# Antidepressant and Antipsychotic Drug Use and Cancer Risk: Protocol for an Overview of Systematic Reviews and Meta-Analyses

## TL;DR

This study will review existing research to determine if antidepressants and antipsychotics affect cancer risk, using both human and AI methods.

## Contribution

The study introduces a hybrid approach combining human and AI (GPT-4o) methods to accelerate systematic review tasks.

## Key findings

- The study will synthesize evidence on cancer risk associated with antidepressant and antipsychotic use.
- It will assess whether these medications are linked to decreased or increased cancer incidence and mortality.
- The study will evaluate the performance of GPT-4o in systematic review processes.

## Abstract

The relationship between cancer and central nervous system disorders has received increasing attention recently. Consequently, antipsychotics and antidepressants, commonly prescribed for conditions such as depression, bipolar disorder, and schizophrenia, have emerged as potential modulators of subsequent cancer risk. Previous studies have suggested that the use of these medications is associated with a decreased risk of cancer incidence and mortality, making them suitable candidates for drug repurposing. However, the potential therapeutic benefits do not extend to all cancer types, as some data suggest an increased risk for specific tumors.

This study aims to conduct a comprehensive review of systematic reviews and meta-analyses (review of reviews) that assess whether exposure to antidepressants or antipsychotics influences cancer incidence and mortality.

To provide a clear overview of this review, we have designed and registered the study protocol. Specifically, we will include systematic reviews and meta-analyses that examine the relationship between previous antipsychotic or antidepressant treatments and the subsequent cancer risk. The primary outcome will be the risk of cancer incidence and mortality (all malignant neoplasms) associated with exposure to psychopharmacological medications. Furthermore, secondary outcomes will include site-specific cancer incidence and mortality (eg, lung cancer). Literature searches will be conducted in multiple electronic databases (from their inception onwards), including PubMed/MEDLINE, Embase, and the Cochrane Database of Systematic Reviews. Three researchers will independently screen all citations, abstracts, and full-text articles. We will perform parallel search, selection, and extraction tasks using a large language model (GPT-4o; OpenAI). Data selection and extraction will involve both human reviewers and GPT-4o, whose performance will be validated through human evaluations. Thus, we will verify whether this type of tool can accelerate or even perform the tasks involved in a systematic review. The risk of bias and the quality of individual studies will be evaluated using appropriate tools. Subsequently, we will extract the summary association measures (eg, pooled relative risk, odds ratio, and hazard ratio) as reported in each included systematic review. Where available, we will summarize subgroup and sensitivity analyses as described by the authors.

Planned searches will be conducted in various electronic databases from their creation until September 2025. No results are available or included in this protocol. The expected results will be published in 2026.

This overview of systematic reviews and meta-analyses will provide an updated synthesis of the cancer risk associated with antipsychotic and antidepressant drugs. Furthermore, this study will examine factors that may explain potential study variations. Ultimately, these findings will be published in a peer-reviewed journal.

OSF Registries 10.17605/OSF.IO/5ACWH; https://osf.io/5acwh/overview

DERR1-10.2196/78596