Dipeptidyl Peptidase-4 Inhibitor-Associated Bullous Pemphigoid

Bullous pemphigoid (BP) is the most common type of autoimmune bullous diseases; drug-induced bullous pemphigoid is a rare variant of it. In the last decade, there is an increasing prevalence of BP, especially dipeptidyl peptidase-4 inhibitor-associated BP (DPP-4i-BP), with the higher prevalence of BP in diabetic patients. Recently, several clinical phenotypes of BP were detected, but there is a tendency in BP cases related to DPP-4 inhibitors to show an atypical noninflammatory form with less distributed skin symptoms, mild erythema, decreased eosinophilic infil-tration in the periblister area, and normal or slightly elevated peripheral eosinophil count. Anti-NC16A BP180 autoantibodies are less frequently detected by ELISA, but they response to other epitopes of BP180. Clinical outcome is similar such as in classical non-DPP-4 BP patients, regardless of withdrawal of DPP-4 inhibitors.


Introduction
Bullous pemphigoid (BP) is a rare autoimmune blistering disease of elderly patients, but in the last decades, it shows increasing incidence [1][2][3][4][5][6][7][8]. Higher prevalence of BP may be according to the increasing global life expectancy of the population, increasing incidence of predisposing neurological diseases, growing numbers of provoking drugs, and improving awareness of newly recognized atypical clinical phenotypes and better diagnostic methods [1]. The role of culprit drugs such as neuroleptics, diuretics, and antidiabetics is reported in several studies [1,2,9]. BP is typically present in elderly with a higher predominance in female patients [4,7]. The classical clinical features of BP are generalized bullous skin eruptions with surrounding erythema and itching; peripheral eosinophilia is also common. Mucosal involvement was observed in 10-30% of patients [2]. Atypical clinical variants may be present up to 20% in BP, including the more common prurigo-like or urticarial type, eczema-like type, dyshidrosiform type, erosive type, and erythema annulare centrifugum-like phenotype [1,7]. The diagnosis is based on the histological findings, including direct and indirect immunofluorescence microscopy and anti-BP180/BP230 enzyme-linked immunosorbent assays (ELISA) [2]. The gold standard for the treatment of the disease according to guidelines is corticosteroid, in topical or systemic administration and in severe cases with adjuvant immunosuppressive medications, such as azathioprine, methotrexate, or mycophenolate mofetil [2]. In the case of drug-induced BP, the most important therapeutic step is the withdrawal of the culprit drug [9].

Demographics of DPP-4 inhibitor-related BP
DPP-4 inhibitor intake was associated with a threefold increased risk for BP [27,31,38]. According to the former investigations [14,23,26,27,29,31,38,39], vildagliptin has the strongest association with BP; the risk was tenfold (ranged between 7.23 and 11.8) in a systemic review and meta-analysis by Kridin et al. [38]. A higher, sixfold risk was also observed with linagliptin [27,38]. Higher risk of DPP-4i-BP was also found with sitagliptin by Lee et al. [40], and they also found in a larger sample size (patients with DPP-4i-BP n = 260) that the risks associated with specific DPP-4 inhibitors were lower than the previous studies [27, 38], 1.81 for vildagliptin, 1.64 for linagliptin, and 1.7 for sitagliptin. However, a significant association was detected with vildagliptin, linagliptin, and sitagliptin in age-and sex-matched controlled population; the association with saxagliptin in BP was not significant [27,31]. Saxagliptin, anagliptin-and alogliptin-induced BP were presented only in a few sporadic cases [29,31,32,40,41]. DPP-4 inhibitors are often given in combination with metformin; the two recent publications reported that the association of BP and gliptins is independent of metformin exposure [27,30]. Despite the BP is more common in females, a multicenter investigation and EudraVigilance data showed that DPP-4 inhibitor-associated BP tends to be more common in men [14,23] [23], while Lee et al. observed 1.76-fold risk in patients younger than 75 years and 1.5-fold risk in patients older than 75 years [40]. The mean latency period between the initiation of gliptin and the appearance of BP is ranging from 6 to 26.4 months [23,27,29,30,33]. That means DPP-4 inhibitors can be suspected in the pathogenesis of BP if the drug has been initiated at least 6 months, but it also has to be considered if the drug intake last for more than 2 years prior to the onset of BP.

Clinical features of DPP-4 inhibitors-related noninflammatory BP
The classical clinical picture of BP is generalized as bullous skin lesions, tense blisters with severe urticarial erythema. Recent publications characterized a noninflammatory form of BP with limited distribution, smaller blisters, and scant erythema   also be found in non-DPP-4-related cases but in a significantly lower manner. Higher frequency of mucosal involvement was reported in gliptin-associated BP in two studies (Kridin et al.,22.2%, n = 36, and Chijiwa et al., 78%, n = 9) [27, 33], but this observation was not supported by Plaquevent (n = 108) [31]. Interestingly, in gliptinassociated mucous membrane pemphigoid (MMP) there is a significantly lower buccal and more common cutaneous involvement [44]. Previous studies demonstrated that eosinophil count is in correlation with the severity of BP [45][46][47] and with BPDAI score [46]. Comparing the Bullous Pemphigoid Disease Area Index (BPDAI) [48], BPDAI scores for urticaria/erythema (U/E) were significantly lower in noninflammatory phenotypes [33,36,42,43], and lower BPDAI U/E was in correlation with decreased eosinophil count in the perilesional skin [33,42]. Significantly decreased peripheral eosinophil count was detected in patients of Kridin et al. [27].

