Dermatitis Herpetiformis and Dietary Gluten: Illuminating a Gut-Skin Connection

Russell P. Hall, M.D.
Lifeline, Spring 2004, Vol XXIV, No 2, pp. 2-3, 8, 10-13


Dermatitis herpetiformis (DH) is an intensely pruritic, chronic blistering disease that often appears quite suddenly, mainly between the ages of 20 and 55, and is much more common in men than women. Overall prevalence—more frequent than the bullous pemphigoid group of blistering diseases—is estimated at somewhere between 10 to 39 cases/100,000 annually. Erythematous papulovesicular lesions are characteristically distributed symmetrically on extensor body surfaces and on the buttocks and back. Granular deposits of IgA (normally involved in immunoprotection of the body’s mucosal surfaces, especially the respiratory and gastrointestinal tracts) at the dermo–epidermal junction (DEJ) are the immunologic marker of this disease, and patients have a gluten-sensitive enteropathy (GSE) that is histologically apparent but generally asymptomatic.

Dapsone effectively controls cutaneous symptoms while leaving the gastrointestinal abnormalities untouched. Yet eliminating gluten from the diet controls not only these abnormalities but resolves the skin lesions as well. This highly intriguing fact suggests that solving the pathogenesis of DH will rest on unraveling the complex pathway connecting the gut’s clinically asymptomatic gluten-sensitive mucosal immune response to the unusual presence of IgA in the skin and the recurrent lesions.

Russell P. Hall III, MD—chief of the Division of Dermatology and professor of immunology at Duke University Medical Center—has been pursuing the secrets of DH intently for more than 25 years, identifying clues to this pathway one by one. He stumbled across this fascinating disease unexpectedly. After completing his medical training, Hall began a research fellowship in the laboratory of Stephen I. Katz, the well-known immunodermatologist who headed the Dermatology Branch at the

National Cancer Institute. “I was interested in immunology and immunology research,” Hall recalls, “and within that context I wanted to explore the pathogenesis of disease. Dr. Katz was working on DH when I arrived, and I became part of that effort.” He has been at it ever since.

DH: An Emerging Portrait
The first description of DH appeared in the Journal of the American Medical Association 120 years ago, back in 1884, but little more was learned until a sudden flurry of investigative activity in the late 1960s set the basic defining elements in place. Cormane discovered immunoglobulin at the DEJ in patient skin which Van der Meer then characterized as IgA—a critical component of the protective immune response in mucosal barrier tissues that is normally found in the mucosal surfaces of the respiratory, gastrointestinal, and urogenital tracts. This discovery allowed identification of two patient groups within the DH clinical phenotype— the 85–90 percent of patients with granular IgA deposits at the DEJ and the small remainder with linear deposits at that site. The appearance of IgA in patient skin was striking—and puzzling.

Another investigator realized that DH patients have an associated— although, intriguingly, most often asymptomatic—gastrointestinal disease that was eventually identified as GSE. The histology showed two basic features; atrophy of the normal villous architecture of the jejunal epithelium, with lamina propria and intraepithelial lympho cytic infiltrates. It was shown to be morphologically identical to, although less severe than, that seen in patients with isolated GSE (celiac sprue). And these changes were confined primarily to the small intestine, most often the jejunum. Despite this, and several laboratory studies documenting abnormal intestinal function, less than 10 percent of DH patients were observed to experience the typical bloating, diarrhea, and malabsorption of GSE.

During the 1970s, a high frequency of certain human leukocyte antigens (HLA-A1, -B8, -DR3, -DQ2) was documented in DH patients—but associated only with granular cutaneous IgA deposits. It became clear that, despite the clinical parallels, patients with linear deposits of IgA represent a different pathogenesis and thus a different disease, and they were recategorized as linear IgA dermatosis. (Linear IgA dermatosis is now known to be one of the autoimmune blistering diseases.)

