Response to: Campbell K, Velazquez O, Sullivan J, Lipshultz L. 8 Finasteride-associated suicide and depression in men treated for hypogonadism and impotence. J Sex Med. 2022 Apr;19(Suppl 1):S4–S5. doi:10.1016/j.jsxm.2022.01.019
- In an analysis of adverse event data, there was an elevated risk of reports of suicidality, anxiety disorders and depressive disorders in men who used finasteride only.
- In cases where finasteride was used with other therapies, these risks were not elevated.
- Inferences about a protective role for combination therapy are unsupported because they are derived from spontaneous reports about unrelated individuals.
- The abstract appears to incorrectly assume that psychological adverse events linked to finasteride are necessarily due to sexual dysfunction. It is unknown whether cases using finasteride only even had sexual dysfunction.
The clear and concerning results in this abstract are statistically significant RORs for suicidal and self-injurious behaviors, anxiety disorders and depressed mood disorders in the finasteride-only group. The suggestion that combination therapy could “decrease” risk or have a “protective” role is unwarranted because the risk ratios are derived from unrelated sets of cases. This study seems to treat pharmacovigilance data as if it were a controlled, prospective study, when the data is uncontrolled and retrospective.
Reporting odds ratios (RORs) of three classes of neuropsychiatric events are elevated for F but null for F+T:
|F: ROR||F: # cases||F+T: ROR||F+T: # cases|
|Suicidal & self-injurious behaviors||1.89 (1.74–2.04)||611||0.80 (0.35–1.78)||6|
|Anxiety disorders||2.51 (2.41–2.62)||2352||1.25 (0.89-1.76)||36|
|Depressed mood disorders||4.60 (4.41–4.80)||2392||1.23 (0.88-1.72)||36|
The authors suggest that testosterone might be “protective” against suicidal and self-injurious behaviors. In doing so, they appear to take a null result as confirmation of the lack of a risk, and distinguish it from the elevated ROR for F cases. This is a spurious distinction, since a null ROR does not confirm the lack of an association. There were, morover, only 6 cases in F+T compared with 611 in F. The F+T group therefore is likely unrepresentative of the population using finasteride.
Antonazzo, et al., 2019 described the limitations of relying on a null finding when exploring potentially protective drug-drug combinations in pharmacovigilance data:
…lack of disproportionality should not be automatically interpreted as an endorsement of drug safety. In other words, a protective drug-related effect cannot be claimed because of inherent limitations: voluntary nature of unsolicited reports, influence of reporting by external factors, lack of certainty on causality, unknown denominator (population exposure), as well as interdependence of disproportionality measures. 
Among users of finasteride and phosphodiesterase 5 inhibitors (PDE5i), RORs were significantly reduced (0.22; CI 0.12–0.39); however, this cannot be compared to an elevated risk in the finasteride-only group, as these are unrelated sets of cases. It is possible that men who seek out PDE5 inhibitors have different characteristics from men using monotherapy. For example, they might be older and taking finasteride for different reasons such as an enlarged prostate. This table shows the difference in the size of each group:
|ROR: F||# cases: F||ROR: F+PDE5i||# cases: F+PDE5i|
|Suicidal & self-injurious behaviors||1.89 (1.74–2.04)||611||0.22 (0.12–0.39)||11|
|Anxiety disorders||2.51 (2.41–2.62)||2352||0.21 (0.16–0.27)||64|
|Depressed mood disorders||4.60 (4.41–4.80)||2392||0.25 (0.19-0.33)||55|
The abstract seems to assume a chain of reasoning: suicidality, anxiety disorders and depressive disorders linked to finasteride are necessarily consequences of sexual dysfunction (SD); PDE5i medications treat symptoms of SD; therefore, if the F group treated SD with PDE5i, their neuropsychiatric symptoms would have been reduced. But it is not clear that individuals in the F-only group even had SD. Finasteride could have a direct causal role in neuropsychiatric events, for which there is a plausible pharmacological basis .
The abstract and a related interview in Urology Times use words such as “decreased” risk and a “protective” effect, implying that these groups have participated in a controlled, prospective study. In fact, no inferences about treatment effects can be made because the statistics are based on spontaneous pharmacovigilance reports from completely unrelated individuals. They represent a wide range of ages, reasons for taking finasteride, dose (1 mg or 5 mg) and duration of use, among other potential confounders.
Even taken at face value, these recommendations would entail younger men adding a PDE5i or testosterone to their regimen, therapies which are not without risks. If a man in his 20s is taking finasteride for hair loss over a 10-year period, is it recommended that he add a PDE5i or testosterone therapy for the duration to avert psychological symptoms? Does the incremental benefit outweigh the risks of these therapies over the long term? Is it desirable to have men taking pills for erectile dysfunction in their 20s, so that they can take another pill to try to stop hair loss, in order to avoid suicidal thoughts, anxiety and depression?
The abstract’s conclusions divert attention away from elevated risks of suicidality, anxiety and depression associated with finasteride only, suggesting that these are treatable based on misinterpretations of highly variable data. Moreover, they seem to rely on an unwarranted assumption that psychological adverse events are a consequence of drug-related sexual dysfunction. The results presented do not support a recommendation of combination therapy, but prompt the question of whether patients are being adequately informed of the risks of psychological adverse events associated with finasteride.
- Antonazzo IC, Poluzzi E, Forcesi E, et al. Signal of potentially protective drug-drug interactions from spontaneous reporting systems: proceed with caution. Acta Diabetol. 2020. doi:10.1007/s00592-019-01441-3
- Walton NL, Antonoudiou P, Maguire JL. Neurosteroid influence on affective tone. Neurosci Biobehav Rev. 2023. doi:10.1016/j.neubiorev.2023.105327