ctDNA Patterns May Predict Outcomes Achieved With mRNA-4157 Plus Pembrolizumab in High-Risk Melanoma

Jeffrey S. Weber, MD, PhD

The addition of a personalized mRNA vaccine to standard-of-care (SOC) pembrolizumab (Keytruda) improved survival outcomes vs pembrolizumab alone in resected high-risk melanoma, and monitoring ctDNA patterns over time can provide valuable insights into the degree of benefit patients can expect to achieve from the approach, according to Jeffrey S. Weber, MD, PhD.

Findings from the phase 2b mRNA-4157-P201/KEYNOTE-942 trial (NCT03897881) of mRNA-4157 (V940) plus pembrolizumab in patients with stage IIIB, IIIC, IIID, or IV cutaneous melanoma were presented at the2023 ESMO Congress. Patients treated with the combination experienced a statistically significant and clinically meaningful improvement in recurrence-free survival (RFS; HR, 0.561; 95% CI, 0.309-1.017; P = .0266) and distant metastasis-free survival (DMFS; HR, 0.347; 95% CI, 0.145-0.828; P = .0063) vs pembrolizumab alone. Notably, higher recurrence rates were observed in molecular nonresponders vs molecular responders; this suggests that ctDNA positivity at baseline or later in the treatment course may serve as a negative prognostic factor for clinical outcomes with the combination.

“KEYNOTE-942 in its initial iteration has a clear signal of efficacy for the combination vs pembrolizumab alone,” said Weber, who serves as the deputy director of the New York University (NYU) Perlmutter Cancer Center, co-director of the Melanoma Research Program, and a Laura and Isaac Perlmutter Professor of Oncology in the Department of Medicine at the NYU Grossman School of Medicine. “We’re going to see many patients on neoantigen vaccine trials in the next couple of years, and this [approach may] change the face of oncology.”

In an interview with OncLive^®, Weber highlighted on key efficacy findings from the mRNA-4157-P201/KEYNOTE-942 trial, expanded on the potential utility of longitudinal ctDNA dynamics as a kinetic marker for clinical outcomes in melanoma, and underscored the importance of continued research on mRNA vaccines paired with standard PD-1 therapy across cancer types.

OncLive: What was the rationale for adding an mRNA neoantigen vaccine to SOC pembrolizumab in patients with high-risk, resected melanoma?

Weber: KEYNOTE-942 was an adjuvant study of a personalized mRNA neoantigen vaccine plus pembrolizumab vs pembrolizumab alone in patients with resected, stage IIIB, IIIC, IIID, and IV melanoma. The rationale for the study was based on investigations done over the past 6 or 7 years. [These studies] have shown that neoantigen approaches, whether it be RNA vaccines or others, can clearly induce neoantigen-specific T-cell responses. Epitope spreading to other antigens can be seen [with these approaches], and responses can be very long-lived. [They also demonstrated that] clinical responses can be seen in a small number of patients who have had metastatic disease, mainly melanoma or lung cancer or other malignancies.

Based on that background, this randomized phase 2 study with a 2:1 randomization of the RNA vaccine with pembrolizumab vs pembrolizumab alone was carried out to test [whether] there was evidence of benefit with the combination for the clinical end points of RFS and DMFS. This would provide the ammunition to proceed on to a large, randomized phase 3 study, which is quite long, costly, and expensive. Generating [positive] data in 150 to 200 patients would encourage us to go ahead with that large study with a very small degree of risk.

What key efficacy findings from this study were presented at the 2023 ESMO Congress?

Both RFS and DMFS were superior in the combination arm vs [the monotherapy] arm. The RFS [improvement] was clinically significant, with a HR of 0.56, reflecting a 44% reduction in the risk of recurrence or death. There was also a clinically and statistically significant 66% reduction in the risk of DFMS, with a HR of 0.35 and a P value of [.0063]. Biomarker data, [which I were presented at the 2023] ESMO Congress, suggested in the past that both tumor mutational burden, PD-L1 staining, and interferon expression signature could be biomarkers associated with a good outcome. Whether [patients are] high or low in those biomarkers, [they] are clinically better off [by receiving treatment with] the combination vs the single agent in terms of RFS and DFMS.

Could you expand on the efficacy observed with mRNA-4157 plus pembrolizumab across BRAF V600E–mutated vs wild-type subpopulations? How does ctDNA status appear to influence the efficacy of this combination acrosssubgroups?

