Tumor and Microenvironment Evolution during Immunotherapy with Nivolumab

Authors

Nadeem Riaz

Jonathan J. Havel

Vladimir Makarov

Alexis Desrichard

Walter J. Urba

Jennifer S. Sims

F. Stephen Hodi

Salvador Martín-Algarra

Rajarsi Mandal

William H. Sharfman

Shailender Bhatia

Wen-Jen Hwu

Thomas F. Gajewski

Craig L. Slingluff Jr.

Diego Chowell

Sviatoslav M. Kendall

Han Chang

Rachna Shah

Fengshen Kuo

Luc G.T. Morris

John-William Sidhom

Jonathan P. Schneck

Christine E. Horak

Nils Weinhold

Timothy A. Chan

Doi

10.1016/j.cell.2017.09.028

PMID: 29033130 · DOI: 10.1016/j.cell.2017.09.028 · Journal: Cell (2017)

TL;DR

Riaz et al. profiled paired pre- and on-therapy biopsies from 68 advanced melanoma patients enrolled on the prospective CA209-038 trial (NCT01621490) of nivolumab (anti-PD-1, 3 mg/kg q2w), stratified into ipilimumab-progressed (Ipi-P, n=35) and ipilimumab-naive (Ipi-N, n=33) cohorts. Combining whole-exome sequencing, RNA-seq, and TCR β-chain sequencing on biopsies obtained pre-treatment and on cycle 1 day 29 (week 4), they show that pre-therapy tumor mutation load and clonal mutation load predict overall survival in Ipi-N but not Ipi-P patients; that responders undergo a marked on-therapy reduction in mutation and neoantigen load (mean 19% of pre-therapy variants retained in CR/PR vs 101% in PD; p = 5.87e–5), consistent with immunoediting; that the ratio of neoantigen-producing to non-antigenic mutations falls on therapy in responders (p = 0.03); that immune-checkpoint genes including PDCD1, CD274, CTLA4, LAG3, TNFRSF9, TNFRSF4, TIGIT, HAVCR2, and VSIR (C10orf54) are upregulated on therapy; and that the number of expanded T-cell clonotypes scales linearly with the number of neoantigens that become undetectable on-therapy in responders (p = 0.03). The paper is the genomic backbone of the mel_iatlas_riaz_nivolumab_2017 dataset PMID:29033130.

Cohort & data

  • Trial: CA209-038 (NCT01621490) — multi-arm, multi-institutional, IRB-approved prospective study of pharmacodynamic activity of nivolumab 3 mg/kg every 2 weeks until progression or up to 2 years; 85 patients accrued, 68 with sufficient material for genomic analysis PMID:29033130.
  • Patients: 68 advanced melanoma patients — 35 previously progressed on ipilimumab (Ipi-P), 33 ipilimumab-naive (Ipi-N); median age 55 (range 22–89); stage distribution 1 M0 / 13 M1a / 10 M1b / 36 M1c / 8 unknown PMID:29033130.
  • Response rates (RECIST v1.1): Comparable across arms — Ipi-N 21% CR/PR, Ipi-P 22% CR/PR PMID:29033130.
  • Biopsies: Paired biopsies from the same anatomic site — pre-therapy (1–7 days before first dose) and on-therapy (cycle 1 day 29, between days 23–29). Tissue divided for FFPE and RNAlater storage PMID:29033130.
  • Assays:
  • Allele-specific copy number by FACETS v0.5.0; recurrent SCNAs from GISTIC Broad Firehose download (Jan 28 2016); cancer cell fraction by PyClone v0.13.0; mutational signatures decomposed with deconstructSigs v1.8.0 against COSMIC; HLA class I typing by SOAP-HLA from exome data; neoantigen prediction with NetMHC v4.0 on 9-mer sliding windows of 17-mer mutated peptides (rank < 2%); pathway analysis with Ingenuity Pathway Analysis and GSEA PMID:29033130.
  • Cancer type: Skin Cutaneous Melanoma (SKCM).
  • Dataset: mel_iatlas_riaz_nivolumab_2017.

