Researchers led by a team from the University of Helsinki in Finland have identified three different evolutionary states in tumors from patients with ovarian high-grade serous carcinoma (HGSC) that are characterized by distinct signaling pathways and associated with treatment response.
HGSC is the most common epithelial ovarian cancer subtype and is “one of the most genomically heterogeneous cancers,” Sampsa Hautaniemi, professor of systems biology at the University of Helsinki, told Inside Precision Medicine. He said that the genetic heterogeneity makes it difficult to examine the tumors in detail. His team therefore developed “several analysis and visualization tools that were crucial in finding the evolutionary states, characteristics, and trajectories” of the tumors.
They carried out whole genome sequencing on 510 samples from the ovaries and fallopian tubes, intra-abdominal metastases, other tissues (lymph nodes, liver), and ascites before and after treatment, from 148 patients with HGSC who were taking part in the prospective, multiregion DECIDER study.
“We used tumor mutations to track how tumor had developed and how it grew in metastases. We identified subclones that shared the same mutations. Then,we compared different metastases and the site-of-origin to estimate if tumor grew in them as complex or simple populations and whether it continued developing within the metastases,” Hautaniemi explained.
The results, reported in Cancer Cell, revealed three evolutionary clusters based on clonal complexity (the number of subclones in the sample) and clonal divergence (the number of unique subclones in the sample).
Cluster 1 was defined as the “evolving state.” This state was estimated to be the youngest in evolutionary time and showed aberrations in MAPK and ERBB2 signaling. Cancers in the evolving state were most stable and exhibited only a few subclones. Patients with tumors in the evolving state have the best response to therapy, with the longest times to relapse or death.
Cluster 2 was called the “maintaining state,” characterized by polyclonal tumors with multiple subclones. Tumors in this intermediate state were enriched for subclonal aberrations in the PI3K/AKT pathway and were associated with the shortest times to relapse and the worst survival odds.
Cluster 3, the most evolved of the three, was referred to as the ‘‘adaptive state.’’ It was distinguished by a high number of clones and the highest clonal divergence as well as mutations in various signaling cascades, including those driving the NOTCH and WNT pathways.
Further analysis showed that the states were linked by two distinct evolutionary trajectories that support transition from the evolving to adaptive states either directly or via the maintaining state. Of note, patients whose tumors evolved directly to the adaptive state had better survival than those whose tumors evolved via the maintaining state, with the latter trajectory initially driven by changes in PI3K/AKT signaling, CD28 activation, and G2-M transition in the cell cycle pathways.
The current standard-of-care for the patients with HGSC is debulking surgery followed by platinum/paclitaxel chemotherapy. “Despite the well-established clinical trials and several PI3K inhibitors being approved by the FDA for other cancers, none of them have yet progressed to clinical use for ovarian cancers,” Hautaniemi said.
To investigate whether PI3K inhibitors could be a potential treatment in patients with HGSC, the investigators exposed five HGSC organoid lines to alpelisib (a PI3K-alpha inhibitor approved for treatment of advanced metastatic breast cancer), idelalisib (a PI3K-delta inhibitor approved for use in relapsed chronic lymphocytic leukemia), and umbralisib (a PI3K-delta inhibitor whose FDA approval was recently withdrawn).
They observed significant cytotoxicity in the cells treated alpelisib and umbralisib. Alpelisib showed a particularly strong effect on cell viability in all samples.
These results “support the relevance of PI3K/AKT pathway in a tumor evolutionary defined subset of patients with HGSC and warrant further investigation with larger numbers of tumor organoids and clinical intervention trials guided by cancer genomics,” the authors write.
Hautaniemi added that as well as discovering the three evolutionary states, his team “also found that tubo-ovarian samples, which are considered the site-of-origin tumor, have 70% more unique subclones than metastasis sites.” He described this as “a surprising finding, [which] suggests that studies predominantly having these samples may be hard to reproduce in another cohorts.”
The current study focused predominantly on treatment-naive samples. Hautaniemi and colleagues are now analyzing samples taken during or after chemotherapy with the goal of characterizing what happens to tumor genomes during chemotherapy and in relapses.
He also pointed out that stratification of the three evolutionary states currently requires multi-site whole-genome sequencing data, which is not feasible in the clinic, and said that the group are “exploring whether it is possible to identify these states from circulating tumor DNA sequencing data.”