Expert Compares Data for Immunotherapy Versus Chemotherapy In GEJ Cancer

In an interview with <em>Targeted Oncology</em>, Daniel Catenacci, MD, discussed the clinical trials supporting immunotherapeutic options for patients with GEJ cancer. He also highlighted how physicians can sequence these agents appropriately in their patients.

Daniel Catenacci, MD,

From the frontline to later-line settings, investigators have evaluated the role of immunotherapy as treatment for patients with gastroesophageal junction (GEJ) cancer in various clinical trials. At the 2019 International Society of Gastrointestinal Oncology (ISGIO) Annual Conference, Daniel Catenacci, MD, reviewed data from these trials and the impact they have had in the GEJ cancer space, be it in the first-, second-, or third-line settings.

Most recently, data from the phase III KEYNOTE-062 trial evaluated pembrolizumab (Keytruda), versus chemotherapy in patients with PD-L1—positive, HER2-negative advanced gastric or GEJ cancer. These data, presented at the 2019 ASCO Annual Meeting, showed that immunotherapy led to an improvement in overall survival (OS) in patients with high levels of PD-L1 expression. However, the data were not statistically significant.

Patients were randomized to 1 of 3 arms in this clinical trial: pembrolizumab monotherapy, pembrolizumab plus chemotherapy, or chemotherapy alone. The median OS was superior in the pembrolizumab monotherapy arm for patients who had a combined positivity score (CPS) ≥10 compared with the chemotherapy arm (17.4 [95% CI, 9.1-23.1] vs 10.8 months [95% CI, 8.5-13.8]; HR, 0.69; 95% CI, 0.49-0.97). However, a longer progression-free survival was seen with chemotherapy (4.1 months) compared with pembrolizumab (1.5 months). The secondary endpoint, objective response rate, was also higher in the pembrolizumab arm (15.8% versus 13.6%).

The rate of serious adverse events (AEs) was 16.9% with pembrolizumab, 69.3% with chemotherapy alone, and 73.2% in the combination arm. The most common AEs included nausea and fatigue, and 22% of patients experienced immune-related toxicities. The safety profile is consistent with previous reports of pembrolizumab.

In an interview withTargeted Oncology, Catenacci, associate professor, University of Chicago, discussed the clinical trials supporting immunotherapeutic options for patients with GEJ cancer. He also highlighted how physicians can sequence these agents appropriately in their patients.

TARGETED ONCOLOGY: What did you discuss at the 2019 ISGIO Annual Conference?

Catenacci:The talk that I gave was an overview of the novel immunotherapy drugs that we have in gastroesophageal cancer and how to implement some of the trials and the data to incorporate [these drugs] into our routine practice. What we talked about was a number of studies in various lines of therapy.

Starting in the third-line setting, we talked about the pivotal KEYNOTE-059 study, which was a nonrandomized study of 259 patients in the third-line or higher setting that received pembrolizumab monotherapy. That study showed a response rate of around 11% in all those patients. But when we used the biomarker for PD-L1 expression by CPS, we were able to enrich for those patients who were positive by that score of a 1 or higher, and those patients had a response rate of 15% to 16%. Ultimately, that led to a conditional approval just over 2 years ago for [pembrolizumab as a] third-line therapy in those that are PD-L1—positive by this measure.

Interestingly, once you look at the microsatellite instability (MSI) subgroup, which is not a common subgroup, but we already knew those patients would have a high response rate in the 40% to 50% range, after you exclude those patients and you just look at those microsatellite stable patients that are PD-L1—positive, the response rate goes down a little bit, as you might expect, to around 13.3%. The reason this led to an approval was because in the patients who are responding, it is a long duration of response. The regulatory bodies took that into account, and that led to an approval.

TARGETED ONCOLOGY: Are there other trials in the third-line setting that you discussed?

Catenacci:After focusing on KEYNOTE-059, we looked at other studies in the third-line setting, including the ATTRACTION-2 study, which was an all-Asian study of nivolumab (Opdivo) in all-comers in the third-line setting or higher, randomized to placebo. That study showed an OS benefit in all-comers and in the PD-L1 analysis of a different biomarker scoring system, the different antibody didn’t seem to help enrich for patients who would do better. Again, I think if this was done with a CPS scoring, it would help to enrich and benefit. To date, in the United States, because that was an Asian study, nivolumab is not approved in the third-line setting here.

We also talked about our third study, the JAVELIN 300 study, which was a randomized study of avelumab (Bavencio), a PD-L1 inhibitor, to physician’s choice of chemotherapy, whether it be irinotecan or taxane, whichever had not been used in prior treatment. Unfortunately, that study did not show benefit, even in a PD-L1—positive subgroup. At this time, including with the novel drug TAS-102 (Lonsurf), the oral agent in the third-line or higher setting, we now have a number of options available in that setting, whether it is chemotherapy, TAS-102, irinotecan, taxane, and now PD-L1–positive, CPS 1 or higher with pembrolizumab. The optimal sequencing of that is unknown, but the 1 randomized trial, JAVELIN 300, suggests that, at least in some patients, chemotherapy is still better than immunotherapy. There is still more work to do in terms of who it is that we should be giving checkpoint blockade to.

TARGETED ONCOLOGY: What trials did you discuss in the second-line setting?