Key findings

  • Pre-therapy mutation/clonal mutation load predicts OS and response in Ipi-N but not Ipi-P patients. Median cohort mutation load 183 (range 1–7360; IQR 44–433); the association of TMB and clonal mutation count with survival held in Ipi-N alone, and Ipi-P patients had significantly more subclonal mutations (p = 0.08 for lower clonal counts) PMID:29033130.
  • No single recurrent mutation associated with response. TCGA mutational subtypes (BRAF, RAS, NF1, triple-WT) did not stratify response, though Ipi-P had more triple-WT than Ipi-N (57% vs 33%, p = 0.06; Fisher’s exact). JAK1 and JAK2 mutations were not associated with resistance in this cohort. One PR patient harbored a frameshift in B2M with LOH (an alteration previously linked to acquired anti-PD-1 resistance) PMID:29033130.
  • SERPINB3/B4 mutations trend with disease control. 5/6 patients with SERPINB3/SERPINB4 mutations had CR/PR or SD, but the small-sample association was not significant (p = 0.21) PMID:29033130.
  • No association of UV/aging mutational signatures with response, in contrast to NSCLC; no recurrent copy-number alteration including chromosome-9p IFN-cluster loss correlated with response. Global genomic instability associated with OS in Ipi-P but not Ipi-N patients PMID:29033130.
  • Mutation and neoantigen load drop on-therapy in responders (n = 41 paired WES). Mean fraction of pre-therapy variants still detectable on-therapy: 19% in CR/PR (range 1–99%), 82% in SD (range 2–140%), 101% in PD (range 33–205%); CR/PR vs PD p = 5.87e–5 (Mann–Whitney). Power calculations confirmed the effect was not explained by tumor purity alone PMID:29033130.
  • Differential clonal evolution between response groups. All CR/PR patients (except one) consistently lost one or more clones on-therapy; SD/PD patients gained novel subclones. Variant loss (>50% of SNVs under contraction) was seen in 5/13 SD samples and 0/19 PD samples; variant gain (>50% of SNVs novel) in 5/19 PD vs 0/13 SD samples PMID:29033130.
  • Net genomic change predicts response and OS better than raw TMB change. Defined as variants undergoing mutational contraction minus those undergoing persistence PMID:29033130.
  • Focal CDKN2A loss emerged on-therapy in 4 PD patients; 3 of these 4 carried co-occurring chromosome-9p deletions encompassing the IFN gene cluster, paralleling reported escape mechanisms PMID:29033130.
  • Novel on-therapy mutations are enriched for COSMIC signature 11 (temozolomide/melanoma signature), suggesting a stress-induced repair-process dysfunction during immunotherapy PMID:29033130.
  • Pre-existing immune programs distinguish responders. 189 DEGs between CR/PR vs PD pre-therapy (q < 0.20), enriched for immune categories (IL17RE, IL17RC, FGFR3 among them). All Ipi-P responders showed a “hot tumor” phenotype pre-therapy; Ipi-N responders showed variable immunological activity. The previously reported IPRES signature did not associate with response PMID:29033130.
  • Genomic-contraction phenotype maps to a stronger pre-existing-immunity signature than clinical response. 695 DEGs (q < 0.10; 565 with >2-fold change) between contraction vs persistence subsets in 26 patients with paired WES + RNA-seq; the gene set stratified survival in independent immunotherapy cohorts. Enriched pathways included PD-1 signaling, CD28 co-stimulation, downstream TCR signaling, IFN-γ, and IL-2 signaling, plus HLA class II and PI3K-γ GPCR signaling PMID:29033130.
  • Pharmacologic on-therapy response: 475 DEGs (q < 0.20) regardless of clinical response. Many immune checkpoint genes upregulated: PDCD1, CD274, CTLA4, CD80, ICOS, LAG3, and TNFRSF9 (4-1BB) PMID:29033130.
  • Responder-specific on-therapy upregulation includes additional checkpoints — TNFRSF4 (OX40), TIGIT, HAVCR2 (TIM-3), and VSIR (VISTA / C10orf54) — plus broader lymphocyte-activation, chemotaxis, cytokine-signaling, and cytolytic gene programs. 2670 DEGs distinguished pre/on changes in responders vs non-responders (q < 0.20). Downregulated programs in responders included neural/melanin pathways, cell cycle, and mitosis PMID:29033130.
  • Immune deconvolution: CD8+ T cells and NK cells increased on-therapy and M1 macrophages decreased in responders PMID:29033130.
  • TIL fraction increased on-therapy in benefiting Ipi-N patients but not Ipi-P (TCR-seq p = 0.040; IHC p = 0.023). Cytolytic-gene activity (GZMA/PRF1 geometric mean) on-therapy associated with benefit in both Ipi-N (p = 0.005) and Ipi-P (p = 0.043) patients PMID:29033130.
  • Diverging TCR-diversity dynamics by prior-therapy status. On-therapy CDR3 richness change associated with benefit in Ipi-P (p = 0.016) but not Ipi-N (p = 0.489). Conversely, evenness change associated with benefit in Ipi-N (p = 0.036) but not Ipi-P (p = 0.594). Per-VJ-cassette analysis confirmed antigen-driven (CDR3-amino-acid-level) rather than VJ-cassette-driven dynamics. CD8-associated V-segment usage correlated with response (p = 0.005, ordinal regression) while CD4-associated V-segments did not PMID:29033130.
  • T-cell clonal expansion scales with neoantigen loss in responders. In CR/PR patients, the number of significantly expanded T-cell clones was linearly related to the number of neoantigens that became undetectable on-therapy (p = 0.03); the relationship did not hold in SD or PD patients PMID:29033130.
  • Selective depletion of antigenic mutations on-therapy. Neoantigen-to-non-antigenic mutation ratio dropped on-therapy in CR/PR vs PD (p = 0.03; Wilcoxon); within individual CR/PR patients, the observed neoantigen count on-therapy was lower than the synonymous-mutation-derived expectation (p < 0.05; chi-squared) PMID:29033130.

