Patients With High PD-L1 Expression in mRCC are Less Sensitive to VEGF-Targeted Drugs

Toni Choueiri, MD

According to a retrospective analysis of the phase III COMPARZ trial, high tumor expression of the protein PD-L1 (programmed cell death 1 ligand 1) is independently associated with shorter overall survival (OS) in patients with metastatic renal cell carcinoma (mRCC) receiving treatment with vascular endothelial growth factor (VEGF)-targeted therapy.

The worst outcomes occurred in patients with high PD-L1 expression and a high level of intratumoral CD8⁺T-cell counts, said Toni Choueiri, MD, at the 2014 Genitourinary Cancers Symposium. The findings could have major implications for future trial designs that involve PD-1 inhibitors, he said.

COMPARZ was a phase III study in which the VEGF inhibitors pazopanib and sunitinib were compared as first-line treatments for mRCC. As part of the current analysis, formalin-fixed, paraffin-embedded tumor samples were analyzed for PD-L1 in 453 COMPARZ participants. Tumor PD-L1 expression was quantified by an H-score based on immunohistochemistry. Intra-tumoral CD8⁺cells were quantified morphometrically.

Some 36.2% of analyzed study tumors expressed PD-L1; the H-scores ranged from 0 to 290. “Interestingly, and not completely in line with reports that are much smaller in this disease, we have found that a higher PD-L1 expression is associated with lower overall survival from sunitinib or pazopanib,” said Choueiri, director of the Kidney Cancer Center at Dana-Farber Cancer Institute/Brigham and Women’s Hospital, in Boston.

H-scores of 50 and 125 were used as separate cut-offs to define low and high H-scores.

Sixty-one patients had H-scores >50, and these patients had significantly shorter median OS than the 392 patients with H-scores ≤50, regardless of treatment arm (P= .046). In the pazopanib arm, median OS was 19.7 months in those with H-scores >50, compared with 31.6 months in those with H-scores ≤50. In the sunitinib arm, the median OS rates were 15.3 months in the patients with high H-scores vs 27.7 months in those with low H-scores.

When an H-score of 125 was used as the cut-off, median OS was just 5.1 months in pazopanib-treated patients with an H-score above this level, compared with 31.6 months in pazopanib-treated patients with an H-score ≤125. The same association was found in the sunitinib arm: Patients with an H-score >125 had a median OS of 8.9 months, compared with 27.4 months in the sunitinib-treated patients with an H-score ≤125. Fifteen patients had scores >125 and 438 had scores ≤125; thePvalue was .017 for those findings.

The same trend existed for progression-free survival (PFS), said Choueiri. PFS was significantly shorter (P= .032) in patients with H-scores >125 compared with patients with H-scores ≤125 (3.1 vs 10.2 months, respectively, in the pazopanib arm and 4.0 vs. 8.4 months, respectively, in the sunitinib arm).

The shortest OS was observed in patients with high PD-L1 expression and a high intra-tumor CD8⁺cell count.

• In the sunitinib arm, patients with H-scores >50 and tumor CD8⁺cell counts >300 had a median OS of just 11.9 months, compared with 28.0 months in those with low H-scores and low CD8⁺counts.
• In the pazopanib arm, patients with H-scores >50 and tumor CD8⁺cell counts >300 had a median OS of just 9.6 months, compared with 28.7 months in those with low H-scores and low CD8⁺counts.

“PD-L1 expression has been actively pursued as one of the main biomarkers behind the benefit to PD-1 inhibitors,” Choueiri said. “The future could be that patients with high PD-L1 expression, because they don’t do as well on a VEGF inhibitor, should be on a PD-1 inhibitor.”

<<<

Back to the GU 2014 conference page