Salivary gland cancers (SGC) are a rare group of tumors with an incidence of about 1.3 cases/100,000 individuals in the United States (1). More than 20 distinct malignant subtypes have been described, many of which are defined by recurrent genetic alterations (2).

Salivary duct carcinoma (SDC) is an aggressive high-grade SGC subtype with a dismal prognosis. SDC most commonly arises in the parotid gland and accounts for about 1.8% of major salivary gland tumors in the SEER database (3–5). SDC can arise de-novo or ex pleomorphic adenoma (ex-PA) (4). Due to the aggressive nature of this disease, metastatic spread and a need for systemic therapy is frequent (6).

In addition to chemotherapy, targeted treatment strategies are increasingly used in SDC. SDC harbors recurrent molecular alterations such as HER2 and androgen receptor (AR) amplification and overexpression. Furthermore, FGFR1 amplification, PIK3CA, HRAS and TP53 mutations and PTEN and CDKN2A loss have been described (2, 4, 7, 8). Some of these alterations have been applied as predictive biomarkers for targeted therapy. Previous prospective studies have shown activity of HER2, AR and HRAS-directed therapy in SDC (9–11). Additionally, a prospective basket study showed a benefit of targeted therapy (targeting HER2 amplification, HER2, BRAF and PTCH1 mutation and high tumor mutational burden) in a large group of SGC, including SDC (12). No prospective trials supporting the efficacy of PI3K-inhibitors in PIK3CA-mutant SDC currently exist. A summary of ongoing and published clinical trials relevant for metastatic salivary duct carcinoma is provided in Table 1 .

These results have led to a recommendation of comprehensive molecular analyses (e.g. next-generation panel or whole-exome sequencing) in patients with advanced SDC. These analyses should be done to assess opportunities for targeted therapy, including HER2- or AR-directed treatment (15, 16). Available data correspond to ESMO Scale of Clinical Actionability (ESCAT) scores of II-B (i.e. investigational therapy, alteration-drug match is associated with antitumor activity but magnitude of benefit is unknown) for AR (>70% positivity by immunohistochemistry, IHC) and HER2 (IHC score 3+ or fluorescence in situ hybridization positivity) in SGC. A participation in clinical trials is strongly recommended (15–17). The use of immune checkpoint inhibition remains investigational (15, 16). However, the FDA-approval of pembrolizumab in tumors with high tumor mutational burden also includes SGC (18).

In SDC, several targetable molecular alterations occur in recurrent patterns. The resulting subgroups of SDC are mainly defined by HER2- and AR-expression. In a retrospective analysis of 63 SDC samples, 34 samples were AR+/HER2- and harbored frequent PIK3CA (50%) and HRAS (41%) mutations (19). In this study, HRAS-mutations were exclusively found in the HER2-/AR+ group and in 93% of cases they co-occurred with a PIK3CA-mutation (19). Additionally, no HRAS mutations were identified in SDC ex pleomorphic adenoma (19). The co-occurrence of three potentially predictive biomarkers complicates selection for targeted treatment decisions in these rare patients. We here present a case series of patients presenting to an institutional molecular tumor board or identified through a systematic search of the literature to assess outcome of AR+, HRAS/PIK3CA SDC patients with targeted treatment.

This is, to the best of our knowledge, the largest clinical case series of AR+, PIK3CA/HRAS co-mutated salivary duct carcinoma. The co-occurrence of these alterations has been described in previous analyses of this disease but was not associated with prognosis (19). The co-occurrence of these alterations poses a challenge for personalized therapy strategies. It is currently unclear, if response to targeted therapy is different in this subgroup. Despite the low number of patients, the heterogeneity of administered treatments in the cohort and the lack of data on mechanisms of secondary resistances, these results still hold important information because of the rarity of this disease. It should be noted that at least 2/13 patients in the cohort did not receive upfront palliative systemic therapy with ADT or chemotherapy (15, 16). This is likely caused by the lack of guidelines at the time of treatment initiation and a lack of data for the specific situation in AR+, HRAS/PIK3CA co-mutated tumors. Yet, despite available data and guidelines, current treatment strategies are not satisfactory and there is a lack of prospective clinical trials. The administration of experimental therapies, including immune checkpoint inhibition, will therefore likely remain a reality in these tumors, thus making sharing of real-world data essential.

