Carcinomas of unknown primary treated with molecularly guided therapies and immune checkpoint inhibitors: A subset from the I-PREDICT N-of-1 Precision Oncology study.

3135 Background: Carcinoma of unknown primary (CUP) is a heterogenous group of aggressive malignancies lacking a detectable primary site and collectively comprises 3–5% of cancer diagnoses. Current treatments rely upon empiric chemotherapeutic regimens with high toxicity, limited efficacy, and historic median overall survival times of 3-11 months. There is emerging evidence to support a precision medicine approach using molecularly guided therapies and tumor-agnostic biomarkers. Methods: We performed a subset analysis of patients previously enrolled in the I-PREDICT (NCT02534675) study who had a CUP that was treated with a combination of targeted therapies and biomarker-driven immune checkpoint inhibitors (ICI), based upon comprehensive genomic profiling and recommendations by a Molecular Tumor Board. Results: Among the 210 evaluable patients, 10 (4.8%) had a CUP from which 9 were molecularly matched to one or more targeted therapies and an ICI based on biomarkers for both types of agents. Baseline characteristics include median age of 67 years (range: 59 – 82), with the majority of patients being male (5/9, 56%), non-Hispanic white (7/9, 78%) and treatment naïve (8/9, 89%). Histological subtypes of CUP included poorly differentiated or unspecified (5/9), adenocarcinoma (3/9) and squamous cell carcinoma (1/9). Administered targeted therapies included bevacizumab (4/9), trametinib (4/9), palbociclib (1/9), crizotinib (1/9), everolimus (1/9), and lenvatinib (1/9), while ICI included nivolumab (6/9), pembrolizumab (2/9) and atezolizumab (1/9). ICI were matched based on positive PD-L1 immunohistochemistry (IHC) (6/9), TMB ≥10 mutations/megabase (4/9), SWI/SNF chromatin remodeling gene alterations (3/9), PD-L1 amplification (1/9), mismatch repair gene alterations (1/9) and microsatellite instability-high (1/9). Most patients were matched to >1 ICI biomarker (5/9). Most patients received treatment doublets (6/9, 67%) while the remaining received triplet regimens (3/9, 33%). The best overall response rate was 44% based on RECIST 1.1. This included partial response (4/9), stable disease (3/9), progressive disease (1/9) and not restaged (1/9). Median progression free survival (PFS) was 10.4 months (95% confidence interval [CI] 4.2 - not estimable [NE]) and median overall survival was 16.9 months (95% CI 6.9 - NE). One patient in particular achieved 36-months of PFS on trametinib + nivolumab as matched to NF1 R440* and PD-L1 IHC 5%. Grade ≥3 treatment related serious adverse events occurred in only one patient. Conclusions: As compared to historic chemotherapy treated cohorts, these results support the safety and efficacy of administering molecularly guided combination therapies including a targeted agent plus an ICI in CUP patients. Larger prospective trials of biomarker-matched N-of-1 combinations in CUP are warranted. Clinical trial information: NCT02534675 .