Add-On Pembrolizumab Misses in Locally Advanced Head and Neck Cancer

PARIS — Combining a PD-1 inhibitor with standard chemoradiotherapy (CRT) failed to significantly improve survival outcomes in unresectable locally advanced head and neck squamous cell carcinoma, results of the phase III KEYNOTE-412 trial indicated.

Median event-free survival (EFS) in the intent-to-treat population was not reached in the group receiving CRT plus pembrolizumab (Keytruda), as compared with 46.6 months for those randomized to CRT plus placebo, reported Jean-Pascal Machiels, MD, PhD, of the Cliniques Universitaires Saint-Luc in Brussels.

This difference in the study’s primary endpoint narrowly missed statistical significance, with EFS rates of 63.2% with pembrolizumab and 56.2% with placebo at 2 years (HR 0.83 95% CI 0.68-1.03), he said at the European Society for Medical Oncology (ESMO) annual congress.

Overall survival (OS) was also no different, with 3-year rates of 71.9% and 70.1% for the pembrolizumab and controls arms, respectively.

Systemic treatment for locally advanced head and neck squamous cell carcinoma has seen no advances in more than 2 years, Machiels explained during an ESMO press briefing, with the standard of care in this patient population involving CRT with concurrent high-dose cisplatin.

Machiels noted that when the trial was initially designed 7 years ago, it was unknown that PD-L1 could predict benefit.

“We learned that 2 years ago in the recurrent/metastatic setting,” he said, when KEYNOTE-048 showed the addition of pembrolizumab to first-line chemotherapy improved OS over chemotherapy alone, as did single-agent pembrolizumab in the subgroup with a combined positive score (CPS) of ≥1, a measure of PD-L1 expression levels.

In the new study, a prespecified subgroup analysis hinted at a potential EFS benefit only for those with a PD-L1 expression level of 1 or greater on CPS:

• CPS ≥1: HR 0.80 (95% CI 0.64-1.00)
• CPS <1: HR 1.09 (95% CI 0.56-2.11)

And post hoc analyses suggested that patients with even higher PD-L1 expression levels (CPS ≥20) had the greatest chance of benefit with pembrolizumab, with 2-years EFS rates of 71.2% versus 62.6% with placebo (HR 0.73, 95% CI 0.49-1.06) and 3-year OS rates of 79.1% versus 73.0% (HR 0.67, 95% CI 0.43-1.04).

The signal of benefit, particularly in the high PD-L1 subgroup, “is consistent with the Javelin-100 study, which was also negative,” said ESMO discussant James Larkin, MD, PhD, of the Institute Of Cancer Research in London.

Larkin posited that the concurrent use of pembrolizumab with CRT in KEYNOTE 412 could have played a role in the negative study findings. “A comparison might be made with the PACIFIC study in non-small cell lung cancer, which is a positive trial, where actually the checkpoint inhibition with durvalumab [Imfinzi] was given immediately after the chemoradiotherapy, leading to benefit,” he said.

KEYNOTE-412 was a phase III trial involving 804 newly diagnosed patients with unresectable locally advanced head and neck cancer. Patients needed to be eligible for curative-intent high-dose cisplatin-based CRT for study inclusion.

Patients received standard doses of CRT and were randomized 1:1 to either pembrolizumab or placebo, given concurrently and then as maintenance every 3 weeks for 17 cycles, or until disease progression or unacceptable toxicity.

Baseline characteristics were well balanced: median patient age was 59 , 82% were men, and about three-fourths were white. About half of the patients were from Western Europe and 10.7% were from North America. Roughly a fourth of patients had HPV-positive tumors.

Tumors in the oropharynx were most common (roughly 50%), followed by those in the larynx (21-23%), hypopharynx (18%), and oral cavity (10%). In terms of PD-L1 status, approximately 85% of patients had a CPS of ≥1, and 36% had a CPS of ≥20.

Treatment exposure was high, said Machiels, with 86% of patients in the pembrolizumab arm and 88% of those in the placebo arm completing CRT and going on to the maintenance portion of the trial. Maintenance treatment was completed by 60% and 63%, respectively. The vast majority, 93% in each arm, received radiation doses of 70 Gy or above.

Looking at components of the EFS endpoint in the total study population, locoregional progressive disease was numerically lower in the pembrolizumab arm (13.2% vs 14.2% in the placebo arm), as was distant progressive disease (12.9% vs 16.7%), Machiels noted.

No new safety signals were observed, with grade ≥3 adverse events (AEs) of any cause slightly higher in the pembrolizumab arm (92% vs 88%). As expected, said Machiels, the most frequent toxicities included stomatitis, radiation skin injury, nausea, and anemia. AEs leading to treatment discontinuation occurred in 41% of the pembrolizumab arm versus 33% of those in the placebo arm. Four and six treatment-related deaths were reported for the two arms, respectively.

Immune-mediated grade ≥3 AEs were reported in 8.3% of patients in the study arm and 2.3% of those in the control arm, with one death in the pembrolizumab arm due to autoimmune cholangitis.