The genesis of autoimmune disorders is recognized to be the result of a complex composite of both genetic and environmental influences, including potential pharmacologic triggers, acting in concert to disrupt immune self-tolerance (1). Pemphigus vulgaris (PV), a rare, potentially fatal autoimmune disorder characterized by the production of autoantibodies against desmosomal proteins, particularly desmoglein 3 (Dsg3) and desmoglein 1 (Dsg1), serves as a classic example of an organ-specific autoimmune disease where significant genetic predisposition is well established. These autoantibodies impair cell-cell adhesion within the epidermis, causing acantholysis and giving rise to the formation of flaccid blisters and erosions on the skin and mucous membranes (2). While Dsg3 and Dsg1 are central to PV pathogenesis, recent studies have identified additional autoantibody targets (3–7), underscoring the multifactorial mechanisms driving disease progression.

The multifaceted interplay of factors influencing susceptibility to PV is reflected in its epidemiological patterns, which demonstrate significant variations in incidence, prevalence, and average age of onset across different geographical regions. These disparities are thought to result from a combination of genetic variability and diverse environmental exposures, and likely further shaped in part by the unique sociocultural and material contexts of each locale. Reported prevalence of PV ranges from a low of 0.38/100,000 individuals in Bulgaria to 30/100,000 in Iran (8). As with many autoimmune diseases, PV is often associated with additional co-morbid autoimmune conditions both in patients themselves as well as among their first-degree family members (9). An overwhelming majority of Caucasian PV patients carry either the DRB1*0402 or DQB1*0503 class II HLA alleles (10–14). However, genetic predisposition alone cannot account for disease onset in any given individual, as evidenced by the incomplete concordance observed in monozygotic twin studies in autoimmune diseases including multiple sclerosis and rheumatoid arthritis (15–17). This brings to the forefront the “exposome”, a term encapsulating the multitude of environmental and lifestyle factors—including medications, infections, psychosocial stressors, dietary factors, immunizations, and physical insults —that collectively contribute to PV etiopathogenesis (1, 18, 19). A thorough review of the literature reveals that drugs and pharmaceuticals are repeatedly identified as critical environmental triggers for PV (20, 21), with a recent analysis by our group determining that they represent the plurality (35%) of all research on environmental triggers in PV (22).

The first report of pemphigus triggered by pharmaceutical use in the literature was made by Degos et al. who in 1969 described a case of pemphigus triggered by penicillamine in a patient being treated for Wilson disease (23). Later, a 1976 study found that about 7% of patients taking penicillamine for at least 6 months went on to develop pemphigus (24). Subsequent work has brought to light a diverse range of agents implicated in the initiation and intensification of pemphigus. These compounds are typically categorized by virtue of common biochemical or structural properties, specifically into the thiol, phenol, and non-thiol non-phenol groups (1, 20, 21). Each group is associated with unique purported mechanisms in provoking pemphigus [illustrated in Figure 1 and expounded upon further elsewhere (1, 20, 21)]. Among those drugs associated with triggering PV are agents as common as lisinopril [non-thiol, non-phenol group (22)] or aspirin [phenol group (22)] and as obscure as gold sodium thiomalate [thiol group (22)]. Presently, the links between most medications and pemphigus are derived from isolated case reports (23–28) which, while valuable, are inherently limited in terms of the evidentiary weight they can offer by virtue of the anecdotal and idiosyncratic nature of case reports.

The World Health Organization (WHO) characterizes pharmacovigilance as the scientific field focused on the identification, evaluation, comprehension, and mitigation of adverse drug reactions and related issues (29). The FDA Adverse Event Reporting System (FAERS), maintained by the Food and Drug Administration (FDA), is a pharmacovigilance resource which collects reports of adverse events associated with drug and therapeutic biologic products and is designed to support drug safety monitoring in the aftermarket period (29). FAERS includes both voluntary submissions from healthcare professionals and consumers, and mandatory reports from manufacturers. As the largest such spontaneous reporting database globally, with over 11 million entries at authorship (30), disproportionality analysis of reporting in FAERS is crucial for untangling whether observed patterns of drug-adverse event reports are merely coincidental or, in fact, drug-induced. This approach is particularly valuable for detecting rare AEs that may not appear during pre-approval clinical trials.

Given the growing evidence in the literature supporting the role of environmental factors in shaping the onset and course of disease in pemphigus, the outsized place of drugs among those implicated environmental factors, and the paucity of population level data on either, we conducted a series of pharmacovigilance analyses with the intention of uncovering additional evidence to support the link between particular pharmaceutical agents previously reported to trigger PV and the development of pemphigus using publicly available, FDA-generated data. We found statistically significant, disproportionately elevated reporting of the drug-adverse event combinations of 11 out of 36 drugs investigated and pemphigus which will be detailed below, each of which constitute pharmacovigilance “signals” worthy of further investigation.

