Imatinib resistant BCR-ABL1 mutations at relapse in children with Ph⁺ALL: a Children’s Oncology Group (COG) study

Dear Editor

Philadelphia chromosome positive acute lymphoblastic leukemia (Ph⁺ALL) occurs in 2-5% of pediatric ALL and is historically associated with a poor prognosis. Although 80-90% of children achieve remission, their event-free-survival (EFS) with conventional chemotherapy prior to tyrosine kinase inhibitors (TKI) was poor with a 32% 7-year event-free survival rate (Arico, et al 2010). The addition of imatinib as monotherapy appeared promising in initial treatment of adults with Ph⁺ALL, despite a high rate of relapse (Druker, et al 2001). Many relapsed adults on imatinib monotherapy were found to have a resistant mutation within the kinase domain of BCR-ABL1 (Jones, et al 2008). Other studies have shown that TKI’s like imatinib or dasatinib as monotherapy can select for TKI resistant clones which may then be overcome by the addition of cytotoxic chemotherapy in the mouse model (Boulos, et al 2011).

The Children’s Oncology Group (COG) clinical trial, AALL0031, used imatinib (340mg/m²/day) in conjunction with intensive chemotherapy to treat children and adolescents with Ph⁺ALL (Schultz, et al 2009). This dosage is equivalent to approximately 600mg/day in adults and was well tolerated with minimal additional side effects as compared to the identical chemotherapy arm without imatinib. AALL0031 differed from adult protocols in several aspects: use of drug combinations not common in adult protocols, intensive dosing of imatinib that was given continuously for the majority of 2.5 years and no continuation of TKI after completion of therapy. Three-year EFS on this treatment was 84% (more than double those of patients treated in the pre-imatinib era). Thus far, it remains unknown whether patients that relapse following this treatment approach have recurred due to development of imatinib resistance.

A two year-old male with Ph⁺ALL and initial white blood cell count of 117 × 10⁹ cells/L was initially treated with standard four-drug induction with vincristine, asparginase, doxorubicin, and prednisone. At presentation he showed no evidence of extramedullary disease. He achieved complete morphologic and cytogenetic remission at the end of induction. He then received post-induction therapy according to COG AALL0031 cohort 5 (not on study). His therapy included the intensive systemic regimen with CNS-directed therapy without cranial radiation. Twenty-four months into treatment he presented with headaches and mental status changes caused by a CNS relapse. BCR-ABL1 sequence analysis of his cerebrospinal fluid (CSF) blasts identified a guanine substitution for adenine producing the missense mutation methionine 244 to valine (M244V) (Figure 1a). Concomitant bone marrow aspiration showed no leukemia by morphology, flow cytometry or by fluorescent in situ hybridization. However, sequence analysis of the marrow sample identified the same mutation found in his CSF. BCR-Abl sequencing of the bone marrow specimen from initial diagnosis identified no mutation (Figure 1a).

Figure 1. Sequence chromatogram of case.

a. Sequence results from the presented case at diagnosis and relapse. Briefly, patient total RNA was isolated from mononuclear cells and cDNA was synthesized with random hexamers using Superscript III (Invitrogen, Grand Island, NY, USA). Patient cDNA (1-μl) was PCR amplified using B2A forward (TTCAGAAGCTTCTCCCTGACAT) and ABL1 4317 reverse (AGC TCT CCT GGA GGT CCT C) in a 20-μl reaction (Table 1). PCR product (1-μl) was used as template in the second round (nested) 50-μl reaction with BCR F4 forward (ACAGCATTCCGCTGACCATCAATA) and ABL1 4307 reverse (GAGGTCCTCGTCTTGGTGG) primers. All PCR reactions were performed with Accuprime TAQ, (Invitrogen) and 20 pmol of each primer. Resulting PCR products were sequenced with 2 forward and 2 reverse sequencing primers. Sequences were aligned with Sequencher (Gene Codes, Ann Arbor, MI, USA) sequence analysis program. b. Sequence results from case #9 from bone marrow samples from diagnosis and relapse.