As in many gastrointestinal cancers, ctDNA is a negative prognostic factor. In this study, we showed that ctDNA is a negative prognostic factor because those who have ctDNA negativity [at baseline and throughout the treatment course] did very well. [Conversely], those who have ctDNA positivity at [baseline] or become positive after having been negative [at baseline] do poorly. The intermediate group [comprises] those who were ctDNA positive and become ctDNA negative, or [those who] are ctDNA negative, become ctDNA positive transiently, and then become negative again. That’s the response group. Whether [they receive] the combination or the single agent, [they are] grouped into those categories. For each biomarker category, [the data suggest that patients are] generally better off receiving the combination vs single-agent pembrolizumab.

Finally, if [a patient had] BRAF-mutated disease, there were clear benefits with the combination vs single-agent pembrolizumab. There was benefit [with the combination] if [they had] BRAF wild-type [disease], but the HR was borderline [in favor of the combination,] at [0.808].

A final unique aspect to the study was the late break in the RFS and DMFS curves. This was approximately 1 year for DMFS, and approximately 10 months for RFS. This reflects the fact that the patients don’t get vaccinated until the vaccine is produced, and that’s at least 6 weeks into treatment with pembrolizumab. Additionally, it probably takes a certain number of vaccines to be able to mount an effective immune response. [These are] unique aspects, which do not take away from the fact that we have a very clinically significant benefit for the first time [with] a neoantigen vaccine approach in patients with resected, high-risk melanoma.

Given the potential relationship between ctDNA dynamics and treatment outcomes, how might this information guide clinical decision making for patients with high-risk resectable melanoma?

The question about a biomarker like ctDNA is how does it impact patient care? You wouldn’t perform an assay or test if it will have no impact on the patient’s management, right? That’s a fundamental concept that I hope is taught to every medical student. If ctDNA is negative at baseline, and stays negative, you can assume the patient will do well. Obviously, a predictive marker is useful only at baseline. Even if baseline ctDNA [is negative], it may become positive over time. That’s a kinetic marker, not a predictive marker, and it limits the ability of ctDNA to be useful. ctDNA is prognostic in that positivity at baseline is bad news, while negativity at baseline is generally good news; if it becomes [positive] at any point, those patients will do poorly.

[Overall,] ctDNA has limited utility right now. It has the greatest utility as a pharmacodynamic biomarker [rather than] just a baseline predictive biomarker, meaning it’s something you monitor over time. [For example], if [a patient’s] ctDNA was negative and [becomes positive], that’s a sign that [we] should be switching therapy. If a patient who was considered low risk elected to go on pembrolizumab alone and their ctDNA goes up, [we would] want to add a vaccine or another drug, provided that the adjuvant LAG-3 antibody plus PD-1 antibody studies are positive.

What next steps are planned to further investigate the activity of this combination in this disease?

We now have 2 years of follow-up data, but we’re looking forward to seeing additional follow-up, which will probably be around 32 months [at the] next data cutoff. The next data cut will be event driven, meaning you must have a certain number of recurrences to be able to interpret the data. The good news is that so far, we haven’t reached that point. When we do reach that point, my hopes are that those RFS and DMFS curves will diverge.

The next step is the [phase 3] registrational study, V940-001 [NCT06077760,] which has already started accrual in Australia and a few other countries and will soon come to the United States. It is the same design as KEYNOTE-942, with a 2:1 randomization, but it will include [patients with] stage IIB, IIC, IIIA, IIIB, IIIC, IIID, and IV resected melanoma. I think that study will [be completed] quickly and I have high hopes that it will be positive. If positive, it will [hopefully] lead to the approval of a new adjuvant treatment in melanoma for the first time in 6 years. This [research] has also spawned a variety of other phase 3 adjuvant studies of vaccines with PD-1 blockade in lung cancer, renal cancer, bladder cancer, and head and neck cancer.

Reference

Weber JS, Khattak A, Carlino M, et al. mRNA-4157 (V940) individualized neoantigen therapy + pembrolizumab vs pembrolizumab in high-risk resected melanoma: Clinical efficacy and correlates of response. Ann Oncol. 2023;34(suppl 2):S1288-S1289. doi:10.1016/j.annonc.2023.10.043