Constitutive inflammation and epithelial-mesenchymal transition dictate sensitivity to nivolumab in CONFIRM: a placebo-controlled, randomised phase III trial

Abstract

Leveraging adaptive tumour immunity to control mesothelioma via immune checkpoint blockade is now a standard therapeutic approach. However, the determinants of sensitivity remain elusive. Low non-synonymous mutation burden and programmed death-ligand 1 expression, an abundance of immunosuppressive immune cell infiltration, and 9p21 deletion should all mitigate responses to therapy. To address this knowledge gap, we conducted a double blind, placebo-controlled, randomized phase III trial of the PD1 inhibitor, nivolumab (ClinicalTrial.gov registration: NCT03063450). After 37.2 months of follow-up, the primary endpoint of progression free-survival, but not overall survival was met. The nivolumab response rate was 10.3%, and related grade 3 or above adverse events occurred in 20.4% versus 7.2% for placebo. Progression-free and overall survival were longer in nivolumab-treated responders versus non-responders. In an exploratory multiomic analysis, blinded whole exome, transcriptome and multiplex immune profiling were used to interrogate R- versus NR-subgroups. Non-synonymous and neoantigen mutation burden were no different between groups, however R-mesotheliomas were infiltrated with activated CD8⁺ T- and CD19⁺ B-lymphocytes, organised into tertiary lymphoid structures. B-cell infiltration correlated with pro-inflammatory chemokines including IL24 and CCL19. Conversely, epithelial-mesenchymal transition and mitosis were associated with resistance to nivolumab. These findings illuminate features which can be leveraged to advance precision immunotherapy in this rare cancer setting.

Fig. 1. Interrogating response heterogeneity via blinded multi-omic analysis of R- versus NR-mesotheliomas.

A Upper panel. Kaplan-Meier curve showing investigator-reported progression-free survival for nivolumab and placebo treatment groups in CONFIRM. Lower panel. Kaplan-Meier curve showing overall survival (proportion alive) for nivolumab and placebo treatment groups in CONFIRM. B Chest computerised tomography scan of the chest of a patient who exhibited an extreme response to nivolumab over the course of 6 months. C Archival formalin fixed tumour blocks corresponding to patients who received nivolumab and had either a partial response (R) or progressive disease (no response, NR) as their best outcome, were subjected to next generation sequencing of the whole exome or transcriptome. These tissues were also spatially phenotyped using multiplex-immunofluorescence microscopy. Created in BioRender. Fennell, D. (https://BioRender.com/sq9aqpm.) D Waterfall plot showing the selection of patients from R- vs NR-groups for blinded multi-omic analysis. E Spider plot showing the relative change from baseline for mesotheliomas in the NR- vs R-subgroups. F Kaplan-Meier plots for investigator reported PFS in the translational research cohorts denoted responder and non-responder respectively. These curves are descriptive only and are not formally comparable. G Kaplan-Meier plot for OS in the translational research cohorts denoted responder and non-responder respectively. These curves are descriptive only and are not formally comparable.

Fig. 2. Genomic correlates of response to nivolumab in the CONFIRM trial.

A Heat map showing the relative burden of driver gene mutations or driver copy number alterations in the R- vs NR-subgroups. B Stacked histogram showing the relative somatic alteration frequency involving 9p21 (left) and BAP1/3p21 (right). BAP1 was not significantly enriched in R versus NR subgroups (NS, not significant). C Upper panel. Somatic copy number alterations (SCNAs) involving amplifications were more frequent in the R- vs NR-subgroups. Lower panel. Copy number alterations specifically affecting DNA damage response genes were not significantly different in the R- vs NR-subgroup. * indicates a Wilcoxon signed rank test two-sided P value equal to or less than 0.05. The boxplots show the median line and interquartile range (IQR, 25^th-75^th percentiles), with the whiskers extending to the maximum and minimum values. D Tumour mutation burden (TMB), clonal or subclonal mutations, intratumour heterogeneity (ITH), neoantigen burden (clonal or subclonal), or somatic copy number alterations (SCNAs). The boxplots show the median line and interquartile range (IQR, 25^th–75^th percentiles), with the whiskers extending to the maximum and minimum values. E Geneset enrichment analysis (GSEA) summary comparing R vs NR transcriptional signature enrichment. F Boxplots comparing geneset enrichment scores (single sample GSEA) in R- vs NR-subgroups. The boxplots show the median line and interquartile range (IQR, 25^th−75^th percentiles), with the whiskers extending to the maximum and minimum values. G Pathway enrichment plot showing inflammatory and chemokine pathway upregulation in R- vs NR-mesotheliomas.

