In patients with acute myeloid leukemia (AML), testing for fms-like tyrosine kinase-3 (FLT3)–internal tandem duplication (FLT3-ITD) and nucleophosmin-1 (NPM1) mutations can allow for further prognostic subclassification, but less is known about the effects of FLT3-ITD allele burden and presenting white blood cell count (WBC) within molecular subgroups.
The authors retrospectively assessed 206 adult patients who had AML with an intermediate-risk karyotype and who received treatment on a uniform induction and consolidation chemotherapy regimen.
The presenting WBC was a prognostic factor for survival only in patients who had an FLT3-ITD mutation. On multivariate analysis, after correcting for age, WBC, secondary AML, and blast percentage, nucleophosmin-1 (NPM1)-mutated/FLT3-ITD–negative patients had superior overall survival compared with patients in the other molecular subgroups. Patients who had FLT3-ITD mutations had an inferior overall survival compared with patients who had NPM1 wild-type/FLT3-negative disease, and patients who had low or intermediate levels of the FLT-ITD of mutant allele had overall and disease-free survival similar to those in patients who had high-level mutations.
Acute myeloid leukemia (AML) with intermediate-risk cytogenetics encompasses a heterogeneous population of patients whose prognosis has recently been further defined by molecular aberrations. Mutations in the nucleophosmin-1 gene (NPM1) are present in 35% to 50% of such patients and confer a survival advantage and lower risk of relapse.1-3 Conversely, an internal tandem duplication (ITD) of the fms-like tyrosine kinase-3 (FLT3) gene (FLT3-ITD) occurs in approximately 25% to 30% of patients and has an adverse impact on the relapse rate and overall survival (OS).4-6 Results of analyses for such mutations often are used to assess prognosis and to direct therapy, most notably the decision regarding whether patients should undergo allogeneic stem cell transplantation (alloSCT).
Several studies have suggested that the FLT3-ITD allele burden also may have prognostic significance.6, 7 A Medical Research Council study indicted that higher FLT3-ITD levels were correlated with higher relapse rates and worse OS, particularly when a high level (>50% allele burden) of FLT3-ITD was present6; however, that analysis included the entire cohort rather than the intermediate-risk cytogenetic group alone. Thus, many questions remain about the prognostic implications of FLT3-ITD levels in patients with intermediate-risk cytogenetics.
Traditionally, it has been held that an elevated presenting white blood cell count (WBC) is a poor risk factor.8-10 However, because FLT3-ITD mutations are associated with a higher WBC, it is unclear whether the WBC remains an important predictor of survival in each of the molecularly defined subgroups. It would be important to know whether the presenting WBC modifies the risk associated with the aforementioned mutations to determine whether postremission therapy needs to be tailored accordingly. Therefore, we sought to clarify the effect of FLT3-ITD level, NPM1 status, and presenting WBC in a population of patients with AML who had intermediate-risk cytogenetics.
MATERIALS AND METHODS
Patients and Treatment
We used a database of adult patients who were treated at Princess Margaret Hospital between 2002 and 2010 to identify 206 patients with adult AML who had an intermediate-risk karyotype based on Southwest Oncology Group (SWOG) criteria, as previously described,11 and who received induction chemotherapy. Approval of the hospital institutional review board was obtained before initiating this retrospective analysis.
For patients aged <60 years, induction chemotherapy consisted of cytarabine 200 mg/m2 daily as a 7-day continuous intravenous infusion and daunorubicin 60 mg/m2 intravenously daily × 3 days. Patients who achieved complete remission (CR) received 2 cycles of consolidation, each of which included intravenous cytarabine 3 g/m2 over 3 hours every 12 hours for 6 doses on days 1, 3, and 5 in combination with intravenous daunorubicin 45 mg/m2 daily × 2 days.
Patients aged ≥60 years received the same induction regimen but with a cytarabine dose of 100 mg/m2 daily. Those who achieved CR then received 2 consolidations, the first of which was the same as the induction regimen; the second consisted of intravenous mitoxantrone 10 mg/m2 daily for 5 days and intravenous etoposide 100 mg/m2 daily for 5 days (NOVE).
Patients who did not achieve CR with induction therapy and who were deemed fit to receive a reinduction regimen received combined mitoxantrone, etoposide, and high-dose cytarabine (NOVE-HiDAC), as previously described.12 The decision to proceed to alloSCT in first CR (CR1) was based on donor availability and the clinical judgment of the treating physician. In the event of relapse, physicians opted for various modalities, including reinduction with or without alloSCT in second CR (CR2), enrollment onto a clinical trial, or supportive therapy alone.