We describe a significantly increased ROR for penicillamine, gold sodium thiomalate, piroxicam, rifampin, hydroxychloroquine, imiquimod, hydrochlorothiazide, irbesartan, lisinopril, nivolumab, and nifedipine and pemphigus above that observed for all other drugs in FAERS, our comparator.

The present work has the advantage of being able to replicate the associations between a subset of the putative PV trigger drugs reported in the literature, largely across isolated case reports, in the context of a population level spontaneous reporting database with significantly larger sample sizes. While pharmacovigilance and disproportionality analysis do not establish causality, they serve to enrich the data pool available to aid clinicians and researchers in their management and study of individuals with PV. Such data is ultimately in service of the task of identifying avoidable triggers of disease and ultimately developing treatment modalities which adequately address the heterogeneous biological and clinical underpinnings and expressions of PV across individuals. While a diverse range of environmental factors such as psychosocial stressors (45), viral infections (46), and, particularly in recent years, vaccinations (47, 48), have been implicated in the onset and modification of disease in PV, our pharmacovigilance-based investigation here focuses on the critical evaluation of drugs previously linked with pemphigus.

Historically, some have proposed that “drug-induced” pemphigus arises primarily due to exogenous triggers, such as medications, with the disease resolving upon cessation of the offending agent, whereas “drug-triggered” pemphigus had been attributed to a complex interplay of endogenous, genetic, hereditary, and immunological factors, with drugs serving as secondary contributors. Previous attempts to distinguish these two etiologically drug-related subtypes of pemphigus from so-called “spontaneous” or “idiopathic” cases (49, 50) based on clinical (51) or histopathological (51) criteria have yielded inconsistent results (44), with features such as pruritus (52) and histopathological patterns (53) failing to reliably differentiate subtypes. Consistent with the views of Ruocco et al., first expressed well over a decade ago (54, 55), we propose that all cases of pemphigus likely result from the dynamic interaction of genetic predisposition with various environmental triggers, including medications, infections, stress, etc. This interplay highlights the complexity of PV’s pathogenesis, suggesting that what has been historically labeled as “spontaneous”/“idiopathic” pemphigus may, in reality, reflect unrecognized environmental factors acting on a susceptible genetic background (54), warranting further investigation into these triggers to enhance disease prevention and personalize management.

Gold sodium thiomalate, the agent found to have the highest ROR in our study, has a paradoxical relationship to pemphigus. Reports dating back to the late 1970s (40), and continuing through the early 2000s (41), document instances of pemphigus exacerbated (40) and/or triggered (42) by gold sodium thiomalate (intramuscular) therapy, as well as oral gold (auranofin) therapy (41). Conversely, the therapeutic potential of gold compounds in pemphigus has also long been recognized (56, 57). More recently, others have cautioned against its use, especially in light of the availability of newer alternatives, owing to its paradoxical ability to exacerbate the very disease it treats (54).

Antihypertensives, the largest group of drugs we found to be associated with PV, represent a therapeutic dilemma in this condition, one magnified by the prevalence of hypertension among the demographic typically affected by PV [average age of onset between 45-65 in various populations (2)] and further compounded by the chronic use of glucocorticoids, still a cornerstone of PV management. Multiple distinct classes of these drugs, including angiotensin converting enzyme (ACE) inhibitors/angiotensin receptor blockers (ARBs), calcium channel blockers (CCBs), diuretics, and beta-blockers, have been implicated in both disease induction and exacerbation in both PV (20–22, 26, 58) as well as other cutaneous diseases, as illustrated by the well-documented association of beta-blockers with psoriasis (59–61).

Of particular interest is a 2010 study of 63 hypertensive adults without dermatologic disease treated with ACE inhibitors where the sera of 33 (52.38%) such individuals was found to contain autoantibodies directed against an antigen of the superficial epidermis (62). Most notably, 7.8% and 6.4% of those same individuals were found to have anti-Dsg 3 and -Dsg 1 autoantibodies at values greater than or equal to 10 IU/mL by ELISA, respectively (1.5% and 3.2% with values above 20 IU/mL for Dsg1/3, respectively). In one individual, who was possibly exposed to multiple drugs previously associated with pemphigus in the literature (reported as captopril (44, 58, 63–65), as well as a beta blocker (66, 67) and a statin (68) – though the specific agents are not listed) autoantibody levels were as high as 26 IU/mL and 69 IU/mL for Dsg3/1, respectively – again, in the absence of any clinically apparent cutaneous disease. The intersection of antihypertensives and PV underscores a critical aspect of managing autoimmune diseases: the necessity of navigating the dual objectives of treating the primary condition while mitigating the risk of exacerbating it through necessary comorbid management strategies.

