October 3, 2023 | Over 900 genes have been implicated in the resistance of tumors to platinum-based chemotherapy that map to many different cellular pathways, thwarting efforts to identify a biomarker to predict refractoriness. Attacking these complex molecular phenotypes requires moving to next-generation molecular diagnostics capable of recognizing different networks of proteins that collectively describe, in advance, all cancer patients who will not respond to the standard first-line treatment, says Amanda Paulovich, M.D., Ph.D., professor and Aven Foundation Endowed Chair at Fred Hutchinson Cancer Center.
In fact, she and her colleagues have succeeded in using mass spectrometry to identify a 64-protein signature to predict the subset of patients with high-grade serous ovarian cancer (HGSOC) who are unlikely to respond to chemotherapy—and describe five new molecular subtypes of refractoriness. As covered in a study that published recently in Cell (DOI: 10.1016/j.cell.2023.07.004), a comprehensive proteomic analysis of 242 HGSOCs turned up the signature identifying 35% of patients with the refractory form of the disease with 98% specificity.
Subtypes of ovarian cancer have been identified previously using RNA-based data (e.g., the Cancer Genome Atlas), but those uncovered by the protein pathway analysis are entirely new. The computational work was led by Pei Wang, Ph.D., professor of genetics and genomic sciences at the Icahn School of Medicine at Mount Sinai.
The protein subtypes are “very stable” and were validated in independent patient cohorts as well as patient-derived xenograft (PDX) models of human cancer, Paulovich reports. Because Wang and her team selectively analyzed pathways that appeared to show differential expression between platinum-sensitive and -refractory tumors, they hypothesize that each of these subtypes represents a different mechanism of platinum refractoriness.
Another finding from the study is the association between lack of chromosome 17 loss of heterozygosity (LOH) and refractoriness, a connection not seen in past genomic studies that didn’t enrich for refractory tumors, says Wang. Ch17 harbors many tumor-suppressive genes, including p53 and BRCA1/2, which get perturbed by the allelic imbalance. Potentially, combining genomic information with the protein panel would produce even more powerful diagnostic results in the clinic, Wang says.
In the future, personalized medicine is going to be driven by “multi-analyte diagnostics coupled with clinical information and patient demographics in algorithmic-driven patient care,” states Paulovich. Patients with ovarian cancer would get two tests—one to analyze the refractoriness of their tumor to chemotherapy and another to identify the disease cluster into which it falls so the women can be triaged to an appropriate clinical trial treatment.
Currently, patients with refractory disease are commonly excluded from participating in clinical trials due to their rapid progression, she explains. This is because, “we don’t have good early-detection screening so many are diagnosed at a late stage, and they then go on and get cycles of futile chemotherapy to which they don’t respond. By the time we figure out their tumors are nonresponsive, they are too sick to be eligible for a study. As a result, the dismal prognosis for patients with refractory disease has not changed in decades.”
Paulovich quickly adds that she remains optimistic that “hope is coming” for these patients and, to that end, she is actively looking for research funding as well as collaborators with access to refractory tumor samples to clinically validate the protein signature. Only about 15% of HGSOC patients have refractory disease, so it will likely take more than 242 patient samples, as with the latest study, to have the statistical power to find the signature that will pick up 100% of them, she notes.
The problem is that academic grants typically don’t fund to the level required to reach statistical significance. And platinum chemotherapy drugs are off patent, so industry has no financial incentive to get involved, says Paulovich. Only patients, and their families and doctors, care. “The barriers are not scientific... it’s a societal issue” that may take government policy changes to solve.
For the latest study, Wang says, she and her team selected a panel of 1,082 proteins predictive of treatment responses by leveraging proteogenomic profiles from HGSOC cell lines and PDX models, three decades of published research on platinum resistance, and a machine-learning-based feature selection procedure. “We first analyzed the pathway activities in the tumor samples and then used these pathway activities to identify the five subtypes of HGSOC.”
The protein data clearly contributes complementary information to the RNA data, says Wang, revealing pathways relevant to refractoriness that have little overlap with known platinum resistant genes. It is also notable how well-preserved pathway activities were in archival tissues, some of them over 20 years old.
It proved difficult to get RNA data from the formalin-fixed paraffin-embedded (FFPE) tissue samples but the SOP for extracting proteomic data from FFPE tissue samples that the team developed a few years ago worked fabulously. “This is probably one of the first studies to use FFPE tissue to generate large-scale proteogenomic data,” Wang says. In the past, researchers have relied on frozen tissue that is not as readily available from biobanks.
Among the chemorefractory subtypes revealed by the analysis is one characterized by highly elevated metabolic activity and another by high immune infiltration, says Wang. Some tumors also clustered around the way TGF-β, alt-EJ, and βAlt proteins were expressed. As drugs are available that target those different pathways, this could prove to be clinically impactful for patients with refractory HGSOCs.
The work is a culmination of a triumvirate of papers that all fed into the analysis by Wang and her team, Paulovich notes. In a published report appearing at the end of 2021 in Cell Reports Medicine (DOI: 10.1016/j.xcrm.2021.100471), the researchers demonstrated that some chemorefractory ovarian cancers are associated with higher fatty acid oxidation pathway expression, which when pharmacologically inhibited can sensitize an initially refractory PDX model to chemotherapy.
The first paper, appearing a few months earlier in Oncogene (DOI: 10.1038/s41388-021-02055-2), was based on a review of 30 years of published research on platinum resistance. It offers a database of all the implicated genes annotated with possible pathways connecting them to chemoresistance and the types of evidence.
Paulovich is a clinically trained oncologist who, for the past two decades, has been running a translational CLIA-certified proteomic lab out of frustration with the inadequacy of available diagnostics, which, as she points out to anyone who will listen, account for a tiny proportion of overall healthcare costs but underlie most healthcare decisions. Diagnostic tests are “woefully under-appreciated and under-reimbursed” because the economic drivers in a capitalistic society favor investment in a potential new blockbuster drug with a one out of 10,000 chance of succeeding over a new diagnostic with one out of two odds.
Precision Assays, acquired by CellCarta last year, was spun out of the Paulovich lab at Fred Hutch to help disseminate the large spectrum of targeted mass spectrometry assays developed there. “It is important that we democratize access to these assays, making them broadly available to all labs, whether or not they have proteomic expertise,” she says.
Platinum-resistant HGSOC is a complex phenotype where influential proteins are crisscrossing many different cellular pathways, Paulovich says. Scientists were historically forced to look at single genes and proteins because that was all technology enabled, but it is a new day. An immunohistochemistry assay simply cannot make the highly multiplex measurements likely to be necessary for implementing the full potential of precision medicine.
Mass spectrometry was used to characterize the proteomic landscape and do the clustering analysis in the latest study. The signatures were identified via a global, untargeted approach with a view across the entire proteome, pared down to signals associated with refractoriness, says Paulovich. To operationalize that finding into a clinically tractable, multiplex assay that can be run in laboratories for testing and validation in additional cohorts, a targeted mode known as multiple action monitoring mass spectrometry is being used.
The targeted method starts with a list of identified proteins that then get rigorously quantified in patient samples. The measurement involves a comparison with isotopically labeled versions of the same proteins, she explains.
Two clinical-grade assays are being developed right now, reports Paulovich. One will indicate the likelihood of chemorefractory disease and the other will identify cases in the cluster having higher expression of metabolic pathways.