Pemphigus vulgaris (PV) is a potentially life-threatening autoimmune blistering skin disorder characterized by a well-defined humoral response directed against specific desmosomal proteins involved in cell-cell adhesion and maintenance of epidermal integrity. The hallmark for autoimmunity in PV is the presence of autoantibodies specific for desmoglein (Dsg)-3, and in some cases Dsg-1, both in serum and in lesional skin. The “desmoglein compensation hypothesis” posits that lesion morphology (mucosal vs. mucocutaneous) can be explained by autoantibody patterns (anti-Dsg3 only vs. a combination of anti-Dsg3 and anti-Dsg1, respectively) (1), however, the validity of this theory has been challenged recently (2). While the target organ damage is ultimately mediated by autoantibodies, it is generally accepted that B cell activation requires the participation of T helper cells necessary for immunoglobulin (Ig) production and class switching. T cell recognition of Dsg3 and/or Dsg1 epitopes appears to be the first step in the disease initiation, leading to B-cell activation and production of Dsg3/1 specific immunoreactants (3–6).

The importance of T cell involvement in disease pathogenesis is underscored by the strong genetic association of PV to the HLA DR4 (DRB1*0402) and DR6 (DQB1*0503) haplotypes (7, 8). T cells play a role in the inception and maturation of the humoral immune response, thus bridging the association of PV with MHC and generation of autoantibodies (9). Of course, these haplotypes are also found in healthy individuals. In fact, the vast majority of healthy individuals who express HLA DR4 (DRB1*0402) or DR6 (DQB1*0503) gene sequences do not progress to a disease state. Why healthy individuals who carry these disease risk elements remain healthy remains a mystery.

While the coordination and regulation of the autoimmune response in PV has not been fully elucidated, cytokines and chemokines are likely to orchestrate the interplay between cellular and humoral responses. Cytokines act as mediators for disease activation and maintenance in autoimmune disorders, but also operate as initiators of disease remission and induction of the tolerogenic state (10). Produced by, and acting on various cellular components of the immune cascade, they participate in both effector and regulatory pathways (10).

Limited attempts have been made to define cytokine and chemokine expression patterns in PV. Apart from a recent publication (11), previous studies typically examined only a handful of cytokines in small sample sizes using disparate methods to quantify cytokine concentrations. Thus, perhaps unsurprisingly, reported results are equivocal. In fact, many cytokines in PV were found to be upregulated in one study, only to be shown to have no significant change or even to be downregulated in another (3, 12–59). Table 1 shows a comprehensive literature review of previous serum cytokine studies, particularly which studies show an elevation, decrease, or no significant change in PV patients versus healthy controls, as well as how this data compares to new findings from this study that will be discussed in detail below. Across studies, heterogeneity amongst patients and control subjects in terms of disease phase, clinical phenotype, gender, HLA-association, and treatment may have contributed to ambivalent results, factors that are particularly difficult to account for in small sample sizes.

To overcome these challenges, we conducted an extensive analysis of serum cytokines/chemokines in PV, both in terms of the range of soluble mediators studied and the number of patient and control samples analyzed. We comprehensively compared multiple serum cytokines in defined clinical categories of PV. Unlike previous studies, we further explored the contribution of HLA haplotype to the cytokine milieu in both patients and controls to pinpoint the role of cytokines in genetically susceptible, but disease-free subjects. We utilized a multiplex bead array platform to evaluate the expression of 20 cytokines and chemokines categorized by function: 1) pro-inflammatory (IL-1α, IL-1β, IL-6, and TNFα), 2) Th1 (IFNγ, IL-2, IL-12), 3) Th2 (IL-4, IL-5, IL-13), 4) Th9 (IL-9), 5) Th17/Th22 (IL-17, IL-21, IL-22, IL-23), 6) Treg (IL-10), 7) NK cell (IL-15), and 8) chemokines (IL-8, MCP-1, Eotaxin) in 130 serum samples obtained from 116 PV patients in different phases of disease activity and remission, along with 15 healthy control subjects that carried the PV-associated HLA alleles DRB1*0402 and/or DQB1*0503 (termed here as HLA-matched controls) and 14 healthy control subjects that did not carry the known PV-associated HLA alleles (HLA-unmatched controls).

