NEW YORK – Agendia's ImPrint immune classifier has demonstrated in two recent studies that it can identify breast cancer patients across subtypes who will respond to the addition of immunotherapy to chemotherapy in the neoadjuvant setting.
Researchers presented data from cohorts of the I-SPY 2 platform trial at the American Association for Cancer Research's annual meeting on Monday and at the European Breast Cancer Conference last month. They reported at both of these meetings that ImPrint, a 53-gene classifier that scores the immune activity of a patient's tumor, could identify which breast cancer patients with hormone receptor (HR)-positive, HER2-negative or triple-negative tumors, who are deemed to be at high risk of recurrence by Agendia's MammaPrint test, are likely to benefit from neoadjuvant chemo and immunotherapy.
Laura van 't Veer, the developer of ImPrint and co-leader of the breast oncology program at the University of California, San Francisco, said the assay was developed to spare patients who won't respond from the potential side effects of immunotherapy. "Immuno-oncology drugs can have severe side effects, and some of those are irreversible such as thyroid or adrenal insufficiency, kidney failure, and diabetes," she said in an interview. "Because of this, within the I-SPY setting, we were very motivated to get this ImPrint signature implemented."
This I-SPY 2 analysis included outcomes data on five different immunotherapies that patients received with a taxane chemotherapy followed by anthracyclines in the neoadjuvant setting: Merck's checkpoint inhibitor Keytruda (pembrolizumab); AstraZeneca's Imfinzi (durvalumab) plus its PARP inhibitor Lynparza (olaparib); TriSalus Life Sciences' TLR9 agonist SD-101 plus Keytruda; Regeneron's checkpoint inhibitor Libtayo (cemiplimab-rwlc); and Libtayo plus Regereron's LAG3 inhibitor REGN-3767. In the control arm, patients received standard neoadjuvant taxane chemotherapy followed by anthracyclines.
The researchers partnered with Agendia to develop ImPrint based on data from the Keytruda arm in I-SPY 2, the first immunotherapy arm in the trial. ImPrint was then used to analyze mRNA in patients' pre-treatment tissue samples and retrospectively gauge their responses to the neoadjuvant immunotherapy regimens they received.
Following this retrospective analysis, the researchers are prospectively validating ImPrint in the next stage of the I-SPY 2 trial, van 't Veer said.
HR-positive, HER2-negative group
At AACR this week, Denise Wolf, a bioinformatics research scientist at UCSF, presented data from the retrospective analysis using the ImPrint classifier in HR-positive, HER2-negative breast cancer patients.
Wolf noted that most immunotherapy trials in this setting have previously excluded HR-positive, HER2-negative breast cancer patients because they were "considered unlikely to respond due to low levels of immune infiltrate."
Across the whole cohort, the pathologic complete response rate on immunotherapy plus chemo was 33 percent versus 13.5 percent on chemo alone. Wolf said this suggested that a subset of HR-positive, HER2-negative breast cancer patients were responding well to the addition of immunotherapy. "However, these [pathologic complete response] rates were much lower than those seen in the triple-negative patients," she said, emphasizing that immunotherapy can come with serious lifelong side effects.
"What we want to do is take the balance between benefit and harm and tip it toward the direction where the benefit greatly outweighs the harm," she continued. "To do this, we need clinical-grade predictive biomarkers to identify the patients with a very high probability of response to immunotherapy."
When the researchers retrospectively explored ImPrint results in this group, 28 percent of patients were found to be ImPrint-positive. Among them, the pathologic complete response rate across all five neoadjuvant immunotherapy arms, comprising 204 patients, was 75 percent, compared to 17 percent among the ImPrint-negative patients. In the control arm, the pathologic complete response rate for the ImPrint-positive group was 33 percent compared to 8 percent in the ImPrint-negative group. This translated to a 52 percent reduction in the risk of breast cancer recurrence on immunotherapy versus chemo in the ImPrint-positive group.
Although patients in the five immunotherapy arms hadn't been on treatment long enough for researchers to evaluate long-term recurrence-free survival, they built a predictive model to estimate what the distant recurrence-free survival would be between the immunotherapy-chemo and chemo arms for the ImPrint-positive group.
They estimated that the five-year distant recurrence-free survival was 91 percent for ImPrint-positive patients with HR-positive, HER2-negative breast cancer who got immunotherapy-chemo versus 80 percent for those on chemo.
Wolf concluded that using ImPrint to guide immunotherapy-chemo treatment in HR-positive, HER2-negative breast cancer patients can result in pathologic complete response rates "similar to what is [historically] seen with the best neoadjuvant therapies in triple-negative and HER2-positive breast cancer" — two subtypes that typically have neoadjuvant immunotherapy response rates near 70 percent.
Triple-negative group
Van 't Veer presented retrospective analysis on ImPrint from the triple-negative breast cancer group in the I-SPY 2 trial at EBCC in March. She also described how the researchers have developed a TNBC-specific ImPrint classifier, called ImPrintTN, to better select this subset of patients for immunotherapy treatment.
Because TNBC patients typically respond well to immunotherapy, van 't Veer noted that the general ImPrint classifier was not sufficient for this group. "Within the triple-negative context, it would have too many patients in the ImPrint-negative group that still would see a benefit," she explained. "That motivated us to improve the ImPrint classifier for the triple-negative population."
In her presentation at EBCC, van 't Veer noted that the pathologic complete response rate in the negative group based on the original ImPrint classifier was too high at 34 percent. So, her team went back to the drawing board and trained a more specific classifier based on the TNBC cohort, she said.
That revamped classifier, ImPrintTN, retrospectively determined that 66 percent of TNBC patients in the I-SPY cohort would respond to immunotherapy. Across the whole immunotherapy arm there were 150 TNBC patients, and the pathologic complete response rate was 54 percent in the immunotherapy-chemo arm compared to 26 percent in the chemo alone arm.
In an independent test set of 54 patients, the pathologic complete response rate on immunotherapy-chemo was 71 percent in the ImPrintTN-positive group compared to 22 percent in ImPrintTN-negative patients. On chemo, ImPrintTN-positive patients had a pathologic complete response rate of 30 percent compared to 15 percent in the negative group.
Across all of the immunotherapy arms, which included both the ImPrintTN training and test sets, the pathologic complete response rate for ImPrintTN-positive patients was 74 percent versus 16 percent for ImPrintTN-negative patients. "Our current data suggest that ImPrintTN may help inform prioritization of which treatment to give these patients [and] to give the immune drugs only when it really will benefit," she concluded.
With prospective validation underway for ImPrint and ImPrintTN, van 't Veer believes the test would have the greatest impact if used in newly diagnosed breast cancer patients who are at high risk of early recurrence.
"One of the rationales of the I-SPY 2 trial to begin with is that you need to give patients the best treatment to which they respond right away so that they never get a metastasis," van 't Veer said. "That's also where this test should land, in the first diagnosis of breast cancer, … [if you] set out to give a patient the best therapy to which their tumor would respond."