Pemphigus Vulgaris (PV) is a rare autoimmune blistering disorder characterized by mucosal and/or skin lesions due to the presence of autoantibodies against desmosomal proteins involved in cell-adhesion. Autoantibody binding leads to suprabasilar acantholysis in the epithelium resulting in superficial blisters that rupture easily and often present as open erosions. Like other autoimmune diseases, the etiology of PV is complex and multifactorial, involving genetic, environmental, and immunologic factors. Moreover, there is considerable clinical heterogeneity within given autoimmune diseases, including PV, that is unexplained and confounds clinical management. Numerous studies have reported a strong association between certain HLA alleles and PV, particularly the class II HLA alleles DRB1*0402 and DQB1*0503 (1). HLA alleles are major players in autoimmune development that predispose individuals to target self-antigens (2). These alleles are highly polymorphic genes of the major histocompatibility complex (MHC) coding for HLA class I and II molecules that are instrumental in antigen presentation for the initiation of immune responses, including those directed against self-antigens in the case of autoimmunity (3). While it is established that specific HLA alleles are a requirement for autoimmune development, the direct links between genetic susceptibility elements, the specificity of the autoimmune response and the ultimate clinical presentation of disease have not been clearly mapped.

There is some evidence in the literature indicating that carrying a particular HLA allele associated with a certain autoimmune condition may also influence disease presentation. For example, the DRB1*0103 allele is associated with the isolated colonic form of Crohn’s disease (4). In rheumatoid arthritis, a shared isotope involving 5 amino acids in the DRB1 locus is associated with earlier disease presentation and more severe bone erosions (5). For PV, the relationship between the expression of specific HLA alleles and disease presentation remains unclear, previous studies were limited by sample size and conflicting results (6–8). Svecova et al. reported that DRB1*0402 and DQB1*0302 were associated with more severe disease, but Le et al. conversely found that DRB1∗04 alleles are likely to be associated with mild and moderate disease. Notably these studies had no more than 50 PV patients per study, and the two conflicting studies are from Slovakia and India, two countries with relatively ethnically homogeneous populations.

Using a large biorepository of PV patients collected over two decades with well-annotated data on disease presentation, we set out to systematically assess for patterns linking a multitude of disease relevant parameters including ethnicity, HLA type, autoantibody levels, and lesion morphology. We confirm previously described HLA associations in well-studied ethnic populations, but also find an unexpected HLA association in the previously underrepresented group of PV patients of African descent. We also uncover associations of specific HLA allele expression with autoantibody profiles and phenotype. This information deepens our understanding of key pathogenetic mechanisms in PV that direct disease activity and underlie clinical heterogeneity, with relevance to improving management strategies targeted to patient subpopulations. This work illuminates previously unrecognized patterns linking genetic susceptibility, autoimmune specificity and disease phenotype.

It is understood that HLA molecules play a crucial role in the HLA-peptide-TCR tripartite that is crucial for mounting adaptive immune system responses to both foreign pathogens as well as self-tissue targets in the case of autoimmune disease (17). By directing the autoimmune response and the effector cells and autoantibodies that drive it, we speculate that HLA alleles also help shape the clinical presentation brought on by these antibodies.

Our data confirms and expands on previously determined associations between specific HLA alleles and PV. For instance, for DRB1, we confirm the well-known association of the 0402 allele and PV (1) in the larger study population analyzed here, where DRB1*0402 is the most common pemphigus-associated allele. The second most common specific DRB1 allele in our population was found to be 1404, which has been suggested as possibly associated with PV in Chinese and Pakistani studies, but its significance has yet to be determined (18, 19). In our population, DRB1*0402 is paired with DQB1*0302 in almost all cases (172 out of 175 total). Our group has previously shown that DQB1*0302 is most likely in itself not a major contributor to disease, but is inherited together with DRB1*0402 because of strong linkage disequilibrium (1). Following DQB1*0302, we found DQB1*0503 as the most common DQB allele, which has been reported for PV previously (1).

It is striking, however, how the distribution of these two PV-associated HLA alleles varies between ethnic groups. PV patients identifying as being of Ashkenazi Jewish descent are predominant carriers of the DRB1*0402 allele (95% of patient within this ethnic group carry at least one DRB1*0402 allele). This is in line with (i) studies on Jewish PV patients from Israel that report around 90% DR4 positivity (20, 21) and (ii) the reported allele frequency of up to 15% of this allele in the general Israeli population (22, 23). Despite the high prevalence of DRB1*0402 in this population, PV is still rare within Israel, highlighting the multifactorial genesis of PV in which other factors in addition to the HLA background ultimately led to disease expression.

