Rationale and feasibility of a rapid integral biomarker program that informs immune-oncology clinical trials: the ADVISE trial

Abstract

Background: ADVISE (ADaptiVe biomarker trial that InformS Evolution of therapy) (NCT03335540) was a biomarker-adapted feasibility clinical trial of immunohistochemistry (IHC) to inform combination immuno-oncology (I-O) treatment.

Methods: To inform I-O combination selection, messenger RNA expression analyses from The Cancer Genome Atlas evaluated associations between programmed death 1/programmed death ligand 1 (PD-1/PD-L1) and other I-O-associated genes. Tumor tissue blocks of melanoma, non-small cell lung cancer, renal cell carcinoma, urothelial carcinoma, squamous cell carcinoma of the head and neck, and gastroesophageal junction/gastric cancer were stained by IHC to assess expression of CD8, colony-stimulating factor 1 receptor, glucocorticoid-induced tumor necrosis factor receptor (GITR), indoleamine 2,3-dioxygenase 1, lymphocyte-activation gene 3, NKp46, forkhead box P3, and PD-L1. These results facilitated an I-O treatment selection algorithm where patient biopsy results dictated allocation into combinations of nivolumab with cabiralizumab, urelumab, linrodostat mesylate, relatlimab, BMS-986156 (anti-GITR), lirilumab, ipilimumab, or irradiation. The primary endpoint was the proportion of patients with qualified baseline tumor biopsy specimens where decision-enabling biomarker analysis was completed within 12 business days to select an I-O combination therapy.

Results: Correlation of PD-1/L1 and I-O-associated genes varied across the spectrum of T-cell-inflamed versus non-inflamed tumors; however, tumors with low/intermediate PD-L1 expression demonstrated distinct upregulation of immune markers grouped by cell type (T cell, macrophage, etc). IHC analyses of I-O naïve tumors corroborated these findings with distinct immune target upregulation in low-to-intermediate inflamed tumors and significant associations between IHC-detected markers and T-cell inflammation score across most markers. In the clinical trial, 20/23 (87%) of eligible patients were successfully allocated and started on treatment within the 12-day window, meeting the primary endpoint. The safety profile appeared to generally align with those reported for the individual combinations from other trials. No treatment responses occurred. Most patients were allocated to the cabiralizumab treatment arm.

Conclusions: Actualization of a patient-specific I-O combination treatment selection strategy is feasible, however, determination of de novo integral biomarker thresholds of novel I-O targets to facilitate effective treatment of PD-1-refractory cancer remains fraught. These data emphasize the difficulty of integral biomarker development for I-O in translating from immunotherapy treatment-naïve biospecimens to the selection of patients in the PD-1-refractory state.

Trial registration number: NCT03335540.

Keywords: Biomarker; Biopsy; Immunotherapy.

Figure 1. ADVISE trial schema. ADVISE, ADaptiVe biomarker trial that InformS Evolution of therapy; CSF-1R, colony-stimulating factor 1 receptor; CTLA-4, cytotoxic T-lymphocyte antigen-4; GEJ, gastroesophageal junction; GITR, glucocorticoid-induced tumor necrosis factor receptor; IDO1, indoleamine 2,3-dioxygenase 1; KIR, killer cell immunoglobulin-like receptor; LAG-3, lymphocyte activation gene-3; NIVO, nivolumab; SBRT, stereotactic body radiation therapy.

Figure 2. Characterization of the relationship between immune and cancer-related gene expression and tumor inflammation status in the TME from TCGA samples. (A) PD-1 expression correlated with expression of 1567 genes across tumor types across all samples regardless of inflammation status. Two clusters of genes are shown. (B) Correlation of genes to PD-1 expression in T cells that are inflamed or non-inflamed tumors is presented here. Less correlation between the expression of PD-1 and other immune genes from the strongly correlated gene cluster (399 genes) was observed in the tumors classified as non-inflamed compared with inflamed (p=0.00017, two-sided Wilcoxon signed-rank test). The most strongly (light yellow box) and less correlated genes (light blue box) are shown in panels A and B of all TCGA tumor samples. Genes were selected from the literature and clinically relevant gene expression panels. HCC, hepatocellular carcinoma; PD-1, programmed death 1; TCGA, The Cancer Genome Atlas; TME, tumor microenvironment.

