What's new?

More than half of all patients with inflammatory bowel disease (IBD) experience at least a single extraintestinal manifestation. One of such manifestations are dermal lesions, which may be caused by a spread of the inflammatory processes from the intestines, another autoimmune disease, or side effects of treatment. We showed that dermal lesions are a common problem in IBD, occurring in 16.4% of patients. In the individuals treated with anti–tumor necrosis factor α drugs, the prevalence was almost twice as high (30.9%). A correlation with treatment was confirmed for the following lesion types: infusion reactions and injection site reactions, cutaneous infection, psorasiform reactions, and lupus‑like symptoms. Understanding the risk factors for their development and clinical consequences may be crucial for determining optimal management of patients with IBD.

Introduction

Patients with inflammatory bowel disease (IBD), such as Crohn disease (CD) and ulcerative colitis (UC), may experience extraintestinal symptoms. Skin symptoms are present even in 55% of CD and 35% of UC cases.^1-6 Dermal lesions may appear at different time points, sometimes even preceding the diagnosis of IBD, while their classification is based on pathophysiology and relationship with the underlying disease.^7,8 Currently, many drugs are used in the treatment of IBD, and the main goal of the treatment is to reduce inflammation.^9-12 The best‑known group of biologic drugs are anti–tumor necrosis factor α (anti–TNF-α) antibodies, specifically infliximab (IFX), adalimumab (ADA), certolizumab, and golimumab.^5-7 These drugs are highly effective; however, they may cause side effects. The most common ones are hypersensitivity reactions and infections, but up to 25% of patients may also experience adverse skin symptoms.^5-8 Their presence often raises concerns among gastroenterologists, mainly because dermal lesions have not been well described, also in terms of their clinical course.

Several types of dermal lesions can occur in the course of IBD. We can distinguish the following: 1) specific (continuous mucocutaneous lesions and metastatic noncaseating granulomas), with histopathologic features similar to those observed in IBD; 2) reactive (erythema nodosum, pyoderma gangrenosum, and Sweet syndrome)—the triggering factor may be antigens common to gut bacteria and the skin, causing an immune reaction; 3) related (psoriasis, hidradenitis suppurativa, vitiligo, phlebitis, erythema multiforme, urticaria, lichen planus, secondary amyloidosis, and various autoimmune bullous disorders), which show a pathogenetic mechanism similar to IBD, but differ from IBD lesions with respect to histopathologic features; 4) related to the treatment (infusion and injection site reactions, dry skin and eczema, skin infections, psoriasis‑like reactions, skin cancers, lupus‑like syndrome, vasculitis, lichenoid drug reaction, granulomatous reactions, alopecia areata / totalis and / or vitiligo erythema multiforme, Stevens–Johnson syndrome, toxic epidermal necrolysis, and dermatomyositis), with unclear immune and nonimmune mechanisms; and 5) manifestations secondary to malabsorption—gastrointestinal failure due to inflammation.^5-8

Types of dermal lesions in patients with IBD are shown in Supplementary material, Figure S1.

First reports on the occurrence of dermal lesions during or after anti–TNF-α therapy appeared only a few years ago and involved mainly psoriasis‑like lesions.^14-22 At the same time, they sparked much controversy, mainly due to the fact that anti–TNF-α drugs are also used in the treatment of psoriasis. However, an analysis of common genetic components offered a possible explanation for this process. In addition, a similar relationship has been demonstrated for atopic dermatitis. Genotype analysis of patients with IBD and psoriasis or atopic dermatitis showed common genetic components at as many as 163 loci identified for IBD, 36 identified for psoriasis, and 11 identified for atopic dermatitis.^23-25

Dermal lesions in patients with IBD are a common phenomenon, yet their impact on the treatment course remains uncertain due to insufficient data. Thus, we aimed to assess the prevalence, risk factors, and clinical sequelae of dermal lesions in IBD patients treated with anti–TNF-α antibodies.

