Responding to a literature of doubt: limitations of three studies of adverse events of finasteride and dutasteride

A fourth study has been published in the ‘literature of doubt’. Read the rebuttal. Background on the co-author of Baas et al, 2018 with a disclosed conflict of interest is here.

Three studies analyzing adverse event (AE) data related to finasteride and dutasteride were published from 2018–2020 (the “AE papers”).^1–3 The first to appear, by Baas et al., included Kevin T. McVary, MD as a co-author. The article disclosed that Dr. McVary had been hired by Merck as an expert for the company’s defense in Propecia litigation¹:

The validity of finasteride AEs was at issue in Propecia litigation. The Discussion in Baas et al. includes this excerpt:

Given the nature of the FAERS reporting, any condition may have been attributed to finasteride. For example, a patient could claim to have worsened sexual function after becoming aware of potential side effects, change in relationship status, other psychosocial events, or social media interactions. In the absence of objective findings, it is impossible to attribute causality to use of 5ARIs.¹

The other two papers bear striking similarities to Baas et al. in their design, findings, discussion and conclusions. All three raise concerns about the validity of AEs after 2012 in light of a 2012 label change by the Food and Drug Administration (FDA), related news coverage, Internet activity, litigation, and publicity by the Post-Finasteride Syndrome Foundation. The AE papers have some blind spots and methological limitations in common. These papers: 1) overlook the role of social context in mediating the flow of AE reports; 2) exhibit a bias for the status quo prior to publicity; and 3) overlook the impact of different regulatory status when comparing AE reporting for finasteride and dutasteride. Finally, Nguyen et al. cites background literature lending mixed support to its arguments. This last point will be covered first.

Mixed support by cited literature in Nguyen et al., 2020. Three of five references on stimulated reporting and a nocebo effect in Nguyen et al.³ described null, transient or irrelevant effects. A paper cited in reference to a nocebo effect concluded (emphasis added): “modern day AE reporting trends do not appear to be substantially affected by FDA alerts.”⁴ Another found increased suicides following news stories about celebrity suicides (irrelevant here), but no increase in suicides after general reporting on suicide.⁵ A third cited paper found a stimulated reporting effect which reverted to baseline after approximately two months.⁶ By contrast, aggregate AE reports of finasteride to the World Health Organization’s (WHO) VigiBase⁷ show an upward linear trend over the 10-year period ending in 2019, the last year included in the analysis of Nguyen et al. Publicly available VigiBase data is not segmented, but FAERS data for men aged 18–40 show a similar upward trend:

Authors of a fourth cited paper do not establish that stimulated AE reports were spurious, and acknowledge that a previous analysis came to the opposite conclusion based on different AE inclusion criteria.⁸ This paper and a fifth cited paper⁹ refer to post-publicity AE reports respectively as “notoriety bias” and “reporting bias.” These terms appear in other literature on AEs; nevertheless, their negative connotation implies post-publicity reports are of lower value, overlooking the possibility that a previously underreported signal has been elicited by publicity.

Omission of social context. The AE papers overlook the mediating role of social context in AE reporting. For reports to reach pharmacovigilance databases, men must be moved to report intangible, embarrassing and stigmatized subject matter such as male sexual dysfunction, penile alterations, depression, anxiety and cognitive problems. Another potential limiting factor is the emergence of dysfunctions after discontinuing the drug. From a drug safety and public health point of view, one might welcome media stories and online discussions that, over a period of years, embolden men to overcome their reticence and report their experiences. The AE papers seem to imply that pharmacovigilance databases are in a preferred state prior to media stories and social media activity, and stimulated reporting would add noise. The opposite is entirely plausible: prior to news coverage, certain AEs may underreported—for example, because of social stigma concerning male sexual dysfunction and psychological AEs. The AE papers suggest that changes in AE reporting after 2012 reflect “reporting bias.” Offsetting this interpretation is the question: “What signal was not getting through prior to 2012?” There is room for debate on these questions, but the AE papers overlook the counterargument.