Genes & alterations

  • B2M — Frameshift with LOH in one PR patient; previously associated with acquired anti-PD-1 resistance, but this case was a responder, indicating the alteration alone is insufficient to predict resistance PMID:29033130.
  • JAK1, JAK2 — Mutations were not associated with resistance to nivolumab in this cohort, contrary to prior reports of JAK pathway disruption mediating IFN-γ-blockade resistance PMID:29033130.
  • SERPINB3, SERPINB4 — Trend toward disease control (5/6 SERPINB3/B4-mutant patients had CR/PR/SD; p = 0.21, not significant due to small n), consistent with prior anti-CTLA-4 association reports PMID:29033130.
  • CDKN2A — Focal homozygous loss emerged on-therapy in 4 PD patients; 3/4 had co-occurring chromosome 9p deletions extending into the IFN gene cluster, suggesting acquired immune evasion at progression PMID:29033130.
  • NF1 — Used in TCGA mutational subtype classification (BRAF/RAS/NF1/triple-WT); NF1 subtype membership did not predict response to nivolumab PMID:29033130.
  • PTEN — Reviewed alongside CDKN2A, NF1, B2M for homozygous CN loss; no significant response association reported PMID:29033130.
  • PDCD1 (PD-1), CD274 (PD-L1) — Upregulated on-therapy regardless of response; pre-therapy expression of either was not differentially expressed between responders/non-responders by RNA-seq, though PD-L1 IHC positivity selected for response among Ipi-P patients PMID:29033130.
  • CTLA4, CD80, CD86, CD28, CD247, CD3D, CD3E, CD3G, PTPN6 — Components of the TCR/co-stimulatory immunological synapse upregulated on-therapy and enriched in the contraction-phenotype DEG set PMID:29033130.
  • ICOS, LAG3, TNFRSF9 (4-1BB) — Immune checkpoint / co-stimulatory genes upregulated on-therapy in all patients PMID:29033130.
  • TNFRSF4 (OX40), TIGIT, HAVCR2 (TIM-3), VSIR (C10orf54 / VISTA) — Additional checkpoint receptors selectively upregulated on-therapy in responders, nominated by the authors as candidate targets for combination immunotherapy PMID:29033130.
  • GZMA, PRF1 — Geometric mean defines the cytolytic activity score; on-therapy increases associated with benefit in both Ipi-N (p = 0.005) and Ipi-P (p = 0.043) patients PMID:29033130.
  • FGFR3, IL17RE, IL17RC — Highly expressed in CR/PR vs PD pre-therapy (189-DEG set, q < 0.20), nominated as immune-environment modulators PMID:29033130.
  • FOXP3 — IHC marker for Treg quantification (clone 236A/E7); contributed to immune-cell-population analyses PMID:29033130.