Patients and methods

Results

A total of 1229 patients with IBD were included in the study. Among them, 541 were treated with anti–TNF-α antibodies, while 688 were not. Characteristics of both groups are presented in Table 1. The groups differed with respect to the proportion of patients with any dermal lesions (30.9% in the anti–TNF-α group vs 16.4% in the group not receiving anti–TNF-α drugs; V = 0.17; 95% CI, 0.11–0.22; P <⁠0.001). Types of dermal lesions with significantly different prevalence included treatment‑induced manifestations (5.5% vs 0%; P <⁠0.001) and the total group of skin‑related adverse events (19.% vs 7%; V = 0.18; 95% CI, 0.13–0.24; P <⁠0.001), within which differences were noted specifically for cutaneous infections, psorasiform reactions, and lupus‑like symptoms (7.8%, 1.5%, and 0.7%, respectively, in the anti–TNF-α group, and no patients in the group not receiving anti–TNF-α antibodies). Dermal lesions with significantly different prevalence are shown in Supplementary material, Figures S2–S5. Additionally, the groups differed in the frequency of dermatology visits, with 98.2% of the patients receiving and 13.4% of those not receiving anti–TNF-α treatment declaring that they regularly attended such appointments (V = 0.85; 95% CI, 0.81–0.87; P <⁠0.001).

A majority of the patients on the first‑line therapy received IFX (n = 319 [59%]), during a median (IQR) time of 175.25 (110–250) months. A total of 9 patients discontinued the first‑line treatment due to developing dermal lesions (lupus‑like symptoms in 4.3%, alopecia in 2.9%, psoriasis in 2.9%, vasculitis in 1.4%, and melanoma in 1.4%). However, the most common reasons for changing the therapy were loss of efficacy in 31 (44.9%) and a hypersensitivity reaction in 29 cases (42%). The most common drug used in the second‑line therapy was ADA in 58.8% of the patients (n = 40), followed by vedolizumab (VDZ) in 32.4% (n = 22), ustekinumab (UST) in 7.4% (n = 5), and IFX in 1.5% (n = 1) of the cases. The median (IQR) duration of the second‑line therapy was 8 (6–10) weeks. The most common reasons for therapy discontinuation were loss of efficacy in 2 cases (66.7%) and psoriasis in 1 case (33.3%). In the patient who developed psoriasis, the treatment was switched from IFX to UST. Only 10 patients received third‑line treatment, which was continued successfully in all the cases until the end of follow‑up, as shown in Table 2.

Characteristics of the patients treated with anti–TNF-α antibodies (n = 541), according to the IBD type, are presented in Table 3. The proportion of cutaneous infections was significantly higher among the patients with UC than among those with CD (43.8% vs 17.6%; V = 0.27; 95% CI, 0.1–0.43; P = 0.01) (Table 4). A comparison of UC patients treated and not treated with anti–TNF-α antibodies is shown in Table 5. In these patients, dermal lesions of any type occurred more often in the group receiving anti–TNF-α treatment than in the group not receiving such treatment (34.5% vs 17.6%; V = 0.19, 95% CI, 0.08–0.3; P = 0.01). The types of dermal lesions with different frequency of occurrence included treatment induced‑manifestations (5% vs 0%; P = 0.01) and the total group of skin‑related adverse events (25.2% vs 9.6%; V = 0.21; 95% CI, 0.09–0.32; P = 0.01), within which a difference was noted specifically for cutaneous infections (15.1% vs 0%; P <⁠0.001). The groups also differed with respect to the frequency of dermatology visits—in the anti–TNF-α group, 96.4% of the patients declared that they regularly attend ed such consultations, whereas in the anti‑TNF-α naive group, only 14.7% did so (V = 0.82; 95% CI, 0.75–0.88; P <⁠0.001). A comparison of patients with CD receiving and not receiving anti–TNF-α treatment is shown in Table 6. In these patients, the proportion of dermal lesions of any type was greater in the group treated with anti–TNF-α antibodies than in the treatment‑naive group (29.6% vs 16.1%; V = 0.16, 95% CI, 0.1–0.22; P <⁠0.001). Types of dermal lesions with different frequency of occurrence included treatment‑induced manifestations (5.7% vs 0%; P <⁠0.001) and the total group of skin‑related adverse events (16.9% vs 6.3%; V = 0.17; 95% CI, 0.11–0.23; P <⁠0.001), among which differences were specifically noted for xerosis and eczema (9.2% vs 5.1%; P = 0.02), cutaneous infections (5.2% vs 0%; P <⁠0.001), and psorasiform reactions (1.2% vs 0%; P = 0.01). A substantially greater proportion of CD patients treated with anti–TNF-α antibodies attended regular dermatology visits, as compared with the patients not receiving anti–TNF-α treatment (98.8% vs 13%; V = 0.85; 95% CI, 0.82–0.88; P <⁠0.001).

