Genetic Basis for Clinical Response to CTLA-4 Blockade in Melanoma

Authors

Alexandra Snyder

Vladimir Makarov

Taha Merghoub

Jianda Yuan

Jesse M Zaretsky

Alexis Desrichard

Logan A Walsh

Michael A Postow

Phillip Wong

Teresa S Ho

Travis J Hollmann

Cameron Bruggeman

Kasthuri Kannan

Yanyun Li

Ceyhan Elipenahli

Cailian Liu

Christopher T Harbison

Lisu Wang

Antoni Ribas

Jedd D Wolchok

Timothy A Chan

Doi

10.1056/NEJMoa1406498

PMID: 25409260 · DOI: 10.1056/NEJMoa1406498 · Journal: New England Journal of Medicine (2014)

TL;DR

Whole-exome sequencing of tumors from 64 patients with metastatic melanoma treated with anti–CTLA4 antibodies (ipilimumab or tremelimumab) showed that high tumor mutational load was significantly associated with long-term clinical benefit (P = 0.01 discovery; P = 0.009 validation). Mutational load alone, however, was not sufficient to predict response — instead, the authors identified a set of shared tetrapeptide neoepitopes (a “neoantigen signature”) in responders that was absent in non-responders and that strongly correlated with overall survival in both the discovery and validation cohorts (P<0.001 by log-rank in each set). Several candidate neoantigens (e.g., TESPFEQHI from a FAM3C mutation and GLEREGFTF from a CSMD1 mutation) elicited polyfunctional T-cell responses ex vivo from patient PBMCs PMID:25409260.

Cohort & data

  • 64 patients with metastatic melanoma (SKCM) treated with anti–CTLA-4 (59 ipilimumab, 5 tremelimumab); 128 exomes (tumor + matched blood) sequenced.
  • Discovery set: 25 patients (11 long-term benefit; 14 minimal/no benefit). Validation set: 39 patients (25 long-term benefit; 14 minimal/no benefit).
  • Cohort registered in cBioPortal as skcm_mskcc_2014.
  • Assay: SureSelect Human All Exon 50-Mb capture (Agilent) → HiSeq 2000 (whole-exome-seq) with mean coverage 103×, >99% of target ≥10×, generated at MSKCC Genomics Core and the Broad Institute.
  • Long-term clinical benefit defined as radiographic freedom from disease or stable/decreased disease >6 months on therapy; minimal/no benefit defined as ≤6 months benefit or progression on every CT after initiation.

Key findings

  • Mutational load correlates with benefit. Median nonsynonymous exome mutation count was higher in long-term benefiters than minimal/no-benefit patients in both the discovery set (P = 0.01, Mann–Whitney) and validation set (P = 0.009, Mann–Whitney) PMID:25409260.
  • Mutational load correlates with survival but imperfectly. In the discovery set, patients with >100 nonsynonymous mutations (N = 17) had significantly improved overall survival vs. ≤100 mutations (N = 8) (P = 0.04, log-rank); a similar trend was seen in validation PMID:25409260.
  • High mutation load alone does not guarantee response. Patients SD7357 (discovery; 1,028 mutations), NR9521, and NR4631 (validation; >1,000 mutations each) had high TMB but did not respond.
  • Tetrapeptide neoepitope signature. Using the NAseek algorithm (HLA-restricted MHC class I binding ≤500 nM), the authors identified 101 tetrapeptide neoepitopes shared exclusively among long-term benefiters in the discovery set; the same signature reproduced in the independent validation set.
  • Signature predicts survival. Patients carrying the neoantigen signature had markedly longer overall survival than those without it in both discovery (signature N = 10 vs. no-signature N = 15; P<0.001, log-rank) and validation (signature N = 20 vs. no-signature N = 19; P<0.001, log-rank) sets.
  • Signature is non-random. Five simulation models confirmed the signature–benefit association was unlikely to arise by chance (P<0.001 for four methods; P = 0.002 for the fifth).
  • Pathogen-homology. Tetrapeptides in responder-shared neoepitopes had higher homology to viral/bacterial antigens in the Immune Epitope Database than non-responder tetrapeptides; e.g., ESSA matches a human cytomegalovirus IE epitope, and ESPF in TESPFEQHI matches the hepatitis D virus p27 large delta epitope.
  • Ex vivo T-cell activation. Two patient-specific predicted neoantigens — TESPFEQHI (from FAM3C c.A577G, p.K193E) and GLEREGFTF (from CSMD1 c.G10337A, p.G3446E) — elicited polyfunctional (IFN-γ+/TNF-α+) CD8+ T-cell responses from patient PBMCs that were absent against the nonmutant peptide and absent in healthy donors. Response to TESPFEQHI in patient CR9306 peaked 60 weeks after ipilimumab initiation; response to GLEREGFTF in patient CR0095 peaked at 24 weeks.