Response and Outcome Criteria
Response to treatment was assessed using International Working Group criteria.13 CR was defined as a postinduction BM with <5% blasts, a peripheral blood platelet count >100 × 109/L, an absolute neutrophil count >1.0 × 109/L, and the absence of extramedullary disease. OS was calculated from the initiation of first induction treatment for AML to the date of either death or last follow-up. Disease-free survival (DFS) was determined only for patients who achieved CR and was measured from the date of CR1 achievement until the date of death, relapse, or last follow-up.
Total cellular RNA was extracted from the peripheral blood using a PerfectPure RNA Blood Kit (5 PRIME, Inc., Gaithersburg, Md), then reverse-transcribed into cDNA using random hexamers and Moloney murine leukemia virus (MMLV) reverse transcriptase. Multiplex polymerase chain reaction analysis for FLT3 and NPM1 was performed using fluorescent primers specific for the FLT3 gene and NPM1 gene. FLT3-ITD and NPM1 insertions were identified using fragment analysis on an ABI Genetic Analyzer 3100/3130 platform (Applied Biosystems, Foster City, Calif). FLT3-ITD was quantified by determining the quantity of mutant allele as a percentage of total allele burden (wild-type plus mutated). A negative FLT3-ITD test was 1 that fell below the threshold of detection (<5%). A low level of mutant allele (FLT3+[L]) was defined as ≥5% but <25%, an intermediate level (FLT3+[I]) was defined as ≥25% but <50%, and a high level (FLT3+[H]) was defined as≥50%. The lower limit of sensitivity of the NPM1 assay was 1% to 10%.
Differences in demographic variables according to mutation status were investigated using the Fisher exact test for categorical variables, such as prior disease and sex, and the Kruskal-Wallis test for age, WBC, and BM blast percentage. OS and DFS probabilities were calculated using the Kaplan-Meier method. Differences in survival curves were compared using a 2-sided log-rank test. Cox proportional hazards regression models were used to examine the effect of mutation status on OS and DFS after adjusting for age, WBC, prior disease, and BM blast percentage. Statistical interactions were investigated to explore whether the effect of WBC on OS and DFS changed depending on mutation status. Simple effect analyses reported the effect of WBC for each mutation group status. The association between concurrent NPM1 and FLT3 mutations and differences in CR rates were investigated using the Fisher exact test. For the patients who had FLT3 mutations, a Cochran-Armitage test for trend was used to detect whether CR rates consistently increased as the level of FLT3 burden increased. All statistical analyses were performed using SAS statistical software (version 9.2; SAS Institute, Inc., Cary, NC) and the open-source R statistical software package (version 2.12.1; R Foundation for Statistical Computing, Vienna, Austria). A 2-sided P value < .05 was used to assess statistical significance.
Demographics for the 206 patients as well as a breakdown according to NPM1 and FLT3 status are provided in Table 1. There were 68 patients (33%) who tested positive for FLT3-ITD mutations (FLT3+). Patients with FLT3 mutations, regardless of level, were more likely to have concurrent NPM1 mutations than patients who were FLT3-negative (FLT3−) (59% vs 37.7%; P = .0082). The median BM blast percentage was similar among the molecular subgroups (P = .313). There was no correlation between the FLT3 allele burden and BM blast percentage (Spearman correlation, 0.22). Cytogenetics were normal in 195 patients (95%); and abnormalities included chromosome 8 gain (+8) (n = 5), chromosome Y loss (−Y) (n = 2), and +Y, +X, +7, and loss of the short arm of chromosome 12 (−12p) (n = 1 each).
Abbreviations: −, negative; + positive; FLT3, fms-like tyrosine kinase-3; FLT3+(H), FLT3-positive with a high level of the mutant allele (≥50% expression); FLT3+(I), FLT3-positive with an intermediate level of the mutant allele (≥25% but <50% expression); FLT3+(L), FLT3-positive with a low level of the mutant allele (≥5% but <25% expression); NPM1, nucleophosmin-1; WBC, white blood cells.
All P values were calculated using the Fisher exact test except as noted.
NPM1 results were missing in 7 patients.
This P value was calculated using the Kruskal-Wallis test across all 5 categories.
Of all 206 patients, 155 (75.2%) achieved CR with 1 induction. Eight patients (3.9%) died during induction. Of the 43 patients (20.9%) who had persistent leukemia after first induction, 37 were reinduced and 24 (64.9%) achieved CR after this second induction. Thus, the total CR rate with 2 inductions was 86.9%. Of the patients who achieved CR, 3 older patients were not candidates for consolidation and either received low-dose cytarabine alone (n = 2) or were observed without treatment (n = 1); all remaining patients received consolidation therapy on protocol.