Discovering the existence of GSE in DH patients prompted some of them to eliminate gluten from their diet. Doing so normalized the morphologic changes in the small intestine and controlled their skin disease— yet another provocative observation. “It became clear that GSE plays a critical role in the pathogenesis of DH,” Hall says, “but in what way the skin disease, the cutaneous IgA deposits, and the associated GSE interact in these patients is still unknown.” The cutaneous IgA presence itself also remains a mystery— its origin, the manner in which it is deposited, and the structures to which this IgA binds.

Hall’s Perspective
“I started out asking a set of simple questions about IgA ,” Hall says. “I wanted to learn what it is doing in the skin. What is its source? How does it get from there to the skin? And, what does it mean to this disease? But my focus has broadened considerably over the years,” he notes, “and I have come to find additional points about DH that make it very intriguing for an investigator.

“First, it is characterized by very severe clinical symptoms,” Hall observes. “Second, this disease has a unique histologic and immunofluorescence picture, with both neutrophils and IgA in the skin. Third,” he continues, “it can be controlled exceedingly well by use of the drug dapsone, but this drug does not help normalize the associated—but asymptomatic —gut disease. However, normalizing the asymptomatic gut resolves the skin disease in a great many patients. And reintroduction of gluten is followed by deterioration of small bowel morphology, recrudescence of cutaneous IgA deposits, and relapse of skin disease.

“And the final point,” Hall notes, “concerns the isolated GSE patients. Their gut disease looks exactly—or very closely —the same by biopsy, they are sensitive to the same wheat protein—the gliadin in gluten, and they possess the same HLA association, but contrary to DH patients, they do gets lots of problems with their gut and they generally have no skin problems at all. So here is a well-characterized dietary protein that a group of people react to immunologically and morphologically in the same way,” he says. “Some of these people develop highly symptomatic gut disease, with diarrhea and bloating, but no skin changes. The others develop extremely severe skin disease but have little to no gut disease.” The burning question is why—why do they go on to develop different diseases in different organs?

The implications extend beyond the DH–GSE association itself. “A number of diseases of the skin are associated with gut diseases—such as the links between pyoderma gangrenosum and erythema nodosum with inflammatory bowel disease. We do not understand the nature of this association,” Hall notes, “and we have come to suspect that elucidating DH may well provide an effective model for understanding what appears to be a more general relationship between inflammation, diseases of the intestinal tract and diseases of the skin. Hopefully we can use the DH–GSE association to understand some basic principles about how the two largest epithelial organs in the body—the skin and the gut—are related.”

And Hall is quick to point out that, despite the years of research that he and others have devoted to DH, we still do not fully understand how IgA ends up in patient skin.

Gut Instincts
With their initial focus on the mechanism for IgA’s entry in to the skin, Hall and his co-workers looked for—and found—the presence of IgA immune complexes in the serum of DH patients. But it wasn’t unique, as IgA immune complexes were also detected in the serum of patients with isolated GSE. “It was obvious that there was something more than the presence of these immune complexes that led to the development of IgA in the skin,” Hall notes, “and that’s when we realized that the gut is where we really need to look to understand this skin disease.”

Hall and his group assessed the production of antibodies in the gut secretions, finding that DH patients develop IgA antibodies against wheat protein, that they are primarily of the IgA1 subgroup and that these distinctive antibodies enter the circulation. This suggested that the IgA in patient skin actually originates in the intestinal tract as a response to gluten, and confirmed Hall’s decision to focus on events there.

This raised additional questions: What is the critical difference between gluten-sensitive people who develop DH and those who develop celiac disease? And how does the gut relate to the development of the actual skin lesions in DH? Hall has approached these questions by studying tissue biopsied from the small bowel of patients in each patient group— and sometimes from normal people as well—and looking at various aspects of the gut’s mucosal immune response.

