Page 192 - 2021_06-Haematologica-web

M. Martin-Izquierdo et al.
2, in green-blue), and these were randomly distributed throughout all the genes without showing a pattern.
The most interesting dynamic patterns were those of newly acquired mutations (Figure 2A, type 3, represented in red) or increased in clonal size (Figure 2A, type 1, in orange) at the time of sAML progression. These muta- tional patterns were mainly found in the cohesin complex and Ras signaling, where they were clustered in the STAG2 and NRAS, KRAS and FLT3 genes. These profiles were also detected in transcription factors and epigenetic modifiers, but in these cases they were randomly distrib- uted among the genes. In fact, the STAG2 VAF median was significantly higher at sAML stage (diagnosis vs. sAML median VAF, P=0.023) and this gene was mutated in eight patients of the discovery cohort, in five of which the VAF had increased by the time they had become sAML (Online Supplementary Figure S3). The increase was statistically significant in three of these five patients, while it was not significant in the other two, although a trend was observed (P<0.08), probably because the VAF at diagnosis was already very high. Moreover, most of the mutations (nine of 12) in the cohesin complex genes were of the frameshift or stop gained (loss of function) type and the cohesin-mutated patients showed a higher num- ber of mutations than wild-type patients (median number of mutations: seven vs. four, P=0.0179). On the other hand, NRAS and FLT3 mutations were newly acquired (diagnosis vs. sAML median VAF, P=0.0029 and P=0.0078, respectively) during the evolution and so were detected at the sAML stage (Online Supplementary Figure S3).
Co-occurrence of cohesin complex and Ras signaling mutations in patients after progressing to secondary acute myeloid leukemia
Within this heterogeneous landscape of mutational dynamics, we focused our study on increasing (type 1) and newly acquired (type 3) mutations because their dynamic patterns suggested that they were positively selected during disease evolution. Moreover, in order to better characterize the mechanisms driving sAML pro- gression, we studied which pathways and combination of them were affected by these types of mutations.
In the discovery cohort, a high proportion of Ras signal- ing-mutated patients at the sAML stage, already harbored cohesin complex mutations. In fact, 26% (11 of 42) of the discovery cohort patients carried mutations in the cohesin complex at diagnosis. On the other hand, 52% (22 of 42) of the patients had at least one Ras pathway mutation at the sAML stage, mainly acquired during the evolution of the disease. Of interest, nine of these cohesin-mutated patients (nine of 11, 82%) carried a co- occurring Ras signaling mutation at the sAML stage. Considering only the most recurrently mutated gene, STAG2 (n=eight of 11), seven patients (seven of eight, 88%) carried another mutation in the Ras pathway, this being a NRAS mutation in five patients. Therefore, there was a statistically significant co-occurrence of these two pathways (P=0.023) and of the most recurrently mutated genes of these pathways, STAG2 and NRAS (P=0.002) (Figure 2B).
In order to confirm these observations and their impact on MDS progression to sAML, the combination of the cohesin complex and Ras pathway mutations was sought in the validation cohort, an independent cohort of 388 patients in which the disease was studied on only one
occasion, at diagnosis. In fact, these co-occurring muta- tions were detected in eleven additional patients: nine of which finally transformed into sAML (nine of 63), while two patients did not evolve during the median follow-up of 19.6 months (two of 325) (Figure 2C). Although all sam- ples of this cohort were studied at diagnosis, these nine patients carried cohesin and Ras co-occurring mutations at an advanced stage of the disease, indeed these were detected in sAML sampling or in patients who trans- formed in a median time of 11 months from sampling.
The discovery cohort included only patients who evolved to sAML, and therefore displayed a very poor outcome. This made it difficult to measure the impact of this co-occurrence in these patients. For that reason and also to further study the clinical consequences of this co- occurrence on outcome, the effects on overall survival and progression-free survival in the validation cohort (median follow-up of 19.6 months) were analyzed. In our validation cohort, where 16.2% of patients evolved to sAML and 44.76% died (Online Supplementary Table S1), those patients harboring both the cohesin complex and Ras signaling mutations had significantly shorter overall survival (16 vs. 60 months, P=0.005) and significantly ear- lier progression to sAML (10 vs. 15 months, P=0.005) (Figure 2D). Moreover, in order to study the contribution of the cohesin and Ras mutations alone to these effects, comparison of median overall survival of the double- mutant and cohesin and Ras single mutant patients was performed and patients harboring double mutations showed shorter overall survival than patients with Ras or cohesin single mutations (16 vs. 25 vs. 37 months, respec- tively, P=0.018, Online Supplementary Figure S4).
Higher proportion of newly acquired or increasing mutations in chromatin modifiers in treated myelodysplastic syndrome patients
As previously mentioned, 48% of the patients in the discovery cohort of this study were treated with 5-azacy- tidine (AZA) (n=16) or lenalidomide (n=4), and pro- gressed to sAML after therapy, whereas the other 52% received no treatment (only supportive care). Thus, we investigated whether the mechanisms of progression could be slightly different between patients who were treated with disease-modifying agents (AZA and lenalidomide) before transformation into sAML and non- treatedpatients.
In order to achieve this aim, the proportions of the dif- ferent mutational dynamics were compared between treated and untreated patients. Thereby, the mutational dynamics featured a significantly higher proportion of newly acquired or increasing mutations in chromatin modifiers at the time of sAML in treated patients (eight of 15 mutations), while in untreated patients the majority of mutations were stable (53% [eight of 15] vs. 19% [four of 21], P=0.031). By contrast, and with respect to the treat- ment, no differences were detected in the dynamics of the cohesin complex (50% [three of six] of newly acquired or increasing mutations in treated patients vs. 50% [three of six] in untreated, P=1.00) or Ras pathway mutations (91% [ten of 11] in treated vs. 76% [13 of 17] in untreated patients, P=0.3299) (Online Supplementary Figure S5). Thus, our study suggests that mutations in chromatin-modifier genes could be related to the evolu- tion of patients who receive disease-modifying treatment before progression to sAML.
2220
haematologica | 2021; 106(8)