Carcinoma of unknown primary (CUP) is a rare and difficult-to-treat malignancy, the management of which might be improved by the identification of actionable driver mutations. We interrogated 54 to 70 genes in 442 patients with CUP using targeted clinical-grade, next-generation sequencing of circulating tumor DNA (ctDNA). Overall, 80% of patients exhibited ctDNA alterations; 66% (290/442) ≥1 characterized alteration(s), excluding variants of unknown significance. TP53-associated genes were most commonly altered [37.8% (167/442)], followed by genes involved in the MAPK pathway [31.2% (138/442)], PI3K signaling [18.1% (80/442)], and the cell-cycle machinery [10.4% (46/442)]. Among 290 patients harboring characterized alterations, distinct genomic profiles were observed in 87.9% (255/290) of CUP cases, with 99.7% (289/290) exhibiting potentially targetable alterations. An illustrative patient with dynamic changes in ctDNA content during therapy and a responder given a checkpoint inhibitor-based regimen because of a mismatch repair gene anomaly are presented. Our results demonstrate that ctDNA evaluation is feasible in CUP and that most patients harbor a unique somatic profile with pharmacologically actionable alterations, justifying the inclusion of noninvasive liquid biopsies in next-generation clinical trials. Cancer Res; 77(16); 4238-46. ©2017 AACR.

Cancer treatments have evolved from indiscriminate cytotoxic agents to selective genome- and immune-targeted drugs that have transformed the outcomes of some malignancies¹. Tumor complexity and heterogeneity suggest that the 'precision medicine' paradigm of cancer therapy requires treatment to be personalized to the individual patient^2-6. To date, precision oncology trials have been based on molecular matching with predetermined monotherapies^7-14. Several of these trials have been hindered by very low matching rates, often in the 5-10% range¹⁵, and low response rates. Low matching rates may be due to the use of limited gene panels, restrictive molecular matching algorithms, lack of drug availability, or the deterioration and death of end-stage patients before therapy can be implemented. We hypothesized that personalized treatment with combination therapies would improve outcomes in patients with refractory malignancies. As a first test of this concept, we implemented a cross-institutional prospective study (I-PREDICT, NCT02534675 ) that used tumor DNA sequencing and timely recommendations for individualized treatment with combination therapies. We found that administration of customized multidrug regimens was feasible, with 49% of consented patients receiving personalized treatment. Targeting of a larger fraction of identified molecular alterations, yielding a higher 'matching score', was correlated with significantly improved disease control rates, as well as longer progression-free and overall survival rates, compared to targeting of fewer somatic alterations. Our findings suggest that the current clinical trial paradigm for precision oncology, which pairs one driver mutation with one drug, may be optimized by treating molecularly complex and heterogeneous cancers with combinations of customized agents.