Clinical implications

  • Pre-therapy TMB / clonal mutation load is a useful response biomarker for ipilimumab-naive but not ipilimumab-progressed melanoma patients. Biomarker validation should account for prior immunotherapy exposure PMID:29033130.
  • Early on-therapy biopsy at 4 weeks is informative. Net genomic contraction at week 4 predicts response and OS more strongly than raw mutation-load change, supporting pharmacodynamic biopsy designs in checkpoint-blockade trials PMID:29033130.
  • Genomic evidence of immunoediting under PD-1 blockade. Selective depletion of neoantigenic mutations on-therapy in responders, coupled to clonal T-cell expansion proportional to neoantigen loss, supports the model that anti-PD-1 efficacy depends on tumor-antigen-driven T-cell selection PMID:29033130.
  • Combination-immunotherapy targets nominated by responder-selective on-therapy upregulation: TNFRSF4 (OX40), TIGIT, HAVCR2 (TIM-3), and VSIR (VISTA), in addition to LAG3, TNFRSF9 (4-1BB), and ICOS PMID:29033130.
  • Mechanism of resistance differs between Ipi-P and Ipi-N patients. In Ipi-P, anti-PD-1 may primarily relieve exhaustion of an extant TIL repertoire; in Ipi-N, anti-PD-1 facilitates intratumoral expansion of tumor-reactive clonotypes — implying different rational combinations and biomarkers in each setting PMID:29033130.
  • PD-L1 IHC retains predictive value for Ipi-P patients. Ipi-P responders required PD-L1-positive or “hot” tumors; Ipi-N patients responded across PD-L1 strata PMID:29033130.
  • Acquired focal CDKN2A and chromosome-9p IFN-cluster loss may signal evolving resistance and could be detected by serial biopsy or potentially circulating tumor DNA PMID:29033130.

Limitations & open questions

  • Single-arm, single-timepoint on-therapy biopsy. Only one on-therapy biopsy per patient (week 4), at one anatomic site; intratumoral and inter-lesional heterogeneity, plus regional sampling bias, may confound clonal-evolution interpretations PMID:29033130.
  • Tumor purity decreased on-therapy but the magnitude of change was insufficient to explain the observed mutation-load decreases (confirmed by 300× re-sequencing and power analysis); residual purity confounding cannot be entirely excluded PMID:29033130.
  • Computational neoantigen identification is limited. The authors cannot unambiguously identify the specific neoantigens driving the immunoediting effect; HLA-class-I-only predictions miss class-II-restricted neoepitopes PMID:29033130.
  • Modest cohort size for stratified analyses. 68 patients split into Ipi-N/Ipi-P arms with further subsetting by response category (CR/PR n = ~13, SD n = ~13, PD n = ~19 in paired-WES analyses) limits power for single-gene biomarker discovery; larger studies needed to confirm Ipi-naive-restricted biomarker associations PMID:29033130.
  • No on-therapy clinical-outcome arm comparisons to ipilimumab-only or other PD-1 agents — observations are descriptive of pharmacodynamics under nivolumab monotherapy, not comparative effectiveness PMID:29033130.
  • TCR-seq is bulk and survey-level (Adaptive immunoSEQ). Specificity of expanded clones (which neoantigen they recognize) cannot be directly demonstrated; D90 and CDR3-evenness inferences are population-level PMID:29033130.
  • Resistance mechanism in PD patients is not pinned down to a single tumor-intrinsic genetic event. The transcriptomic signature is more consistent with immune ignorance / immune exclusion than with adaptive resistance, but exact drivers remain open PMID:29033130.
  • Mutational signature 11 emergence on-therapy is provocative (typically temozolomide/MMR-defect related) but its mechanistic link to immunotherapy stress is untested PMID:29033130.
  • Generalization to non-melanoma tumor types is unestablished. Findings derive from cutaneous melanoma; whether the same on-therapy contraction/expansion dynamics hold in lower-TMB tumors is open.

