Multi-Agent Chemotherapy Overcomes Steroid Resistance Conferred by a BIM Deletion Polymorphism in Pediatric Acute Lymphoblastic Leukemia (ALL)

We have shown previously that a novel BIM intronic deletion polymorphism was significantly associated with resistance to tyrosine kinase inhibitors (TKIs) in a chronic myeloid leukemia (CML) cohort (OR=2.94, 95% CI = 1.17-7.43, p=0.02) (Ng et al. Nature Medicine, 18: 521-528, 2012). BIM is a potent pro-apoptotic protein that binds pro-survival BCL-2 family members via its BH3 domain, and is required for apoptosis induced by a number of stimuli, including TKIs. Mechanistically, the deletion favors the generation of non-apoptotic exon 3-containing BIM splice variants that lack the BH3 domain found in exon 4. Furthermore, we found that introducing the deletion into a TKI-sensitive CML cell line, K562, was sufficient to confer resistance, and that substituting BIM function with a BH3 mimetic drug could restore sensitivity.

Because BIM has also been implicated in steroid-induced apoptosis, we hypothesized that the polymorphism would contribute to response heterogeneity in ALL, where response to steroid treatment is an important prognostic factor. Hence, we used ALL as a model to study the effects of the deletion on steroid resistance and treatment outcome.

First, we employed zinc finger nucleases to recreate the 2.9kb intronic deletion in subclones of the ALL cell line CCRF-CEM. Using qPCR, we confirmed that the BIM exon 3: exon 4 ratio was increased in the heterozygous and homozygous clones in a polymorphism dose-dependent manner. When treated with dexamethasone, the deletion-containing clones expressed lower pro-apoptotic BIM isoforms, and were rendered resistant to steroid-induced apoptosis, as determined by MTT assays and PARP and caspase 3 cleavage. These data demonstrate that the polymorphism is sufficient to confer resistance to dexamethasone, and also suggest that it might predict for inferior responses in ALL patients treated with steroid-containing regimens.

To test this prediction, we genotyped 411 pediatric ALL patients from the Ma-Spore ALL 2003 study for the BIM polymorphism (Yeoh et al. J Clin Oncol. 30: 2384-92, 2012), and correlated genotype with clinical outcome. These patients were uniformly treated with a modified Berlin-Frankfurt-Münster protocol, which consisted of steroids as well as multi-agent chemotherapy including L-asparaginase and vincristine. When we compared outcome between patients with (n=50) and without (n=361) the polymorphism, there was no significant difference in either event-free survival or overall survival. Further, there was no significant enrichment in the intermediate and high Day 33 minimal residual disease group for patients with the polymorphism.

To account for the lack of clinical predictive value, we hypothesized that the chemotherapy agents administered in the study overcame steroid resistance conferred by the polymorphism. Indeed, we found that either L-asparaginase or vincristine resensitized the deletion-containing clones to steroid-induced apoptosis, and that this was associated with MCL-1 downregulation, but not BIM induction. Together, our data demonstrate that BIM deletion-mediated steroid resistance can be overcome by co-administration of drugs that induce cell death via a BIM-independent manner. Importantly, such drugs include standard chemotherapy agents that lower the cellular apoptotic threshold by decreasing pro-survival molecules like MCL-1.

1. The BIM deletion polymorphism confers steroid resistance to ALL cells in vitro.

2. In vitro resistance can be overcome with L-asparaginase or vincristine at clinically relevant doses.

3. The BIM deletion polymorphism does not predict for poorer clinical outcome in patients treated with standard protocols, and this is likely to be because co-administration of L-asparaginase or vincristine results in BIM-independent cell death.

4. Our findings suggest that selected chemotherapy agents may be used to overcome BIM deletion-associated inferior responses in cancer.

5. Finally, our experience highlights the difficulties of identifying associations between genetic variants and clinical outcome in cancers treated with multiple agents, since these may employ different mechanisms of killing and cause confounding. Our earlier success in identifying the polymorphism, and its role in mediating TKI resistance in CML, is likely attributable to the fact that CML is treated with TKI mono-therapy.

No relevant conflicts of interest to declare.