Association of pathologic complete response with the 27-gene IO score and week 3 IOpath response following neoadjuvant pembrolizumab +/- intralymphatic cytokines in the neoIRX trial

Introduction: Novel biomarker strategies are needed to identify immunotherapy (IO)-sensitive early-stage triple-negative breast cancers (TNBC). We recently reported a phase II trial (neoIRX, NCT04373031) evaluating induction IO (pembrolizumab +/- intralymphatic cytokines, IRX-2, comprised of physiologic doses of IL-2, IFNγ, and other cytokines derived from activated donor lymphocytes) preceding de-escalated neoadjuvant chemotherapy (NAC). Here, we report associations of response with the 27-gene IO score (DetermaIO), a commercially available RT-qPCR assay shown to predict IO response in the neoPACT, I-SPY2, and neoTRIP trials. We also propose a novel surrogate endpoint, entitled “IOpath” response, that characterizes radiographic and pathologic response following single-cycle IO and may be useful for guiding NAC de-escalation or omission. Methods: Subjects with stage II-III TNBC were randomized 1:1 (n=12) to receive induction pembrolizumab (200mg IV) +/- peri-areolar IRX-2 (IRX-2 arm: 1 ml SQ x2 daily for 10 days + cyclophosphamide 300 mg/m2 IV x 1) preceding initiation of pembrolizumab + NAC. The primary endpoint was pathologic complete response (pCR) rate. Post-induction IO response (IOpath) was assessed by week 3 ultrasound (US) and biopsy. IOpath response was designated as IOpath-NR (non-response: radiographic progression and/or biopsy non-response [neither treatment effect nor TILs expansion]), IOpath-PR (partial response: viable tumor with either treatment effect or TILs expansion), IOpath-CR (complete response: treatment effect with no viable tumor), or IOpath-NE (not evaluable). Results: The IRX-2 arm achieved 83% pCR (n=5/6, CI 36-100%) compared to 33% pCR with pembrolizumab alone (n=2/6, CI 4-78%) and was well tolerated (67% grade I skin bruise). Baseline 27-gene IO score predicted pCR (AUC 0.77, IO+: 5/6 pCR; IO-: 2/6 pCR) and week 3 IOpath response (AUC 0.85, IO+: 3/5 IOpath-CR/PR; IO-: 1/4 IOpath-PR), and outperformed TILs (AUC for pCR: 0.66; AUC for IOpath: 0.68) and CPS score (AUC for pCR: 0.63, AUC for IOpath: 0.58). Week 3 IOpath response was associated with 100% pCR (IOpath+: 4/4 pCR; IOpath-: 2/5 pCR). Conclusion: A subset of TNBCs experience brisk IO response following single-cycle IO (IOpath response), which is associated with 100% pCR in this preliminary dataset, highlighting IOpath as a potential surrogate biomarker to guide NAC de-escalation or omission in early-stage TNBC. Baseline 27-gene IO score predicts pCR and week 3 IOpath response. A future study is planned (neoINBRX) to evaluate the feasibility of combining baseline IO score with week 3 IOpath response to identify and treat IO-sensitive tumors with chemotherapy-sparing IO combination therapy. Acknowledgements: The study was funded in part by Merck, Sharpe, & Dohme as part of the Merck Investigator Studies Program, and by BrooklynImmunotherapeutics.