Preclinical data identify the serotonin transporter gene SLC6A4 (encoding SERT) as an immune checkpoint on CD8+ T cells, and mouse experiments suggested that selective serotonin reuptake inhibitors (SSRIs) can enhance antitumor immunity when combined with anti‑PD‑1 therapy. Whether this mechanism translates into clinical benefit for patients receiving immune checkpoint inhibitors (ICIs) has been uncertain. Because depressive and anxiety disorders are common among patients receiving immunotherapy and are treated with different psychotropic classes, the choice of psychiatric medication offers an opportunity to examine real‑world associations between SSRI exposure and ICI outcomes.
The investigators emulated a target trial using the TriNetX federated electronic health record network, which contains records for more than 187 million patients. Eligible adults had a solid tumor, a first ICI administration between 2015 and 2025, and a recorded diagnosis of depressive or anxiety disorder within the prior 365 days. Exposure classification required an active SSRI or benzodiazepine (BZD) prescription overlapping the ICI index date or starting within the preceding 90 days; a patient was categorized according to the psychiatric drug class in use at ICI initiation.
The primary estimand was an intention‑to‑treat analogue comparing SSRI versus BZD users at ICI initiation. Propensity score matching (1:1) balanced 49 baseline covariates to create matched pairs. The primary clinical outcome was 2‑year overall survival. Twenty secondary outcomes were prespecified across four mechanistic pathways, and three negative control outcomes (nephrolithiasis, cholelithiasis, cataract) were included to assess residual confounding. Hazard ratios (HRs) with 95% confidence intervals (CIs) were the principal effect measures. Transcriptomic analyses were conducted separately using The Cancer Genome Atlas (TCGA) bulk tumor data.
After propensity score matching, the analysis included 1,567 matched pairs (3,134 patients). Over two years, all‑cause mortality was lower in the SSRI group: 23.5% versus 34.4% in the BZD group (HR 0.627; 95% CI 0.549–0.716; p < 0.001). The authors reported an E‑value of 2.57 for the point estimate to quantify the potential impact of unmeasured confounding.
Among secondary outcomes, newly coded distant metastases occurred less frequently in SSRI users (HR 0.629; 95% CI 0.487–0.811; p < 0.001). The three prespecified negative control outcomes showed null associations (all p > 0.76), supporting specificity of the main associations. The study also reported that each of the five SSRIs analyzable individually was associated with lower mortality (HR range 0.61–0.71).
Composite immune‑related adverse events (irAEs) were more frequent with SSRI use (HR 1.162; 95% CI 1.039–1.299; p = 0.008). This increase was driven primarily by thyroid dysfunction (HR 1.199; 95% CI 1.053–1.366; p = 0.006). There were no statistically significant increases in hepatitis, pneumonitis, or colitis (all p > 0.09). The pattern of irAEs was described as organ‑selective, with thyroid dysfunction accounting for most of the observed safety signal.
Separately from the clinical EHR analyses, the authors interrogated bulk tumor transcriptomes from TCGA (8,272 tumor samples from 8,172 patients across 20 tumor types). They examined associations between SLC6A4 expression and measures of tumor immune‑cell infiltration and T‑cell inflammation. In 13 of 20 cancer types, higher SLC6A4 expression was inversely correlated with T‑cell inflammation, suggesting lower immune infiltration where the transporter gene was more active. The organ‑level patterns of SLC6A4 expression reportedly aligned with organ‑specific patterns of irAEs observed in the clinical analysis.
The transcriptomic code and derived data files used for these analyses were made publicly available by the authors and archived in a GitHub repository and Zenodo, and the underlying TCGA projects used are publicly accessible through the National Cancer Institute Genomic Data Commons.
The study authors emphasize that residual confounding is the main limitation. Key clinical variables predictive of cancer outcomes—most notably performance status (for example ECOG)—are not captured in TriNetX, and benzodiazepines and SSRIs are prescribed in different clinical circumstances. BZDs are often used for acute distress or in more severe illness, which could bias comparisons even after extensive matching. The TCGA transcriptomic analyses used bulk tumor tissue from patients not treated with ICIs; therefore, these data cannot localize SLC6A4 expression to specific cell types (tumor cells versus infiltrating immune cells) nor demonstrate that SSRI exposure alters tumor microenvironments in humans.
All aggregate values for event counts, risks, HRs, CIs, and p values for matched analyses are reported in the article and supporting information; patient‑level TriNetX records remain within participating institutions and cannot be redistributed by the authors.
In this large observational target trial emulation, SSRI use at ICI initiation among patients with solid tumors and coexisting depression or anxiety was associated with lower 2‑year mortality and fewer newly coded distant metastases compared with benzodiazepine use, alongside a modest increase in organ‑selective immune‑related adverse events largely limited to thyroid dysfunction. Parallel TCGA transcriptomic analyses provide biological plausibility by showing inverse correlations between SLC6A4 expression and T‑cell inflammation in many tumor types.
Because patients were not randomized to SSRI versus BZD, the results demonstrate association rather than causation. The authors conclude that these findings justify prospective randomized evaluation of SSRI repurposing as an adjuvant to ICI therapy to determine whether SSRIs can safely and effectively improve outcomes in patients receiving immune checkpoint blockade.