Figure 3. Supervised clustering of IHC data identified discrete inflammation profiles in low/intermediate groups from commercially acquired tumor samples. IHC samples shown are sorted by inflammation status (CD8 IHC or IFN-γ expression signature with low, medium and high tertiles indicated) in individual tumors: (A) NSCLC, (B) RCC, and (C) UC, as well as (D) sample histology of UC with low and high inflammation by IFN-γ gene expression. Examples of tumors with generalized high inflammation or with outlier biomarker expression are marked by vertical rectangles in the heat map. ^aSamples were ordered by low-to-high inflammatory marker expression (CD8 for NSCLC and RCC, IFN-γ for UC), with the rest of the markers ordered based on highest to lowest Spearman correlation to CD8 or IFN-γ. CD, cluster of differentiation; CSF-1R, colony-stimulating factor 1 receptor; FoxP3, forkhead box P3; GITR, glucocorticoid-induced tumor necrosis factor receptor; IC, immune cell; IDO1, indoleamine 2,3-dioxygenase 1; IFN-γ, interferon-gamma; IHC, immunohistochemistry; LAG-3, lymphocyte-activation gene 3; NSCLC, non-small cell lung cancer; PD-1, programmed death 1; PD-L1, programmed death ligand 1; RCC, renal cell carcinoma; TC, tumor cell; UC, urothelial carcinoma.

Figure 4. LAG-3 expression by IHC correlates with tumor inflammation in validation samples. (A) LAG-3 expression by tertiles in all tumors (B) LAG-3 expression by tertiles in individual tumors. IHC, immunohistochemistry; LAG-3, lymphocyte-activation gene 3; NSCLC, non-small cell lung cancer; RCC, renal cell carcinoma; SCCHN, squamous cell carcinoma of the head and neck; UC, urothelial carcinoma.

Figure 5. Decision matrix for combination therapy selection. (A) Schema, (B) decisional IHC biomarkers, (C) case example of IHC biopsy results in patient with NSCLC informed I-O combination treatment selection in ADVISE (image presented previously at European Society for Medical Oncology (ESMO) 2018; Poster 1135PD. Munich, Germany). (Panel A notes) ^aCD8 and PD-L1 expression >1%. ^bAll six biomarker levels are low. ADVISE, ADaptiVe biomarker trial that InformS Evolution of therapy; CD8, cluster of differentiation 8; CSF-1R, colony-stimulating factor 1 receptor; FOXP3, forkhead box P3; GITR, glucocorticoid-induced tumor necrosis factor receptor; IC, immune cell; IDO, indoleamine 2,3-dioxygenase; IHC, immunohistochemistry; I-O, immune-oncology; LAG-3, lymphocyte-activation gene 3; NIVO, nivolumab; NSCLC, non-small cell lung cancer; PD-L1, programmed death ligand 1; SBRT, stereotactic body radiation therapy; TC, tumor cell.

Figure 6. Assessment of the feasibility of real-time biomarker analysis in the ADVISE trial. (A) Planned timeline. (B) 21 patients received biopsy results (orange circles), 20 of whom were successfully assigned to combination therapy using the ADVISE algorithm (blue circles), with 19 of them receiving that assignment within 12 days of initial biopsy, and 15 initiating treatment. (Panel A notes) Steps involved with 12 business days timeline from biopsy collection to results communicated to clinical site. ^aDepending on whether fixation time was 24 or 48 hours. ^bPathology peer review was initiated beginning with patient #16. (Panel B notes) ^aEquivalent to 16–18 calendar days depending on the day of the week the biopsy was obtained. ^bPatient withdrew consent for treatment due to decline in status. ^cBiopsy was evaluated within 15 business days. Screen failed patients due to no tumor biopsy. ADVISE, ADaptiVe biomarker trial that InformS Evolution of therapy; CRO, clinical research organization.