Impact of early molecular response in children with chronic myeloid leukemia treated in the French Glivec phase 4 study

Although imatinib is effective for the majority of children with chronic myeloid leukemia (CML),^1-3  13% to 25% of children have a poor response.^3-5  Reports in adults indicate that measurement of BCR-ABL1 transcripts 3 months after starting imatinib can predict outcome with better overall survival in patients with a BCR-ABL1/ABL ratio of <10% at this time point.⁶  In order to address the relevance of this cutoff in the pediatric population in terms of cytogenetic and molecular responses and outcome, we reanalyzed the French Glivec phase 4 study data.³ 

This study was approved by the institutional review board of the University Hospital of Poitiers (Poitiers, France) and participants provided informed consent in accordance with the Declaration of Helsinki. Between March 2004 and December 2008, 44 consecutive patients younger than 18 years of age with newly diagnosed CML in chronic phase were enrolled in the prospective Glivec phase 4 study from 16 French university hospitals. Patient eligibility and trial design were previously described.³  Patients received imatinib 260 mg/m² daily.¹  Bone marrow assessment including morphology, cytogenetics, and molecular analysis was performed at diagnosis, 3 months, and 12 months after the start of imatinib. Complete cytogenetic response (CCyR) was defined by the absence of Ph-positive metaphases among at least 20 analyzed metaphases in bone marrow. Determination of BCR-ABL1 transcript level in the blood was performed at 3-month intervals using quantitative reverse transcription–polymerase chain reaction independently aligned on the International Scale (IS) by specific conversion factors as described previously.^7-9  Major molecular response (MMR) was defined as a BCR-ABL1 IS ≤0.1%. molecular response (MR)^4.0 and MR^4.5 were defined as either detectable disease with BCR-ABL1 IS ≤0.01% and ≤0.0032%, respectively, or undetectable disease with ≥10 000 and ≥32 000 ABL1 copies, respectively, in the same volume of complementary DNA used to test for BCR-ABL1. Complete molecular response (CMR) was defined by no detectable transcripts in 2 consecutive samples, with ≥32 000 ABL1 copies. BCR-ABL1 transcript levels (≤1%, >1% to ≤10%, >10%) 3 months after starting imatinib were used to assess the association between early molecular response and the rates of CCyR and MMR 1 year after the start of imatinib as well as the association between early molecular response and progression-free survival (PFS). PFS was defined by absence of accelerated phase, blast crisis, and death from any cause.¹⁰ 

Comparisons between groups were performed using the χ² test or the nonparametric Fisher exact test and Wilcoxon test, as appropriate. PFS from the onset of imatinib was estimated by the Kaplan-Meier method and was compared within groups by log-rank test. All analyses were 2-sided and were performed using SAS, version 3 (SAS Institute).

Four of the 44 patients enrolled in the Glivec phase 4 study had a missing polymerase chain reaction assessment 3 months after starting imatinib and were excluded from landmark analyses performed on BCR-ABL1 transcript levels. Hence, the patient population consisted of 40 children. Because of severe side effects, 1 of them was switched to a second-generation tyrosine kinase inhibitor (TKI) 2 months after starting imatinib and was included in the intention-to-treat analysis. Patient characteristics are reported in Table 1. A high Sokal risk score (for patients younger than 45 years of age) was observed in the majority of the patients.¹¹  Three months after the start of imatinib, a BCR-ABL1/ABL ratio of >10% was found in more than one-third of the patients (15 of 40; 37%) whereas 18 of 40 of the patients (45%) had a BCR-ABL1/ABL ratio of 1% to 10% and 7 of 40 patients (18%) had a BCR-ABL1/ABL ratio of ≤1%. Children and adolescents with a BCR-ABL1/ABL1 ratio >10% 3 months after starting imatinib had a similar Sokal score distribution at diagnosis but a larger spleen size and a higher white blood cell count compared with patients with a BCR-ABL1/ABL ratio of ≤10% (Table 1). The dose of imatinib administered within the first 3 months of treatment was similar (P = .209) in patients with a BCR-ABL1/ABL ratio >10% (median, 242 mg/m²; range, 132-381) and those with ≤10% at 3 months (median, 250 mg/m²; range, 300-342). Two children progressed to blast phase 25 months and 42 months after the start of imatinib (1 of them was switched for a second-generation TKI before onset of progression) and 1 of them died. Both children had a BCR-ABL1/ABL ratio of >10% at 3 months after starting imatinib. No other death was reported.