Cytokine Patterns
Hall and his co-workers first turned their attention to cytokine expression in the gut. Cytokines mediate immune responses, and the pattern of local cytokine expression can determine the immunopathogenesis of an inflammatory disorder. Cytokines were already known to function in the normal gastrointestinal mucosa. Their presence often increases greatly in inflammatory enteropathies, which was suspected to play an important role in the observed tissue damage. T cells are among the cytokine-secreting leukocytes, and there was already considerable speculation due to evidence from isolated GSE patients that a T-cell-mediated immunologic response to ingested gliadin is central to their disease pathogenesis. Hall and his colleagues hypothesized that systematic variations in this gut mucosal cytokine pattern may account for the different clinical manifestations of these two gluten-sensitive diseases. A number of investigators had already identified cytokines that appear to predominate in the gut of isolated GSE patients— primarily interferon-gamma (IFN- ã), with some evidence as well for IL-2, IL-4, IL-6, IL-10, and TNF- á—but no one had yet described the gut cytokine profile in DH patients.

Hall and his team took duodenal biopsies from 3 healthy controls, 8 DH patients (7 on medication controlling their skin lesions, 1 on a partially gluten-free diet because of moderate gastrointestinal symptoms, and about to start on dapsone), and 9 isolated GSE patients (4 were asymptomatic on a glutenfree diet, 5 ingesting gluten with active gastrointestinal symptoms). The research team looked for the expression of two cytokines in particular. One was IFN-ã, already identified as prominent in GSE, and the hallmark cytokine produced by Th1 helper T cells— and thus, prominent in Th-1–associated diseases (eg, delayed-type hypersensitivity). The other was IL-4, the hallmark cytokine produced by Th2 helper T cells—and thus, prominent in Th-2–associated diseases (eg, allergic disease). Each type of helper T cell tends to involve a different constellation of additional cytokines, and thus a different type of inflammatory response. Th0 cells—which express both of these cytokines in relative balance—are immune-activated but have not yet differentiated into Th1 or Th2.

Hall and his team found mRNA levels to discriminate between the groups. Symptomatic GSE patients showed a high level of IFN-ã, in line with earlier reports, while IL-4 expression was low—a Th1 pattern. Compared to these GSE values, tissue samples from DH patients showed lower IFN-ã levels and higher IL-4, although within the DH group these values were comparable—a Th0 pattern. The asymptomatic GSE patients fell in between. Their IFN-ã and IL-4 expression were equivalent, reflecting Th0 T cells. But they expressed substantially less IFN-ã than the symptomatic GSE group, yet still higher than for the DH patients. Their IL-4 levels were higher than in both of the other groups.

Reflecting on these cytokine patterns, Hall remarked on the similar Th0 pattern in both the DH group and the asymptomatic GSE patients despite continued gluten consumption in the former and a gluten-free diet in the latter. And regarding the clear cytokine difference between the DH and symptomatic GSE patients, he found it “of considerable interest that these different patterns occurred in the face of continuing gluten ingestion by both groups.” Although the root of these cytokine similarities despite different diets, and cytokine differences despite the same diet—all in patients with equivalent gluten sensitivity—remained to be explained, Hall believed that these differing cytokine responses to dietary gluten may play an important role in the puzzling clinical differences between DH and isolated GSE patients.

Looking at T-Cell Receptors
As these results were being prepared for publication, evidence was appearing from other labs pointing even more strongly to a significant T-cell–mediated immune event in the gut. Although both varieties of T cells—á/â T cells and ã/ä T cells—were found to be increased in the gut of DH and isolated GSE patients groups, only á/â T cells returned to normal counts when dietary gluten was eliminated, paralleling normalization of the small bowel mucosa.

