Detection of NFKB2 3’end Loos By Quantitative PCR (QPCR) or Detection of NFKB2 Rearrangements Correlate with Bortezomib Response in Multiple Myeloma

Background: Multiple myeloma is a malignant plasma cell cancer that is increasing in frequency in today’s aging population. While proteasome inhibitors such as bortezomib have improved the outcome of treatment, many patients fail to achieve a complete remission and ultimately the disease progresses. With these new treatments come higher associated toxicities and significant healthcare costs for the drugs. Several new cancer therapies with companion diagnostics have been approved in parallel with the drug to guide physicians in selecting those patients who would specifically benefit from the drug. No such test exists for the use of bortezomib in multiple myeloma. We have previously demonstrated that the 3’ end of the nuclear factor kappa-B 2 (NFKB2/p100) gene, a member of the NFKB/Rel gene family, functions as a mediator of bortezomib apoptotic effect through the induction of caspase 8 activation.

Methodology: We have investigated the loss of NF-kB2 3’end loss by quantitative PCR (QPCR) in 56 MM patients and by break-apart Fluorescent In-Situ Hybridization in the whole bone marrow of 31 patients as well as in sorted cells of 24 MM patients treated with bortezomib containing regimens. Quality of response was assessed after 4 cycles of treatment initiation. QPCR was measured in peripheral blood lymphocyte and CD138 positive cells from the bone marrow to generate a QPCR ratio.

Results: 37 of the 56 newly diagnosed patient with MM achieved >partial response (PR), while 16 remained stable or progress to treatment. The mean QPCR ratio in responders was 1.69 ± 0.13, while in non-responder was 0.43 ± 0.1 (p<0.001). Based on this finding, we assessed the influence of the loss of NF-kB 3’end on the clinical response. Univariable analysis demonstrated that low CD38(+) NF-kB2 3’end mRNA levels (odds ratio [OR] of 10.8; 95% CI 1.99 to 58.16, p=0.0057) and history of MGUS (OR: 7.6; 95% CI of 1.2 to 47.6, p<0.05) were significantly associated with poor outcome. However, multivariate analysis indicated that low CD38(+) NF-kB2 3’end levels was an independent risk factor for poor response to bortezomib and dexamethasone (VD, HR:21.47, 95% CI: 2.99 to 154.399, p <0.01, supplemental Table 2). To further evaluate for the presence of a structural rearrangement, we designed a NF-kB2 break-apart FISH assay. We first performed FISH on whole bone marrow for 31 patients treated with VD. As expected the percentage of broken signals in unsorted samples was low, but when we considered the proportion of plasma cells in each sample, the total percentage of broken signals was statistically significantly higher in the nonresponder (nonresonder: 9.12± 4.2 vs Responder:2.37 ± 0.7, p<0.035). Using a cut off detected by AUC curve, we identified NF-kB2 structural rearrangement in 9 patients (30%) patients of which 6/10 were present in VD-non responders and 3/21 of VD-responders. To validate these findings we performed FISH on CD38+ sorted cell from 24 new patients treated with VD or carfilzomib, an irreversible inhibitor of the proteasome, and found consistency with our previous results, in which 7 patients demonstrated a NFKB2 positive break signal. Six out of 7 samples of non-responders patients demonstrated a positive break signal while 1/16 were positive in the responders, indicating a chromosome translocation or inversion had disrupted the NFKB2 locus and that this disruption predispose tumors to respond poorly to proteasome inhibitors. We tested multiple cell lines carrying NF-kB2 rearrangement and identified by target sequencing that HUT88 presents a balance translocation between chromosomes 10 and 3. Taking advantage of this finding we compared the sensitivity of detection between 3’end NFKB2 QPCR and the break-apart FISH. To this end, we mixed cell line with normal NF-kB2 as it is in MM1S with titrating doses of HUT88. Quantitative PCR was able to detect a minimum of 3% HUT88 cells in the mix, while NFKB2 break-apart FISH was able to detect when the percent of HUT88 reach 10%. The correlation between both test was r²:0.98, P<0.02.

Conclusion: loss of NF-kB2 3’ end is a potential companion diagnostic test for bortezomib. QPCR is more sensitive to detect NFKB2 3’end abnormalities.