In the first step of analysis, IBD patients on anti–TNF-α medications with and without dermal lesions were compared. Significant differences were found for body mass index (BMI), which was higher in the patients with dermal lesions (mean difference, 0.86 kg/m², 95% CI, 0.21–1.52; P = 0.01). The percentage of patients with CD located in the small intestine was higher (14.4% vs 8%; V = 0.1; 95% CI, 0.01–0.19; P = 0.03), while the percentage of patients with CD located in the ileocecal region was lower in the individuals with dermal lesions (54.5% vs 64.4%; V = 0.09; 95% CI, 0.02–0.17; P = 0.04). Patients with and without dermal lesions also differed in terms of the reasons for termination of the first‑line therapy. In the group with dermal lesions, the most common reasons were hypersensitivity reactions (59.2%) and loss of effectiveness (22.4%). Only 9 patients (12.9%) required change of the biologic agent due to developing skin lesions. These included lupus‑like symptoms and psoriasis in 3 cases each (4.3%), alopecia in 2 cases (2.9%), and melanoma in 1 case (1.4%). The level of association between the reason for discontinuation of the first‑line therapy and presence of dermal lesions was weak (V = 0.06; 95% CI, 0–0.16; P <⁠ 0.001), as shown in Tables 7 and 8.

A 2‑step logistic regression analysis was performed to identify the risk factors for dermal lesions in the group receiving anti–TNF-α treatment. The first step consisted in building univariable logistic regression models for each predictor variable. Statistical significance levels were the main filter for selecting variables for the second step. It was assumed that only the variables with a P value below 0.25 in the univariable models can enter a multivariable model. In the second stage of the logistic regression analysis, the multivariable model based on the stepwise approach was run. The stepwise approach was used in the second phase of the logistic regression analysis to select final predictors for the multivariable model. At the stage of stepwise selection, BMI, presence of CD in the small intestine, duration of the first‑line therapy, and regular dermatology visits were indicated as the variables with the highest statistical quality for predicting occurrence of dermal lesions. Presence of CD in the ileum was rejected from the multivariable model at the stage of stepwise selection, despite its significance in the univariable model, as the multivariable model including this predictor did not yield accurate predictions.

In patients treated with anti–TNF-α medications, higher BMI increased the odds of developing dermal lesions. An increase in BMI by 1 kg/m² was associated with a 7% higher likelihood of dermal lesions (OR, 1.07; 95% CI, 1.02–1.13; P = 0.01). Presence of CD in the small intestine increased the likelihood of dermal lesions by 92% (OR, 1.92; 95% CI, 1.08–3.4; P = 0.03). The likelihood of dermal lesions was 34% lower for the patients with ileocecal location of CD (OR, 0.66; 95% CI, 0.46–0.96; P = 0.03). The other predictor variables did not reach the level of statistical significance.

Multivariable regression analysis showed that a 1‑kg/m² increment in BMI increased the likelihood of dermal lesions by 7% (OR, 1.07; 95% CI, 1.02–1.13; P = 0.01). Additionally, CD location in the small intestine was associated with doubled odds of skin lesion occurrence (OR, 2; 95% CI, 1.11–3.57; P = 0.02). Longer duration of the first‑line therapy also increased the risk—extending the therapy by 1 week was associated with 2% higher odds of developing skin lesions (OR, 1.02; 95% CI, 1–1.04; P = 0.04). The other predictors included in the multivariable regression analysis did not have a significant impact on the risk, when considered at the same time (due to the multivariable model step), as shown in Table 8.

