Neoadjuvant Cemiplimab Hit 36% pCR in High-Risk CSCC Trial

Two cycles of neoadjuvant cemiplimab produced a pathological complete response in 36% of patients with high-risk, resectable cutaneous squamous cell carcinoma of the head and neck — and an exploratory 10-gene signature called NeoSquamos identified recurrence risk with an AUC of 0.838. That's the core result of the phase II NEO-CESQ trial (NCT04632433), led by Paolo A. Ascierto and colleagues across seven Italian centers.

The trial enrolled 25 patients between May 2021 and October 2022 with surgically resectable, high-risk stage III or IV (M0) CSCC of the head and neck. Median age was 82 years; 88% were men. Patients received cemiplimab 350 mg IV every three weeks for two neoadjuvant cycles, then resection, followed by planned adjuvant cemiplimab for up to one year. Postoperative radiotherapy was optional — only three patients received it. The primary endpoint was major pathological response (MPR), defined as under 10% residual viable tumor.

What were the response and survival numbers?

In the intention-to-treat population, 11 patients achieved MPR (44%) and nine achieved pCR (36%). Among the 23 patients who underwent surgery, the descriptive rates were 47.8% and 39.1%. Radiology told a different story: the objective response rate was 52%, but only two of six radiological complete responders had a true pCR — fibrosis and immune infiltration can mimic eradication on imaging.

At a median follow-up of 42 months, three recurrences occurred, all in patients without MPR. No patient with pCR relapsed, and there were no disease-related deaths. MPR was associated with event-free survival (HR 0.16; 95% CI 0.03–0.75); recurrence-free survival showed a favorable but nonsignificant trend (HR 0.17; 95% CI 0.02–1.47). Safety was manageable: no grade 3 or 4 treatment-related adverse events, though the cohort was small and limited to ECOG 0–1 patients.

Why did some tumors resist PD-1 blockade?

Transcriptomic profiling (NanoString PanCancer IO 360) succeeded in 21 patients. Responders showed enrichment of interferon-γ-associated gene signatures — evidence of pre-existing immune activation. Nonresponders expressed more WNT5A, linked to immune exclusion, and CD276 (B7-H3), an immunoregulatory target of growing interest for antibody–drug conjugates. Higher CD276 also tracked with less favorable event-free survival.

Circulating IL-6 added a longitudinal signal. Before the first adjuvant cycle, patients with pCR had median IL-6 of 5.5 pg/mL versus 13.5 pg/mL for those without (p=0.003); the gap persisted at cycles 5 (4.3 vs 7.6) and 9 (3.7 vs 8.1). But baseline IL-6 predicted nothing — no association with pCR, MPR, or survival — so it's a candidate monitoring marker, not a selection tool.

Can NeoSquamos predict recurrence?

The investigators compared pretreatment expression between patients who recurred and those who didn't, selecting ten genes upregulated in recurrence: MFGE8, TGFB3, LAMB3, TPM1, ITGAV, THBS1, VCAN, COL5A1, LAMC2 and PKM — a stroma-and-metabolism profile rather than a pure immune signature. In the discovery cohort, AUC for recurrence was 0.838.

In an independent cohort of 18 patients with advanced CSCC on PD-1 therapy, the signature separated recurrence-free survival significantly (log-rank p=0.037): one-year recurrence-free survival was 100% in the low-risk group versus 30.3% in the high-risk group. Combined analysis gave a hazard ratio of 16.01 (95% CI 3.12–82.13) for high-risk classification.

Caveats matter here. The signature came from 21 patients with just three recurrence events — real overfitting risk. Notably, NeoSquamos did not predict MPR or pCR, suggesting it captures long-term recurrence biology, not immediate drug sensitivity.

How does this fit the broader picture?

Context is strong. Gross and colleagues' phase II study of neoadjuvant cemiplimab in stage II–IV CSCC reported a 51% pCR rate; its final analysis (Journal of Clinical Oncology, October 2026) showed a 54-month event-free survival of 76.2% and zero recurrences among 40 pCR patients. The phase III C-POST trial, which drove FDA approval of adjuvant cemiplimab in October 2025, showed 24-month disease-free survival of 87.1% versus 64.1% for placebo (HR 0.32).

So what does this change for clinicians? Not practice — not yet. No biomarker here is prospectively validated, and a single-arm design can't isolate the neoadjuvant contribution. But NEO-CESQ makes the case that integrating tumor transcriptomics, circulating markers like IL-6, pathological response and possibly circulating tumor DNA could eventually individualize postoperative management. Larger prospective studies will have to test that.