Carrying this to the next step, Hall speculated that the lack of gastrointestinal symptoms in DH patients might reflect a more limited T-cell response to gluten in the gut. This, in turn, would result from a more limited repertoire of T-cell families present in the gut of these patients. Isolated GSE patients, on the other hand, may rapidly expand their glutenassociated T-cell response, leadingn to symptomatic gastrointestinal disease. Variation is a function of the T-cell receptor (TCR), which comes in a vast array of alternatives, each variant slightly different in behavior. The great many possibilities are organized into families, and the families into larger groupings, based on certain common receptor elements. Because it was the á/â T cells showing the responsiveness to the presence and absence of gluten, Hall and his co-workers compared the range of TCR families within the variable (V) region of the â- chain (ie, TCR Vâ families)—the appropriate TCR component in this context—between the DH and

isolated GSE gut.

They obtained small bowel biopsies from 11 DH patients and 17 patients with isolated GSE. All of the DH patients were on medication to control their skin disease, 9 were free of GI symptoms and still consuming gluten, 1 without GI symptoms was avoiding gluten to lower medication, 1 with GI symptoms was on a partial glutenfree diet. Of the isolated GSE group, 7 were symptom-free and following a gluten-free diet. Four of the 10 with symptoms were poorly compliant, and 4 had not modified their diet. Reverse transcriptase PCR was used to assess the range of TCR Vâ for each individual patient. DH patients and the GSE gluten free/symptom-free subgroup had a comparable picture—they expressed 6.6 and 5.6 of 20 Vâ families, respectively, with no one family predominating. The individual DH patient range was from 1 to 16 families (median = 4), with 16 identifying the single DH patient with GI symptoms, and the asymptomatic GSE patients had a range of 1 to 15, also with a median of 4. The remaining GSE patients expressed 12.6 Vâ families, with no single family predominating. They ranged from 6 to 20 families per patient gut, with a median of 13.

So Hall’s suspicion was borne out. GSE patients without GI symptoms— whether DH patients or isolated GSE patients who religiously avoid gluten— express a small number of TCR Vâ families, while those with symptoms express a significantly larger number.

Number of Vâ families is correlated with presence/absence of GI symptoms, and clearly not with gluten consumption or characteristic small bowel mucosa changes, as all but one DH patient consumed gluten and showed the impact in gut mucosal abnormalities. These results do not constitute an answer, as “the role of restricted TCR repertoire in human disease has proven controversial,” Hall notes. “But they suggest that further investigation into the TCR utilization in DH patients may provide important clues into the initiating events in GSE and the mechanisms involved in the different clinical phenotypes associated with gluten sensitivity.”

Neutrophils: Ready for Action
The skin lesions that develop in DH have a characteristic intense inflammatory infiltrate composed almost entirely of neutrophils that localize fairly exclusively to the dermal papillary tips—contiguous to the cutaneous IgA deposits. Since the pathology begins in the gut, which is also the source of the cutaneous IgA deposits, it was logical to hypothesize that changes in the gut also play a fundamental role in this neutrophil presence in the skin.

Before resting circulating neutrophils can respond fully to inflammatory or infectious events, they have to be primed. This means equipping them with the molecules enabling them to react to subsequent activation signals by extravasating and migrating to the site of chemoattractant action. Priming occurs when specific cytokines (or agents such as lipopolysaccharides)— induced by the initial pathologic event— then induce the expression of these response molecules. Activation of primed peripheral blood neutrophils by inflammation at distant sites had already been demonstrated in other situations. “So we hypothesized that the presence of an ongoing mucosal immune response in the gut of DH patients,” he explains, “may result in a priming of circulating neutrophils that would potentiate their ability to bind to vascular endothelial cells and transmigrate to the IgA deposits already present in the skin.”

Hall also knew from work elsewhere that primed neutrophils upregulate their expression of the cell surface adhesion molecule CD11b (which helps neutrophils adhere to the E-selectin molecule on endothelial cells) and minimize their expression of the adhesion molecule L-selectin (which is a lymph node homing molecule). So he and his team analyzed neutrophils from 12 normal subjects, 10 DH patients with active, ongoing disease, and 14 DH patients with quiescent disease activity. They quantified the relative expression of both relevant adhesion molecules. They also quantified the expression of Fc IgA receptor, the molecule that binds with IgA (also referred to as CD89), and they assessed the affinity with which it binds to this antibody. The results were in line with expectation. The CD11b adhesion molecule was present in much higher numbers on the neutrophils from patients with active disease, and L-selectin was reduced in both patient groups compared to neutrophils from healthy subjects. Fc IgA receptor expression did not change, but its binding affinity did. When neutrophils from each group were incubated with monoclonal human IgA, binding capacity was significantly greater for those taken from DH patients with active disease.