The Hosmer–Lemeshow test yielded a P value of 0.88, indicating a good model fit. The fit of the model was additionally assessed with the Nagelkerke R², which was equal to 4.7% and indicated limited ability of the model to predict the total odds of developing dermal lesions. The collinearity of independent variables was assessed with VIF parameters, all of which were equal to 1, indicating no collinearity.

Discussion

The rate of IBD patients who develop dermal lesions during anti–TNF-α therapy varies significantly, ranging between 22% and 62%, depending on the study.^7,8,31-48 The reported percentage of patients requiring a change in or discontinuation of biologic treatment also varies, reaching as much as 34% in some studies.^7,8,31 In our cohort, dermal lesions were present in almost one‑third of the patients (30.9%), which is consistent with other studies. A surprising finding is the extremely low number of individuals who needed a change in or discontinuation of treatment due to dermal lesions (only 10 patients [5.9%]). The most likely explanation for this seems to be the strict policy of skin monitoring in the patients receiving biologic treatment in our center. This policy seems justified, as it allows for prompt implementation of local treatment of the lesions, in which case the anti–TNF-α therapy can be continued (provided the lesions are not advanced). This is best illustrated by an example of psoriasis, which occurred in 9 patients. A majority of them were able to continue the anti–TNF-α therapy after the implementation of local treatment, and only 3 patients with advanced lesions needed changes in their biologic therapy. There are no unequivocal guidelines for skin monitoring during immunosuppressive therapy in IBD; however, since such recommendations exists for other patient groups (eg, organ transplantat recipients), it seems most reasonable to extrapolate them.^32,33

In 2016, Cleynen et al³⁴ published a retrospective study that examined the prevalence of dermal lesions in a cohort of 917 patients with IBD treated with anti–TNF-α antibodies. Over a median follow‑up of 3.5 years (vs 3.65 years in our study), dermal lesions were found in 29% of patients, which is consistent with our results (30.9%). Among the observed lesions, the most common ones were psoriasiform eczema in 30.6% of the affected patients, eczema in 23.5%, xerosis cutis in 10.6%, palmoplantar pustulosis in 5.3%, and psoriasis in 3.8%. Other types of lesions were reported in 26.1% of the affected individuals. In our study, treatment‑induced skin manifestations were the most common, with 3 leading types: xerosis and eczema (8.5%), cutaneous infections (7.8%), and infusion and injection site reactions (5.5%). In contrast to our study, Cleynen et al³⁴ also examined location of the lesions—they were most often located around the folds, genitals, and scalp. In our study, there were no sex‑related differences in the frequency of skin lesions, while Cleynen et al³⁴ reported a female predominance. Interestingly, the authors observed a lack of correlation between the cumulative dose of IFX, IFX trough levels (TLs), and the development of anti‑IFX antibodies (ATIs) and the duration of treatment or the presence of skin lesions. In contrast, we found a significant correlation with the duration of therapy—an increase in therapy duration by 1 week was associated with a 1% higher prevalence of dermal lesions (OR, 1.02; 95% CI, 1–1.04; P = 0.04). The rate of therapy discontinuation or change due to skin lesion occurrence was almost twice as high in the cohort analyzed by Cleynen et al³⁴ (11% vs 5.9% in our study), which may have been related to easier access to other drug classes in Belgium.³⁴

In 2016, Hellström et al³⁵ conducted a prospective study in which nurses performed a skin screening based on a questionnaire prepared by doctors. A group of 118 patients with IBD receiving IFX infusions were evaluated, and noninfectious skin lesions were found in 27 patients (22.9%). Of those, 8 individuals (29.6%) had new lesions, while in another 8 (29.6%) the lesions had exacerbated. There were also more skin lesions than in our cohort, as they occurred in 30.9% of the patients, which could be related to the fact that the analysis was performed by gastroenterologists specializing in IBD and dermatologists. The identified risk factors were significantly different between the 2 reports. In the study by Hellström et al,³⁵ the factor most strongly associated with skin symptoms was CD, which occurred in almost 80% of the affected patients. In contrast, our study showed no effect of CD. The authors concluded that noninfectious skin lesions can be treated locally and do not require discontinuation of anti–TNF-α therapy, which is consistent with our findings.³⁵