Immunological characterization of DPP-4 inhibitor-related BP
In BP, there are two targets for autoantibodies, the hemidesmosomal BP180 and BP230 and the juxtamembranous extracellular non-collagenous 16A (NC16A), both of them can be easily detected by commercially available ELISA tests [2, 49,50]. The domain of BP180 (also called COL17) is a major target epitope in 80-90% of cases [49]. In several investigations, noninflammatory BP patients did not show reactivity against the NC16A domain of BP180, but they were positive for full-length BP180 and its ectodomain midportion with ELISA [34,36,42]. The midportion of BP180 is more likely to be recognized than the NC16A domain in DPP-4i-associated noninflammatory BP patients; they are presented with localized symptoms and mild erythema [32,35,36,42,51]. Although there was a positive reaction to anti-NC16A in DPP-4i-BP cases, but they were mainly presented with prominent erythema and inflammation, concurring with classical phenotype of BP [31, 35, 42]. Kawaguchi et al. showed that the rate of ELISA positivity and antibody titers for anti-BP180 NC16A was significantly lower in DPP-4i-BP than the non-DPP-4 group [22], and this lower titer was also observed by García-Díez et al. [52]. Kawaguchi has also emphasized that patients with DPP-4i-BP tended to have noninflammatory phenotype of BP and presented with negative ELISA for BP180 NC16A domain [22]. Some studies reported noninflammatory DPP-4-induced BP patients who were negative for anti-NC16A domain initially but responded to the full-length BP180 and became positive for NC16A during the course of the disease. This epitope spreading was observed in several cases, after the prolonged use of DPP-4 inhibitors after the onset of BP [17,37,52]. Other investigations also demonstrated that multiple epitopes of BP180 are targeted in DPP-4i-BP (midportion, C terminus, and LAD1) [35,52], and it may suggest that epitope spreading is more common in DPP-4i-BP than in classical BP cases [35]. García-Díez et al. suggested the major role of the midportion of BP180 in DPP-4i-BP, while other BP180 regions are involved later by epitope spreading [52]. Indirect IF positivity was also detected in DPP-4i-BP patients [13,27,34], and anti-BP230 autoantibodies were present [34, 35], but the sensitivity of the test was only 38%, which is lower than usually reported [51].

Clinical outcome
Based on several investigations, withdrawal of the DPP-4 inhibitors was the first therapeutic step in most cases in the treatment of the disease [23,34]. Regarding to these data, discontinuation of DPP-4i treatment seems to have a favorable

Conclusions
Dipeptidyl peptidase-4 inhibitors (also called gliptins) are widely used drug in the treatment of type 2 diabetes mellitus. There is an increased risk of BP in patients during DPP-4 inhibitor treatment. The exact pathomechanism of DPP-4i-associated BP is still unclear. Dipeptidyl peptidase 4 (also called CD26) is a 110 kDa transmembrane glycoprotein, which is expressed on the surface of several cells, such as T cells [53,54]. DPP-4 has antihyperglycemic effect and enzymatic activity; it plays a major role in glucose metabolism by blocking incretin [54]. DPP-4 is a plasminogen receptor that activates plasminogen resulted in plasmin formation [55,56]. Plasmin, a serine protease, which has a high level in lesional skin and in blister fluid in BP [57], cleaves BP180 within the NC16A domain [58]. Cleavage of BP180 in the NC16A region can induce neoepitopes with altered antigenicities [42,59]. The antifibrotic effect of DPP-4 inhibitors in the skin also supports the role of DPP-4 in collagen metabolism [56]. DPP-4 is involved in immune cell activation, and its inhibition can modify the immune response, which may increase the activation of eosinophil recruitment into the dermis, which is considered to be essential in blister formation in BP [60]. In contrary to these findings, in patients with gliptin-associated noninflammatory BP, both peripheral and perilesional skin eosinophil counts are significantly lower than in classical BP [27, 33], so the exact pathognostic role of eosinophils in DPP-4-related BP needs further investigations. It is also not elucidated why vildagliptin has the strongest association with BP, but it is known that vildagliptin has the lowest selectivity among gliptins with strong inhibition of DDP8 and DPP9 isozymes [22,61]. Some results suggest that DPP-4 inhibitor has immunomodifier effect mainly in  genetically susceptible individuals, and they have detected that HLA-DQB1*03:01 allele has higher prevalence in DPP-4i-BP patients [22,62].
In conclusion, DPP-4 inhibitor-related BP tends to be presented with noninflammatory phenotype of BP, with limited extension, smaller blisters, scant perilesional erythema and eosinophilic infiltration, and normal or slightly elevated peripheral eosinophil count (Table 1). Anti-NC16A BP180 positivity is less common, and ELISA titers are slightly elevated, similarly to anti-BP230, but positivity for full-length BP180 or other epitopes of BP180 may be detected. Response to therapy is similar such as in classical non-DPP-4i-BP patients, regardless of withdrawal of DPP-4 inhibitors.

Conflict of interest
There is no conflict of interest.