However, she noted that most AEs, even severe ones, can be successfully managed. “What we’ve learned from several years of trials now is if you find these problems and you treat them aggressively, usually with steroids in more significant situations, they’re typically very manageable and patients can have reversal of symptoms and feel very well once they’re under control,” she said.
The panel discussed exciting recently published data suggesting these agents can treat brain metastases.3
“In patients with brain metastases that have not been radiated, or those who have been radiated and then have progression, using pembrolizumab instead of another local therapy can be effective in the brain,” said Goldberg.
Goldberg’s ongoing study included 18 patients with melanoma and 18 with lung cancer, with brain metastasis response achieved in 4 (22%) with melanoma and 6 (33%) with NSCLC. There were 4 complete responses in the brain, and the responses achieved appear to be durable.
Although the trial is small, if no other options are available to patients (eg, a patient had radiation and there are toxicity concerns), Goldberg suggested this treatment could be considered outside of a clinical trial with close monitoring and careful multidisciplinary collaboration, including with neurosurgeons and radiation oncologists. “We’ve created a tumor board to discuss these patients [because] they can be complicated,” she said.
Value of PD-L1 Testing
Despite the difference in nivolumab and pembrolizumab approvals, the panelists generally thought these drugs were interchangeable and not all routinely perform PD-L1 testing before prescribing these agents, although this may change with the release of CheckMate 026.
“I think some people think of the two drugs as nivolumab is the one where you don’t have to check for PD-L1 and pembrolizumab is the one you do, but I don’t think there’s an inherent difference in the drugs,” said Goldberg.
Others try to follow the FDA testing recommendations whenever possible. “If we have the biopsy specimen and we’re able to do it, we do test it,” said Chandra P. Belani, MD.“If the PD-L1 is positive, we give pembrolizumab, [and] if not, we give nivolumab.”
However, this discussion took place before Bristol-Myers announced on August 5 that nivolumab failed to improve progression-free survival (PFS) compared with physician’s choice of chemotherapy as first-line therapy in PD-L1–positive NSCLC.
Additionally, the developer of atezolizumab (Genentech) said on August 31 that the agent improved OS compared with docetaxel in a second-line phase III trial regardless of PD-L1 status. No details were released for either study.
The reviewers noted several challenges with PD-L1 testing. A key challenge is that the criteria used to define PD-L1 status have varied greatly. In nivolumab trials, a cutoff ranging from 1% to 10% of positive cells has been used, whereas the cutoff for pembrolizumab has ranged from 1% to 50%, making it unclear what level of expression is needed to qualify a tumor as being PD-L1 positive.
“We don’t routinely test for PD-L1 because we don’t know which is the antibody, which is the cutoff,” said Marina Garassino, MD. “The eyes of the pathologists are also different,” she noted, indicating a need for better training and standardization. Even the use of different PD-L1 reagents can lead to different sensitivities.
Additionally, even when PD-L1 expression is observed, it is not always predictive of response. “What we’ve seen in trials and from personal experience treating patients is when you have a patient where you test for PD-L1 and it’s negative, they can clearly still respond,” said Goldberg. “It seems, in several trials now, the response rate is lower, so there’s less of a chance of response, but they can respond.”
Govindan pointed out that another major confounding factor of PD-L1 testing is that expression of this biomarker is not stable or consistent across tumors. “This is a very dynamic marker, so the PD-L1–negative tumor can become PD-L1–positive and it can be present in one area and not in other areas,” he said. Subsequently, PD-L1 expression can be difficult to define and easy to miss, which might also explain why some patients show response despite testing negative.
Levy summed up that PD-L1 testing is complicated and that it will take some time to determine how best to proceed on this front. “There’s a lot of good work going into trying to define those patients most likely to respond, but, just as important, trying to look at why patients aren’t responding, and I think that’s where combination strategies come in handy,” he said.
Several trials are assessing whether combining immunotherapy with other agents can improve treatment response. One such approach involves combining checkpoint inhibitors with agonists. In the CheckMate 012 trial, this approach was assessed by combining ipilimumab, a CTLA-4 inhibitor, and nivolumab as a frontline treatment for NSCLC.4