High intratumoral tryptophan metabolism is a poor predictor of response to pembrolizumab (pembro) in metastatic melanoma (MM): Results from a prospective trial using baseline C11-labeled alpha-methyl tryptophan (C11-AMT) PET imaging for response prediction.

Abstract

3556 Background: Molecular imaging of metabolic pathways critical for effector T cell response other than glucose could predict response to PD1 inhibitors. We have previously shown that high expression of tryptophan metabolic pathway enzymes in stage III/IV melanoma correlates with reduced abundance of tumor-infiltrating lymphocytes (TILs, ASCO 2019, e21014). Here we investigated C11-AMT, a PET tracer that images tryptophan metabolism, as a predictor of response to pembro in patients (pts) with PD1 inhibitor-naïve MM. Methods: In this trial (NCT03089606) pts must have had measurable MM by RECIST, have undergone IV contrast CT, FDG-PET, and C11-AMT PET scan (30-40 min dynamic imaging), plus a mandatory tumor biopsy prior to pembro treatment. Results: 21 pts (16 males; 15 stage IV; median age 61) had all 3 baseline scans and evaluable research biopsies. 13 pts were non-progressors (CR = 4, PR = 6, and SD = 3). At a median f/u of 13.7 mons (2.8-25.1+), 6 pts are dead from MM, 11 are alive without MM and 4 are alive with MM. 46 tumor lesions were assessed by all 3 scans. Of the pts with tumor lesion, C11-AMT PET SUVmax < 7 and skewness < +0.3 of the tumor lesion with the highest C11-AMT SUVmax was associated with non-progression by RECIST (SUVmax < 7/skewness < +0.3 in progressors vs. non-progressors; Fisher’s 2-tail test p= 0.055). The corresponding association between baseline FDG-PET (SUVmax < 14 and skewness < +0.3) with treatment response was insignificant ( p= 0.08). There was a weak (Spearman ρ= 0.33) but significant correlation ( p= 0.001) in SUVmax between FDG-PET and C11-AMT among the 46 tumors analyzed. There was no significant correlation between melanoma-specific expression of the tryptophan- (TPH1, TPH2, IDO1, TDO2, LAT1) and glucose-metabolizing enzymes (GLUT1, HK1, HK3) by immunohistochemistry. Response to pembro trends to associate with present TILs (Fisher’s p= 0.087). Conclusions: Baseline C11-AMT PET imaging using simple radiomics measures (highest metabolic activity, SUVmax, and tumor heterogeneity, skewness) may better predict clinical benefit from pembro in MM than FDG PET. Variability in C11-AMT’s SUVmax cannot be solely explained by FDG-PET’s SUVmax, suggesting that these two imaging modalities may provide complementary information about intratumoral metabolic dysregulation that may relate with pembro response. Texture analysis using LifeX v4.0 will be presented at the meeting. Clinical trial information: NCT03089606 .