DQB1*0503, on the other hand, is the predominant allele in the South Asian population (in our study 82% of all South Asian patients carry this allele). An overabundance of this allele has been shown in this population before (allele frequency of up to 13% in a study from India) (24, 25). Despite the fact that both DRB1*0402 and DQB1*0503 are fairly rare amongst European populations, these are the two leading alleles within non-Jewish Caucasian PV patients who show a fairly equal distribution of DRB1*0402 and DQB1*0503 (55% and 36% of all non-Jewish Caucasian patients, respectively).

In addition to the accepted HLA alleles, our study has uncovered a previously underappreciated allele in PV, namely DRB1*0804. While only 6% of our entire patient population are DRB1*0804 positive, it should be noted that this allele is most commonly seen in African American PV patients, of which 64% are carriers. Since only one of these African-American patients carried the previously PV-associated allele DQB1*0503 and none carried DRB1*0402, the data suggest that DRB1*0804 can be considered a PV susceptibility allele in its own right. DRB1*0804 has been associated with African descent in several studies (26, 27). A comprehensive analysis of HLA allele frequencies worldwide found that DRB1*0804 is by far the most frequent in northwestern Africa, around modern Burkina Faso, and is more prevalent throughout Africa than other continents (24). While DRB1*0804 has not been studied as extensively in the context of PV, previous studies have noted that DRB1*0804 is increased in PV patients in Brazil and Egypt (28, 29).

Our findings highlight how closely PV is associated with different HLA alleles in different ethnic populations. Importantly, beyond establishing associations of HLA expression in PV patients of different ethnic backgrounds, our data further shows that the expression of certain HLA molecules in the context of PV is highly associated with, and likely shapes antibody selection and ultimately lesional phenotype. In this context, it has previously been shown that HLA DRB1*0402 contains a binding motif that can selectively bind Dsg3 specific peptides (30) and present them to DRB1*0402-restricted autoreactive CD4(+) T cells to initiate the production of Dsg3-reactive IgG autoantibodies (31). In line with these findings, patients carrying DRB1*0402 were found to express relatively higher levels of anti-Dsg3 autoantibodies. DRB1*0402 positive patients, including those with ethnicities in which the allele is prevalent, such as Ashkenazi Jewish and non-Jewish Caucasian, were predisposed to developing mucosal PV lesions over mucocutaneous and cutaneous only lesions.

Patients that carried DQB1*0503, including South Asian patients who are heavily DQB1*0503 positive, had higher rates of mucocutaneous disease than other patients. Moreover, expression of the DQB1*0503 allele was found to predispose patients to having relatively higher levels of anti-Dsg1 autoantibodies, perhaps a reflection of differing peptide binding capacities between DQB1*0503 vs DRB1*0402 molecules that link to anti-Dsg3 and anti-Dsg1 responses. Interestingly, when DQB1*0503 positive patients were subdivided by ethnicity, South Asian patients (n=16) had average anti-Dsg1 levels almost twice that of patients of other ethnicities (n=30) ( Supplemental Figure 1 ), indicating that South Asian ethnicity, beyond reasons linked to HLA type, in itself is associated with elevated anti-Dsg1 expression, implicating additional thus far unknown genetic and/or environmental factors. While these results did not reach statistical significance (p=0.11), repeating this analysis with a larger sample size may elucidate the true significance of this discrepancy. Nevertheless, our findings are consistent with a study by Harman et al. in which 75% of Indian PV patients were positive for anti-Dsg1, compared to only 45% of White patients and 56% of patients of “other” ethnic background (16).

It is noteworthy that patients typing as DRB1*0804, despite displaying the highest mean levels of anti-Dsg3 and the lowest mean levels of anti-Dsg1 when compared to all other HLA-associations, did not show a clear predilection for morphology. In fact, this population split equally into patients carrying mucosal and patients carrying cutaneous lesions, indicating that autoantibody selection in itself it not an absolute factor in determining lesion location (skin vs. mucosal). This point is further underlined by the fact that the Hispanic population, despite being predominantly DRB1*0402 positive, displays relatively low levels of anti-Dsg3 and also did not have a predisposition towards mucosal disease, but instead was the only population with more cutaneous disease than mucosal or mucocutaneous. These data indicate that while DRB1*0402 appears to favor the selection of anti-Dsg3 antibody specificities, correspondingly leading to the predominantly mucosal pattern predicted according to the Dsg-Compensation hypothesis (32, 33), supplemental genetic and/or environmental elements must also play a role in determining lesional fate, again underscoring the multifactorial basis of disease.

