Where Does LAG-3/PD-1 Inhibition Fit In PD-1–Refractory Hodgkin Lymphoma?

PD-1 blockade has become a cornerstone of treatment for patients with relapsed or refractory classic Hodgkin lymphoma (cHL), yet most patients will eventually progress on this approach, leaving the optimal post–PD-1 therapy undefined. The LAG-3 inhibitor favezelimab combined with pembrolizumab (Keytruda) demonstrated promising efficacy in anti–PD-1–refractory cHL in a phase 1/2 study, establishing the rationale for the randomized phase 2 KEYFORM-008 trial (NCT05508867).

CancerNetwork® spoke with Guilherme Fleury Perini, MD, PhD, first author of the KEYFORM-008 study and investigator at Instituto Américas and Einstein Hospital Israelita in São Paulo, Brazil, about the design and results of the trial, which were presented at the European Hematology Association (EHA) Congress 2026.

Perini began by describing the scientific rationale for combining favezelimab with pembrolizumab in PD-1–refractory cHL and outlining the study design and patient population. He then reviewed the key efficacy findings, noting the seemingly paradoxical relationship between the higher complete response (CR) rate observed with chemotherapy and the superior progression-free survival (PFS) and duration of response (DOR) seen with favezelimab/pembrolizumab.

Perini next addressed the safety profile of the combination and the experience of patients who crossed over from chemotherapy to favezelimab/pembrolizumab upon progression. He concluded by contextualizing the sponsor's decision to discontinue favezelimab's clinical development program and sharing his perspective on the future of dual LAG-3/PD-1 inhibition in this population.

CancerNetwork: What was the rationale for initiating KEYFORM-008, and what were the key design features and patient characteristics of the trial?

Perini: KEYFORM-008 was a trial that compared the combination of pembrolizumab with favezelimab in patients who had already [progressed after] PD-1 inhibition. We know that checkpoint inhibitors are standard of care for relapsed/refractory Hodgkin lymphoma, and these agents are now utilized even in earlier lines of therapy. As a result, most patients will eventually be exposed to a PD-1 inhibitor, and when it is used as monotherapy, as continuous therapy, or for at least 2 years, some patients will eventually [experience progression] and lose response to pembrolizumab. One of the mechanisms of resistance to checkpoint inhibitors is upregulation of other immune checkpoints, including LAG-3. LAG-3 leads to T-cell exhaustion, so the idea of the trial was to inhibit the LAG-3 axis and thereby regain sensitivity to pembrolizumab. This concept was first tested in a phase 1/2 study, and those results were very promising, which led to KEYFORM-008, a randomized, open-label, international phase 2 trial.

In KEYFORM-008, patients were [randomly assigned] 1:1 to coformulated favezelimab 800 mg/pembrolizumab 200 mg intravenously every 3 weeks for up to 35 cycles, or investigator's choice of chemotherapy, includingeither bendamustine [Treanda] or gemcitabine as monotherapy, with flexible schedules and doses. The primary endpoint was progression-free survival [PFS] by investigator assessment, with key secondary end points including overall survival [OS], duration of response [DOR], and safety.

This was a heavily pretreated population. The median number of prior lines of therapy was 5 [range, 2-12], all patients were refractory to PD-1 inhibition, approximately 90% had prior exposure to brentuximab vedotin [Adcetris], and approximately half had [progressed following] prior autologous stem cell transplant [ASCT]. These were truly patients who were running out of options.

What were the key efficacy findings, and how should clinicians interpret the differences between the CR data favoring chemotherapy and the PFS and DOR data favoring favezelimab/pembrolizumab?

This was a positive trial, the primary end point of PFS was met, with favezelimab/pembrolizumab demonstrating superior PFS compared with chemotherapy [median 6.4 months vs 5.7 months; HR, 0.757; P = .0462]. What was interesting is that we observed a higher rate of complete responses with chemotherapy [36.4% vs 13.5%]. That said, the overall disease control rate, which includes CRs, partial responses, and stable disease, was similar between the 2 groups [67.3% vs 71.7%]. We saw more partial responses with favezelimab/pembrolizumab, which is consistent with what we typically observe with checkpoint inhibition.

Duration of response, however, was meaningfully longer with favezelimab/pembrolizumab [median 11.0 months vs 5.4 months]. This tells us that while chemotherapy can produce deep responses in these patients, those responses are short-lived. The longer DOR with favezelimab/pembrolizumab is a proof of concept that when you block the LAG-3 axis, you can regain sensitivity to checkpoint inhibition. For this population — heavily pretreated patients with a median of 5 prior lines of therapy who have truly exhausted most options — PFS and DOR are the most meaningful endpoints. A partial response that does not progress is more valuable than a complete response that progresses after 2 or 3 months.

What toxicities were observed with favezelimab/pembrolizumab, and what did the crossover data show?

The toxicity profile was consistent with what we see with checkpoint inhibition alone. We did not observe any new safety signals with the addition of favezelimab. The most common treatment-related adverse [effect] was hypothyroidism [15.4%]. We did not see a higher rate of immune-mediated adverse [effects] overall [grade 3 or 4: 4.8%], the rate of discontinuation due to treatment-related adverse [effects] was low [12.5%], and there were no treatment-related deaths in the favezelimab/pembrolizumab arm. This demonstrates that you can regain efficacy without adding meaningful toxicity.

Regarding the crossover component: patients in the chemotherapy arm who experienced progression per blinded independent central review were offered the option to cross over to favezelimab/pembrolizumab. Of 99 patients in the chemotherapy arm, 52 crossed over and received the combination. In that crossover population, the objective response rate was 42.3%, consistent with the results observed in the primary favezelimab/pembrolizumab arm. Patients coming out of chemotherapy after multiple prior lines of therapy did experience somewhat more diarrhea and some additional hematologic adverse events in the crossover phase, but these findings are interpretable in the context of that treatment history and should not be attributed to the combination itself.

Given the decision to discontinue favezelimab's clinical development program, what is your perspective on the future of dual LAG-3/PD-1 inhibition in this population?

As an optimist, I believe the data are meaningful. We are talking about a truly heavily pretreated population, some patients had received up to 13 prior lines of therapy before enrolling in this trial, and the disease control rate was genuinely good. The safety profile was something that stood out: you do not add meaningful toxicity when you combine LAG-3 and PD-1 inhibition. The decision to discontinue favezelimab was a business decision by the sponsor, not driven by safety concerns or lack of efficacy signal.

That said, we still have work to do for these patients. The population of patients with heavily refractory cHL is becoming smaller over time as we cure more patients in the first-line setting and achieve better responses with first salvage and transplant. However, we will still encounter these very refractory patients, and we owe them effective options. Regaining sensitivity to PD-1 inhibition is a compelling strategy, and dual LAG-3/PD-1 blockade has demonstrated it can work, but making this approach available to patients remains the challenge.

Reference

Perini GF, Lavie D, Bröckelmann PJ, et al. Coformulated favezelimab/pembrolizumab versus chemotherapy in participants with relapsed or refractory classic Hodgkin lymphoma: results of the randomized phase 2 KEYFORM-008 study. Presented at: European Hematology Association 2026 Congress; June 11-14, 2026; Stockholm, Sweeden. Abstract S220.