Molecular Predictors of Response in Patients with Myeloid Neoplasms Treated with Lenalidomide

Up to 70% of patients with del(5q) MDS may respond to Lenalidomide (LEN). However, the success rates in non-del(5q) cases, while substantial, are much lower (ranging from 20-40% depending on selection criteria). Aside from the presence of del(5q), up front identification of potentially responsive patients is difficult, particularly as the mechanistic underpinnings of LEN response are still under investigation. Initial attempts to prospectively predict LEN sensitivity resulted in a description of response expression signatures, but they have not been robust enough to serve as an actionable diagnostic test. In the outset of this study, we stipulated that apart of clinical selection (low risk MDS, transfusion-dependence, normal/low risk cytogenetics, etc.), analyses of molecular lesions including somatic mutations and chromosomal defects may help to predict LEN responsiveness. To that end, we performed deep targeted NGS (using multiamplicon panel of the top most commonly mutated genes in MDS). In total we analyzed 143 cases of myeloid neoplasms (MDS, MDS/MPN, or MPN) treated with LEN (median duration 6 months) for whom annotated clinical outcomes were available (83 responders vs. 60 refractory cases). Clinical parameters including IPSS-R, cytogenetics (FISH, SNP-array or metaphase cytogenetics) were used to characterize patients, whose responses were assessed by 2006 IWG criteria.

Initially, in a combined analysis, we included both del(5q) (N =37) and non-del(5q) patients (N =106). Very low/low, intermediate, high/very high IPSS-R scores were found in 47%, 23%, 34% cases, respectively. Of 143 LEN-treated patients, regimens included LEN (80%), or LEN+5-Aza (20%). Any hematologic improvement (HI), partial response (PI), and complete response (CR) were achieved in 44%, 14% and 42%, respectively. Responses were associated with a better survival (median survival time 6.2 yrs. vs. 3.7 yrs. in refractory cases; P =.003). Non-responders showed significantly lower platelet levels compared to responders (median 169 vs. 89 K/uL; P =.007) but intricate analysis of clinical parameters (age, other blood counts, blasts and IPSS-R score) failed to identify other factors that would help to select potential responders. As expected, when sub-analysis of patients with del(5q) was performed, combined response was achieved in 78% (OR 13.14 [4.34-39.75]; PR 16%, CR 35%) of patients, respectively, while in non-del(5q) the responses were as predicted lower at 51% (P =.004). Of note is that both del(5q) involving and excluding commonly retained regions (CRR; q11.1-q14.2 and/or q34-qter) also was associated with sensitivity (CRR affected; OR=9.9 [1.4-102], vs. CRR not affected; OR=6.3 [1.3-37.6]). When we analyzed impact of karyotype on LEN sensitivity, -7/del(7q), -20/del(20q), complex karyotype and normal cytogenetics did not correlate with response, but in addition to del(5q); the presence of +8 (7/10 responded; OR 12.25 [1.33-113.06]) was significantly associated with responsiveness. Using targeted deep NGS, we confirmed 168 somatic mutations in responders vs. 142 mutations in non-responders (OR .85; .67-1.07). The number of mutational events per patient did not correlate with responses (P =.38).

Among genes sequenced mutations in DDX41 (100% vs. 0%; OR infinity [6.7-infinity]) and RUNX1 mutation+deletion (75% vs. 25%; OR=8.1 [1.1-84.6]) were overrepresented in responders vs. refractory cases while in U2AF1 mutationswere more common among non-responders (20% vs. 80%; OR=.075 [.004-.76]). When reverse analysis was performed DDX41 mutations correlated with LEN response (10% mutant cases among responders vs. 0% in refractory; P =.009), while mutations in U2AF1 correlated with LEN failure (2.4% vs. 13.3% of mutant cases in responders and refractory, respectively; P =.02). The presence of all combined or any of the other spliceosomal mutations (SRSF2, SF3B1, ZRSR2, LUC7L2, and PRPF8) did not influence the results of the therapy. All TP53 mutations were found with complex karyotype with del(5q) and 5/7 (71%) TP53 mutant cases were treatment failures (OR .19 [.01-2.41]).

In conclusion, in addition to the presence of del(5q), low platelet count and the presence of various molecular lesions (+8 and RUNX1, DDX41 mutations or wild type status of U2AF1) may help to predict responses to LEN.