsed patients in the German Multicentre Study Group for Adult ALL (GMALL). The OS at 5 years from relapse in patients who had received chemotherapy alone in CR1 was significantly better than that in patients who had received Allo-HSCT in CR1 (28% vs. 15%, P < 0·001). This difference might be attributed to the high rate (75%) of Allo-HSCT after relapse in patients who had received chemotherapy alone in CR1. In our study, Allo-HSCT was performed in 55% of the patients who relapsed after they had received chemotherapy alone in CR1. Allo-HSCT in CR2 was associated with a better prognosis after the achievement of CR2, and in addition, some of the patients who received Allo-HSCT even in NR after salvage chemotherapy showed long-term survival (the OS was 20·7% at 5 years). These findings suggested that we should consider Allo-HSCT as much as possible for relapsed patients.
|Patients (n)||Rate of CR2 (%)||5-year OS||Prognostic factors for improved OS|
|LALA-94 (Tavernier et al, 2007)||421||44%||8%|| |
Allo-HSCT after relapse
CR1 > 1 year
Platelet count at relapse >100 × 109/l
|MRC UKALL12/ECOG2993 (Fielding et al, 2007)||609||N.M.||7%|| |
Age at relapse <20 years
CR1 > 2 years
|PETHEMA trials (Oriol et al, 2010)||263||45%||10%|| |
Age at relapse <30 years
CR1 > 2 years
|GMALL trials (Gokbuget et al, 2012)a||547|| |
42% after CTx aloneb
23% after Allo-HSCTc
28% after CTx aloneb
15% after Allo-HSCTc
No Allo-HSCT in CR1
Extramedullary relapse (other than CNS)
CR1 > 18 monthsd
Age at relapsed
(15–25 years vs. 26–45 years vs. 46–55 years)
CR after 1st salvage CTxd
Allo-HSCT at any staged
|This study||332|| |
53% after CTx aloneb
54% after Allo-HSCTc
16% after CTx aloneb
11% after Allo-HSCTc
HSCT in CR2e
WBC count at relapse <10 × 109/le
Age at relapse <35 yearse
Prognostic factors that were associated with a better OS from CR2 were younger age at relapse, lower WBC count at relapse, and Allo-HSCT in CR2 among patients who relapsed after chemotherapy alone in CR1 and achieved CR2 following salvage chemotherapy. None of the factors at diagnosis was associated with OS after relapse. Age and Allo-HSCT were the common factors observed in other studies (Table 3). The duration of CR1 was not associated with better OS in the multivariate analysis in our study. In other studies, the duration of CR1 was associated with not only better OS but also a high rate of achieving CR2 (Thomas et al, 1999; Tavernier et al, 2007; Oriol et al, 2010). Given that our analysis was limited to patients who achieved CR2, the duration of CR1 might not be a significant factor for OS. Although the WBC count at relapse was not analysed in the other studies, it should be considered as an important factor.
We should note that there might be selection bias regarding the performance of Allo-HSCT in CR2, because it depended on each institution's decision. By a multivariate analysis using logistic regression for all covariates, relapse year (after 2003) and younger age at relapse (younger than 36 years) were significantly associated with the performance of Allo-HSCT in CR2. However, a multivariate analysis for OS including relapse year, age at relapse, and Allo-HSCT in CR2 (treated as a time-dependent covariate) as covariates demonstrated that Allo-HSCT in CR2, as well as age at relapse, was still significantly associated with better OS. In addition, there was no significant interactions between Allo-HSCT in CR2 and relapse year and between Allo-HSCT in CR2 and age at relapse (P = 0·36 and P = 0·97, respectively).
Comparisons of different salvage chemotherapy regimens have been limited (Thomas et al, 1999; Tavernier et al, 2007; Oriol et al, 2010). The selection of salvage regimens often depends on the condition of the relapsed patient (Garcia-Manero & Thomas, 2001). In our analyses, the duration of CR1, intensity of chemotherapy at diagnosis, and relapse year were factors that were associated with the selection of salvage regimens. AdVP-type salvage chemotherapy was related to a better OS in patients who had a longer duration of CR1 and a worse OS in those who had a shorter duration of CR1. If we consider that many patients had received induction chemotherapy including doxorubicin, vincristine, and steroids, the duration of CR1 might reflect the sensitivity of ALL to the AdVP-type regimen. Patients who had a longer duration of CR1 might have had ALL that was sensitive to an AdVP-type regimen, and those who had a shorter duration of CR1 might have had ALL that was refractory to an AdVP-type regimen. In patients who relapsed late, the toxicity of moderate-intensity regimens, such as those including high-dose cytarabine, used as the first salvage chemotherapy might offset their effectiveness. Unlike in a previous study (Thomas et al, 1999), the type of chemotherapy at diagnosis did not influence OS following each salvage regimen.
In conclusion, the prognosis of adult patients with relapsed Ph-negative ALL is poor. However, Allo-HSCT after the first relapse could improve the prognosis, especially if performed in CR2. The efficacy of different types of salvage chemotherapy might depend on the duration of CR1, and this should be considered in the selection of the salvage regimen.
The authors thank all of staff of the participating institutions that provided excellent patient care and reported clinical data for this study. The list of participating centres is included in Appendix S1.
Author contributions: S.K. wrote the paper. All authors contributed to writing the paper and checked the final version. S.K. and Y.K. designed the study and analysed data. All authors participated in data collection.
Conflict of interest disclosures
The authors declare no competing financial interests.