BCL2-Inhibitors Target a Major Group of Newly-Diagnosed and Relapsed/Refractory Acute Myeloid Leukemia Ex Vivo

BACKGROUND:

BCL-2 family members play a critical role in the regulation of apoptosis. BCL-2 and BCL-XL promote cell survival by preventing mitochondrial apoptotic pore formation. BH3 mimetic drugs such as venetoclax (ABT-199) promote apoptosis by inhibiting BCL-2 while navitoclax (ABT-263) inhibits both BCL-2 and BCL-XL. In AML, the expression of anti-apoptotic proteins is highly variable. In a recent study venetoclax showed single-agent activity in 6/12 AML cell lines and 20/25 patient samples. (Pan et al. Cancer Disc 2014). The samples with complex cytogenetics were largely resistant. Sensitivity correlated with increased BCL-2 protein levels and negatively correlated with BCL-XL and MCL-1 protein levels. We aimed to expand these data to both newly diagnosed and relapsed AML patients and to isolate biomarkers for patient selection.

METHODS:

We assessed the ex vivo sensitivity of fresh leukemic cells from 16 diagnosed and 36 relapsed/refractory AML patient samples to venetoclax (25 samples) and navitoclax (52 samples). Exome sequencing was performed on 32 samples and gene expression of BCL2 family members (BCL-2, BCL-XL, MCL-1, BIK, BAX, BAK1, BID, BCL2L12, BIM, BCL2A1, PUMA and BAD) was determined on 31 samples by qRT-PCR. Samples from primary cells of healthy individuals (n=10), and CMML (n=7) or CLL (n=2) patients were used as controls. Drug sensitivity was determined over a 10,000-fold concentration range (1-10 000 nM). A leukemia-specific drug sensitivity score (sDSS) derived from area under the dose response curve calculations was used as the efficacy variable by comparing leukemia results with those from normal bone marrow cells (Bhagwan et al., Sci Rep 2014, Pemovska et al., Cancer Disc 2014).

RESULTS:

Compared to healthy controls, CMML samples were largely non-sensitive, whereas CLL samples were highly sensitive to BCL-2 inhibitors ex vivo. The AML samples exhibited heterogeneous responses. 15/25 (60%) of AML samples were sensitive to venetoclax and 35/52 (67%) to navitoclax.

Both diagnostic (12 of 16 samples, 75%) and relapsed/refractory samples (24 of 36 samples, 64%) were sensitive to navitoclax. Similarly, 6/7 (86%) of diagnostic samples and 9/18 (50%) of relapsed/refractory samples were sensitive to venetoclax. We observed responses to venetoclax and navitoclax in each patient to be similar, although navitoclax showed efficacy at lower concentrations: in 25 samples tested with both agents, mean sDSS values were lower in navitoclax-treated samples (paired t-test, p=0.02). All except one patient sample exhibited a difference in resistance between the two drugs showing sensitivity to navitoclax but not to venetoclax.

We observed responses across all mutational profiles, including samples with mutations to FLT3-ITD, NPM1, TP53, NRAS and IDH1 and IDH2, as well as in samples with complex karyotypes.  Intriguingly, three of four samples with mutated TP53 exhibited sensitivity to BCL2 inhibition. No single mutation predicted sensitivity or resistance. At the RNA level, no statistical correlation between BCL2 or BCL-XL expression for BCL2 inhibitor response was observed. Instead we observed high levels of beta-2-microglobulin (B2M) mRNA expression in BCL2 inhibitor-resistant samples with a strong negative correlation to navitoclax sensitivity (r₌-0.60, P=0.0008).

DISCUSSION:

We did not observe BCL2 and BCL-XL mRNA expression to be optimal predictors for BCL2 inhibitor response. On the other hand, we observed high expression of B2M mRNA expression in resistant samples suggesting that it could serve as a biomarker for sensitivity to BCL2 inhibitors. The high B2M expression has been previously linked to poor prognosis in solid tumors and in AML (Albitar et al., Leukemia 2007). In cell line models B2M leads to phosphorylation and inactivation of proapototic protein BAD (Nokura et al., J Urol 2007). This may affect the balance between pro- and antiapoptotic proteins and thus offer a escape route from BCL2 inhibition.

To conclude, we observed BCL2 inhibition to be effective ex vivo in over half of all AML samples, tested both in primary and relapsed/refractory state as well as across different subgroups defined by AML driver mutations. Of the potential biomarkers that were assessed, B2M was the best mRNA-level indicator for anti-BCL-2 drug efficacy.