The Miller Lab The Miller Lab The Miller Lab
  • Our Approach
  • Meet The Team
  • Publications
  • Presentations
  • Posts
  • Software
  • Funding
  • Slides
  • Donate

Comparing First-Line Immunotherapy in Advanced Melanoma: No Clear Winner?

Yoshine Saito

David M. Miller

Isaac Brownell

2025-02-15

Perspectives on the Science · Journal of Cutaneous Oncology

Comparing First-Line Immunotherapy in Advanced Melanoma: No Clear Winner?

An indirect comparison of nivolumab plus relatlimab and nivolumab plus ipilimumab suggests similar efficacy but markedly different toxicity—while leaving treatment selection dependent on disease biology, intracranial activity, patient preference, and the limitations of cross-trial inference.

Advanced Melanoma Nivolumab + Relatlimab Nivolumab + Ipilimumab Indirect Treatment Comparison Treatment Selection
Journal Journal of Cutaneous Oncology
Published February 15, 2025
Article type Perspectives on the Science · Volume 3, Issue 1

Citation

Saito Y, Miller DM, Brownell I. Comparing first-line immunotherapy in advanced melanoma: No clear winner? Journal of Cutaneous Oncology. 2025;3(1). doi:10.59449/joco.2025.02.15.

View DOI Read Full Article

The featured comparison

The featured study used patient-level data from RELATIVITY-047 and CheckMate 067 to compare first-line nivolumab plus relatlimab with nivolumab plus ipilimumab in advanced melanoma. Because no head-to-head randomized trial exists, the investigators used propensity score weighting and inverse probability of treatment weighting to balance measured baseline differences between the two trial populations.

The analysis evaluated progression-free survival, overall survival, confirmed objective response, duration of response, melanoma-specific survival, and treatment-related adverse events.

Central Perspective

The analysis did not identify a clear efficacy winner. Nivolumab plus relatlimab appeared substantially less toxic, but the absence of randomization means the study should refine—not replace—clinical judgment.

Weighted populations

Nivo/Rela 339

Effective sample size

Patients from RELATIVITY-047 after inverse probability weighting.

Nivo/Ipi 297

Effective sample size

Patients from CheckMate 067 after covariate balancing.

Primary method IPTW

Cross-trial adjustment

Measured baseline characteristics were reweighted to improve comparability.

Validation Nivo

Monotherapy-arm check

The nivolumab-alone arms were compared internally to test the analytic approach.

Efficacy findings

36-month PFS 36% vs 39%

Nivo/Rela versus Nivo/Ipi

The hazard ratio was 1.08, with confidence intervals spanning no difference.

Confirmed ORR 48% vs 50%

Nearly identical response rates

The odds ratio did not demonstrate a statistically significant difference.

36-month OS 57% vs 57%

Equivalent observed survival

Overall survival appeared similar after weighting.

36-month MSS 65% vs 62%

No clear melanoma-specific advantage

The confidence interval again included no difference.

The Efficacy Read

Across response and survival outcomes, the two regimens looked more similar than different. The study supports therapeutic equipoise rather than a definitive hierarchy.

The toxicity difference

Grade 3–4 TRAEs 23% vs 61%

Lower with Nivo/Rela

High-grade treatment-related toxicity was substantially less frequent.

Discontinuation 17% vs 41%

Fewer toxicity-related stops

Any-grade adverse events leading to treatment discontinuation were less common with Nivo/Rela.

Lower any-grade toxicity 5 systems

Endocrine, GI, hepatic, renal, skin

Multiple organ systems showed lower treatment-related adverse-event rates.

Clinical tradeoff Less toxic

But not automatically preferred

Tolerability is only one component of first-line treatment selection.

Why the study was not practice-changing for everyone

  • The findings largely aligned with how many clinicians were already approaching treatment selection.
  • Indirect comparisons cannot eliminate unmeasured differences between trials.
  • Clinical choices depend on disease tempo, metastatic distribution, comorbidity, patient preference, and treatment goals.
  • Evidence for intracranial activity remains much stronger for nivolumab plus ipilimumab.
  • The CheckMate 067 ipilimumab dose may overstate toxicity relative to lower-dose contemporary regimens.

When the regimens may diverge clinically

Brain metastases

Journal club participants overwhelmingly favored nivolumab plus ipilimumab because of its established intracranial activity and the limited evidence for nivolumab plus relatlimab in this setting.

High toxicity concern

Nivolumab plus relatlimab may be more attractive when comorbidity, frailty, or patient preference makes severe immune toxicity especially consequential.

Rapidly progressive disease

Some clinicians may still prefer nivolumab plus ipilimumab when seeking the most intensive established immune regimen for aggressive disease.

Lower-dose ipilimumab strategies

The comparative safety advantage of nivolumab plus relatlimab may narrow when nivolumab plus ipilimumab is delivered with lower-dose ipilimumab.

The Brain Metastasis Exception

“No clear winner” applies most comfortably to extracranial first-line disease. When active brain metastases are present, the evidence base still favors nivolumab plus ipilimumab.

The biomarker question

LAG-3 expression could theoretically help distinguish patients more likely to benefit from nivolumab plus relatlimab, but its predictive value remains uncertain. Routine LAG-3 testing has not been adopted broadly, and the CheckMate 067 cohort did not provide a directly comparable biomarker framework.

Future treatment selection may depend less on choosing one regimen for all patients and more on integrating molecular, immunologic, and anatomic features into individualized decisions.

What the methodology adds

  • Patient-level data permit more credible adjustment than published aggregate comparisons alone.
  • Inverse probability weighting can improve balance across independently conducted trials.
  • Sensitivity analyses and monotherapy-arm validation strengthen confidence in the analytic pipeline.
  • Indirect treatment comparisons can help address questions unlikely to receive a dedicated randomized trial.
  • They remain observational analyses and cannot fully reproduce the protection of randomization.

Why transparency matters

The statistical methods were rigorous but difficult for many clinicians to evaluate directly. Most journal club participants had never personally performed a propensity score analysis, and a substantial minority reported limited comfort with the methodology.

Open code, expanded methods, and synthetic replication data could make analyses like this easier to interrogate, reproduce, and translate into practice without compromising patient-level confidentiality.

The Open-Science Opportunity

Complex methods gain clinical influence when readers can see how the model was built, test its assumptions, and reproduce the analysis. Transparency is not an accessory to indirect comparison—it is part of its credibility.

A practical treatment framework

  • Favor nivolumab plus ipilimumab when intracranial activity is a dominant concern.
  • Favor nivolumab plus relatlimab when reducing severe toxicity is especially important.
  • Consider disease tempo, metastatic pattern, BRAF status, comorbidity, and patient priorities.
  • Recognize that lower-dose ipilimumab regimens may alter the toxicity comparison.
  • Avoid treating an indirect comparison as equivalent to a randomized head-to-head trial.
  • Use shared decision-making when efficacy appears similar but treatment burden differs substantially.
The Practical View

Both regimens remain reasonable first-line options. The best choice is not determined by a single hazard ratio, but by how efficacy uncertainty, toxicity, central nervous system activity, treatment logistics, and patient preferences intersect in an individual case.

Authors

Yoshine Saito, David M. Miller, and Isaac Brownell.

This page highlights selected elements of the published perspective rather than reproducing the complete article. See the original publication for the full journal-club survey, figures, statistical discussion, methods, disclosures, and references. This site represents our opinions only. See our full Disclaimer and Terms of Use Agreement.