With a median follow-up of 71 months (range, 22-96 months), a transcript level ≤10% correlated with better PFS (Table 2, BCR-ABL1/ABL transcript level at 3 months) (supplemental Figure 1, available at the Blood Web site). The rate of cytogenetic response 12 months after starting imatinib was: CCyR, 72.5%; partial cytogenetic response (PCyR), 10%; less than PCyR, 12.5%. Twelve months after starting imatinib, the rates of MMR, MR^4.0, MR^4.5, and CMR were 28%, 3%, 3%, and 0%, respectively. Patients with a BCR-ABL1/ABL of ≤10% 3 months after starting imatinib had a higher rate of CCyR and MMR 12 months after starting imatinib compared with those with a BCR-ABL1 transcript level >10% (Table 2, BCR-ABL1/ABL transcript level at 3 months). When children were classified according to the transcript level at 3 and 6 months using the 10% cutoff (Table 2, molecular response rate outcome over 12 months) according to the European Leukemia Net (ELN) response criteria,¹²  those with BCR-ABL1/ABL of <10% at these 2 time points (60% of the patients) had a higher rate of MMR but comparison with the other 3 subgroups could not be performed because of the small number of patients.

Early molecular response defined by a BCR-ABL1 transcript level ≤10% at 3 months has been reported to predict survival in adults treated with imatinib.^6,13-15  Recently, Jabbour et al reported significant improvement in PFS in patients who achieved a transcript level ≤1% 6 months after starting imatinib.¹⁶  It is important to note that a BCR-ABL1 transcript level of >10% at 3 months and/or of a level between 1% and 10% at 6 months after the start of imatinib are now considered as a warning according to the recent ELN recommendations.¹²  Our data show that a high proportion (37%; 95% confidence interval [CI], 23-54) of the children in our study failed to achieve the ≤10% cutoff at 3 months. This is higher than the 24% to 28% of adults who fail to reach the ≤10% cutoff.^6,14  This might be explained in part by the higher proportion of patients with bulky disease (large spleen, high leukocyte count) and a higher proportion of patients with a high Sokal risk score in our pediatric population compared with adult series.⁶  Consistent with this, Hughes et al showed that adult patients who failed to achieve a BCR-ABL1/ABL ≤10% 3 months after the start of imatinib also had a larger median spleen size and higher Sokal risk score than those who achieve an early molecular response.¹⁷  In the present study, the dose of imatinib administered during the first 3 months of treatment was similar in children who did or did not achieve an early molecular response, suggesting that treatment exposure was not an independent influencing factor.

In summary, given the limitation of the small number of patients and events, we show that BCR-ABL1 transcript level 3 months after starting imatinib can identify children with a lower rate of progression who might benefit from an alternative treatment strategy. The reliability of a cutoff of 10% at 3 months after the start of imatinib as a surrogate marker of response at 1 year and PFS remains to be confirmed in a larger cohort of children.

The authors gratefully acknowledge Prof Irene Roberts (Centre for Haematology, Imperial College London) for helpful comments on the manuscript and Sophie Zin-Ka-Ieu for the data monitoring.

Contribution: F. Millot coordinated the study and wrote the manuscript; J.G. performed statistical analysis; A.B., A.P.,Y.B., F. Mazingue, P.L., C.V., C.B., C.G., N.S., K.Y., C.S., V.G., Y.R., G.C., and F. Mechinaud provided patients; and J.M. performed the molecular review.

Conflict-of-interest disclosure: F. Millot received speaker fees from Novartis and Pfizer. The remaining authors declare no competing financial interests.

Correspondence: Frédéric Millot, CIC 1402 Inserm, 2 rue de la Miletrie, Poitiers, France; e-mail: f.millot@chu-poitiers.fr.