In total, 24 patients underwent alloSCT in CR1, 15 from human leukemic antigen (HLA)-matched sibling donors and 9 from HLA-matched unrelated donors. Most of these patients (n = 20) were aged <60 years and received myeloablative conditioning using a variety of regimens. Patients aged ≥60 years (n = 4) received reduced-intensity conditioning. Among the patients who underwent transplantation in CR1, 2 were in the FLT3+(H) subgroup, 6 were in the combined low-intermediate FLT3-ITD (FLT3+[L-I]) subgroup, 12 were in the wild-type NPM1 (NPM1−)/FLT3− subgroup, and 4 were in the NPM1+/FLT3− subgroup.
The median follow-up of the entire group was 26.4 months. Of the 179 patients who achieved CR, 98 patients (54.7%) relapsed. Among these, 70 patients were reinduced and 41 patients (58.6%) achieved CR2; 16 of these patients were consolidated with alloSCT in CR2 or later. The median OS of the entire cohort of 206 patients was 21.2 months (95% confidence interval, 16.9-30.5 months), and 117 patients (56.8%) died.
Influence of Molecular Status on the Complete Remission Rate
Patients who were NPM1+ were more likely to obtain CR with up to 2 inductions than those who had wild-type NPM1 (CR rate 95.5% vs 81.3%; P = .0023). Conversely, FLT3-ITD status did not significantly affect CR rates for the negative, low, intermediate, and high groups, which had CR rates of 91.3%, 70%, 76%, and 86.9%, respectively (P = .09).
Influence of Molecular Status on Survival
Figure 1a indicates that patients in the NPM1+/FLT3− subgroup had superior OS compared with all other subgroups, with a 3-year OS rate of 61%. Patients who were both NPM1− and FLT3− (defined as the double-negative group) had a 3-year OS rate of 37%. Patients who had a low, intermediate, or high levels of FLT3-ITD, regardless of NPM1 status, had 3-year OS rates of 25%, 25%, and 12%, respectively. Comparing the low and intermediate level FLT3-ITD groups, the hazard ratio for OS was 0.898 (P = .759), indicating that these 2 groups did not differ in their outcomes. Thus, these 2 groups were combined for further analysis. This combined FLT3+(L-I) subgroup had OS that was similar to that in the FLT3+(H) subgroup but significantly worse OS than the double-negative group (Table 2).
Table 2. Univariate Analysis Demonstrating the Effect of Mutation Status on Overall and Disease-Free Survival
Abbreviations: −, negative; + positive; CI, confidence interval; FLT3, fms-like tyrosine kinase-3; FLT3+(H), FLT3-positive with a high level of mutant allele; FLT3+(L-I), FLT3-positive with low to intermediate levels of mutant allele; HR, hazard ratio; NPM1, nucleophosmin-1.
This P value indicates a statistically significant difference.
The OS results were similar for the cohort of patients aged <60 years who did not undergo alloSCT in CR1 (n = 119), as indicated in Figure 1b. In this cohort, patients in the NPM1+/FLT3− subgroup continued to have significantly superior OS (3-year OS rate, 75%). The OS rate in the FLT3+(H) subgroup, again, was significantly inferior to that in the double-negative subgroup (hazard ratio [HR], 2.34; P = .017), but it was not significantly different from the OS in the combined FLT3+(L-I) subgroup (HR, 1.42; P = .330).
The DFS according to molecular subgroup is illustrated in Figure 2. The low and intermediate level FLT3-ITD groups again had similar DFS and, thus, were combined into an FLT3+(L-I) subgroup for further analysis. The NPM1+/FLT3− subgroup had a 3-year DFS rate of 47%, which was significantly superior to that in the other subgroups; the DFS rate for the double-negative, FLT3+(L-I), and FLT3+(H) subgroups were 22%, 26%, and 10%, respectively (log-rank P = .000035 comparing all 4 subgroups). DFS was significantly inferior in the FLT3+(H) subgroup compared with both the double-negative subgroup and the FLT3+(L-I) subgroup. Unlike OS, DFS did not differ between the double-negative and FLT3+(L-I) subgroups (Table 2).
In the cohort of patients aged <60 years who did not undergo transplantation in CR1, DFS remained significantly better in the NPM1+/FLT3− subgroup; similarly, the FLT3+(H) subgroup continued a trend toward worse DFS compared with double-negative subgroup (HR, 1.86; P = .076) and the combined FLT3+(L-I) subgroup (HR, 1.63; P = .20). There was no significant difference in DFS between the double-negative and FLT3+(L-I) subgroups.
