ImPrint Biomarker Predicts Immunotherapy Response in Patients With HR+/HER2– Breast Cancer

Denise M. Wolf, PhD

The ImPrint immune signature predicted the likelihood of pathologic complete response (pCR) following neoadjuvant immunotherapy treatment in patients with hormone receptor (HR)–positive, HER2-negative breast cancer, as well as those with triple-negative disease, according to Denise M. Wolf, PhD.

Data from the phase 2 I-SPY2 trial (NCT01042379) presented by Wolf during the 2023 ASCO Annual Meeting demonstrated that among 200 total patients with HR-positive, HER2-negative breast cancer across 5 immunotherapy arms, 29% of patients were ImPrint positive, and 71% were ImPrint negative. The pCR rate was 76% among patients who were ImPrint positive vs 16% among those who were ImPrint negative. Across all 5 immunotherapy arms, patients with triple-negative breast cancer (TNBC; n = 142) who were ImPrint positive (51%) and ImPrint negative (49%) achieved pCR rates of 75% and 37%, respectively.

“This clinical trial is a good example of translation that could potentially change the way patients are categorized and the way they’re treated, and [that could] lead to higher response rates, saving people’s lives,” Wolf said in an interview with OncLive^®.

In the interview, Wolf, a bioinformatics research scientist in the Department of Laboratory Medicine at the University of California, San Francisco (UCSF), UCSF Health, discussed the methods used to conduct the I-SPY2 study and the implications that immune biomarkers could have on selecting treatment for patients with HR-positive, HER2-negative breast cancer in the future.

OncLive: What was the rationale for initiating I-SPY2?

Wolf: I-SPY2 is a neoadjuvant platform trial for early-stage, aggressive breast cancer. There is a single control arm where patients get standard-of-care [SOC] therapy, as well as up to 4 different experimental arms. The [primary] end point is pCR, defined as no invasive cancer cells in the breast or the lymph nodes.

The goal of the trial is to find the most effective treatment for each patient subset as defined by biomarkers. The experimental agents ‘graduate’ for efficacy if they achieve a level of 85% predicted probability of success in a [subsequent] phase 3 trial with 300 patients. The trial has evolved, but the data I presented were based on the trial as it ran for approximately 10 years. The idea is to have a pipeline of different classes of agents and to get early, strong signals of improved levels of efficacy in patient groups, and then to come up with pairs. Within [a given] patient subset, as defined by biomarkers, [the goal is figuring out] the best class of agents, [agents that have] the highest probability of response and permanent cure. These are curable patients with early stage, though aggressive, disease.

The biomarker study investigated different immune arms. I-SPY2 has now tested more than 8 immuno-oncology agents. [This biomarker analysis] gives us an opportunity to understand which patients respond, which patients don’t respond, and the biology behind those responses. In more pragmatic terms, it’s also helping to develop a biomarker that predicts response. The reason that’s important is because immuno-oncology agents are associated with their [own set of] adverse effects [AEs]. Some of these immune-related AEs are permanent, such as adrenal insufficiency and thyroid disorder, so [the AEs are] not just a temporary discomfort; [they] last for life.

We as a field, not just in breast cancer but in all cancer types where immunotherapy is being used, need to be able to identify patients who have an extremely high likelihood of response [with immunotherapy] and only give the immune agents to them to spare patients with a low probability of response the possible long-term toxicities of an immune agent.

How was I-SPY2 conducted?

We had a set of approximately 30 continuous, immune-related biomarkers, which represented different genes that are part of the checkpoints that prevent a patient’s immune system from recognizing and destroying cancer. There were immune cell subpopulation signatures, immune cancer signaling signatures, and a few other classes. We looked to see which ones [were] associated with response as a continuous variable, and in which group and arm.

One of our main observations was that the group of biomarkers that can be classified as immune tumor signaling—which are enriched for chemokines and cytokines, not so much for T-cell receptors and the other cell population signatures—are highly associated with response in both HR-positive breast cancers and TNBC. [We noted this finding] across arms. That was our most universal signal.

We’ve been working with the immunologist at UCSF, Michael J. Campbell, PhD, to perform multiplex immunofluorescence analyses, which take tumor sections from patients and stain them with different antibodies. This allows you to know within a single patient what kinds of immune cells are present, how many, and what their relative spatial orientation is compared with the tumor cells.