Possible mechanisms of pemphigus induction for each of the drugs found to have a positive pharmacovigilance signal in the present analysis are summarized in Table 2 . Interestingly, all 11 of the agents found to have a positive signal in this analysis exert either direct or indirect immunomodulatory activity. Direct immunomodulators include imiquimod (an immune response modifier) and nivolumab (an immune checkpoint inhibitor). Additionally, piroxicam, penicillamine, and gold sodium thiomalate (anti-inflammatory/anti-rheumatic agents) are employed clinically for their ability to alter immune homeostasis. Hydroxychloroquine, used in the treatment of a number of inflammatory and autoimmune diseases, interacts with and alters the immune system in a myriad of ways including alteration of lysosomal acidification and inhibition of toll like receptor signaling, among others, and is discussed in greater detail elsewhere (94). In terms of indirect immunomodulation, rifampin may exert its effects via the induction of pro-inflammatory cytokine expression in keratinocytes (1, 95) and/or the downstream consequences of alterations in the microbiome. Finally, the largest single group among the drugs found to have positive pharmacovigilance signals, the anti-hypertensives, have also been shown to elicit changes in immune function, specifically: lisinopril inhibits the production of Th1 cytokines including IFN-γ and IL-12 which may facilitate an immunological shift towards Th2 dominance (96, 97), Irbesartan decreases the counts and cytokine production of Th1 and Th17 cells lines and increases the number of Th2 cells (96, 97), HCTZ promotes an increase in the number of activated B-cells, and nifedipine has been shown to reverse the uremia-associated inhibition of B-cell proliferation (96). Collectively, the drugs that are statistically linked to pemphigus share, as a unifying factor, the propensity to alter immune function, which may explain their ability to trigger the onset of disease in genetically susceptible individuals. Figure 4 attempts to place the limited available evidence regarding potential immunologic sites of action of the drugs found to have significant associations with PV in this study in the context of known factors relevant to PV pathogenesis. Future mechanistic and translational investigations are necessary in order to examine direct impacts of specific medications on the immunome.

While the drugs with significant pharmacovigilance signals identified in this work are unified broadly by a capacity for immunomodulation, those drugs are, of course, not the only agents capable of exerting such effects on immune function. For the overwhelming majority of PV-associated drugs reported in the literature there is a lack of substantive experimental evidence regarding the potential immune mechanisms at play. Where evidence is available, a diverse range of pathways including biochemical drug-self-antigen interactions resulting in neoantigen formation and subsequent immune activation as well as drug induced alterations of cytokine expression patterns have variously been implicated (1, 20–22, 63, 95, 98, 99) (see Figure 1 and Table 2 ) as potential contributors to the breakdown of immune tolerance that is the sine qua non of all autoimmune phenomena.

We note that the temporal relationship between drug exposure and disease onset varies widely in the literature. For instance, one study reported an average latency of 154.27 days between drug exposure and PV onset, with ranges of 172–1140 days depending on the specific drug involved, though this study only reported data on a small fraction of all previously reported putative PV trigger drugs (21). Other reports suggest that environmental exposures occurring years, and in some cases decades (15 years (100), 53 years (49), etc.), prior to symptom onset may also contribute to disease development (49, 100–102). These findings underscore the complexity of PV pathogenesis, where both proximal (<1 year) and remote (>1 year) exposures may act as triggers, likely in the context of cumulative immune dysregulation over time.

Despite the lack of definitive mechanistic explanations linking exposure to the drugs examined here to the development of clinically significant disease, the strong statistical evidence for the association between exposure to the subset of drugs validated in this work and the development of pemphigus supports our hypothesis, and corroborates previous case report-based evidence, that those drugs likely represent trigger factors for the development of pemphigus. It must be noted that, insofar as the subset of drugs which did not produce significant pharmacovigilance signals, i.e. those that were not validated by the present approach, are concerned this work does not rule out the possibility that some or all of those agents may well represent bona fide triggers of PV which simply failed to be detected by our present pharmacovigilance-based approach. Given the severity of PV, the burden associated with both the disease itself and its still primarily glucocorticoid based treatment, and the outsized role of drugs in triggering PV as evidenced by previous systematic reviews, future studies capable of establishing causality are warranted for all reasonably suspect putative trigger drugs.

We acknowledge several limitations of this observational, retrospective, pharmacovigilance analysis. Reports in FAERS are not FDA-validated prior to database entry, and the quality of individual reports can be inconsistent, with some missing details about dosage, demographics, or the adverse event itself (30, 103). A given adverse event reported in FAERS cannot be definitively attributed to any particular drug exposure at the level of the individual (103). Sakaeda et al. highlighted the fact that while individual reports may be anecdotal, in aggregate, they can reflect a more accurate picture, writing that “[one] report in the FAERS database is a story, sometimes only a rumor, but numerous reports can reflect reality” (104). An additional limitation of this study is the potential underrepresentation of drugs that did not show significant associations with pemphigus in our analysis. The absence of a significant signal for certain drugs, despite reported associations in the literature, may reflect factors such as reporting bias, underreporting, or variability in prescription frequency rather than a true absence of etiological significance in PV. Ultimately, this pharmacovigilance disproportionality research serves as a foundation for hypothesis generation for future work on establishing conclusive evidence of causality.