Implication that stimulated reports are invalid. The AE papers make a credible case that some combination of news coverage, internet activity, advocacy and litigation contributed to stimulated reporting. From a pharmacovigilance perspective, is this good, bad or neutral? The AE data provides no prima facie basis for determining whether a stimulated AE report is more or less valid than a non-stimulated report. The veridicality of stimulated reports vs. non-stimulated reports is fair game for debate, but the AE papers do not take up this question.

Omission of regulatory context in comparison with dutasteride AEs. The AE papers point out differences in AE reporting for finasteride vs. dutasteride, suggesting the low number of dutasteride reports calls the validity of finasteride AE reports into question. Baas et al. wrote: “Interestingly there were no PFS [post-finasteride syndrome] symptoms reported for dutasteride in the 1 monotherapy case in the FAERS database.”¹ In a comment on Nguyen et al., Ho wrote (emphasis added):

Third, and most importantly, reporting bias can similarly occur in these adverse event databases. In particular, the quality and quantity of reports in these databases can be influenced by reporting behavior, such as false reports submitted by patients who are dissatisfied or involved in litigation, or stimulated reporting affected by mass media. In fact, the sensitivity analyses by Nguyen and colleagues did reveal such reporting bias. The lack of suicidality signal observed for dustasteride, a drug that is similar to finasteride in mechanism of action but has not attracted as much media attention, suggests that there may be something unique about finasteride that may have resulted in the observed suicidality signals, but it also hints at a potential reporting bias unique to finasteride. Considering that the publication of the first clinical study suggesting an association between suicidality and finasteride, the founding of the Post-Finasteride Syndrome Foundation, and the start of Google Trends showing an increase in searches for “post-finasteride” all took place in 2012, their ultimate stratified analysis showing a significant signal after 2012 but none before 2012 supports the potential presence of a reporting bias or stimulated reporting accounting for the observed suicidality signal for finasteride.¹⁰

Not mentioned is the possible impact of different regulatory status of finasteride and dutasteride on AE reporting. Dutasteride is only approved for the treatment of androgenetic alopecia in a handful of countries such as Japan and South Korea (I have been unable to identify other countries where it is approved for this indication). Off-label use in most countries would tend to limit the number of prescriptions written to young men seeking treatment for hair loss in the first place, and may introduce hesitancy of health care providers to file AE reports because of liability concerns regarding off-label prescriptions. Difference in regulatory status may also introduce a geographic confound, since AE reports for dutasteride may come from different countries or regions than those of finasteride. Attitudes towards psychological AEs, and propensity to report them, may differ in these countries compared to the geographic sources of finasteride AE reports.

Nguyen et al. did not include counts of dutasteride AE reports.³ When these data were requested from the corresponding author, Dr. Trinh declined to do so, describing it as an “additional analysis” which was “outside the scope of this initial project.” An extrapolation from public VigiBase data produces an estimate of 16.2 reports per year concerning dutasteride taken by men ages 18–45 (the number of these which included psychological AEs is unknown but may be lower). As further context on dutasteride AE reporting, the table below shows an average of 4.4 AE reports submitted to FDA per year for dutasteride taken by men 18–40. In a discussion of limitations, Nguyen et al. state that underreporting of “some adverse events…is mitigated by the breadth of data collection (153 countries) and high volumes of reports.” Publicly available information suggests this was not the case for dutasteride AEs. Taken together, the different regulatory status of these drugs, low baseline volume of reports for dutasteride taken by younger men, and a potential geographic confound call into question whether a comparison of finasteride and dutasteride AEs is appropriate or meaningful.

