TKI resistance in Ph-like ALL

In this issue of Blood, Zhang et al identify and characterize an imatinib-resistant mutation (PDGFRB C843G) in a B-progenitor acute lymphoblastic leukemia (B-ALL) patient harboring a novel AGGF1-PDGFRB fusion.¹ 

Advances in cancer genomics and large-scale sequencing studies have significantly increased our understanding of the genetic landscape of ALL through the classification of new subtypes and, importantly, have identified therapeutic targets for the basis of precision medicine trials. One such subtype is Philadelphia chromosome-like ALL (Ph-like or BCR-ABL1–like), so termed because these patients have a similar gene expression profile to Ph⁺ ALL but lack the BCR-ABL1 fusion.²  Ph-like ALL is associated with high rates of treatment failure and poor survival in both children and adults³  and is characterized by a diverse range of genetic alterations that dysregulate cytokine receptor and signaling pathways,²  suggesting that treatment with tyrosine kinase inhibitors (TKIs) is an attractive therapeutic strategy for improving the outcome of these patients, as has been observed for childhood Ph⁺ ALL.⁴  Indeed, compelling preclinical evidence and anecdotal reports demonstrate potent activity of the ABL1 inhibitors, imatinib or dasatinib, against fusions involving ABL1, ABL2, CSF1R, PDGFRA, or PDGFRB (termed “ABL1 class”).^2,5 

Despite the clinical success of TKI therapy in chronic myeloid leukemia (CML) and Ph⁺ ALL, a substantial number of patients experience TKI resistance caused by the acquisition of secondary mutations in ABL1.⁶  To date, over 70 different ABL1 imatinib-resistant mutations have been identified, with the gatekeeper residue T315I being the most common.⁷  In contrast, very few mutations in PDGFR fusions have been identified, with fewer than 10 cases of T674I reported in patients with hypereosinophilia harboring FIP1L1-PDGFRA, possibly due to a limited number of residues that mediate critical drug binding.⁸  Although the T681I gatekeeper mutation in PDGFRB confers imatinib resistance in vitro, no patients have been identified with this mutation to date.⁹  Whether Ph-like ALL patients harboring ABL1-class fusions will develop TKI resistance is largely unknown. Zhang et al provide insight to this question by reporting the emergence of an imatinib-resistant mutation (PDGFRB C843G) acquired in a novel AGGF1-PDGFRB fusion during imatinib therapy. By performing whole-genome sequencing and sensitive droplet digital polymerase chain reaction of longitudinal samples obtained from diagnosis to relapse, they demonstrate that the PDGFRB C843G mutation was present in a very minor subclone (0.03%) 2 months prior to clinical relapse, which evolved into the major relapse clone. This suggests that predicting TKI response at diagnosis will be difficult but raises the possibility of molecular monitoring during therapy to identify emerging TKI-resistant mutations.

It is important to note that imatinib was provided to this patient as a monotherapy. Treatment of Ph⁺ ALL in adults suggests that TKI-resistant mutations are more likely to occur when administered as a monotherapy, whereas fewer TKI-resistant mutations are identified in children in whom TKI treatment is provided with a backbone of chemotherapy.¹⁰  This is also supported by the fact that although 90% of patients with early-stage CML (chronic phase) achieve hematologic responses with imatinib, the vast majority of patients with advanced-stage CML (blast crisis) experience relapse, suggesting that TKI therapy alone is less effective at treating acute leukemia.⁷  As such, ongoing clinical trials assessing the efficacy of targeted therapy with intensive chemotherapy in large cohorts of Ph-like ALL patients will provide important information on the clinical utility of these agents.

To provide causal evidence that the PDGFRB C843G mutation conferred resistance to imatinib, the authors expressed both the wild-type and mutated version of AGGF1-PDGFRB into experimental murine cell lines, including Ba/F3 and primary Arf^−/− cells and demonstrated reduced in vitro efficacy of imatinib, dasatinib, and a third-generation inhibitor, ponatinib, in mutant cells, although responses were observed at higher concentrations. Ponatinib targets the notoriously difficult-to-treat T315I mutation in ABL1, so it is noteworthy that the PDGFRB C843G mutation conferred resistance to this agent. The inhibition of oncogenic signaling pathways, including JAK-STAT, PI3K, and Ras, by dasatinib was also greatly attenuated in the presence of the mutant PDGFRB fusion, providing further confirmation that this mutation contributed to TKI resistance. The C843 residue is located within the activation loop of the PDGFRB kinase domain. Structural modeling indicates that the substitution of cysteine to glycine increases the flexibility of the binding pocket and interferes with dasatinib binding to the kinase domain. Whether this change also influences the confirmation of the PDFGRB activation loop, similar to activation loop mutations in ABL1,⁶  was not explored. Future studies investigating the specific biochemical and molecular properties of this mutation are required to determine the mechanisms of resistance. To identify alternative therapies, the authors show that the multikinase inhibitor, CHZ868, may overcome the TKI resistance observed with imatinib and dasatinib. Although this is an interesting strategy, preclinical studies are required to validate these conclusions.

The identification of targetable kinase alterations in Ph-like ALL has raised much interest and excitement within the leukemia community, with the potential of improving patient outcome through the use of TKIs. This report by Zhang et al provides a reality check on the clinical utility of targeted agents by demonstrating TKI resistance in a Ph-like ALL patient harboring a PDGFRB fusion. This information prepares us for the possibility of mutation-mediated TKI relapse in Ph-like ALL and highlights the need to continue identifying and developing improved therapeutic strategies for ALL patients at highest risk of relapse.