Background: The fact that high dose of testosterone (HDT) elicits positive responses in a subset of prostate cancer patients (pts) is surprising and puzzling. Genomics data differentiating responders (Rs) from non-responders (NRs) is sparse. Pts with mutations in DNA repair pathway genes may be particularly sensitive to HDT (see BA Teply et al. Eur Urol 71:499, 2017). Herein we perform exploratory analyses to better understand the role of pathogenic mutations (muts) in ctDNA as a predictive biomarker for patients treated with HDT. Methods: ctDNA essays were performed with the Guardant360 methodology pre-HDT. Point mutations were classified by cancervar (http://cancervar.wglab.org). Truncating mutations (frameshift and nonsense) were manually curated to assess for pathogenicity. All patients had CRPC and were pre-treated with abiraterone and/or enzalutamide. HDT was typically administered as 400 mg testosterone cypionate q 3-4 weeks. Rs were compared to NRs. All Rs had >3 more cycles of HDT and a >50% PSA decline (N = 17). Non-responders had <3 cycles and no PSA decline (N = 23). Only muts with an allelic fraction of >0.5% were analyzed given dubious importance of mutations with lower allelic fractions. Results: AR muts (4/17 vs 6/23 in Rs and NRs) and TP53 muts (10/17 vs 11/23 in Rs and NRs) were similar (P= 0.85 and 0.49, respectively) but the number of pts with non-AR/non-TP53 muts was distinct (3/17 for Rs, and 12/23 for NRs; P= 0.026). The average number of non-AR/non-TP53 muts (Rs = 0.23 and NRs = 0.83) was higher in the NRs (P = 0.046). When analyzing DNA repair alterations, no differences were noted in those with BRCA1/BRCA2/ATM mutations in the Rs and NRs (1/17 vs 6/23 respectively; P= 0.09). Conclusions: AR and TP53 pathogenic mutations are common in both Rs and NRs but other pathogenic mutations are more common in non-responders. We hypothesize that genetic pathways outside of the AR/TP53 axis drive resistance to HDT. Additional studies are warranted to assess whether or not these pathways drive resistance to HDT. Limitations are acknowledged with regard to the Guardant assay gene selection for CRPC pts.