Genomic, Transcriptomic Profiling Lays Groundwork for Personalization in RCC

A. Ari Hakimi, MD

RNA-based analyses of prospectively-collected tumor specimens from patients with metastatic renal cell carcinoma (mRCC) in the phase III COMPARZ trial revealed 4 molecular subgroups, of which a high angiogenesis expression and a low macrophage infiltrate were indicative of response to frontline TKI therapy.¹

Based on the analysis, which was recently published in Cancer Discovery, these molecular subgroups showed significant differences in the rates of median progression-free survival (PFS; P = .0002) and overall survival (OS; P = .03) observed in the international COMPARZ trial, which randomized patients with advanced clear-cell mRCC to frontline therapy with either pazopanib (Votrient) or sunitinib (Sutent).

“We identified 4 distinct molecular subgroups that differed significantly in response and survival,” first author A. Ari Hakimi, MD, a urologist of Memorial Sloan Kettering Cancer Center, and investigators wrote. “Detailed characterization of these clusters emphasized the central role of the tumor microenvironment (TME) for the outcome with TKI therapy and identified angiogenesis and macrophage infiltration as critical determinants of TKI response.”

For the trial, cluster analysis was performed on 212 patients in the sunitinib arm and 197 patients in the pazopanib arm by way of unsupervised consensus nonnegative matrix factorization. Cluster 4 was associated with a substantially worse OS (HR, 2.09; 95% CI, 1.47-2.97; P = .0000458) and PFS (HR, 1.54; 95% CI, 1.13-2.09; P = .00572) relative to clusters 1 to 3.

Notably, patients categorized as poor risk according to the IMDC risk model were enriched with cluster 4 (45.7%) compared with clusters 1-3 (P = .009). Further, patients with progressive disease (PD) were enriched with cluster 4 (P = .017) as opposed to those who achieved stable disease, a complete response (CR) or a partial response (PR) to TKI monotherapy.

In order to compare PD with CR and PR, investigators evaluated differently expressed genes and ingenuity pathways analysis. They observed an upregulation of the IFN-γ, IFN-α, inflammatory response, interleukin-6, and TNF-α signaling genes involved in inflammatory pathways in the patients with PD compared with those who had achieved a PR or better.

The team then analyzed differences in the frequency of mutations in common cancer driver genes between the clusters. Although cluster 4 had a similar frequency of mutations common to clear cell cancer drivers compared with clusters 1 to 3 (P = .12), it had a lower abundance of favorable somatic PBRM1 mutations and a higher load of poorer prognostic TP53 and BAP1 mutations (P = .003).

“With an increasing understanding of tumor biology and our growing ability to decipher the details of such on a molecular level, one should argue that integration of such information is the logical next step to improve our ability to prognosticate patients, possibly to guide the rational choice of agents,” first author A. Ari Hakimi, MD, a urologic surgeon at Memorial Sloan Kettering Cancer Center, and colleagues wrote.

After observing significant differences in angiogenesis gene expression between clusters, the team evaluated the impact of that on TKI response and survival outcomes. Higher angiogenesis (Angio^high) gene expression was associated with an improved objective response rate among the entire population in COMPARZ trial.² Moreover, patients in the COMPARZ² Angio^high group showed an improved OS and PFS compared with those in the lower angiogenesis (Angio^low) group. However, patients within cluster 4 classified as Angio^high did not demonstrate the same benefit in PFS or OS relative to those classified as Angio^low.

In defining the relationship between somatic mutations and angiogenesis expression, tumors harboring PBRM1 mutations were shown to reflect higher angiogenesis gene expression (P =.0004), whereas tumors harboring BAP1 mutations were shown to reflect lower angiogenesis gene expression (P = .01).

Pathway analyses were also performed to discern the impact of a patient’s TME on response to treatment. Surprisingly, cluster 4 had the highest immune infiltrate among the clusters (P = 2.20 x 10^-16) and the greatest expression of PD-L1 positivity (P = 1.71 x 10^-8). Specifically, cluster 4 was enriched for immune and macrophage-related genes, of which a high infiltration was linked to a significantly worse OS in the COMPARZ trial (HR, 1.54; 95% CI, 1.17-2.03; P = .00198).

“These findings need to be validated and explored in immunotherapy trials,” wrote investigators. “Further, with several agents being applied in combinations trials with checkpoint inhibitors, it is important to consider these notable differences in the TME interaction as we interpret trial results and conduct correlative analyses.”

Patients were categorized as either Angio^low/Angio^high and Macrophage^low/Macrophage^high to analyze the collective impact of angiogenesis and immune TME on response to TKI therapy. Patients in the Angio^lowMacrophage^high group showed the worst survival outcomes compared with the Angio^highMacrophage^low group, which demonstrated the best survival outcomes with regard to PFS (HR, 2.27; 95% CI, 1.51-3.42; P = 8.58 x 10^-5) and OS (HR, 3.12; 95% CI, 1.93-5.03; P = 2.91 x 10^-6). Cluster 4 was enriched with the Angio^lowMacrophage^high groups, which was consistent with prior observations.

Moreover, patients who achieved a PR or better with sunitinib had a higher angiogenic, cytotoxic T cell and CD8-positive T-cell infiltrate versus those who had PD. Conversely, patients who attained the same parameters of response with pazopanib had a lower macrophage infiltrate versus those who had PD.

As expected, Angio^lowMacrophage^high patients treated with pazopanib experienced the worst PFS (HR, 6.01; 95% CI, 2.73-13.20; P = 8.00 x 10^-6) and OS (HR, 1.86; 95% CI, 1.00-3.46; P = .05) compared with the Angio^highMacrophage^low group. However, no difference was observed between endpoints in either group of patients treated with sunitinib.

When evaluating angiogenesis alone by drug type, the investigators found that the Angio^high group was associated with improved PFS (HR, 1.56; 95% CI, 1.10-2.22; P = .012) and OS (HR, 1.54; 95% CI, 1.07-2.25; P = .022) in the sunitinib arm of the COMPARZ trial. Additionally, the Macrophage^high group experienced a significantly worse PFS (HR, 1.85; 95% CI, 1.30-2.63; P = 6.74 x 10^-4) and OS (HR, 2.62; 95% CI, 1.71-4.00; P = 8.56 x 10^-6) compared with the Macrophage^low group in the pazopanib arm. Survival did not differ between the Macrophage groups in the sunitinib arm.

“These findings warrant further investigation and should motivate separate analyses for other TKIs routinely used in this disease; ultimately, they could inform the rational choice of agent based on baseline analysis of tumor tissue,” the study authors concluded. “Our results give biologic insights into possible mechanisms by which the TME might affect the efficacy of specific drugs.”

References

1. Hakimi A, Voss M, Kuo F, et al. Transcriptomic profiling of the tumor microenvironment reveals distinct subgroups of clear cell renal cell cancer - data from a randomized phase III trial [published online January 8, 2019]. Cancer Discov. doi: 10.1158/2159-8290.CD-18-0957.
2. Therasse P, Arbuck SG, Eisenhauer EA, et al. New guidelines to evaluate the response to treatment in solid tumors. European Organization for Research and Treatment of Cancer, National Cancer Institute of the United States, National Cancer Institute of Canada. J Natl Cancer Inst. 2000;92(3):205-216. academic.oup.com/jnci/article/92/3/205/2965042.