HER2 overexpression or amplification is a well-defined therapeutic target in SDC (11). HER2 positivity has been found to be mutually exclusive with the AR+, HRAS/PIK3CA co-mutated subgroup (19). In this cohort, concurrent HER2 expression or amplification was reported in 6 patients with mostly low to moderate staining intensity. One patient was reported to harbor a concurrent HER2 amplification and received trastuzumab in combination with chemotherapy, achieving a partial response. These results suggest, that HER2-positivity is not entirely mutually exclusive with the here described subgroup and HER2-directed treatment might be an additional option in some of the patients. In addition to HER2-directed antibodies, efficacy of HER2-directed antibody drug conjugates was shown in salivary gland cancer (24).

In the group of patients treated with ADT, a clinical benefit was observed in about a third of the evaluable patients, which is less than the clinical benefit rate of about 75% in previously reported results (10). However, clinical benefit was found in some patients, thus providing evidence of activity and the low number of patients does not allow further conclusions. Furthermore, previous work did not show an impact of oncogenic drivers, including HRAS and PIK3CA mutations, on the efficacy of ADT but studies on larger cohorts are warranted (10). Therefore, these data do not provide evidence against the use of ADT in the AR+, HRAS/PIK3CA co-mutated subgroup. Importantly, no data currently exist for the use of abiraterone in castration-resistant AR+, HRAS/PIK3CA co-mutated SDC. Abiraterone is active as a second line ADT in AR+ salivary gland cancer and might represent an additional treatment strategy for the here reported subgroup (13). Additional data on limited activity with enzalutamide were reported previously (14). In this phase 2 trial, tumor regressions were also noted among patients with prior ADT (14).

In patients receiving HRAS-directed therapy, 1 objective response and clinically meaningful disease stabilizations in about half of the patients was reported. These results are similar to previously reported results in SGC (9). In the same trial, co-occuring PIK3CA-alterations or the type of HRAS mutation (more common Q61 or less common G13) did also not seem to impact treatment efficacy (9). Again, these data further support the investigation of tipifarnib in the here reported disease subgroup.

Two patients received combination therapy after prior progression of disease. One patient achieved disease stabilization with tipifarnib and ADT after prior progression with tipifarnib monotherapy. In this patient, tipifarnib treatment was continued because of low toxicity and improvement in local symptoms. However, disease stabilization might be mediated by ADT alone. Another patient achieved a partial response with the PI3K-inhibitor alpelisib and ADT. The impact of the individual drugs in this combination therapy can also not be assessed. Further investigation of combination therapies is warranted.

A single agent PI3K-inhibitor was only used in 1 patient and no response data were available. The published results from the NCI-MATCH subprotocol Z1F of Copanlisib in PIK3CA-mutated cancer did show activity of the drug in several cancer types, but no SDC were enrolled (25). Additional research is needed to establish the activity of single-agent PI3K-inhibitors in SDC.

The activity of immune checkpoint inhibition also remains to be determined. One patient in the reported cohort achieved a mixed response with single-agent PD-1 blockade for 7 months, followed by disease progression. The same patient then experienced disease progression with combined PD-1 and CTLA-4 inhibition. Immune checkpoint inhibition was administered in this patient because of high PD-L1 expression and the co-occurrence of driver alterations, potentially complicating single-agent targeted treatment. These results suggest, that immune checkpoint inhibitors might be an additional treatment option in some patients. An analysis of 109 patients with advanced SGC in the Keynote-158 study showed an overall response rate of 4.6% in the overall population and 10.7% (n=3/28) in the PD-L1 positive population (26). Only 3 patients in this trial were found to be TMB-high, among which 1 had an objective reponse (26). PD-L1 expression or a high TMB > 10 mut/Mb was reported in 2 patients in the here reported cohort, accordingly.

For untargeted therapies, best data currently exist for carboplatin/paclitaxel use, which is also supported by previous analyses and yields clinically meaningful benefit (27).

In summary, the here provided data show multiple targeted treatment strategies for patients with AR+, HRAS/PIK3CA co-mutated SDC. Best available evidence, expected toxicities and patient factors need to be considered for a choice of treatment in this rare subgroup. These results support comprehensive molecular profiling of SDC. Additional molecular analyses might help with further establishing active signaling pathways for treatment stratification. Further research is required to establish optimal treatment combinations and sequences, which is a challenge in this rare disease.

The original contributions presented in the study are included in the article/ Supplementary Material . Further inquiries can be directed to the corresponding author.

The studies involving human participants were reviewed and approved by Charité - Universitätsmedizin Berlin. Written informed consent for participation was not required for this study in accordance with the national legislation and the institutional requirements.

DR, SS, PS and KK provided patient data, DR performed a systematic review of the literature, DR, EZ, BE, DB, EB, UK (8^th Author), UK (9^th Author) and KK analyzed data. DR, UK (9^th Author) and KK wrote the manuscript. All authors contributed to the article and approved the submitted version.