Genes & alterations

  • CTLA4 — target of ipilimumab and tremelimumab; blockade activates effector and helper T cells and depletes regulatory T cells, enabling antitumor immunity PMID:25409260.
  • BRAF, NRAS — used as baseline driver-mutation stratifiers (BRAF or NRAS mutation status by cohort arm; no significant enrichment by benefit category was claimed).
  • FAM3C — missense c.A577G (p.K193E) yields neoepitope TESPFEQHI, which bound patient HLA-B4402 (predicted 472 nM vs. 18,323 nM for nonmutant TKSPFEQHI) and drove a polyfunctional T-cell response in patient CR9306.
  • CSMD1 — missense c.G10337A (p.G3446E) yields neoepitope GLEREGFTF, which drove a polyfunctional T-cell response in patient CR0095; the peptide shares 80% homology with a known Burkholderia pseudomallei antigen (IEDB ID 1027043).
  • No single recurrently mutated gene was universally mutated across responders — the signal is at the level of shared tetrapeptide neoepitopes encoded across diverse genes, not at the level of individual driver genes.

Clinical implications

  • Biomarker. Provides a proof-of-principle that tumor exomes can be used to anticipate response to CTLA-4 checkpoint blockade — the authors explicitly recommend “examining exomes of patients for whom anti–CTLA-4 agents are being considered.” The neoantigen tetrapeptide signature outperforms raw mutational load alone for stratifying long-term benefit.
  • Biology of response. Supports the model that somatic missense mutations create neoepitopes that the immune system recognizes as nonself, and that the released brake on T cells (via CTLA4 blockade) is most effective when such neoepitopes — particularly those resembling pathogen-derived sequences — are present.
  • Reframing “passenger” mutations. The authors argue that some mutations previously classified as passengers should be reconsidered as “immune determinants” because they confer increased MHC class I binding and contribute to the antigenic landscape recognized under checkpoint blockade.

Limitations & open questions

  • Sample size, while large for a 2014 genomic study (128 exomes), is still limited; patients had heterogeneous prior treatments and biopsies were collected at varying time points.
  • The relative in vivo immunologic contribution of each neoepitope in the signature is unclear — the panel may include the most important ones but the dose-response of any single epitope is not established.
  • Role of MHC class II–restricted neoantigens was not addressed and is called out as future work.
  • Generalizability beyond melanoma (a high-TMB tumor type) to other cancers requires further study.
  • In vitro T-cell activation assays are insensitive (positive pools in 3 of 5 patients with sufficient PBMCs); the absence of a detectable T-cell response does not rule out a given neoantigen.
  • Validation cohort included 8 patients with non-responding tumors who had systemic disease control, which may confound the load–survival relationship.

Citations from this paper used in the wiki

  • “Mutational load was associated with the degree of clinical benefit (P = 0.01) but alone was not sufficient to predict benefit.” — Abstract.
  • “There was a significant difference in mutational load between patients with a long-term clinical benefit and those with a minimal benefit or no benefit, both in the discovery set (P = 0.01 by the Mann–Whitney test) and in the validation set (P = 0.009 by the Mann–Whitney test).”
  • “We identified a number of tetrapeptide sequences that were shared by patients with a long-term clinical benefit but completely absent in patients with a minimal benefit or no benefit.”
  • “Simulation testing with five different models showed that the association between the neoepitope signature and a long-term clinical benefit was highly significant and was unlikely to have resulted from chance alone (P<0.001 for four methods and P = 0.002 for a fifth method).”
  • “The neoepitope signature derived from the discovery set correlated strongly with survival in both the discovery and validation sets (P<0.001 for both comparisons by the log-rank test).”
  • “TESPFEQHI had a predicted MHC class I affinity for B4402 of 472 nM, as compared with 18323 nM for TKSPFEQHI.” (Patient CR9306; FAM3C c.A577G;p.K193E.)
  • “GLEREGFTF arises from a mutation in CSMD1 (c.G10337A;p.G3446E)… the peptide has 80% homology to a known Burkholderia pseudomallei antigen.”
  • “These data suggest a need for an expanded definition of the previous categories of driver and passenger mutations… some mutations formerly categorized as passengers may in fact represent ‘immune determinants.’”

This page was processed by crosslinker on 2026-05-14.