A comprehensive view of the immune mechanisms leading to tissue damage in Pemphigus vulgaris has not been elucidated. Although the presence of B cell-produced autoantibodies directed against specific cell adhesion molecules including Dsg3 and Dsg1 (and perhaps other non-Dsg targets) is critical to lesional development, T cells are certain to play a central role in the pathogenesis of disease. While the rules that govern the coordinated T and B cell response in PV remains unclear, cytokines and chemokines are likely key modulators of the effector and regulatory pathways at play in disease.

It is unlikely that changes in a single cytokine is sufficient to cause the pathological phenomena in any given autoimmune disease (28). Accumulating literature suggests a model in which the action of multiple cytokines ultimately promotes immune-mediated tissue damage (15, 32, 33, 38, 39, 65). A recent study from China found generally elevated levels of Th1, Th2 and Th17 cytokines in PV patients (11). However, it remains clear that a more in-depth knowledge of cytokine profiles in patients is still required for a better understanding of disease mechanisms, and to facilitate the discovery of novel targets for future therapies. To address the mechanisms that underlie the variable and heterogeneous disease expression in PV, we undertook a non-reductionist strategy to comprehensively survey global changes in multiple cytokines/chemokines.

We unexpectedly found that PV patients show a wide variation of cytokine levels across all cytokines tested with levels ranging from undetectable to highly elevated for each individual cytokine regardless of disease phase or therapy status. Interestingly, patients tended to be either ‘high-expressors’ or ‘low-expressors’ of cytokines with ‘high expressors’ exhibiting high serum cytokine concentrations across numerous cytokines (albeit not necessarily all), while ‘low-expressors’ tended to have uniformly lower cytokine levels. Thus, cytokine expression appears to operate relatively, at differing scales dependent on the person. Intriguingly, our data show that if a patient displays elevated cytokine levels, they typically span multiple cytokines within a given pathway (e.g. one or two of the T helper pathways analyzed or generally pro-inflammatory cytokines), but not the entire set of cytokines analyzed. The reasons for this remarkable cytokine pathway heterogeneity among PV patients are intriguing but remain opaque and in need of further exploration, including the investigation of genetic factors such as HLA. Nevertheless, our data have implications for understanding the immunologic basis of clinical heterogeneity, disease operative mechanisms, and have the potential for guiding treatment decisions for patients.

Despite these inter-individual differences, our analysis uncovers a pronounced dysregulation of multiple cytokine immune pathways in PV compared to healthy controls, namely with elevation of the Th2 cytokines IL-4 and IL-5, the Th17 cytokines IL-22 and IL-23, as well as IL-2, IL-9, IL-10 and IL-15. However, a more composite picture emerges when we divide our healthy control population into those controls that carry PV-associated HLA-susceptibility alleles (“HLA-matched”) vs. those that do not (“HLA-unmatched”). It is well established that PV-associated HLA alleles such as DRB1*0402 are necessary for T cells to recognize Dsg3 (40), and that Dsg-reactive Th2 cells can be detected in patients with PV (5). Accordingly, PV has been considered a Th2 driven disease (66). Although disease-free, we found that HLA-matched controls express the pro-inflammatory cytokines IL-1α, IL-1β, IL-6, and TNFα, and Th2 cytokine IL-13, and chemokine IL-8 at similar concentrations to that seen in PV patients. These cytokines, however, are significantly higher in PV when compared to HLA-unmatched controls, suggesting that healthy controls carrying the PV associated HLA-haplotypes have a heightened ability to mount a pro-inflammatory and Th2 cytokine response similar to PV patients, while those lacking expression of the PV-associated HLA alleles do not. These findings echo other work in our lab using CyTOF technology to interrogate immune cell distribution in PV patients vs. healthy controls. Using CyTOF, we found an increase in Th2 cells in PV patients compared to “HLA-unmatched” controls, while “HLA-matched” controls exhibited similar numbers of Th2 cells to those found in PV patients (manuscript in preparation). Our CyTOF studies also showed an increase in Th17 cells in PV patients as compared to HLA-matched controls and HLA-unmatched controls which is supported by our observation of higher levels of the Th17 cytokines IL-21, IL-22 and IL-23 in this study. Table 1 summarizes differences in cytokine concentration between PV patients and controls that carry PV susceptibility alleles (“HLA-matched”) and those who do not carry HLA susceptibility alleles (“HLA-unmatched”).