We used those data to try to understand the biology we’re capturing in these most predictive signatures. We found that our strongest correlations were between the spatial proximity measures that indicate a high level of co-localization, or mixing, between PD-1–positive T-cells and PD-L1–positive tumor cells. That was especially true in TNBC.

There’s a subtype specificity in immune predictors of response. HR-positive breast cancers and TNBC share tumor immune signaling signatures [that] are associated with response. There are differences too, one [being] that in the HR-positive group, there are negative signals, meaning that high levels [of these signals are] associated with resistance. To get good [immunotherapy] performance in the HR-positive group, you need to combine the positive and the negative [signals]. In the TNBC group, [all the signals are] positive, and you only need the positive ones, which is great, and you get good performance.

From these continuous biomarkers, in a paper published in Cancer Cell in June 2023, we developed response-predictive subtypes, a new way of subtyping breast cancer. Instead of just classifying patients according to their HR and HER2 status, we bring in biomarkers, including an immune dichotomous biomarker, based on the continuous results, that classify patients as immune high or immune low. That’s incorporated into the subtyping schema that, if applied to prioritize different treatment types in patients, is predicted to result in a higher likelihood of each patient getting the treatment they most need, and longer survival.

What were the key findings from the study?

We partnered with a company to translate our most promising and robust immune biomarker results in the research setting into a clinical biomarker called ImPrint. We applied for and received Investigational Device Exemption from the FDA, and ImPrint is now being used in the I-SPY2 trial.

We took the 5 immune arms and assessed the performance and prevalence of this clinical biomarker. In the HR-positive group, we are especially excited about ImPrint. It was developed entirely in the first immune arm of the trial and then tested for validation in the other arms; the performance was phenomenal.

In the HR-positive group, the typical level of response in the immune-oncology arms was approximately 30%, which is better than [the response rates seen with the] control which [were] more like 15%. However, 30% is not that high. Whereas when you apply ImPrint, in the ImPrint-positive patients who are HR-positive and [HER2-negative], the average pCR rate across all the immune arms was 76%. These are patients who will do well when they have a CR. In the ImPrint-negative group, response rates were quite low, [at 16%], so it’s a huge gap. We’re excited about this [research because] we feel this work has identified a subpopulation of patients with HR-positive disease for whom immunotherapy agents are exactly the right [treatments] and will save their lives.

How might these findings potentially affect clinical practice?

This was the first trial to test an immune agent in the HR-positive subtype, and big trials are ongoing now to finally test [this]. Right now, immunotherapy is the SOC for triple-negative disease. It’s been [studied] in TNBC extensively, but people have not been [studying] it in HR-positive [disease]. With the trial results from I-SPY2 and this biomarker, we are well poised to show that [immunotherapy is how patients with HR-positive disease] should be treated. This should be the new SOC for HR-positive disease [when the patient is either] ImPrint positive or [positive for] another immune biomarker—it doesn’t have to be ImPrint, but we know ImPrint works.

In the triple-negative group, high pCR rates [were seen] in the ImPrint-positive group, at 75% [compared with] 37% in the ImPrint-negative group. We’re currently working on a refinement that will [lower] the response rate in the ImPrint-negative group to be closer to approximately 25% or 30%.

What is your main message for colleagues based on this research?

My main message for colleagues and the whole breast cancer treatment and research field is that gene expression is sufficient to predict which patients with breast cancer will respond to immunotherapy and which will not. Although it’s nice to have multiplex immunofluorescence to understand [which patients may respond], you don’t need it [because] the gene expression is enough. These tumor immune communication signals are robust, and patients with HR-positive, HER2-negative disease should not be left out of the immuno-oncology revolution—we have a way to identify which patients will respond [to immunotherapy].

Reference

Wolf DM, Yau C, Campbell MJ, et al. Biomarkers predicting response to 5 immunotherapy arms in the neoadjuvant I-SPY2 trial for early-stage breast cancer (BC): evaluation of immune subtyping in the response predictive subtypes (RPS). J Clin Oncol. 2023;41(suppl_16). doi:10.1200/JCO.2023.41.16_suppl.102