Personalizing Immunotherapy in NSCLC Using PD-L1 and Tumor Mutation Burden as Biomarkers

San Francisco, CA—Tumor expression of PD-L1 has consistently predicted ­response and survival outcomes in non–small-cell lung cancer (NSCLC), whereas the role of PD-L1 in immune cells is unclear, said Edward B. Garon, MD, Director, Thoracic Oncology Program, David Geffen School of Medicine, University of California, Los Angeles, at the 2019 ASCO-SITC Clinical Immuno-Oncology Symposium.

Tumor mutation burden (TMB) also consistently predicts objective response rate, as well as progression-free survival (PFS) in NSCLC.

PD-L1 Expression

In the phase 1 KEYNOTE-001 study, patients with PD-L1 expression on at least 50% of the tumor cells were 3 to 4 times more likely to have a response to pembrolizumab (Keytruda) than those with no PD-L1 expression. This finding was true in the total study population, for those who had received previous treatment, and in the treatment-naïve population.

In addition, the PFS and overall survival (OS) were better among patients who had PD-L1 of at least 50% compared with the other patients.

This correlation was validated as part of the KEYNOTE-024 study, which enrolled 305 patients with NSCLC, all of whom had high PD-L1 expression. Patients were randomized to pembrolizu­mab or to a platinum-based dual-chemotherapy treatment. The PFS favored pembrolizumab, and although crossover from chemotherapy to pembrolizumab was allowed at disease progression, the OS still favored the pembrolizumab arm.

“There has been a great deal of debate as to the generalizability of PD-L1 expression,” Dr Garon said.

In the CheckMate-017 study, which compared nivolumab (Opdivo) with docetaxel (Taxotere) in patients with NSCLC, patients benefited from nivolumab versus docetaxel, regardless of PD-L1 expression, although numerically outcomes “did look better in the PD-L1–expressing patients,” he said.

Among patients with nonsquamous NSCLC in the CheckMate-057 study, however, a clear difference in outcomes was seen by PD-L1 expression. In patients with PD-L1 expression >10%, nivolumab demonstrated superior OS to docetaxel. “Nearly the entirety of the OS benefit was in patients who had staining for PD-L1 in ≥10% of cells,” Dr Garon said.

In the randomized, open-label pivotal OAK trial, which compared atezolizu­mab (Tecentriq) with docetaxel in patients with NSCLC, the OS superiority of atezolizumab was similar in the intent-to-treat population and in the subset of patients with PD-L1 expression.

“One thing to note about the atezo­lizumab biomarker is it’s different from the other PD-L1 biomarkers that have been assessed,” Dr Garon said. “They looked for PD-L1 not just on the tumor cells, but also on the immune cells.”

PD-L1 expression may be used to select responders to atezolizumab only when PD-L1 expression is in the top 15%, he added.

In the OAK study, a strong positive result was observed with atezolizumab in patients with PD-L1 expression of ≥50%. It showed no advantage over chemotherapy in patients with lower PD-L1. Including these patients with those with high PD-L1 expression, therefore, weakens the predictive capability of the PD-L1 biomarker, Dr Garon argued.

One concern with the use of immune-infiltrating PD-L1 cells is that “if you show multiple pathologists the same slide, they generally do not agree,” Dr Garon said. “That obviously is a problem for a biomarker.”

The 50% cutoff point for PD-L1 expression in NSCLC is reasonable, he suggested. In an exploratory analysis of the KEYNOTE-024 study, pembroliz­umab improved OS versus chemotherapy in patients with untreated advanced or metastatic NSCLC and a PD-L1 expression of ≥50%.

The difference in OS between groups was not significant in those with a PD-L1 expression of ≤49%. “Certainly in my practice, this is not a group where I would look for single-agent PD-1 inhibition, and instead would look at the immunochemotherapy combinations,” Dr Garon added.

Tumor Mutation Burden

In the KEYNOTE-001 study, higher TMB was associated with durable clinical benefit in patients with NSCLC who received treatment with pembro­lizumab. In the CheckMate-026 study, which compared front-line nivolumab versus chemotherapy, a retrospective analysis showed that patients with high TMB had superior PFS in the nivolu­mab arm; by contrast, in those with low or medium TMB, chemotherapy outperformed nivolumab.

Despite the improvement in PFS by TMB, TMB had no effect on the OS. This finding was replicated in a study of nivolumab plus low-dose ipilimu­mab (Yervoy) compared with chemotherapy, in which the hazard ratio for OS in favor of the combination therapy was 0.77 in patients with TMB ≥10 mutations compared with 0.78 in patients with TMB <10 mutations. TMB as a biomarker “won’t be coming to the clinic any time soon,” Dr Garon predicted.

Nearly 50% of the patients who received treatment in clinical trials did not have enough tissue available for ­assessment of TMB, Dr Garon noted. Furthermore, TMB values obtained by different laboratories are not necessarily reproducible. Blood TMB addresses some of the issues with a lack of tissue availability, but it does introduce additional variability based on shedding of DNA from tumors, he concluded.