Looking more broadly at the genetic basis of autoimmune development, it is striking to note that the pattern of HLA-linked partial activation of autoimmune-linked pathways is not limited only to cytokine expression as revealed in this study, but extends to also gene expression (67), auto-antibody profiles (68), and anti-oxidant status (69) as revealed in published work from our group. These parallel patterns across multiple datasets exploring intersecting biological levels promote a new hypotheses regarding the immunological currents that either promote or prevent autoimmune development (67). Specifically, there may be HLA-independent pathways required for patients to progress to the disease state in PV, but also protective counter-regulatory mechanisms in genetically susceptible (HLA-matched) controls, restraining them from disease despite their genetic susceptibility.

Along these lines, we found that HLA-matched controls had significantly lower levels of IL-10 and IL-15 than PV patients, while HLA-unmatched controls did not, setting up the intriguing hypothesis that the downregulation of these cytokines affords those healthy people that carry genetic susceptibility a certain protection. IL-10, produced by T regulatory cells (amongst others), is thought to be a predominantly inhibitory cytokine. It is known to inhibit production of IFNγ, and activation of macrophages, NK cells, and neutrophils (10, 70). IL-10 plays a critical role in slowing the progression of autoimmune diseases (70). Nevertheless, multiple studies found elevated levels of IL-10 in serum, blister fluid, and supernatants of T cell and B regulatory cell cultures from PV patients compared to controls (25, 40, 45, 71, 72). Also, recent studies suggest that IL-10 is ineffective at accomplishing its downstream regulatory effects in PV (41, 72, 73). Without reaching its downstream target, IL-10 concentration may increase because feedback inhibition does not occur. Perhaps this seemingly contradictory data on IL-10 in PV could be explained by other immune-promoting functions of IL-10, such as induction of B-cell maturation, survival, and differentiation, as well as immunoglobulin production (74–76). Kowalski et al. suggested IL-10 may be higher in PV due this effect on B-cell production of IgG autoantibodies (19). Further, IL-10 can inhibit Th1 cytokines, and thus shift Th1/Th2 balance and promote Th2 cells in the context of PV (25).

Similarly to IL-10, we find that for IL-15, HLA-matched controls show significantly lower serum levels than PV patients, while HLA-unmatched controls do not. IL-15, produced by epithelial cells, fibroblasts, and peripheral blood mononuclear (monocyte enriched) cells promotes immunoglobulin switching, memory immunoglobulins, and NK cell differentiation and proliferation (77, 78). In support of these cytokine data, CyTOF analysis found that the balance of early vs. late NK cells was shifted towards late NK cells in PV patients when compared to all controls, as well as when compared to HLA-matched controls (manuscript in preparation). Downregulation of IL-15 in matched controls, as shown in this work, could explain why we also see a shift towards early NK cells in these same controls when compared to patients. Also, it is conceivable that a specific downregulation of IL-15 could lead to a reduction in immunoglobulin switching and memory immunoglobulin production and thus mediate disease protection in healthy individuals carrying PV susceptibility alleles.

Surprisingly, we do not see an appreciable difference in cytokine levels between different phases of disease activity, neither in patients regardless of therapy status nor when looking at the subgroup completely off therapy. It is possible that differences between groups in this study may be obscured by the wide variations in cytokine levels across patients. Thus, it will be important in future work to analyze samples taken from patients in a longitudinal study design to truly appreciate if disease activity is paralleled by changed in cytokine levels. In our study, we had three patients with longitudinal samples. Of these three, two exhibited generally higher cytokine levels in active disease, while one had higher levels in remission. Follow up studies are needed to include more patients followed longitudinally across different phases of disease to discern meaningful differences. Similarly, we observed only minimal differences based on disease morphology (mucocutaneous vs. mucosal) and based on sex (male vs. female). Meaningful differences between these subgroups may again be obscured by the wide variation of cytokine expression between patients.

