Background: MSGT are rare with limited systemic treatments. This single institution, prospective study in recurrent or metastatic (RM) MSGT involved 2 phases: genomic profiling followed by treatment with either genomically-matched or unmatched therapy. Here we present the results of the unmatched arm for patients (pts) treated with S an oral selective inhibitor of XPO1 that leads to activation of tumor suppressor proteins and retention of oncoprotein mRNAs in the cell nucleus, inducing cancer cell apoptosis. Methods: Patients (pts) with RM-MSGT had archived paraffin embedded tumor samples profiled with targeted next generation sequencing, immunohistochemistry for androgen receptor (AR) and fluorescent in-situ hybridization for HER-2 and ALK. If no actionable mutations were identified or if no matched agents were available, pts with progressive disease could receive S (60mg given twice weekly Q28 days). The study had a simon-2 stage design; 1 partial response in the first 18 pts treated with S, would trigger an additional 7pts to receive S in stage 2. Results: Between July 2014 and April 2019 85 pts were enrolled on study: 73 had sequencing which identified 41 with no actionable mutations and 32 with actionable mutations. 18 pts (10F/8M, median age 61 years [40-79]) were treated with S and included adenoid cystic (n = 8), salivary duct (n = 4), acinic cell (n = 2) and other subtypes (n = 4). Of these 18, 4 pts had actionable aberrations: AR amplification (n = 2), mutations in SMARCB1 (n = 1) and CDKN2A (n = 1). 13pts were treatment naïve, 3pts and 2pts received 1 and 2 or more lines of treatment respectively prior to enrollment: androgen deprivation therapy (n = 2), chemotherapy (n = 3), early phase clinical trials (n = 3). The median number of cycles of S received were 3 (range: 1-19). The best response by RECIST was SD in 13pts (72%) (SD > 6 months (range: 6-18 months) in 5pts (28%); tumor reduction measured in 7pts (39%)), no PRs, PD in 3pts (17%), and 2pts (11%) were not evaluable for response due to insufficient duration of treatment coming off early due to toxicity. The median PFS (95% CI) was 7.6 (3.5-NA) months and the median OS (95% CI) was 15.4 (7.3-NA) months. The most common drug-related toxicities were grade 1-2 fatigue 14pts (78%), nausea 13pts (72%) and dysguesia 10pts (56%). 5 (28%) pts had a dose reduction and 6 (33%) in total had a dose interruption due to toxicity. Conclusions: Single agent antitumor activity was limited and the side effect profile was tolerable. No specific genomic aberration was associated with response to S. Clinical trial information: NCT02069730.