Molecular Predictors Of Response To Lenalidomide In Myeloid Malignancies

Lenalidomide (LEN) is effective in alleviating anemia in patients with MDS. Patients with del5q show particularly good responsiveness to LEN, but this drug has a significant activity in MDS patients without del5q and those with MDS/MPN and MPN. Currently, apart from del5q no routine molecular biomarkers have been identified to better predict response to LEN. Here, based on ability to perform a broad, unbiased mutational screen using generation sequencing (NGS) we hypothesized that somatic mutational patterns may help identify patients with myeloid malignancies who would benefit from LEN.

To that end, we identified a cohort of 92 patients with myeloid malignancies who were treated with LEN. Response criteria were by the 2006 International Working Group and only those who received at least 3 months of LEN were selected for outcome analysis. Sequencing was performed using targeted NGS for the 60 most common mutated genes (frequency >2%; 87% CI) identified by whole exome sequencing cohort (see another abstract from our group). Our approach focused on previously described and/or confirmed mutations, acknowledging that there may be some false negatives and our spectrum may not be entirely complete. For most of the cases confirmatory targeted Sanger sequencing for the top 19 mutations has been performed, in addition to NGS. For some analyses we also divided mutations into functional gene families; e.g., DNMT family (DNMT1/3A/3B), PRC2 family (EZH2/EED/SUZ12/JARID2/RBBP4/PHF1), IDH family (IDH1/2), RAS family (NRAS/KRAS/HRAS/NF1/RIT1/PTPN11), cohesin, various spliceosomal gene families among others.

Median age was 68 years (range, 20-84); according to IPSS there were 66 were lower (Low and Int-1) and 24 higher risk (Int-2; RAEB1/2 and some sAML) patients. According to cytogenetic risk groups there were 60% low, 18% intermediate, 22% poor risk patients; 31% had del5q. Overall response rate was 61%; rate of CR (including marrow/cytogenetic CR) was 19.5%, PR 5%, any HI 37%, SD 4%; 34% of patient were refractory. The cohort was then dichotomized into “responders” and “refractory,” with responders classified as those achieving CR, PR or any HI for the purpose of this analysis. Response rate was higher among del5q compared to non del5q (69% vs. 57%). Within the whole cohort the most frequently mutated genes were SF3B1 (24%), ASXL1 (18%), TET2 (17%), DNMT3A (14%), RUNX1 (12%), and U2AF1 (10%). At least one of the spliceosomal gene mutations was found in 37 (40%) cases and these seemed to be mutually exclusive.

We compared the mutations frequencies in responders vs. non responders and conversely response within mutant or wild type (WT) cases and prioritized them according to p value for subsequent analyses. For instance, ASXL1 (29% vs. 13%; p= .052), U2AF1 (20% vs. 4%; p= .025), DNMT3A (22% vs. 10%; p= .1), LUC7L2 (12.5% vs. 0%; p= .02), SETBP1 (12% vs. 2%; p= .071) and RAS family (12% vs. 4%; p= .2) mutations seem overrepresented among refractory cases. Albeit rare, mutations in ASXL1, U2AF1, and LUC7L2 were always associated with refractoriness. Among responders, mutations in RUNX1 (17% vs. 6%; p= .2), EZH2 (7% vs. 3%; p= .6), TP53 (6% vs. 3%; p= 1) were more prevalent. SETBP1 mutant cases were also more likely to be refractory. Interestingly, all 3 DDX41 mutant cases were responders. The results were augmented when mutations were grouped according to pathways showing e.g., that mutations in ASXL1/ASXL2 and SRSF2/ZRSR2/LUC7L2 were significantly associated with refractoriness (32%; p= .025 and 26%; p= .009, respectively). We next adopted the algorithm when multiple genes were combined in “either-or” fashion. Such an analysis revealed that patients with mutations in either/or ASXL1/U2AF1 (39% vs. 14%) were significantly associated with refractoriness (p= .007; OR 1.4). Similarly, patients harboring either/or RUNX1/EZH2 (22% vs. 6%) were significantly associated with response (p= .041, OR 3.8). Multivariate analyses confirmed U2AF1 mutation to be associated with refractoriness to LEN (OR 6.64; CI: 1.18-37.34; p= .032). Of importance is that presence of TP53 (4/5 mutant cases were responders) and RAS gene family mutations (4/6 refractory cases), usually associated to refractoriness to conventional chemotherapy, did not preclude response to LEN.

Our results demonstrate that analysis of somatic mutational patterns may be helpful in predicting patients who most likely benefit or those who will be refractory to LEN therapy.