Genomic landscapes of ovarian clear cell carcinoma from latin countries reveal aberrations linked to survival and progression

In the present study, the median age of diagnosis was 50 years, consistent with the earlier diagnoses of OCCCs compared to HGSC in the United States (median 55 vs. 64 years) [3]. Moreover, OCCC was detected earlier in women with endometriosis (median = 38 years, range 31–59), which is probably connected with the fact that most of those patients were diagnosed at the initial stage [40, 41]. To date, the most important prognostic factor for overall survival is the sensitivity to platinum-based chemotherapy, regardless of the stage of the tumor [2]. However, for predicting progression-free survival (PFS), an advanced stage of cancer has been identified as the main prognostic factor, with the occurrence of thromboembolic events (TEEs) also contributing to the prognosis. In our cohort, we found that endometriosis did not significantly impact either OS or PFS. This is in contrast to a previous report, which suggested that endometriosis could be a prognostic factor [42].

PIK3CA mutations are common in OCCC; accordingly, they were detected in 30% of our samples with a relatively lower frequency than reported in other populations [37, 43‐50]. Our cohort also showed the concomitant loss of ARID1A expression associated with PI3KCA mutation, as previously observed [16]; those alterations are thought to occur at an early stage in the development of OCCC due to its presence in endometriosis precursor lesions [6]. In fact, we observed that endometriosis was more frequent in the presence of PIK3CA mutations and that both were associated with diagnosis at the initial stage. The cohort displayed an incidence of abnormal TP53 which is consistent with previously reported incidences in other populations [51, 52].

The genome-wide analysis of alterations by OncoScan was never used to detect CN alterations for this particular ovarian tumor type; the first studies of OCCCs were performed with aCGH, and currently, whole-exome sequencing is being performed. Despite the different reported techniques, the most common CN alterations were easily detected in our samples, including chromosome 8q and 20q13.2 amplification. The latter contains the ZNF217 oncogene that was not associated with shorter PFS or OS (log-rank P-value = 0.3125 and 0.5571, respectively), as suggested before [53]. The OCCC samples were clearly separated into clusters depending on their genomic architecture (FSl, STl and Sxl) rather than on individual CN alterations. The finding that most of the polyploid samples were inside the FSl cluster and none were classified as Sxl suggests that the genomic instability associated with the FSl pattern can be related to the loss of the diploid state [54]. Interestingly, ploidy alterations are associated with outcome in ovarian cancer, including OCCC [55].

Amplification of MYC was identified in 57.7% of the samples, which is consistent with previous reports (40–64%) [20, 44, 46] and was associated with better survival. However, a concomitant loss of the chromosome 13 region containing BRCA2, in the MB group, was associated with outstanding good OS prognosis, suggesting that a synergistic interaction might exist. In fact, loss of BRCA2 was a recurrent alteration in the samples within our cohort with high levels of HRD. A recent article by Wang et al. proposed a classification of ovarian tumors based on genomic signatures of aberrant DNA repair mechanisms rather than on histology [49]. In Wang et al., the subgroup H-HRD (high-HRD), characterized by enrichment of HRD signatures, showed better survival even in ovarian tumors without BRCA1/BRCA2 mutations. Importantly, the H-HRD subgroup showed amplification of MYC and MECOM genes (3q26.2) [49], the latter being within a region we identified as recurrently amplified in the HRD-sum high group of OCCC patients. In this manner, the MB OS group of OCCCs we defined is reminiscent of the H-HRD; however, in the combined study of Wang et al., most H-HRD samples were HSGC, and no OCCC was present [49]. Our hypothesis for this observation is that when the different histotypes are studied together, the greater levels of alterations found in HGSC mask similar alterations at lower levels in the other histotypes. This is clear from the differences observed in genomic scar levels between the FSl and STl clusters, where the latter showed exaggerated levels of all HRD signatures, in particular of HRD-LOH, relative to the FSl cluster. A recent study by Pesenti et al. classified a cohort of Italian-origin stage I epithelial ovarian cancer, including OCCCs, based on their genomic instability patterns, regardless of histological subtype [56]. The three genomic instability patterns defined by Pesenti et al. - “stable”, “unstable”, and “highly unstable” - are analogous to the Sx, FS, and ST genomic patterns, respectively, previously defined by Tan et al. [24]. OCCCs with lower levels of genomic alterations in our cohort were more prevalent than in the study by Pesenti et al. However, differences in sample classification between the two studies may have contributed to this discrepancy.

Synthetic lethality induced by PARP inhibitors in tumor cells with HRD is supposed to be a groundbreaking therapeutic strategy [57] in particular, for HGSC where ~ 50% of the cases have HRD [36]. Recently, mutations in 16 HR-associated genes were tested in a Japanese cohort, and 28% of the OCCCs showed alterations, which suggests that more patients could be selected for treatment with PARP inhibitors [58]. We observed that loss of BRCA2 and gain of PIK3CB genes were present in OCCCs with higher HRD-sum levels, which includes most of the cases in the FSl and STI clusters. In this manner, it would be interesting to verify what level of HRD-sum is necessary for the use of PARP inhibitors in OCCCs and to verify whether CN loss of BRCA2 is associated with that. Furthermore, the observed CN gain of the PIK3CB gene in HRD-sum high OCCCs might be linked to the promising results observed with the combination of inhibitors for the PI3K pathway and PARP [59, 60]. It has been reported that ovarian cancer patients with mutations in HR-associated genes have higher platinum sensitivity and prolonged overall survival [61]. In our cohorts, HRD-sum and CN loss of BRCA2 were not necessarily associated with longer OS or platinum sensitivity. In fact, all the patients with concomitant gain in MYC and loss of BRCA2 were sensitive to platinum-based treatment, and all of them showed a better prognosis.

The majority of the cases of the poorest OS group (PS group) showed amplification of the ASH1L gene on chromosome 1q22. This gene codes for a histone lysine methyltransferase that can mono- or di-methylate histone H3 lysine 36 (H3K36) [62] and is part of the Trithorax group of chromatin proteins that act as epigenetic regulators [63]. Recently, ASH1L function has been linked to leukemogenesis in mixed-lineage leukemia [57] and acute myelogenous leukemia [64]. Furthermore, ASH1L gene was identified as a driver gene liver cancer [65]. Also, congruent ASH1L gene amplification and mRNA up-regulation was reported in hepatocellular carcinoma [66]. Additionally, ASH1L is overexpressed in anaplastic thyroid cancer (ATC), contributing to its aggressiveness [67]. As in other tumor types, ASH1L might be associated with OCCC biology, and since epigenetic regulators are considered important targets for cancer treatment, our observation opens a new research opportunity to define the role of ASH1L in OCCC [68].

An important challenge in OCCC is to predict recurrence in patients diagnosed at initial stages. We identified that gains in MAPK8 and MKL1 genes were associated with fast progression. The MAPK8 gene encodes the stress-activated kinase JNK1, and its activated form is associated with shorter PFS in epithelial ovarian cancer [69]. The importance of our finding is linked to the use of JNK1 inhibitors in clinical trials for other cancers [70] and to the eventual use of these inhibitors to control progression in OCCC. Interestingly, MAPK8 gene gain was observed after progression of the Occ53 OCCC. At recurrence of Occ53, few alterations were conserved compared to the primary tumor, including amplification of WNT7B and MAPK1 oncogenes (Supplementary Fig.S8).