Efforts continue to personalize radiation therapies with predictive biomarkers

Silvia C. Formenti, MD
Silvia C. Formenti, MD

As more and more cancer therapies are tailored to individual tumors, the search for predictive biomarkers in radiation oncology remains ongoing.

A comprehensive approach is needed to look at multiple components that can influence response to radiotherapy, said Silvia C. Formenti, MD, Weill Cornell Medicine. Targeted approaches could include looking at the tumor microenvironment, the host microbiome, the tumor stage, the time of radiation, and the genetic characteristics of the host.

Formenti chaired an Annual Meeting symposium on Monday, April 11, that included three presentations exploring current research in some of these areas. The session, Predictive Biomarkers for Precision Radiation Oncology, can be viewed on the virtual platform by registered meeting participants through July 13, 2022.

Tumor microenvironment
Anna Wilkins, PhD, Royal Marsden Hospital Institute of Cancer Research, London, United Kingdom, opened the session with a discussion about how different features of the tumor microenvironment, including non-cancerous cells, might aid in tumor survival after radiation treatment.

Anna Wilkins, PhD
Anna Wilkins, PhD

Wilkins described previous work where she evaluated a panel of proteins using immunohistochemistry markers in patients with prostate cancer receiving radiotherapy with curative intent. The analysis showed that both PTEN loss and proliferative markers predicted recurrence after radiotherapy. These were independent of each other and of standard clinical prognostic factors.

“There was clear potential for clinical impact here,” Wilkins said.

While conducting this research, Wilkins noticed a fascinating range in the biology outside of the tumor and her interest was piqued by the role of the stroma and cancer-associated fibroblasts. With a focus on rectal tumors, Wilkins and colleagues hypothesized that they might be able to identify factors in the tumor microenvironment that predict radiotherapy response.

Since then, ongoing research has shown that in patients with rectal cancer with good response to radiotherapy the tumors have low numbers of cancer-associated fibroblasts and specific myeloid populations compared to poor responders.

Additionally, tumors with poor response to radiotherapy show high inflammatory signaling at baseline but remain immunologically dormant after treatment, whereas tumors that show good response transition to an immunologically hot phenotype. Wilkins’ preclinical experiments showed that tumors that were similar to good-responding tumors in patients appeared to get stuck halfway and were not able to make the transition to an immunologically hot phenotype. This could be associated with the rapid expansion of specific fibroblast populations in response to radiation, Wilkins said, which appeared to impact T cell entry into these tumors.

“With improved understanding of this biology we can achieve two things,” Wilkins said. “We can improve treatment stratification and—if we can understand the biology and apply it right—cancer-associated fibroblast-targeting therapeutic strategies with radiotherapy hold real potential for therapeutic gain.”

Julie K. Schwarz, MD, PhD
Julie K. Schwarz, MD, PhD

Radiation sensitivity
Julie K. Schwarz, MD, PhD, of Washington University School of Medicine, discussed the use of imaging and genomics to identify markers of radiation sensitivity with a focus on the microbiome and cervical cancer.

The standard of care for cervical cancer is radiation combined with systemic cisplatin chemotherapy. HPV-positive cervical cancer is a radiation-sensitive disease and is treated for cure. Despite the existence of powerful and effective tools against this disease, there is a failure rate of about one-third, Schwarz said.

Over the last decade, Schwarz’s lab has used a translational approach to identify patients who will not benefit from treatment. Schwarz and colleagues have collected more than 300 matched pair/triplet specimens from patients before and during standard-of-care chemoradiation treatment.

Ongoing experiments support the idea that HPV phenotype matters when it comes to outcomes of cervical cancer, and Schwarz and colleagues found that it may also have an effect on radiation sensitivity. A series of experiments ultimately led to the publication of a study showing that higher expression of E6*I and E6 in cervical tumors with HPV genotypes other than HPV 16 may represent novel biomarkers of chemoradiotherapy resistance. Patients with high expression levels may benefit from alternative treatments.

“It is time to start thinking about new biologically targeted agents for this disease,” Schwarz said.

Sean P. Pitroda, MD
Sean P. Pitroda, MD

Sean P. Pitroda, MD, University of Chicago, discussed integrated molecular subtyping as a way of personalizing radiation therapy for oligometastatic disease.

He shared his vision for improving the treatment of metastatic disease by challenging the prevailing view that metastases are uniformly widespread and incurable, and advancing treatment paradigms that lead to optimal integration of localized therapy, such as radiation therapy, and systemic therapies, to potentially cure patients with metastatic disease who were previously deemed incurable.

The basis for his work is the idea that metastases occur on a spectrum and that integrated clinical and molecular staging can be used to predict the risk of metastatic dissemination and subsequently guide treatment.

Pitroda discussed collaborative work he has done looking at curative-intent treatment of primary disease and liver metastases in 134 patients with de novo colorectal cancer (CRC) without microsatellite instability and with liver oligometastases. Through integration of RNAseq, whole-exome sequencing, and microRNA profiling, Pitroda and colleagues investigated whether there was a biologic basis for favorable outcomes in this context.

First, they looked at the consensus molecular subtypes (CMS) of primary CRC, which have distinct biological underpinnings. They found that CMS4, the mesenchymal subtype, was associated with significantly worse overall survival compared with other subtypes. An analysis of the cancers with liver metastases showed a significant enrichment for CMS2, the canonical subtype, and a large fraction of unclassifiable tumors.

Then, through integration of mRNA and microRNA data, utilizing an approach called similarity network fusion clustering, Pitroda and colleagues were able to identify three optimally distinct molecular groups within their cohort—labeled canonical, immune, and stromal—with vastly different overall survival. An analysis of the molecular underpinnings of these subgroups showed hallmark signatures in each. The immunogenic subtype had the best prognosis.

By taking the biologic features and integrating them with clinical features from the clinical risk score, Pitroda and colleagues were able to identify risk categories. The low-risk group—about 25 percent of the cohort—had almost 95 percent survival at 10 years and consisted of patients with immune or canonical tumor type and low-risk clinical factors. The intermediate risk population had a survival of about 50 percent at 10 years and included the immune tumor type with adverse clinical factors, or the stromal tumor type with favorable clinical factors. The group with the poorest prognosis, which constituted half of the patients, had adverse clinical factors combined with a stromal or canonical tumor type. This group had 19 percent survival at 10 years.

Looking at patterns of recurrence, patients at low risk had a 50 percent recurrence frequency with a maximum of one to three hepatic recurrences. These patients underwent consecutive resections and some of them could be cured. At the other end of the spectrum, the high-risk group had recurrences 90 percent of the time, often outside of the liver, and recurrence was mostly widespread. Therefore, integrating clinical and biological factors allowed the researchers to predict the number and site of recurrences. These findings were then validated in another dataset.

“By understanding the clinical and genetic determinants underlying metastatic spread, we can potentially inform concepts of clinical-molecular staging of metastasis, therapeutic targeting of metastatic disease, and personalized medicine,” Pitroda said.