Taken together, our study reveals that compared to healthy controls (regardless of HLA association), PV patients harbor elevated cytokine expression levels in generally pro-inflammatory- and more specifically Th2 and Th17 pathways, albeit at a very patient-specific, individualized level. This sets up the intriguing future management paradigm that based on an individual patient’s cytokine pattern, cytokine-targeted therapies could be aligned to be maximally effective. Cytokine profiling of individual patients could be leveraged in the future to personalize treatment plans.

Currently, the mainstay of treatment of autoimmune disease remains dependent on potent immunosuppressants that inhibit multiple immune pathways. In 2018, Rituximab was the first biologic to be FDA-approved for treatment of PV. Rituximab is now considered first line for treatment of this disease, although not without side effects. While achieving disease control in a majority of cases, these treatments also place the patient at risk for opportunistic infections and untoward side effects. Front runners of anti-cytokine biologic therapies for various autoimmune diseases include drugs targeting TNFα, IL-1, IL-6, IL-12/IL-23, and IL-17 (79). In the past years, several therapies that target specific cytokines, eg. etanercept (TNFα receptor), adalimumab (anti-TNFα), ustekinumab (anti-IL-23/IL-12), have been developed, with many more in the pipeline. To date, there is limited literature on anti-cytokine therapy in PV. However, multiple recent case reports show Dupilumab, an inhibitor of IL-4 and IL-13 to be effective in treating refractory or severe cases of disease (80–82). Larger studies and clinical trials are necessary to elucidate the true efficacy of dupilumab in PV, as well as explore additional anti-cytokine therapies. Our finding of high cytokine concentrations across one or two T-helper cell pathways may be relevant to predicting which PV patients are likely responders to dupilumab.

While this study emphasizes the importance of cytokines in disease pathogenesis, naturally occurring anti-cytokine autoantibodies are becoming an increasingly prevalent topic in autoimmunity. Though anti-cytokine autoantibodies can be found at low levels in disease-free, healthy subjects, they are found at higher titers in certain autoimmune diseases and immunodeficiencies (83). There are limited reports on the presence of these anti-cytokine autoantibodies in pemphigus and pemphigoid diseases (84, 85). Interestingly, rituximab has been shown to decrease levels of anti-IL-8 autoantibodies in rheumatoid arthritis (86). Given the significance of cytokines in PV pathogenesis, and increasing use of rituximab in PV, further exploration of anti-cytokine autoantibodies is warranted.

Our data support the role of the Th2 pathway in PV, while also lending credence to the increasing literature on the importance of the Th17 pathway. Thus, multiple cytokine pathways appear to be operative in the development of PV. However, in any given patient, not all pathways are simultaneously active; some patients are more Th2 and/orTh17 dominant, while others lean more towards pro-inflammatory or Th1 pathways. We propose a paradigm shift in our understanding of the pathogenesis of PV, specifically that there seem to be different ways individuals can reach disease, either through a majority Th2 and Th17 response, a pro-inflammatory response, or a Th1 response or a combination thereof. The precise molecular means by which patients are directed to different paths leading to PV are unknown, but perhaps there is a genetic basis for which HLA, especially, might play a role. Future mechanistic studies are needed to gain insight into disease risk and prognosis. For now, stratifying patients based on cytokine profiles has the potential to change our approach to therapy. By matching cytokine profiles to more targeted anti-cytokine/anti-pathway treatment, we can envision personalizing care to improve outcomes based on patients’ specific inflammatory profile.

Overall, our studies shed light on the complexities of critical cytokine networks impacting PV on a personalized level. This work can be expected to deepen our understanding of underlying pathogenetic mechanisms relevant to autoimmunity, help to resolve the biologic underpinnings of clinical heterogeneity that confound our ability to prognose and manage patients, and ultimately serve to advance efforts to align appropriately targeted therapies in select patients while reducing side effects associated with current approaches reliant largely on general immunosuppression.