Page 143 - Haematologica March 2020

Risk factors for post-HCT outcomes in CMML
with previous reports on mutation profiles in patients with CMML.16,17 Among the mutations in ASXL1, 67% were nonsense mutations, while the remainder were either frameshift or stop/gain mutations. Mutations were also frequent in WT1 (27%), RUNX1 (17%), DNMT3A (17%), SMC1A (17%), EZH2 (12%), and ATRX (12%), the high frequency presumably being related to the fact that more than 85% of our cohort (104 of 129 patients) had intermediate-2 or high-risk disease according to the CPSS (Table 1).
Incorporation of mutations into the overall analysis
Among mutations with prognostic weight reported in previous studies,16 such as ASXL1, NRAS, RUNX1, and SETBP1, only mutations in NRAS (n=6) were significantly associated with relapse (HR, 4.7; 95% CI: 1.4-16; P=0.03). In addition, mutations in ATRX (n=6) and in the WT1 gene (n=10) were significantly associated with relapse (HR, 17.3; 95% CI: 4.1-73; P=0.0005; and HR, 6.3; 95% CI: 1.6- 24; P=0.01, respectively) and inferior survival (HR, 4.9; 95% CI: 1.8-13; P=0.005; and HR, 4.3; 95% CI: 1.7-11; P=0.004, respectively).
The impact of mutations was not affected by adjust- ment in the multivariate analysis, and results remained unchanged.
We then grouped mutations and clinical parameters that co-occurred by unbiased clustering of a correlation matrix across individual mutations and clinical features, including prognostic scoring systems (MDAPS, CPSS, and the CMML-specific cytogenetic classification), leukocyte counts (MP-CMML), and blast counts (Figure 2 and Online Supplementary Figure S3). This approach identified two groups: group 1 mainly included mutations in signaling pathways (e.g. RAS or JAK2) and TP53, associated with high-risk cytogenetics and blasts (Figure 2A and Online Supplementary Figure S3A), while group 2 included muta- tions in epigenetic regulatory genes and splicing factors (Figure 2A). Within this group, mutations in DNMT3A, SETBP1, and EZH2 tended to co-occur, and mutations in TET2 and WT1 were associated with mutations in ATRX. The total number of mutations [≥10 mutations (n=15): HR, 3.4; 95% CI: 1.2-9.6; P=0.02] and greater number of mutations in genes regulating epigenetic processes [≥4 mutations (n=10): HR, 5.4; 95% CI: 1.9-16; P=0.003], but not mutations in signaling pathways and tumor suppres- sor genes (TP53 and PPM1D), were associated with relapse (Figure 2B, C, Table 3, and Online Supplementary Table S1). When cytogenetics, total number of mutations, established prognostic scoring systems and mutations in epigenetic processes and signaling pathways were consid- ered for their impact on relapse (Figure 2D and Online Supplementary Figure S3B), mutations in epigenetic processes [≥4 mutations; odds ratio (OR)=8.8], high blast count (≥20%; OR=4.7), high-risk MDAPS (OR=4.0) and high-risk cytogenetics (OR=3.2) were the major factors contributing to the probability of relapse (details in Online Supplementary Table S3). Pairwise association analysis showed that high blast counts, high-risk by CPSS, and high-risk cytogenetics were closely associated, and muta- tions in epigenetic regulators tended to occur predomi- nantly in disease classified as high-risk by the MDAPS (Figure 2D and Online Supplementary Table S3). The m- CPSS risk model did not allow for further differentiation of risk groups as virtually all patients became classified as intermediate-2 or high risk. All patients were upgraded to
higher risk groups due to their prevalent high-risk muta- tions. (Online Supplementary Figure S3B). As illustrated in Figure 2E, the delay between diagnosis and HCT was longer (median, 536 days) among patients with higher numbers of mutations than among those with fewer mutations (median delay, 309 days), including epigenetic regulators (522 vs. 344 days), but not mutations in signal- ing pathways (475 vs. 366 days). These observations were consistent with a previous report on non-transplanted patients showing increasing mutations with longer disease duration.15 Overall, the data indicate that molecular anno- tation uncovered distinct subgroups of CMML that were not distinguished by conventional risk classification. Specifically, a very high-risk group (independent of high- risk cytogenetics and high blast counts) with a long delay to HCT was characterized by a higher number of muta- tions in epigenetic regulators.
Discussion
We analyzed the long-term outcomes of allogeneic HCT in 129 patients with CMML in relation to clinical, patho- logical, and molecular characteristics. The results confirm the curative potential of HCT, since transplanted patients had 3- and 10 -year relapse-free survival rates of 37%, and 29%, respectively. The outcomes were superior in lower- risk patients, who had survival probabilities of 40% to 50% at 10 years. However, the relapse incidence and NRM rates were high. As in our previous studies21 and in reports from other centers,25,26,35 high-risk cytogenetics, dis- ease transformation (to CMML-T), and risk classification in scoring systems such as IPSS-R, CPSS and MDAPS, were strongly correlated with post-transplant relapse. As in earlier analyses,36 there was a suggestion that intensive cytotoxic therapy prior to HCT was associated with infe- rior outcome, presumably related to the fact that patients considered to be at high risk were more likely to receive induction-type chemotherapy. Nevertheless, it was of note that the intensity of conditioning did not significantly affect overall transplant outcome, although it must be noted that a broad spectrum of conditioning regimens was given to this cohort. Furthermore, as in a previous analy- sis21 and in other trials in patients with myeloid malignan- cies,37 patients with high HCT-CI scores had inferior sur- vival, although the probability of relapse was not altered. The source of stem cells did not affect outcome signifi- cantly, and results with related and unrelated donors were similar.
The present results indicate that providing HCT at an earlier disease stage and in patients with a low comorbid- ity burden who have not received cytotoxic therapy results in superior transplant outcomes, as also suggested by others.5 These three parameters are likely interrelated as prolonged pre-HCT observation and therapy tend to be associated with the acquisition of new co-morbidities, and likely new mutations, a concept supported by results in our subcohort of patients with mutational data. However, disease characteristics such as high-risk cytoge- netics and evidence of measurable residual disease by cytogenetics, still had a profound impact on relapse and survival, even with high-intensity conditioning regimens.
We hypothesize that mutational profiling assists in defining prognosis more narrowly. The analysis of muta- tions in 75 genes uncovered distinct risk groups defined by
haematologica | 2020; 105(3)
657