Cytokine and angiogenic factor (CAF) profiling for prediction of outcome to first line sorafenib (SR) versus SR plus low-dose interferon-α (IFNα) in patients with advanced renal cell carcinoma (mRCC)

Abstract

LB-218

Background: SR, an oral inhibitor of Raf kinase, VEGFR-1, 2, and 3, PDGFR-β, FLT-3, c-KIT, and RET-receptor tyrosine kinases, has been shown to prolong progression-free survival (PFS) in pts with mRCC after cytokine failure. In a randomized phase II trial in mRCC, pts were randomized to first line treatment with SR or SR plus IFNα to assess whether combination therapy improved PFS compared to single agent SR. Our preliminary results suggested similar outcomes for both treatment arms (Jonasch et al, ASCO 2007). Here we performed exploratory analyses of the plasma levels of 45 CAFs to investigate potential biomarkers predictive of response or prognosis.
Methods: Pts received SR 400 mg PO BID or same dose SR plus IFNα 0.5 million units SC BID. Plasma was collected from 69 pts at baseline (BL; SR 34, SR+IFNα 35), day (D) 28 (n = 58), and D56 (n = 56). We used multiplex bead suspension arrays to measure 42 plasma CAFs, including VEGF, PDGFbb, EGF, HGF, E-selectin, MMP-9, and multiple chemokines and interleukins (IL). Osteopontin (OPN), soluble carbonic anhydrase 9 (sCA9), and sVEGFR-2 were measured by ELISA. The primary objective of this exploratory analysis was to assess the predictive effect of biomarker expression level (above vs. below median) on PFS. Multivariate Cox models adjusting for treatment arm were fitted on PFS at baseline or on treatment (D28 or D56).
Results: Among 45 individual CAFs at BL, only OPN showed a significant interaction with treatment arm (P<0.01), suggesting that the predictive effect of OPN on PFS may be significantly different between the two arms. Pts with high BL OPN had a lower risk for disease progression when treated with SR as compared to SR+IFNα (7.74 vs. 3.93 mos, P=0.007). Higher BL levels of VEGF, PDGFbb, and IL-2 correlated with longer PFS, whereas EGF levels predicted the opposite effect, independently of treatment arm (P ≤0.025 for all). We next assessed the association between marker modulation and PFS per treatment arm. Soluble CA9 (D28, P=0.01) and GRO-alpha (D56, P=0.03) showed a significant interaction effect with treatment in predicting PFS. Greater increases in sCA9 and GRO-alpha correlated with a worse outcome for pts treated with SR (HR=2.37 and 9.64, respectively) in comparison to SR+IFNα (HR=0.68 and 1.43, respectively).
Conclusions: Several CAFs showed potential value in predicting differential benefit from single agent SR vs. SR+IFNα in this phase II study. Broad-based screening of circulating cytokines and angiogenic factors may identify predictive and prognostic biomarkers in the context of clinical trials.