“Our observations suggest that limited priming of neutrophils may be occurring in the gut of DH patients who are eating glutencontaining diet,” Hall said at the time, “but further studies will be needed to confirm our observation.” And he concluded by adding that “further study of neutrophils and cutaneous endothelial cell expression of adhesion molecules in sites of minor trauma may provide important insights into the pathogenesis

of these skin lesions.”

The Impact of Minimal Trauma
This mention of “sites of minor trauma” foreshadows the most recent study to be published from Hall’s lab. “One of the most intriguing issues for me at this point,” Hall says, “is why DH skin lesions develop, why they are uniquely localized on the elbows, knees, buttocks, across the shoulders, and on the extensor areas of the body— and how all of this relates to what is occurring in the gut.” A 1997 study from the Harvard-associated laboratory of Thomas Kupper, MD, had seemed to offer relevance. Kupper had found that simple mechanical deformation of human keratinocytes leads to rapid release of the primary cytokine IL-1á, that the amount released is dependent on the amplitude of strain, and that this release of IL-1á holds the potential to activate vascular endothelium. Kupper viewed this as a pathophysiologic mechanism that may play a role in the anatomic localization of some inflammatory skin diseases—such as psoriasis—which occur more commonly in locations where the dermis is subjected to repetitive stretch or trauma.

For Hall, Kupper’s observation offered the possible sequence of steps between the DH patients’ circulating primed neutrophils, their skin lesions, and the typical appearance in areas subject to stretch and/or friction. He knew that primed neutrophils cannot migrate into tissue unless the vascular endothelium has also been altered, upregulating its expression of the selectin family of adhesion molecules (P-selectin and E-selectin in this instance) that enable primed neutrophils —as they prepare to leave the circulation —to tether to the endothelium and then extravasate. E-selectin expression in particular requires induction by local cytokines such as IL-1 or TNF-á. Looking at the full\ sequence, Hall hypothesized that when the DH patient—in the absence of GI symptoms— continues to eat wheat protein, this continuous presence of gluten in an environment that produces anti-gluten antibodies leads to a chronic low-grade immune response in the gut. This cytokine milieu partially activates inflammatory cells that traffic through the gut. Based on Kupper’s observations, Hall theorized that keratinocytes in the areas where DH lesions develop—which regularly stretch or rub—release IL-1, stimulating inflammatory mediators and initiating the local endothelial changes enabling migration of the gut-primed neutropils. These neutrophils would home to thesource of this IL-1 release and bind with the IgA antibodies already deposited there, and the conditions for initiating lesions would all be in place.

Hall and his co-workers carried out the initial steps to test this conception. “We hypothesized that minor trauma to the skin, which does not result in any perceptible clinical or histologic change, would lead to increased expression of the adhesion molecules E-selectin and ICAM1 in the skin, which are events critical to the development of an inflammatory response,” he says. “We also hypothesized that pro-inflammatory chemokines and cytokines in the skin that are known to be involved in the development of skin lesions—such as IL-8 and TNF- á—would also be expressed in the skin after such minor trauma. Such an expression of both endothelial adhesion molecules and proinflammatory cytokines would make skin a permissive site for the development of an inflammatory infiltrate,” he adds.