The AE papers are the product of high-profile people, institutions and journals. One was published in the highest-impact journal in its field, JAMA Dermatology, with authors affiliated with Brigham and Women’s Hospital and Harvard Medical School.³ Another paper’s co-author is the current Chair of the Office of Research of the American Urological Association.² Given one co-author’s conflict of interest as an expert for a drug maker in litigation, the stature of the authors, the similarity of these papers and their emergence in a three-year period, this leaves the impression of a professional guild mobilizing to defend its interests.¹¹ Claims of life-altering harms from an approved drug represent a challenge to the specialties which regularly prescribe it: urology and dermatology. The AE papers arguably show these fields generating a “literature of doubt,” pressing one set of arguments while overlooking or giving short shrift to context and counterarguments. These articles come across as an assertion of power to decide the ontological status of men’s reports of harms from finasteride. On this basis I conclude that they primarily serve guild and industry interests, with the interests of science and the public coming in second.

References

1. Baas WR, Butcher MJ, Lwin A, et al. A Review of the FAERS Data on 5-Alpha Reductase Inhibitors: Implications for Postfinasteride Syndrome. Urology. 2018;120:143-149. doi:10.1016/j.urology.2018.06.022
2. Harrell MB, Ho K, Te AE, Kaplan SA, Chughtai B. An evaluation of the federal adverse events reporting system data on adverse effects of 5-alpha reductase inhibitors [published correction appears in World J Urol. 2021 Nov;39(11):4293]. World J Urol. 2021;39(4):1233-1239. doi:10.1007/s00345-020-03314-9
3. Nguyen DD, Marchese M, Cone EB, et al. Investigation of Suicidality and Psychological Adverse Events in Patients Treated With Finasteride. JAMA Dermatol. 2021;157(1):35-42. doi:10.1001/jamadermatol.2020.3385
4. Hoffman KB, Demakas AR, Dimbil M, Tatonetti NP, Erdman CB. Stimulated reporting: the impact of US Food and Drug Administration-issued alerts on the Adverse Event Reporting System (FAERS). Drug Saf. 2014;37(11):971-980. doi:10.1007/s40264-014-0225-0
5. Niederkrotenthaler T, Braun M, Pirkis J, et al. Association between suicide reporting in the media and suicide: systematic review and meta-analysis. BMJ. 2020;368:m575. doi:10.1136/bmj.m575
6. MacKrill K, Gamble GD, Bean DJ, Cundy T, Petrie KJ. Evidence of a Media-Induced Nocebo Response Following a Nationwide Antidepressant Drug Switch. Clin Psychol Eur. 2019;1(1):1-12. doi:10.32872/cpe.v1i1.29642
7. WHO Collaborating Centre for International Drug Monitoring, Uppsala Monitoring Centre (UMC). VigiAccess. Accessed February 5, 2022. http://vigiaccess.org/
8. Fadini GP, Sarangdhar M, Avogaro A. Pharmacovigilance Evaluation of the Association Between DPP-4 Inhibitors and Heart Failure: Stimulated Reporting and Moderation by Drug Interactions. Diabetes Ther. 2018;9(2):851-861. doi:10.1007/s13300-018-0408-2
9. De Bruin ML, van Puijenbroek EP, Egberts ACG, Hoes AW, Leufkens HGM. Non-sedating antihistamine drugs and cardiac arrhythmias — biased risk estimates from spontaneous reporting systems? Br J Clin Pharmacol. 2002;53(4):370-374. doi:10.1046/j.1365-2125.2002.01569.x
10. Ho RS. Ongoing Concerns Regarding Finasteride for the Treatment of Male-Pattern Androgenetic Alopecia. JAMA Dermatol. 2021;157(1):25-26. doi:10.1001/jamadermatol.2020.3384
11. Chervenak FA, McCullough LB, Hale RW. Guild interests: an insidious threat to professionalism in obstetrics and gynecology. Am J Obstet Gynecol. 2018;219(6):581-584. doi:10.1016/j.ajog.2018.09.007

Table: Adverse event reports of dutasteride and finasteride for men 18–40