A biologic correlate study to AALL0031 was developed to determine whether or not BCR-ABL1 kinase domain mutations were present in medullary relapse samples from Ph⁺ALL patients. COG AALL0031 enrolled 93 patients with Ph⁺ALL aged 1 – 21 years from 2002 to 2006 (Schultz, et al 2009). From this study, nine relapsed bone marrow samples were available for sequence analysis (Table 1). Eight of the 9 samples from imatinib-treated patients showed no BCR-ABL1 kinase domain mutation (Table 1). One sample from a patient who relapsed 15 months after diagnosis, carried the histidine 396 to proline (H396P) mutation (Figure 1b). A bone marrow sample from initial diagnosis of this child identified no mutation (Figure 1b).

Table I.

Patient Samples

#	time to relapse*	imatinib exposure	BCR-ABL1
Case$	24 months	>1 yr	M244V
1	26 months	<1 yr (BMT)†	WT
2	10 months	<1 yr	WT
3	refractory	<1 yr	WT
4	12 months	<1 yr	WT
5	23 months	<1 yr	WT
6	19 months	<1 yr (BMT)†	WT
7	39 months	>1 yr	WT
8	45 months	>1 yr	WT
9	15 months	>1 yr	H396P

^*As defined from initiation of treatment

^$Not on study, treated according to AALL0031 cohort 5

^†patients who underwent BMT were scheduled to receive 6 months of imatinib after engraftment.

These results further validate that BCR-ABL1 kinase domain mutations can occur after treatment of Ph⁺ALL with imatinib and intensive multiple chemotherapeutic agents. From these ten samples we identified two resistant mutations from patients who received imatinib and combination chemotherapy for more than one year. This mutation rate appears to be less than previously published in adults treated with imatinib monotherapy (Jones, et al 2008) (15 of 17 relapsed patients) or with hyperCVAD combination therapy (Ravandi, et al 2010) where mutations were observed in 3 of 5 relapsed patients.

Neither mutation was detected in samples obtained at diagnosis suggesting that the vast majority of the leukemic cells did not have the mutation. This does not preclude the concept of a low level of mutations at diagnosis as previously shown (Pfeifer, et al 2007). M244V and H396 mutations have been shown to be more resistant to imatinib but both have been shown to be sensitive to second generation TKI’s such as nilotinib and dasatinib (O’Hare, et al 2005). Treatment with dasatinib has been shown to overcome H396R resistance in CML (Talpaz, et al 2006).

Our results are the first to describe BCR-ABL1 kinase domain mutations in pediatric patients with Ph⁺ALL treated with intensive chemotherapy and imatinib. We are also the first to report an imatinib resistant BCR-ABL1 kinase mutation from a CNS recurrence in a pediatric patient. It has been previously shown that imatinib has low penetrance into the CNS, which implies that selective pressure occurred systemically followed by expansion in the sanctuary of the CNS.

The two mutations we identified are predicted to be sensitive to second generation TKIs, suggesting that these TKIs may be effective reinduction therapy for relapse following treatment of Ph⁺ALL with chemotherapy and imatinib. Importantly, dasatinib penetrates the CNS, a property that may help to decrease the risk of CNS recurrence in Ph⁺ALL. These next generation TKI’s may further decrease relapse rates when used for initial therapy of Ph⁺ALL. Many patients who experience a relapse with combination chemotherapy and a TKI do not appear to carry a TKI resistant mutation, suggesting that other BCR-Abl independent pathways play critical roles in leukemia cell survival. Other tyrosine kinases such as HCK, FGR, and LYN are essential for Ph⁺ALL transformation (Hu, et al 2004). Therefore, less selective inhibitors like dasatinib may play an important role in salvage therapy for these patients. The current and planned COG Ph⁺ALL trials combine dasatinib rather than imatinib with intensive chemotherapy. Future studies will address whether dasatinib resistant BCR-Abl mutants develop in patients who relapse on these studies.