In May, we reported on an FDA alert that came after the data monitoring boards of two phase III trials — KEYNOTE-361 (Merck) and IMvigor130 (Genentech) — found that previously untreated metastatic urothelial cancer patients with low expression levels of programmed death ligand 1 (PD-L1) had worse survival with single-agent checkpoint inhibition compared with those on standard chemotherapy. In this report, we follow up on what has happened since this initial communication.
Subsequent to the alert, the agency rolled back first-line indications for pembrolizumab (Keytruda) and atezolizumab (Tecentriq), warning against their use in untreated PD-L1-low metastatic urothelial cancer patients who would otherwise be eligible for a standard chemotherapy regimen. The labels of the two drugs were updated to reflect these findings.
In interviews with MedPage Today, Petros Grivas, MD, PhD, clinical director of the genitourinary cancers program at the University of Washington in Seattle, and Guru Sonpavde, MD, bladder cancer director at the Dana-Farber Cancer Institute in Boston, discussed the clinical ramifications of the FDA’s alert and highlighted questions that have since arisen.
“The initial approval by the FDA was in all comers regardless of expression of PD-L1, at the same time the contingency was that the companies have to run phase III trials to confirm the very promising results and compare those agents with standard platinum-based chemotherapy,” explained Grivas. “So the label changed based on review of unreleased data, which were viewed only by the data monitoring boards of the two trials and the regulatory agencies.”
The three-arm KEYNOTE-361 trial is testing a chemotherapy-alone regimen of gemcitabine plus platinum-based therapy (cisplatin or carboplatin) vs a combination of pembrolizumab plus chemotherapy vs single-agent pembrolizumab (now just in PD-L1-high patients). IMvigor130 is a similarly designed three-arm trial, but instead uses atezolizumab as the checkpoint inhibitor.
In cisplatin-ineligible patients, upfront immunotherapy is still indicated for those with high PD-L1 expression levels, though testing this is not always straightforward, timely, or feasible.
“I’m having to think about doing a PD-L1 assay before considering first-line pembrolizumab or atezolizumab, because these two drugs lost to platinum-gemcitabine in PD-L1-low patients in this first-line setting,” said Sonpavde.
Delays in obtaining results can present an obstacle for physicians looking to start a patient on a given treatment, especially if the tumor needs to be sent out to an off-site lab. Inadequate tissue samples requiring the need for repeat biopsy can be a further source of delays.
Another potential hurdle is that the two checkpoint inhibitors each require a different companion assay, specific to the agent; it is critical not to use a different assay instead.
“Institutions don’t do any antibody assay you want,” Sonpavde said. “They have one antibody they kind of get familiar with and do that for everyone, and so that’s a challenge too.”
The Ventana SP142 PD-L1 assay is used for atezolizumab and Agilent/Dako 22C3 with combined positive score (CPS) interpretation is used for pembrolizumab.
“Some patients are going downhill fast and you don’t have a couple of weeks to get these assays — sometimes it can take more than a couple weeks to get this test done if the tumor is sitting in a different facility,” Sonpavde said. “Practically speaking, if the patient is going downhill fast and I cannot get the PD-L1 result quickly — meaning within a week — then I might just end up instituting the default, or the standard of gemcitabine-carboplatin for patients who are cisplatin ineligible.”
In its series of alerts, the FDA created a new group of patients where checkpoint inhibition is still indicated in first-line regardless of the tumor’s PD-L1 expression levels, those that are “not eligible for any platinum-containing therapy” — i.e., cisplatin or carboplatin.
“If someone in the U.S. is not fit enough to get even carboplatin for whatever medical reason, in that scenario you do not need to check PD-L1 status in tumor tissue,” explained Grivas.
But while patients who are cisplatin-ineligible are a well-defined group based on previously defined consensus criteria, including those with one or more of the following factors — ECOG performance status ≥2, poor kidney function (creatinine-clearance <60 mL/min), grade ≥2 neuropathy, among others -- Sonpavde said the agency did not define what specifically makes a patient platinum ineligible.
“Basically, it’s up in the air, we don’t have a uniform definition for defining platinum ineligibility,” he said, but noted that experts are trying to form a consensus on who these patients are.
Sonpavde said that presumably, patients with multiple factors or comorbidities that individually would have ruled out cisplatin might be considered “bad enough” that they won’t tolerate gemcitabine-carboplatin either.
“It’s a very gray space and not very well defined yet,” agreed Grivas.
Grivas suggested that patients that might fall into this category would be those that are too frail and at high risk for complications — those with ECOG 3 performance status, grade ≥2 neuropathy, multiple comorbidities, very poor kidney function, and with a lack of social and nutritional support, among other factors.
“So far we never had the need to objectively define that in urothelial cancer,” said Grivas.
He pointed to a recent American Society of Clinical Oncology (ASCO) guideline on chemotherapy eligibility in geriatric patients that can help identify who might be unfit for chemotherapy.
“It may be broader than platinum, it may apply to any chemotherapy, and raises issues about the frailty, resilience and robustness of patients regarding their ability to get chemotherapy, so I think those criteria are very relevant to this discussion beyond the platinum-particular context,” said Grivas, adding that if a patient lacks the fitness for carboplatin then this would probably apply to most chemotherapy agents as well.