The kind of minor trauma he envisioned refers to the stretching and friction that occur every day as we move and sit and reach. Hall mimicked this kind of minor trauma to the skin by gently rubbing the inner arm skin of 11 healthy adults (6 women, 5 men) with a pencil eraser for 2 minutes. Mild redness had disappeared within the first hour, and there were no evident changes of any kind in the rubbed skin by 4 hours afterward, at which point punch biopsies were taken from both rubbed and untouched skin. This 4-hour point was chosen because of previous studies in humans showing maximum expression of E-selectin mRNA 4 hours after chemical irritation of the skin. Assessing the normal and rubbed skin for the level of mRNA for both E-selectin and ICAM-1, and for the cytokines IL-8 and IL-10, showed that the message for both adhesion molecules had increased significantly in 10 of the 11 subjects, and for the two cytokines in all 11 subjects. IL-8, a pro-inflammatory cytokine that is a neutrophil chemotactic factor, may possibly be produced by the keratinocytes themselves. IL-10—an anti-inflammatory cytokine that inhibits E-selectin expression—would appear to be the simultaneous mechanism for restraining the inflammatory reaction. Looking at E-selectin protein at 4 hours after rubbing, staining was seen in the dermal blood vessels of 3 of the 4 subjects whose tissue was evaluated with immunohistochemistry. There were no differences between rubbed and unrubbed skin when it came to Pselectin (also assessed) and ICAM-1.

Hall points out that “these studies demonstrate that even minor trauma to the skin—trauma that does not induce any clinical or histologic changes—can result in the expression of E-selectin, ICAM-1, and IL-8 mRNA, and the production of Eselectin protein by endothelial cells. These events would appear to predispose the skin to the development of an inflammatory infiltrate,” he adds, “and if activated or partially activated inflammatory cells are traversing the skin in these areas of trauma, the skin would be poised to develop an inflammatory response that could not be blunted with the low levels of IL- 10 induced by this minor trauma.”

Looking Forward
Hall is continuing to characterize Tcell receptor differences between DH and isolated GSE. He finds his most recent results extremely interesting, but currently raising more questions than they have answered. He and his team are also continuing to explore the consequences of minor trauma on keratinocytes and the inflammatory process. In addition, they are taking a much closer look at the specific Tcell responses to the different regions of the gluten protein gliadin to see if DH and isolated GSE may actually represent immune reactions to different segments of the protein molecule, and thus help to determine which organ—the gut or the skin—becomes symptomatic when the gut is inflamed.

Reflecting on the therapeutic impact for DH skin lesions that occurs from a gluten-free diet, Hall speculates that “dietary regulation of inflammation of the gut, or regulation of inflammation there by other means, may actually be a more broadly important way to look at controlling skin disease.” This is something he has given thought to for some time, and intends to explore in the future. “We think that there may be a real potential for developing a model for the interaction of inflammation in the intestinal tract and the development of skin disease. And that will lead the way to new therapeutics. ”


Suggested Readings
Smith AD, Bagheri B, Streilein RD, et al. “Expression of interleukin-4 and interferon-ã in the small bowel of patients with dermatitis herpetiformis and isolated gluten-sensitive enteropathy.” Digestive Diseases and Sciences. 1999;44:2124–32.

Hall RP III, Owen S, Smith A, et al. “TCR Vâ expression in the small bowel of patients with dermatitis herpetiformis and gluten sensitive enteropathy.” Experimental Dermatology. 2000;9:275–82.

Smith AD, Streilein RD, Hall RP III. “Neutrophil CD11b, L-selectin and Fc IgA receptors in patients with dermatitis herpetiformis.” British Journal of Dermatology. 2002;147:1109–17.

Takeuchi F, Streileiln RD, Hall RP III. “Increased E-selectin, IL-8 and IL-10 gene expression in human skin after minimal trauma.” Experimental Dermatology. 2003;12:777–83.


CSA Library Series

CSA Library Series is a collection of articles that pertain to celiac disease and dermatitis herpetiformis. Most of these articles have appeared in CSA’s quarterly newsletter, Lifeline, which all CSA members receive. Historic articles included in these resources may or may not include updated notes. Updated information indicated in red type. Articles represent the work of the author.