Fig. 3. Pro-inflammatory pathway transcriptionally upregulated in R- vs NR-mesotheliomas.

A Box plots showing relative chemokine expression corresponding to CCL5, CCL16, CCL17, CCL19, CCL21 and CCL23 transcripts in R- vs NR-mesotheliomas (* indicates a Wilcoxon signed rank test, two-sided p-value equal to or less than 0.05, ** equal to or less than 0.01). Y-axis units correspond to fragments per kilobase of feature per million mapped reads (TPM) normalised counts. The boxplots show the median line and interquartile range (IQR, 25^th−75^th percentiles), with the whiskers extending to the maximum and minimum values. B Boxplots showing Interleukins IL17C, IL23A and IL24 transcripts in R- vs NR-mesotheliomas. The boxplots show the median line and interquartile range (IQR, 25^th−75^th percentiles), with the whiskers extending to the maximum and minimum values. C Scatter plot showing CCL19 versus B lymphocyte lineage as determined by transcriptome deconvolution. Blue dots correspond to R-mesotheliomas and purple to NR (spearman’s rank correlation coefficient= 0.85, p = 1.1 × 10⁻⁷). D Scatter plot showing IL24 versus B lymphocyte lineage as determined by transcriptome deconvolution. Blue dots correspond to R-mesotheliomas and purple to NR (spearman’s rank correlation coefficient= 0.66, two-sided p = 0.0006). E Box plot comparing IL24/EMT ratio in R- vs NR- mesotheliomas (*** indicates a significance p < 0.0001). F Receiver operating characteristics plot for the ratio of IL24 transcript levels to EMT enrichment score in R- vs NR-mesotheliomas. AUROC was 0.9. G Left. Box plot showing the relative enrichment of CD8 immune cells in R (blue) vs NR (purple) mesotheliomas deconvoluted using MCP-Counter, Wilcoxon two-sided p = 0.024). This boxplot shows the median line and interquartile range (IQR, 25^th−75^th percentiles), with the whiskers extending to the maximum and minimum values. Right. Scatter plots showing CD8 + T cell enumeration by deconvolution versus multiplex immunofluorescence (mIF) respectively. Spearman’s r = 0.6018, p = 0.002. H Correlation matrix showing a positive (blue) correlation between immune deconvolution algorithms (CIBERSORT, MCP Counter) and multiplex immunofluorescence for CD8 T cell enumeration. I Kaplan Meier curves showing progression free survival of CD8 T cells, EMT, or IL24/CCL19 ratio (dichotomised by the medians) in an independent (MEDUSA) cohort.

Fig. 4. Tertiary lymphoid structures are enriched in nivolumab responders.

A Box plot showing a significantly higher number of infiltrating CD8⁺ T- lymphocytes in R-subgroup mesotheliomas compared with NR. The boxplots show the median line and interquartile range (IQR, 25^th−75^th percentiles), with the whiskers extending to the maximum and minimum values. B Geneset enrichment plot showing germinal centre formation transcriptional signature enrichment in R- vs NR-mesotheliomas. This boxplot shows the median line and interquartile range (IQR, 25^th−75^th percentiles), with the whiskers extending to the maximum and minimum values. C Micrograph showing a representative lymphoid aggregate (TLS) using a CD8, CD4 CD19, PD1, TIM-3, TIGIT multiplex immunofluorescence panel. D Box plots showing the relative TLS abundance in R- vs NR-groups with respect to TLS area, T-cell density both within and outside TLS areas. B-cell density per TLS area and B-cell density within and outside of the TLS areas (* Wilcoxon signed rank test P value < 0.05, **p < 0.01). The boxplots show the median line and interquartile range (IQR, 25^th−75^th percentiles), with the whiskers extending to the maximum and minimum values. E Scatter plot showing the correlation between %TLS area and OS (r = 0.47, p = 0.02), R is shown in blue and NR in purple. F Endogenous retroviral gene expression showed no difference between R- vs NR-subgroups (NS, not significant). G. Schematic showing the relative balance between stemness (EMT) and inflammation, reflected by chemokine expression and TLS formation. IL24:EMT ratio predicts response, PFS and OS. This figure was created using BioRender. Fennell, D. (https://BioRender.com/4dwlhk1).