Comprehensive Analysis of Genotype and Prior Exposures in Therapy-Related Myeloid Neoplasms (t-MNs)

Background: Therapy-related myeloid neoplasms (t-MNs) occur after exposures to chemotherapies and/or radiation therapies (CRT). However, true etiological association between CRT exposures and t-MN development is not clearly understood. We and others have recently revealed that selection of preleukemic mutant clones (TP53, PPM1D and others) under the selective pressure of CRT might play an important role in t-MN development (Wong et al. Nature 2014, Takahashi et al. Lancet Oncology 2017, Hsu et al. Cell Stem Cell 2018). For instance, PPM1D-mutated clone exhibited preferential expansion over wild-type clone only when exposed to cisplatin or etoposide, likely due to reduced apoptosis of mutant cells upon exposure to DNA damaging agents. In support of these findings, PPM1D-mutated t-MN patients had significantly frequent prior exposures to platinum and etoposide (Hsu et al. Cell Stem Cell 2018). These data led us to systematically analyze correlations between t-MN genotypes and prior exposures.

Methods: We studied 471 patients (pts) with t-MNs (t-AML N = 188 and t-MDS N = 283) whose bone marrow or peripheral blood samples were analyzed by targeted next-generation sequencing. We performed extensive chart review to obtain history of prior exposures and analyzed correlation between CRT exposures and t-MN genotype.

Results: Median age at diagnosis was 68 years (IQR:17-91) and the median latency from CRT exposures was 6 years (IQR: 0.1-45). (Table 1). The most frequent prior malignancies were breast cancer (20.8%), non-hodgkin's lymphoma (NHL) (20.3%) and prostate cancer (11.6%). Pts with NHL and multiple myeloma (MM) had significantly increased likelihood of developing t-MDS than t-AML (t-MDS vs. t-AML, 73.4% vs. 28.1%, P = .008 for NHL, 94.2% vs. 5.7%, P = < 0.001 for MM), whereas breast cancer pts developed t-AML more frequently than t-MDS (t-MDS vs. t-AML 43.8% vs. 56.1%, P < 0.001). Two hundred and eight (44.1%) pts had exposures to chemotherapy alone, 185 pts (39.2%) to chemo-radiotherapy, and 78 pts (16.5%) to radiotherapy alone. Eighty (16.9%) pts underwent autologous hematologic stem-cell transplant (auto-SCT).

At least 1 driver mutation was detected in 392/471 (83.2%) t-MN pts and TP53 (36.3%), PPM1D (19.4%), TET2 (14.9%) and DNMT3A (14.6%%) were among the most frequently detected mutations. TP53 mutations were significantly more frequent in t-MDS than t-AML (43.8% vs. 25%; P=.000), while FLT3, NPM1, IDH1/2 mutations were more frequent in t-AML (14.9%, 7.4%, 9% respectively) than t-MDS (1.8%, 0%, 5.3%, 3.5% respectively)

Correlation analysis of genotype and exposures confirmed previously known associations such as PPM1D mutations and platinum (log odds ratio [OR] 1.1, P = 0.01, false discovery rate [FDR] = 0.10), 11q23 rearrangement and topoisomerase II inhibitors (log OR 0.40, FDR = 0.71), and chromosome 7 abnormalities and alkylating agents (log OR 0.48, FDR =0.08). We found significant associations between TP53 mutations and immunomodulatory imide drugs (IMiDs, log OR = 0.98, , FDR =0.03); negative associations of alkylating agents with TET2 (log OR -0.93, FDR= 0.01) and NPM1 (log OR -3.31, FDR=0.003); and SRSF2 mutations with auto-SCT (log OR -2.24, FDR=0.03) and topoisomerase II inhibitors (log OR -2.42, FDR= 0.02). (Figure 1 Univariate analysis)

We adjusted for the confounding effect from multiple exposures by multivariate logistic regression analysis (MLR) with modelbuilding by stepwise AIC or BIC selection criteria. MLR revealed a significant association of TP53 mutations with IMiDs (log OR = 1.07, P = 0.001), platinum (log OR = 0.61, P = 0.02), and vinca alkaloid (log OR = 0.62, P = 0.01). As well as a significant association between EZH2 mutations and vinca alkaloid (Log OR = 1.68, P = 0.009, Table 2).

Conclusions: Comprehensive analysis of genotype and prior exposures in a large cohort of t-MNs revealed previously unknown associations, such as EZH2 mutations and vinka alkaloids, and TP53 mutations and IMiDs. The connection between TP53 mutations and IMiDs is consistent with the clonal selection of TP53 mutant clone with lenalidomide in 5q- syndrome (Jädersten et al. JCO 2011). These data provide a rationale for studying the mechanism of clonal selection of TP53-mutant or EZH2-mutant cells under IMiDs or vinka alkaloids. Further, this knowledge might be clinically useful for cancer pts with specific clonal hematopoiesis in reducing the risk of t-MNs by avoiding exposures to certain drugs.

View largeDownload slide

View largeDownload slide

Close modal