I-SPY2 Graduates a Two-Immunotherapy Pre-Surgery Breast Cancer Regimen, With a Gene Test to Pick Who Gets It

The I-SPY2 platform published a graduate in JAMA Oncology on July 30: a two-immunotherapy pre-surgery regimen that raised no-cancer-at-surgery rates across every HER2-negative subtype it tracks. The mechanism behind the gain, and the question of who actually benefits, lives in a tumor immune-gene signature called ImPrint (primary publication).

The two checkpoint drugs in the regimen are cemiplimab, an anti–PD-1 antibody, and fianlimab, an anti–LAG-3 antibody. PD-1 is the well-known brake on tumor-killing T cells; LAG-3 is a second, less familiar brake on the same cells. Hitting both at once is the bet the new regimen is making. The trade-off is added immune-related toxicity, especially in the glands that regulate the body's stress hormones.

In the PCF arm of I-SPY2, 78 patients got cemiplimab and fianlimab layered onto standard pre-surgery chemotherapy, against 350 historical controls who got chemotherapy alone. Across all HER2-negative patients, pathologic complete response (pCR), meaning no invasive cancer found in the tissue removed at surgery, rose from 21% in controls to 44% with the dual checkpoint. In triple-negative breast cancer the rate was 53% versus 29%. In hormone-receptor-positive, HER2-negative disease it was 36% versus 14%. The platform "graduated" the arm in every clinical signature it tracks, the formal jargon for crossing the 85% Bayesian threshold the trial uses to flag a regimen as worth moving to definitive testing.

The platform's main job is the biomarker split. ImPrint is an FDA-cleared mRNA classifier that grades a tumor's immune activity. Across five prior I-SPY2 immunotherapy arms, ImPrint-positive HR-positive, HER2-negative patients had a 75% pCR rate versus 17% in ImPrint-negative tumors (prior I-SPY2 IO analysis). The PCF arm sits on top of that pattern. Dual checkpoint plus chemo in ImPrint-positive patients produces the largest no-cancer-at-surgery rates the platform has recorded for HR-positive, HER2-negative disease, and that is the patient population in which the regimen most clearly justifies its toxicity.

The cost is real. Twenty-one percent of patients on the PCF arm developed adrenal insufficiency, including hypophysitis, the inflammation of the pituitary that controls the body's stress-hormone axis. Eleven percent of those events were grade 3 or 4, severe enough to need treatment, and most surfaced after immunotherapy ended, not during chemotherapy. For a high-risk patient weighing pre-surgical options, the trade is between a higher chance of no invasive cancer at surgery and a roughly one-in-five chance of needing lifelong hormone replacement.

The same drug combination had a different ending in another tumor type. On May 15, Regeneron reported that fianlimab plus cemiplimab missed its primary progression-free survival endpoint in first-line unresectable or metastatic melanoma, with median PFS of 11.5 months versus 6.4 months for pembrolizumab (hazard ratio 0.845, p=0.0627) (Regeneron release). The breast-cancer graduation is a counterweight in the same program, and the open question is whether the patients ImPrint identifies are the ones the melanoma arm tested.

I-SPY2 uses pCR as a surrogate endpoint, not overall survival. Graduates of the platform have not always held up in phase 3, and the trial authors are explicit that the result "warrants further definitive trials" rather than practice change. The next step is a phase 3 study in ImPrint-selected patients, with long-term survival, endocrine outcomes, and the durability of the adrenal events as the open questions. For now, the platform has done what it was designed to do. It flagged a candidate regimen, identified the patients who should get it, and did both years before a phase 3 readout could answer either question.