Catenacci:The next thing we did was talk about second-line therapy, where we had the KEYNOTE-061 study, which unfortunately was a negative study overall in all-comers. In PD-L1—negative patients, about a third of patients did worse than chemotherapy, so that arm of the study was closed early and terminated. For CPS 1 or higher patients, there was what I’ve been calling now a yin-yang plot because there’s a bunch of patients at the beginning who do worse with immunotherapy compared with chemotherapy, but then, later on, the curves flip and you see benefit in subsets of patients with the immunotherapy compared with chemotherapy. That suggests that we have our work cut out to find out who those patients are by better biomarkers. We do know from that study that MSI-high (MSI-H) tumors are in that subgroup, so that is a novel biomarker that we already know to use in the second-line setting.

The next study we looked at was the KEYNOTE-181 study, which was an esophageal cancer study with squamous cell and adenocarcinoma. That study had 3 primary endpoints, including the whole all-comers, both histologies in any PD-L1 status, and then looking specifically at squamous cell tumors, then looking at PD-L1 of 10 or higher for CPS. That study, overall, in the intent-to-treat population didn’t make its endpoint, but in the squamous cell and the above 10 CPS score, that is where the benefit was seen and most pronounced. Recently, a couple months ago, we had an approval for esophageal squamous cell with 10 or higher. Unfortunately, the benefit was not seen in the adenocarcinoma and even in the 10 or higher group.

TARGETED ONCOLOGY: Did you consider any data for immunotherapy in the frontline?

Catenacci:The last study we looked at with respect to immunotherapy was the first-line study KEYNOTE-062, which was a long-awaited 3-arm study of chemotherapy plus placebo versus chemotherapy plus pembrolizumab versus a third open-label arm of pembrolizumab alone. Again, unfortunately, that study of chemotherapy with or without the pembrolizumab was negative. It trended to better [with pembrolizumab], but it just wasn’t statistically significant.

In the pembrolizumab monotherapy arm, we had this yin-yang plot again like with KEYNOTE-061 and so many other studies where we see patients at the beginning are doing worse compared with chemotherapy, but there is a subset that benefits and the curves cross. We’ve seen this persistently across different studies, suggesting again that we need to focus on identifying who those patients are that should be getting the immunotherapy and who should be getting chemotherapy.

TARGETED ONCOLOGY: How do you make decisions regarding sequencing?

Catenacci:I make sequencing decisions based on a number of factors. These include what line of therapy we are referring to in the first place, the patient’s characteristics, the burden of disease, their symptoms, and their performance status. Something we focused the talk on was many of these studies that are selecting patients based on a given biomarker that requires central confirmation of this biomarker. It takes several weeks to get that answer for these patients in the first-, second-, and third-line settings. You are already imposing a selection bias by having patients waiting for several weeks before you start treatment. Patients on these studies are not necessarily representing the full body of patients that we see, and in fact, the patients needing therapy almost immediately in any line are better served getting chemotherapy first. We’ve seen this over and over again. This is the yin in those plots where the patients aren’t doing as good in the beginning because there’s not enough to rescue them.

These are things to help us sequence the immunotherapy and optimize every individual patient based on what they are doing, what their characteristics are, and what the biology of the tumors are. For example, if their tumors are MSI-H, these are the patients that we want to move the treatment with immunotherapy to as early as possible because the benefits are large. We don’t want to miss the chance of not being able to get it to them because they progress rapidly on first- or second-line treatment.

In contrast, in people who are negative for PD-L1, we would not recommend using monotherapy checkpoint blockade. If immunotherapy is being considered, we would recommend that to be in the context of a clinical trial. If you’re a low-level PD-L1—positive patient with a CPS from 1 to 10, for example, I would use it judiciously in patients that have a good performance status, don’t have a heavy burden of disease, and don’t need an immediate response, for example, as opposed to patients that are in 1 of those categories, where they have a heavy burden of disease or they have symptomatic disease. They are better served with chemotherapy.

TARGETED ONCOLOGY: What do you hope was taken away from this talk at ISGIO?

Catenacci:I hope that the audience took away that patients should be looked at as individuals for all of the aspects of their care, such as the characteristics of their tumor, characteristics of the patient, other comorbidities. What we’ve learned is what to apply from these various studies to them individually. I think that the vast majority of patients are going to benefit most from palliative chemotherapy. There are subsets of patients like MSI-H, and there are others like that, that we are continuing to try to identify. Some include EBV-positivity, some include extremely high levels of PD-L1 above 50 that may be best served by immunotherapy. These are the patients that we want to get immunotherapy to quickly. There are other novel biomarkers, includingEGFRamplification,METamplification,FGFR2amplification, and further studies are looking into that. What we want everyone to take away from our talk is that this is a heterogenous disease with multiple different molecular biomarkers and drivers. It’s not a one-size-fits-all approach. Everyone’s treatment plan has to be personalized to them.


Fuchs CS, Ozguroglu M, Bang YJ, et al. Pembrolizumab (pembro) vs paclitaxel (PTX) for previously treated advanced gastric or gastroesophageal junction (G/GEJ) cancer: Phase 3 KEYNOTE-061 trial.J Clin Oncol. 36, 2018 (suppl; abstr 4062).