Influence of Presenting White Blood Cell Count
The WBC at presentation was analyzed both as a continuous variable and using cutoffs values of 50 × 109/L and 100 × 109/L. Table 3 indicates that the WBC had no influence on either OS or DFS in both the NPM1+/FLT3− and double-negative subgroups when analyzed either as a continuous variable or using specific cutoff values. When analyzed as a continuous variable, a high WBC was associated with significantly inferior OS in the FLT3+(H) subgroup and the FLT3+(L-I) subgroup. When analyzed as a dichotomous variable, a WBC ≥50 × 109/L was a negative prognostic factor in the FLT3+(L-I) subgroup for both OS and DFS, and a WBC ≥100 × 109/L was a negative prognostic factor for OS in the FLT3+(H) and FLT3+(L-I) subgroups.
Table 3. Influence of Presenting White Blood Cell Count on Overall and Disease-Free Survival According to Molecular Subgroup
OS, n = 206
DFS, n = 179
Abbreviations: −, negative; + positive; DFS, disease-free survival; FLT3, fms-like tyrosine kinase-3; FLT3+(H), FLT3-positive with a high level of mutant allele; FLT3+(L-I), FLT3-positive with low or intermediate levels of mutant allele; HR, hazard ratio; NPM1, nucleophosmin-1; OS, overall survival; WBC, white blood cells.
This P value indicates a statistically significant difference.
This effect was maintained in the cohort of patients aged <60 years who did not undergo alloSCT in CR1 (data not shown), in which a higher WBC as a continuous variable had a significant negative effect on both OS (HR, 1.01; P = .0077) and DFS (HR, 1.01; P = .035) in the FLT3+(H) subgroup. Similarly, a WBC ≥50 × 109/L remained a negative prognostic factor for both OS (HR, 3.40; P = .001) and DFS (HR, 3.01 P = .0075) in the FLT3+(L-I) subgroup; whereas a WBC ≥100 × 109/L became a significant negative prognostic factor in the FLT3+(H) subgroup for both OS (HR, 3.79; P = .028) and DFS (HR, 4.349; P = .015).
In a multivariate Cox regression analysis that evaluated mutation status, age, WBC, prior disease (de novo vs secondary), and BM blast percentage, all variables except BM blast percentage remained significant predictors of outcome for both OS and DFS, as indicated in Table 4. The NPM1+/FLT3− subgroup continued to fare better than all other molecular subgroups. Correcting for age, WBC, and disease status, there remained only a nonsignificant trend toward inferior DFS in the FLT3+(H) subgroup compared with the FLT3+(L-I) subgroup, but OS did not differ between these 2 subgroups. Both the FLT3+(L-I) and FLT3+(H) subgroups continued to have worse OS compared with the double-negative subgroup, whereas DFS remained similar between the FLT3+(L-I) and double-negative subgroups.
Table 4. Results of Multivariate Analysis According to Mutation Status, White Blood Cell Count, Age, Prior Disease, and Bone Marrow Blast Percentage on Overall and Disease-Free Survival
Abbreviations: −, negative; + positive; BM, bone marrow; CI, confidence interval; FLT3, fms-like tyrosine kinase-3; FLT3+(H), FLT3-positive with a high level of mutant allele; FLT3+(L-I), FLT3-positive with low to intermediate levels of mutant allele; HR, hazard ratio; NPM1, nucleophosmin-1; WBC, white blood cells.
This P value indicates a statistically significant difference.
Within the FLT3+ subgroup, multivariate analysis was also performed to assess survival using FLT3 allele burden as a continuous variable. In this model, FLT3 burden did not significantly influence OS or DFS within the FLT3+ population when adjusting for presenting WBC, age, and prior disease (Table 5). The WBC remained a significant predictor of OS within the FLT3+ subgroup. Furthermore, correcting FLT3 burden for the BM blast percentage did not alter the results; the ratio of FLT3 allele burden to BM blast percentage did not significantly predict for OS or DFS (data not shown).
