ASCO 2018: Characterizing Tumor Immune Microenvironment and Outcomes for 409 Patients Treated On COMPARZ: Distinct Clusters Emphasize Immune Infiltration Vs. Angiogenesis
The phase III COMPARZ trial compared sunitinib and pazopanib in the first line setting for advanced/metastatic renal cell carcinoma, and found that pazopanib was non-inferior to sunitinib, but had a better safety and QOL profile.1 Using this patient population, including tissue collected during the trial, the authors of this abstract assessed the TME of patients using mRNA expression profiling (Affymetrix GeneChip HTA 2.0) from archival paraffin specimens. Tumor samples were all obtained at baseline, prior to treatment. RNA from the tissue samples were utilized, as was DNA from primary tumors and normal matched tissue. They utilized samples from 437 patients – the original study had 1110 patients. Of these 437 patients, 412 contributed tumor RNA, 377 contributed tumor DNA and 352 contributed both.
Characteristics of patients included was not provided – unclear why there did not have access to all 1110 patients.
Using these patients and mRNA data, they found that there were 4 biologically distinct clusters, which they labeled C1-4. In terms of clinical outcomes, each of the clusters had significant differences in median overall survival (OS;P= 2.00E-4) and progression free survival (PFS;P= 0.03). Median OS was 31.5, 31.6, 36.7 and 21.8 months for Clusters 1-4, respectively. Median PFS was 10.9, 11.1, 11.1 and 8.2 months for Clusters 1-4, respectively. Clearly Cluster 4 does the worst.
Patients in Cluster 4 displayed the worst outcomes and highest rate of IMDC poor risk features (45.7% pts); in this same population, gene signatures showed enrichment for inflammation signatures, e.g. IFN-γ responses. In particular, cluster 4 patients demonstrated the most immune-infiltrated TME with enrichment of many immune subsets, especially macrophages (compared to C1-3, P= 0.0015). C4 also had the highest rate of PD-L1 expression on tumor cells and macrophages (P= 3.50E-7).
In contrast, C3 had the most favorable outcomes and displayed the highest angiogenesis gene expression levels (P= 2.20E-16).
One thing they did not do that would have been interesting is comparing response to pazopanib or sunitinib in the trial based on these clusters.
When accounting for the clustering and clinical variables in a multivariable model, the clusters will still independently significant. When combined with IMDC clinical variables (compared to clinical variables alone), the predictive c-index for OS improved from 0.63 to 0.69 and PFS from 0.60 to 0.65. Hence, the cluster data provided some information above and beyond clinical data.
As we know from other malignancies, molecular signatures may add to the ability to risk stratify patients. It would appear that 4 distinct TME exist for RCC, which contribute to clinical outcomes. Obviously, further study needed.
Limitations:
1. Obviously, these are all treated patients. All treated with TKI. Responses may vary for other treatment modalities.
2. Despite having the data, they do not look at whether clusters predicted response to pazopanib or sunitinib.
Presented by: Martin Henner Voss, MD
Written by: Thenappan Chandrasekar, MD, Clinical Fellow, University of Toronto, Twitter: @tchandra_uromd at the 2018 ASCO Annual Meeting - June 1-5, 2018 – Chicago, IL USA