While Grivas said that immunotherapy is considered more tolerable than chemotherapy and that there’s certainly a role in patients unfit for platinum-containing regimens, he pointed out that the clinical trials for checkpoint inhibitors did not test the agents in very frail populations, such as those with an ECOG 3 performance status or very poor kidney function with estimated creatinine clearance <30 mL/min.
“My note of caution is that we have to be mindful of which patients we give checkpoint inhibitors to, in terms of the degree of frailty and the very poor performance status, where exactly they stand in the spectrum of terminal-stage disease,” he said.
For a patient in the intensive care unit setting, for example, some clinicians might attempt to use these agents in a “Hail Mary” approach.
“If someone is in that terminal stage, pre-hospice, it’s very unlikely that the trajectory of the disease might meaningfully change with a checkpoint inhibitor,” said Grivas. “I think that’s another layer of discussion about proper utilization of those agents and, unfortunately, data is lacking in this setting.”
He noted that relevant discussions may occur in future meetings (e.g., the 2019 Genitourinary Cancers Symposium), which would help further define who the platinum-ineligible patients are. He also emphasized that the FDA alert and label change for atezolizumab and pembrolizumab applied only to the frontline setting of cisplatin-unfit patients, and not in platinum-refractory patients, where there remains no need to check PD-L1 expression before offering a checkpoint inhibitor.
Need for More Data
While treating PD-L1-low patients with single-agent checkpoint inhibitors in the frontline setting has to a degree been clarified by the FDA’s alert, the best option for PD-L1-high patients has not yet been defined.
Four phase III trials will help answer this question and others, the already-described KEYNOTE-361 and IMvigor130 trials, and two trials of anti-CTLA-4 plus anti-PD-1/PD-L1 combinations compared with standard chemotherapy: Checkmate 901 with ipilimumab (Yervoy) plus nivolumab (Opdivo) and DANUBE with tremelimumab plus durvalumab (Imfinzi).
The oncology community will have to wait until these trials read out — the first of which will be DANUBE — to find out whether adding PD-1/PD-L1 to chemotherapy or whether a chemo-free anti-CTLA-4 plus anti-PD-1/PD-L1 combination can improve outcomes over standard upfront gemcitabine plus platinum therapy, said Sonpavde.
“Then we have to see what regimen might be of more benefit for which patient,” he said. “It’s going to be rapidly evolving over the next year.”
Other data to watch for include phase II and III trials of the antibody-drug conjugate enfortumab vedotin in the post-chemotherapy and post-immunotherapy setting, and a phase III trial of the FGFR inhibitor erdafitinib for patients with FGFR3 alterations. Both of these agents received FDA breakthrough designation for urothelial cancer, showing responses in the 40% range among pretreated patients. And two switch maintenance trials are testing pembrolizumab (Keytruda) and avelumab (Bavencio), respectively, in patients without progression after frontline platinum-based chemotherapy.
Implications for Future Trials
Grivas pointed out that the FDA alert also has implications for future clinical trials.
“If you have a trial that is a single-arm phase II study with a combination of a PD-L1 or PD-1 inhibitor plus something else, I personally do not feel that we need to select patients based on PD-L1 because it’s a combination,” he said. “We do not know the relevance or clinical utility of PD-L1 in these combination therapies.”
But he said that for any study with a single-agent checkpoint inhibitor in the first-line setting, regulatory agencies would likely mandate PD-L1 testing.
“There’s definitely a need, an urgent need, to design those studies to improve upon checkpoint inhibitors alone in that setting, because most patients do not have an objective response,” he said.
Grivas also said that better biomarkers are needed to help clarify which patients should get immunotherapy, and that clinical trials present an opportunity to do this. He suggested that while PD-L1 might serve as a backbone, a composite biomarker in the future might include various putative biomarkers, such as tumor mutational burden, gene-expression profiling, DNA-repair gene mutation, circulating tumor DNA, microsatellite instability (MSI) status, and T-cell clonality and diversity, among others, to better guide treatment selection.
“Except for MSI status across tumor types, none of those other biomarkers have shown great clinical utility in urothelial cancer, as of yet, so they’re not being used in clinical practice, but there’s definitely a need to discover and prospectively validate biomarkers; that’s what we need to do better,” he said.
Grivas has done consulting with Genentech, Merck & Co., Pfizer, Bristol-Myers Squibb, AstraZeneca, Biocept, Clovis Oncology, EMD Serono, Seattle Genetics, Foundation Medicine, Driver Inc., QED Therapeutics, Heron Therapeutics, and Janssen within the past 2 years. He was previously involved in an educational, unbranded program with Genentech and Bristol-Myers Squibb. His institutions have received research support from Genentech, Bayer, Merck & Co., Mirati, OncoGenex, Bavarian Nordic, Pfizer, AstraZeneca, Clovis Oncology, and Immunomedics.
Sonpavde has reported consultant relationships with and/or institutional research support from Amgen, AstraZeneca, Astellas, Bavarian Nordic, Bayer, Boehringer-Ingelheim, Bristol-Myers Squibb, Eisai, EMD Serono, Exelixis, Genentech, Janssen, Merck, Novartis, Pfizer, and Sanofi.