Table 5. Results of Multivariate Analysis Using fms-Like Tyrosine Kinase-3 (FLT3) Allele Burden as a Continuous Variable Within the FLT3-Positive Group: Overall and Disease-Free Survival
In this retrospective study, we analyzed patients with AML who had intermediate-risk cytogenetics and identified 4 molecularly defined subgroups: 1) an NPM1+/FLT3− subgroup, which that had significantly superior OS and DFS compared with all other subgroups; 2) a double-negative cohort, in whom OS and DFS were inferior to those of the NPM1+/FLT3− subgroup but superior to those of the FLT3+(H) subgroup (this subgroup had comparable DFS but significantly better OS compared with the FLT3+[L-I] subgroup); 3) the combined FLT3+(L-I) cohort, which had similar OS but significantly better DFS (on univariate analysis) compared with the FLT3+(H) subgroup; and 4) the FLT3+(H) subgroup, which had the worst prognosis of the 4 cohorts of patients.
The difference in DFS between the FLT3+(L-I) and FLT3+(H) subgroups was no longer significant after correcting for age, WBC, prior disease status, and BM blast percentage on multivariate analysis. This finding was confirmed using traditional, predefined cutoff values for FLT3 burden and using FLT3 burden as a continuous variable. Thus, it is likely that the differences observed on univariate analysis were related to the higher median WBC in the FLT3+(H) subgroup. However, it is also possible that the absence of a detectable difference on multivariate analysis was related to the small numbers in each cohort as well as a confounding influence of NPM1 mutations in these FLT3+ groups. We also cannot rule out the possibility that some of the less significant differences observed were related to the number of statistical comparisons done.
In a recent study, de Jonge et al observed that the WBC did not significantly affect outcome in either the NPM1+/FLT3− or the double-negative subgroup.14 Our results confirm their findings, because we observed no significant effect of the presenting WBC, either as a continuous variable or as a dichotomous variable using several predefined cutoff levels. That study by de Jonge et al demonstrated that using a cutoff of 100 × 109/L, the WBC was a significant prognostic factor for FLT3+ patients who were NPM1+ but not in FLT3+ patients with wild-type NPM1. In our study, we chose to analyze patients within the FLT3+ subgroup according to FLT3 allele burden rather than NPM1 status. With this approach, we observed that a high presenting WBC was an independent, adverse prognostic variable in both the FLT3+(L-I) and FLT3+(H) subgroups, and this effect was maintained on multivariate analysis. This influence of WBC on OS remained significant when FLT3 allele burden was assessed on a continuum. In our study, we did not have enough patients to analyze the influence of WBC according to NPM1 status within these FLT3+ subgroups.
Clinically, these results can assist in decision making regarding alloSCT and its timing. Schlenk et al demonstrated that patients who are either FLT3-ITD+ or who are negative for both FLT3-ITD and NPM1 mutations have a superior outcome if they have a potential alloSCT donor compared with those without a donor.15 Our results suggest that the presenting WBC should not be used to modify the decision to undergo alloSCT in either the NPM1+/FLT3− cohort or the double-negative cohort: Those with NPM1+/FLT3− findings should not be referred for transplantation on the basis of a high WBC alone; conversely, patients in the double-negative group should not have a transplantation withheld simply because they presented with a normal or low WBC. Although our double-negative cohort had a longer CR1 duration than the FLT3+ subgroups, these patients continued to relapse as late as 3 and 4 years postremission and, ultimately, their DFS remained unfavorable.
In contrast, patients who had a high FLT3 allele burden tended to relapse earlier, usually in the first 6 to 12 months after achieving CR. After correcting for the WBC on multivariate analysis, the DFS between the FLT3(H) and FLT3(L-I) subgroups became similar. This suggests that members of either subgroup, particularly those with higher WBC, should undergo transplantation as quickly as possible and that perhaps these patients should remain on consolidation therapy for a longer period until a suitable donor is identified. Several published reports have argued against the prognostic value of a low-intermediate level of FLT3.6, 16 Our results support the Medical Research Council study6 and indicate that a low-intermediate level of FLT3-ITD positivity confers a worse prognosis than the absence of mutated FLT3-ITD. Furthermore, after correcting for WBC, patients who had a low-intermediate allele burden fared as poorly as those with a high allele burden, indicating that the presence of FLT3-ITD, rather than its level, is what confers prognostic importance.
The outcomes in our study population were somewhat inferior to those in other published reports, particularly with respect to DFS in the aged <60 years, double-negative cohort. It is possible that our results are a better reflection of treatment outcomes in an unselected AML population than what may be observed in a controlled trial setting. It is also possible that the use of 2 consolidation cycles, compared with the 4 cycles used in previous reports,17 may have contributed to a higher relapse rate. Finally, outcomes in this subgroup may have been influenced by the presence of additional, recently identified mutations that reportedly affect prognosis, such as CEPBA,18TET2,19IDH1 and IDH220, 21 and RUNX1.22 Further studies will be required to determine the potential roles of such mutations in decision-making regarding alloSCT.