Citations from this paper used in the wiki

  • “We assessed genomic changes in tumors from 68 patients with advanced melanoma, who progressed on ipilimumab or were ipilimumab-naive, before and after nivolumab initiation (CA209-038 study)” (Summary, p. 1).
  • “In responding patients, mutation and neoantigen load were reduced from baseline, and analysis of intratumoral heterogeneity during therapy demonstrated differential clonal evolution within tumors and putative selection against neoantigenic mutations on-therapy” (Summary, p. 1).
  • “Median mutation load in the patient cohort was 183 mutations (range: 1–7360; interquartile range: 44–433)” (p. 3).
  • “Tumor mutation load associated with OS in Ipi-N but not Ipi-P patients … Stratification of patients by the number of clonal mutations improved the ability to predict survival and response of Ipi-N but not Ipi-P patients” (p. 4).
  • “JAK1 and JAK2 mutations were not associated with resistance in this cohort” (p. 4).
  • “One patient with PR had a frame-shift alteration in B2M with corresponding loss of heterozygosity, alterations previously associated with acquired resistance to anti-PD-1 therapy” (p. 4).
  • “Significant differences between pre- and on-therapy biopsies correlated with response (p = 5.87e–5; Mann–Whitney test for CR/PR versus progressive disease [PD])” (p. 4).
  • “the mean fraction of variants in on-therapy samples was 19% for CR/PR (range: 1%–99%), 82% for SD (range: 2%–140%) and 101% for PD (range: 33%–205%)” (pp. 4–5).
  • “There were four cases of focal loss of CDKN2A that appeared in on-therapy samples, all in patients with PD. In three of these patients, chromosome 9p deletions also included the nearby IFN gene cluster” (p. 5).
  • “novel subclonal variants were often due to mutational signature 11, which has previously been associated with melanoma and exposure to temozolomide … suggests that a specific type of repair process becomes dysfunctional due to immunotherapy-mediated stress” (p. 5).
  • “Many immune checkpoint genes increased in expression, regardless of response to therapy, including PDCD1 (PD-1), CD274 (PD-L1), CTLA-4, CD80 (CTLA-4-L), ICOS, LAG3, and TNFRSF9 (4-1BB)” (p. 6).
  • “additional checkpoint-related genes (TNFRSF4 [OX40], TIGIT, HAVCR2 [TIM-3], and C10orf54 [VISTA])” upregulated in responders (p. 7).
  • “An increase in number of CD8+ T cells and NK cells, and a decrease in M1 macrophages, associated with response to therapy” (p. 7).
  • “the median fold change in the number of unique CDR3 sequences (richness) was significantly associated with benefit in Ipi-P but not Ipi-N patients (p = 0.016 versus p = 0.489) … the median change in T-cell evenness on-therapy was associated with benefit in Ipi-N but not Ipi-P patients (p = 0.036 versus p = 0.594)” (p. 7).
  • “In patients with CR/PR there was a linear relationship between the number of expanded T-cell clones and the number of neoantigens that became undetectable on-therapy (p = 0.03)” (p. 9).
  • “Patients with CR/PR had lower neoantigen ratios on-therapy than patients with PD (p = 0.03; Wilcoxon rank-sum test)” (p. 9).
  • “Eighty-five patients were accrued to a multi-arm, multi-institutional, institutional-review-board-approved, prospective study (CA209-038; NCT01621490) … All patients received Nivo (3 mg/kg every 2 weeks) until progression or for a maximum of 2 years” (p. 12).

This page was processed by crosslinker on 2026-05-15.