Of note, patients that did not express the typical disease-associated HLA alleles, i.e. those typing negative for DRB1*0402, DQB1*0503, or DRB1*0804, had very low levels of anti-Dsg antibodies and the highest levels of cutaneous only lesions among all HLA-subgrouping analyzed. We and others have suggested before that non-Dsg antibodies or other thus far underappreciated factors may help explain lesion development in the absence of the accepted PV-associated antibodies (13, 34–36). Our group has also previously reported that in a slightly smaller patient set than the one used in this study, anti-TPO and anti-Tg have higher activity rates in patients not expressing DRB1*0402 and/or DQB1*0503. This study supports the notion that non-Dsg antibodies such as those directed against anti-TPO and-Tg may be of relevance in patients lacking the typical PV-associated alleles, since we show the highest levels of both antibodies in either DRB1*0804 positive patients or those lacking all 3 susceptibility alleles (DRB1*0402-, DQB1*0503- and DRB1*0804-triple negative). It should be noted that a number of non-Dsg autoantibodies besides anti-TPO and anti-Tg have been reported in PV (13, 35). While not investigated in this report, these specificities may similarly play a role in fine-tuning the autoimmune response in select patients.

Overall, our analysis reveals and reinforces an overarching theme: there are clear variances in HLA types found in PV patients of differing ethnic origin (including an unexpected and novel HLA association in PV patients of African descent, a historically understudied population in the context of PV), and these variances in HLA type link to variances in autoantibody profiles and clinical morphology. Taken together, our data provide strong support for the novel hypothesis that HLA allele expression not only predisposes to disease development in PV, but is also crucial in shaping autoantibody profiles and phenotypic expression, and that HLA variance undergirds disease heterogeneity. We report, for the first time, clear distinctions regarding associations of various disease associated HLA alleles (which are unevenly distributed across various ethnic groups) with key immunological and clinical parameters of disease ( Table 2A , Supplemental Figure 2A ). While the relative contributions of as yet unknown genetic and environmental factors to autoimmune activity will need further investigation, several previously unrecognized patterns and trends linked to HLA type have emerged ( Table 2B , Supplemental Figure 2B ): 1) Patients typing as HLA DRB1*0402 (found most commonly in Ashkenazi Jewish and Caucasian non-Jewish populations) had higher levels of anti-Dsg3 antibodies than patients without DRB1*0402 and had mucosal only lesions more often than cutaneous only or mucocutaneous lesions. 2) Patients typing as DQB1*0503 (found most commonly in patients of East Asian and South Asian descent) had higher levels of anti-Dsg1 antibodies compared to other groups and higher rates of mucocutaneous disease than other lesion types. 3) PV patients of African descent carried predominantly the DRB1*0804 allele, presented with highly elevated levels of anti-Dsg3, but did not link to any specific lesion morphology. 4) Patients that carried neither DRB1*0402, nor DQB1*0503 or DRB1*0804 had the lowest levels of anti-Dsg3 antibodies and the highest rate of solely cutaneous disease compared to carriers of these alleles.

Interestingly, PDAI scores available for a limited number of patient visits did not reveal any significant differences when analyzed by gender, age of onset, HLA, or ethnicity, suggesting that disease severity is not predetermined by any of these factors individually. However, PDAI scores were not available for all patients or all visits and will have to be reassessed in a larger patient population later.

Future studies will be needed to elucidate the details of (auto)antigen recognition by each PV linked HLA allele that drive T/B cell activation and autoantibody selection and, in turn, disease morphology (lesion location). Defining these connections can be expected to deliver a fuller understanding of autoimmune induction and disease manifestation in pemphigus, and will advance the path towards precision medicine where individual genetic and immune response elements are incorporated into clinical decision making and patient management. Detailed information on the mechanisms by which HLA genes direct and determine disease-defining features is required to unravel the complexities of disease heterogeneity can be expected to improve the ability to diagnose, track and treat.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

The studies involving human participants were reviewed and approved by Institutional Review Board, University at Buffalo Institutional Review Board, Michigan State University Institutional Review Board, Weill Medical College at Cornell University. The patients/participants provided their written informed consent to participate in this study.

AS and KS-S devised the project and performed data collection. JB performed data selection, data analysis. All authors contributed to writing and critical revision of the manuscript.