Burtness Breaks Down Pembrolizumab Data in HNSCC

Barbara Burtness, MD, professor of Medicine at Yale Cancer Center,

Barbara Burtness, MD

As investigators attempt to move immunotherapy earlier in the management of head and neck squamous cell carcinoma (HNSCC), questions are arising regarding who should receive single-agent therapy, a chemoimmunotherapy approach, or the current standard of care.

Interim findings from the phase III KEYNOTE-048 trial showed that frontline pembrolizumab (Keytruda) improved overall survival (OS) and duration of response compared with standard therapy for patients with recurrent or metastatic HNSCC as a single agent. However, the benefit was not observed in progression-free survival (PFS) or overall response rate.

OncLive: Could you provide some background on this study?

In an interview with OncLive, lead study author Barbara A. Burtness, MD, professor of Medicine (Medical Oncology); Disease Aligned Research Team Leader, Head and Neck Cancers Program; co-director, Developmental Therapeutics Research Program; of Yale Cancer Center, explained these data, and the impact they could have on the treatment of patients with HNSCC.Burtness: We know that immunotherapy has activity in head and neck cancer, and both pembrolizumab and nivolumab have been approved because of survival advantages in the second-line setting. The question then became, “Could you use immunotherapy in the first-line setting?” This is partly because that is where it would have the biggest impact on survival, and partly because many patients with HNSCC are quite sick and are not often fit for second-line therapy.

This trial addressed 2 questions. The first was the superiority of pembrolizumab alone in patients with the biomarker profile that is associated with response. Secondly, given that chemotherapy is beneficial in head and neck cancer, would there be an advantage to incorporating pembrolizumab with chemotherapy?

This was a phase III study that had 3 arms. It compared pembrolizumab alone to a new regimen, which added either cisplatin or carboplatin together with 5-fluorouracil (5-FU) and combined it with pembrolizumab. The chemotherapy was given for 6 cycles, and if patients were still progression free, they could continue on pembrolizumab alone. The third arm was the EXTREME regimen—6 cycles of a platinum agent with 5-FU and cetuximab (Erbitux) followed by cetuximab alone.

Patients who were eligible were those with squamous cell cancer with recurrent metastatic disease, could not have had a curative intent local therapy, and they had to have tissue available to do biomarker testing. If [a patient] had oropharynx cancer, we had to know their p16 status so we could stratify by that. Patients were also stratified by their performance status and PD-L1 stain level, dichotomized at 50% of the tumor cell staining. They were randomized 1:1:1 to pembrolizumab, pembrolizumab plus chemotherapy, or EXTREME.

The primary endpoints of the study were OS and PFS, and those analyses were conducted for 3 separate populations. The first was the overall population, and then there were the 2 biomarker-enriched populations. The way that the biomarker-enriched populations were determined were based not only on tumor cell staining, but on a different way of scoring PD-L1 expression called the combined positive score (CPS). We looked at the patients who expressed this richly, which was a CPS score ≥20, and those who expressed any PD-L1 expression, which was a CPS ≥1 (CPS 1). CPS was defined as the number of cells that were PD-L1—positive—looking at tumor cells, lymphocytes, and macrophages—divided by the tumor number of cells counted and multiplied by 100.

The first analysis was OS for the CPS ≥20 enriched population. In that population, pembrolizumab was superior to combination chemotherapy with cetuximab. It raised the median OS from 10.7 months to 14.9 months, which is the longest survival that has ever been described for this subpopulation. The hazard ratio was 0.61, the P value was .0007, and the effects were durable.

There were significantly more people alive at the 2-year mark on the pembrolizumab arm than on the chemotherapy arm. The interesting thing is that the response rate and PFS were not superior for pembrolizumab. There were patients who remained progression free at 2 years, but there were a substantial number of patients who had crossed over to second-line therapy. It appeared that the early use of pembrolizumab conveyed a survival advantage that was quite durable.

