Skip to main content
Premium Trial:

Request an Annual Quote

Gene Signature Beats TILs as Biomarker of Response to Anti-HER2 Therapy


NEW YORK – An immune-related gene expression signature may have advantages over tumor-infiltrating lymphocytes (TILs) as a predictive or prognostic biomarker in early-stage HER2-positive breast cancer patients receiving neoadjuvant therapy.

Detection of TILs is a standard method for assessing immune system activation within the tumor in the research setting and is under investigation as a potential clinical biomarker of prognosis following targeted therapies such as Genentech's Herceptin (trastuzumab). A higher number of TILs in the breast tumor is generally an indicator that the patient will likely do well with anti-HER2 therapies.

Immune activation in the tumor can also be measured using immune-related gene expression signatures. In a study published this month in JAMA Oncology, researchers compared TILs to immune-related gene expression signatures in a set of samples from two breast cancer clinical trials.

"In breast cancer and many other human cancers, the involvement of the immune system is critically important for predicting prognosis," said Charles Perou, a genetics professor at the University of North Carolina at Chapel Hill and a study author. "Immune system features predict survival and response to HER2-positive patients receiving the monoclonal antibody trastuzumab, suggesting that engaging the immune system is important for that particular biologically directed drug."

In embarking on the study, Perou's group used a common dataset to compare the ability of the two measurements to predict patients' survival. "We did this head-to-head comparison, and the gene expression signatures proved to be the best prognostic and the best predictive features that we saw," he said.

Perou and his collaborators carried out gene expression profiling by RNA sequencing and measuring TILs on pre-treatment tumor samples from two breast cancer trials — the CALGB 40601 trial of paclitaxel and Herceptin with or without Novartis' Tykerb (lapatinib) and the PAMELA trial of the PAM50 HER2-E subtype as a predictor of response to neoadjuvant dual anti-HER2 blockade.

"What was unique about this paper was on two different datasets we had both the gene signatures and the TILs, and we had good power to compare the predictive value of two ways of measuring the immune system," Perou said. He explained that the results showed that both types of biomarkers were predictive and prognostic of pathologic complete response on anti-HER2 treatment.

However, when taking into account other variables including standard pathology features, tumor size, nodal status, and immune features, the gene signature performed better than TILs. This finding, according to Perou, tracks with other research showing that immune-related gene expression signatures can predict response not only to anti-HER2 therapies but also immune checkpoint inhibitors.

Looking further into the gene expression characteristics, they found tumors high in TILs tended to be infiltrated with T cells, whereas those testing high for the immune-related gene expression signatures showed more B cells. That was a surprise, according to Perou, as TILs could have reasonably been expected to capture the presence of B cells.

"The B-cell signatures were actually some of the most prognostic and predictive features," said Perou. "That might explain why TILs didn't perform as well." He added that this finding highlights the likely biologic importance of B cells in the response to Herceptin.

As a next step, Perou and his colleagues are now conducting retrospective studies to try to establish the clinical utility of using immune-related gene signatures as biomarkers of response and survival. If successful, Perou said these signatures could be used to make treatment decisions for patients outside of the research setting.

As a clinical tool, when choosing between TILs and gene expression signatures, TILs have the advantage of being easier and more cost-effective, Perou noted. "However, if it's not a very good biomarker, then I don't care if it's easy and cheap," he said, noting that precision medicine by definition requires precise, quantitative, objective biomarkers. Since gene expression signatures are already used clinically in breast cancer, Perou said the new immune-related signatures could be added to the menu once clinical utility is established.

For the physician and the patient, Perou said gene expression testing could be carried out on the tumor biopsy sample and the results could inform treatment decisions, such as choosing the drug or the combination of drugs a patient is most likely to respond to. These immune-related gene signatures are also not unique to breast cancer. Such signatures, many of them prognostic, Perou noted, have been identified in other tumor types.

For the recently published study, Perou and his collaborators used RNA sequencing to gauge gene expression signatures. That's a technology that is not yet as widely used clinically as DNA sequencing. However, according to Perou, the gene signatures are platform agnostic and could be adapted to any standard clinical testing platform such as the NanoString nCounter, quantitative PCR, or microarray.

Perou saw these same signatures in another study published in Nature Cancer last year. In that study, his group set out to identify molecular features associated with metastasis in breast cancer patients using RNA sequencing, exome and whole-genome sequencing, and DNA methylation microarrays. In that research, investigators were able to observe changes in immune-related gene expression signatures associated with metastases.

Studying those signatures also allowed them to observe granular differences between patients and between primary cancers and metastases in a single patient. "There were a number of patients where we had the primary tumor and [about] five different metastases," said Perou. That revealed that the sites of the metastases were significant. "With regard to these immune signatures, certain sites like the liver and the brain, on average, tended to be immune cold, and other sites like the lung tended to have immune involvement," he said.

This research is a precursor to the present work in HER2-targeted therapy and provides insights on immune gene signature dynamics in cancer and gauging treatment response. Perou explained that the site the tumor sample is drawn from is important as it could influence the decision to give immune checkpoint inhibitors. For example, if the sample is taken from the liver metastasis, it could give a false-negative result, suggesting the patient would have a limited immune response to treatment, whereas a sample from a lung metastasis is more likely to show a positive immune-related gene signature, giving the patient a better shot at immunotherapy.

Overall, by studying immune-related gene expression signatures, Perou and colleagues concluded that about 30 percent of the time, a tumor is changing its subtype from the primary to the metastatic site. "The suggestion there is that you should be making treatment decisions based on the metastasis, not the primary [tumor]," Perou said.