乌鸦传媒

Adjuvant pembrolizumab (Keytruda) after surgery for early lung cancer improved disease-free survival (DFS) by almost a year, a large randomized, placebo-controlled trial showed.

Patients who received the PD-1 inhibitor, with or without chemotherapy, had a median DFS of 53.6 months as compared with 42.0 months for the placebo arm. A subgroup analysis showed a consistent advantage for pembrolizumab with a few exceptions. Patients who received adjuvant chemotherapy and those with squamous histology had numerically better DFS with placebo.

Additionally, patients with high PD-L1 expression (total positive score [TPS] ≥50%) did numerically better with pembrolizumab, but the difference did not achieve statistical significance, reported Luis Paz-Ares, MD, of the University Hospital October 12 in Madrid, during a European Society for Medical Oncology (ESMO) virtual plenary presentation.

"We believe that pembrolizumab has the potential to become a new adjuvant treatment option for patients with stage Ib of more than four centimeters, stage II, and stage III non-small cell lung cancer (NSCLC) following complete resection and adjuvant chemotherapy when recommended, regardless of PD-L1 expression," he said.

The findings from the trial follow the positive outcome of the IMpower010 randomized trial of adjuvant atezolizumab (Tecentriq), which led to the latter drug's approval for stage II-IIIa NSCLC with PD-L1 expression ≥1%. IMpower010, which compared atezolizumab and best supportive care, led to a 34% reduction in the DFS hazard, whereas pembrolizumab reduced the hazard by 24%.

Study Details

Investigators in the multinational PEARLS/KEYNOTE-091 trial enrolled patients with stage Ib-IIIa NSCLC and negative margins after surgical resection. After centralized assessment of tumor tissue PD-L1 expression status, patients were randomized to pembrolizumab or placebo. Adjuvant chemotherapy was not mandated by the protocol, but was "recommended" for patients with Ib tumors and "strongly encouraged" for those with stage II and IIIa disease. In the IMpower010 trial, all patients received investigator's choice of chemotherapy.

The trial had dual primary endpoints: DFS in the overall population and in the subgroup of patients with PD-L1 TPS ≥50%. Investigators randomized 1,177 patients, and the trial had a median follow-up of 35.6 months.

The study population had a median age of 65, men accounted for 68% of all patients, and a slim majority (~51%) of participants came from Western Europe. Additionally, 87% of the patients were current or former smokers, about 6% had EGFR mutations, and 1.2% had ALK translocation. Almost two thirds of the patients had nonsquamous histology, 56% had stage II disease (29% stage IIIa), and almost 86% of patients received adjuvant chemotherapy. A third of patients had PD-L1 TPS 1-49%, and 28% had TPS ≥50%.

The primary analysis of DFS in the total population showed that adjuvant pembrolizumab met criteria for statistical superiority (HR 0.76, 95% CI 0.63-0.91, P=0.0014). In the TPS ≥50% subgroup, median DFS had yet to be reached in either treatment arm. At the time of the analysis, patients in the pembrolizumab arm had an 18% lower DFS hazard (95% CI 0.57-1.18). A preliminary analysis of overall survival showed no difference between groups (HR 0.87, 95% CI 0.67-1.15).

Grade 3-5 adverse events (AEs) occurred more often with pembrolizumab (34.1% vs 25.8%), and four treatment-related fatal AEs occurred in the pembrolizumab arm (two related to myocarditis and one each from pneumonia and sudden death). Serious AEs (24.5% vs 15.5%) and AEs leading to discontinuation (19.8% vs 5.9%) or treatment interruption (38.1% vs 25.0%) also occurred more often in the pembrolizumab group.

'Cadre of Elephants'

Invited discussant Martin Reck, MD, of Hospital Grosshansdorf in Hamburg, Germany, highlighted what he characterized as "a cadre of elephants" in the room related to the PEARLS/KEYNOTE-091 trial.

He began with a disconnect in the relationship between PD-L1 expression and pembrolizumab activity. In the IMpower010 trial, patients with PD-L1-positive tumors did better than the overall population, whereas patients with PD-L1 TPS ≥50% did no better with pembrolizumab than the overall population. The 18-month DFS was 73% in the overall population versus 72% in the TPS ≥50% subgroup.

"Do we have to rethink all the work that we have done in recent years ... defining the cutoff of 50% as the cutoff of efficacy for pembrolizumab?" Reck asked at the presentation.

Moreover, patients in the placebo arm did numerically better if they had tumors with TPS ≥50% (18-month DFS of 70% vs 64% in the overall population).

"This is also something which is in contrast to our previous clinical experience," he continued.

In summarizing the seeming contradictions regarding the predictive and prognostic value of PD-L1, Reck added, "We have to be very honest. We are talking about very early data. We are talking about an interim analysis and we are talking about very early cutoffs for efficacy. We have to wait for more mature data."

A second elephant: PEARLS/KEYNOTE-091 showed a numerical advantage with pembrolizumab in the small group of patients with EGFR mutations, a clear contrast with historical data that showed a detrimental impact of immunotherapy in such patients.

A third statistical pachyderm: The appearance of a detrimental impact of pembrolizumab in the small group of patients who did not receive adjuvant chemotherapy (HR 1.25 vs placebo).

Reck described this as a "very large confidence interval and very immature results, but this is something of clinical relevance because we need to know how to proceed with accurate immunotherapy in those patients who are not capable of adjuvant chemotherapy."

What about the "somewhat convincing data" that DFS is a good surrogate endpoint for OS?

"This data of correlation has been generated only for adjuvant chemotherapy. We do not have this quality of correlation for perioperative immunotherapies or targeted therapies," Reck noted.

Alluding to the recent approval of nivolumab (Opdivo) as neoadjuvant therapy in early NSCLC, Reck cautioned that comparisons between neoadjuvant and adjuvant, or perioperative, immunotherapy have yet to be performed, and "there are multiple arguments for one or the other approach." Beyond that data void, a switch to neoadjuvant treatment for early NSCLC would require a "fundamental change of our oncologic treatment schedule," said Reck, pointing to a recent analysis of the National Cancer Database showing that fewer than 3% of patients with early lung cancer receive neoadjuvant treatment.

Finally, the German lung cancer specialist joined a long list of cancer researchers calling for more research to identify better biomarkers to guide treatment decisions.

"I do think we are entering the times of perioperative immunotherapies," he said. "We are seeing the first signals of efficacy for adjuvant immunotherapy. We do need more follow-up and maturation of data, and we have got new questions from the PEARLS trial. I think the future of perioperative immunotherapy might become very, very challenging."

Comment