We then looked at the CPS ≥1 population. The initial use of pembrolizumab led to a superior OS compared with the EXTREME regimen. OS went from 10.3 to 12.3 months. The hazard ratio was 0.78, which was statistically significant, and was associated with response rate and PFS being a little lower. But, it was a durable benefit.

We then looked at the OS benefit of using pembrolizumab with chemotherapy compared with the EXTREME regimen in the biomarker-unselected population. There was also a benefit for the early use of pembrolizumab observed. The OS was also improved, going from 10.7 with the EXTREME regimen to 13 months in the pembrolizumab arm, with a hazard ratio of 0.77. The response rates were comparable, the PFS were not significantly different, and the amount of grade 3 to 5 and overall toxicity was quite similar. Patients who saw a response saw a more durable response, with a visible survival benefit at 2 years.

This demonstrates that early use of pembrolizumab in patients with recurrent metastatic cancer leads to an improvement in OS in HNSCC. For patients who have the biomarker enrichment and express PD-L1, it appears that pembrolizumab alone as first-line therapy is superior to anything that we've done before. For patients who are not biomarker-enriched, the combination of pembrolizumab and chemotherapy is superior to previous standard treatment.

This is an interim analysis of this study, and there are several planned. It has a sort of complicated statistical design, and there were a number of primary hypotheses that were tested in parallel. If those were positive, the analysis could roll down to a subsequent analysis. We are still awaiting OS analysis of the comparison of pembrolizumab alone versus the EXTREME regimen in the biomarker-unselected population. At this interim analysis, that was seen to be noninferior. We will also be seeing the final analysis of pembrolizumab plus chemotherapy in the biomarker-enriched populations.

This is a major change in the way that we approach metastatic recurrent HNSCC. People were expecting to see a benefit for immunotherapy, given how potent it was in the second-line setting. However, to see that there are some patients who may not ever need to see chemotherapy for metastatic recurrent disease is a remarkable change.

What unmet need could pembrolizumab fill in the current treatment of these patients?

The question is, “Are we going to be using pembrolizumab alone?” For the CPS ≥20 and CPS ≥1 population, it appears to lead to a superior OS. The response rates were 23% and 19% in those populations, which were a little less than the EXTREME regimen. For patients who are highly symptomatic and need a rapid response, even if they have the biomarker enrichment, you might prefer to treat with pembrolizumab and chemotherapy. In other patients who are asymptomatic with lung metastases who have a CPS ≥20, you might feel comfortable treating them just with pembrolizumab alone.The EXTREME regimen has been the best that we have, but I don't think anyone has enjoyed using it. It is quite toxic, there is a risk for neutropenia, and patients can develop a rash; there are days where they really don't feel well. This regimen has a response rate of 36% and a median survival of 10.7 months; for anyone who has metastatic HNSCC, you don't want to hear that this is their average life expectancy. The unmet need is the fact that we are using a toxic regimen, and although it does lead to responses, it has been a modest effect. Pembrolizumab clearly looks better.

What is your take-home message about this study?

We are also anticipating forthcoming trials incorporating PD-1—directed therapies with chemoradiation in the curative setting. Perhaps down the road, we won't be seeing as much recurrent metastatic disease. For now, seeing a regimen that is no more toxic than what we have with the combination of chemotherapy and pembrolizumab—in addition to the less-toxic pembrolizumab monotherapy—meets our unmet needs. We are looking for novel immunotherapy combinations that will get even more patients into complete response and be even more durable.One of the things that is going to come up in the clinic is the use of this CPS ≥20 biomarker. Many of our institutions are now doing PD-L1 testing, and are reporting it out in terms of either the proportion of tumor cells that stain for PD-L1 or the proportion of stromal cell that stain for PD-L1. This novel way of calculating the positive biomarker, by looking at the proportion of positive cells that either tumor cells or immune cells, does not require the pathologist to do anything different when looking through the microscope. It is going to require a different reporting algorithm. If it is going to go forward as a companion biomarker at registration, that will add some complexity.