In previous studies, the reported risk factors conducive to dermal lesion development included female sex, smoking, and the presence of antinuclear antibodies and / or anti–double‑stranded DNA antibodies. In our cohort, sex did not significantly differentiate the groups, while smoking and the presence of antibodies were not analyzed. At the same time, other risk factors were found—BMI, small intestinal lesions in CD, and duration of therapy. The latter seems to be a predictable risk factor for development of dermal lesions; however, conflicting results have been published in this regard. Cleynen et al³⁴found no relationship between the occurrence of skin lesions and longer duration of therapy. At the same time, they confirmed a lack of correlation between the cumulative dose of IFX, IFX TLs, and development of ATIs and the presence of skin lesions. Moreover, in a similar study, although performed in a population of patients with rheumatoid arthritis, psoriasis developed faster in individuals with a shorter duration of therapy and those receiving a lower dose of IFX (3 vs 5 mg/kg of body weight), which suggests that the cumulative dose is not involved in skin lesion pathogenesis. The influence of higher BMI on development of dermal lesions has already been demonstrated,^36,37,48and was further confirmed in our analysis. Overweight and obesity are associated with increased secretion of proinflammatory cytokines, which include interleukins (ILs) 17 and 23.^36,37 It is likely that their elevated levels may be responsible for the occurrence of skin lesions, which has already been shown by Włodarczyk et al.³⁸ Presence of inflammatory lesions in the small intestine in CD is considered to be one of the risk factors for a severe disease course. Recurrence of the disease after surgical treatment is observed in up to 80% of patients. Therefore, a correlation between such location of inflammatory lesions and development of dermal lesions is not a surprising finding. At the same time, there are no data on the profile of specific cytokines for a given lesion location in CD. However, it should be remembered that cytokines associated with development of skin lesions, such as IL‑17 and IL‑23, are those circulating in blood, and are not tissue‑specific. To support this hypothesis, we can refer to a study by Li et al,⁵⁰ who reported an increased concentration of IL‑23 in CD patients with inflammatory lesions localized in the small intestine and its positive correlation with psoriatic arthritis. A very important aspect of that work is the inability to assess the impact of thiopurines, the use of which is associated with development of some skin cancers.^5-8,26-30 In our study, both patient groups (receiving and not receiving anti–TNF-α treatment) included individuals who used thiopurines, as shown in Table 1.

We observed a difference in the prevalence of dermal lesions between UC and CD patients (43.8% vs 17.6%) only in the group treated with anti–TNF-α. This finding may be related to different treatment of these diseases—in UC, the only anti–TNF-α agent used is IFX, which causes more infections than ADA, including cutaneous ones.¹³

Dermal lesions are a common adverse effect of anti–TNF-α therapy. Occurrence of treatment‑induced lesions typically coincides with drug administration, and the lesions regress after therapy discontinuation. In unclear situations, biopsy may be useful; however, the diagnosis should be made on the basis of the entire clinical picture, mainly due to a large diversity of histopathologic findings. The most common dermal lesions, that is, psoriasis‑like eruptions, can be confirmed in up to 80% of cases. However, it is not possible to diagnose psoriasis‑like lesions induced by anti–TNF-α therapy based on histopathology alone. In some cases biopsy is necessary, especially when specific manifestations are suspected (continuous / contiguous CD or metastatic CD) or in disorders associated with IBD (epidermolysis bullosa acquisita), mucocutaneous conditions secondary to IBD treatment (cutaneous malignancies), and reactive manifestations (Sweet syndrome, pyoderma gangrenosum, pyodermatitis‑pyostomatitis vegetans, aseptic abscess ulcers, and PAPA syndrome).^7,8,31

Treatment of dermal lesions in patients receiving anti–TNF-α therapy, should be local in mild cases, and systemic in severe or generalized ones. At the same time, it should be remembered that there is no perfect algorithm for terminating anti–TNF-α treatment. The decision should be made primarily based on the experience of the gastroenterologist, in consultation with a dermatologist and / or a rheumatologist. However, it seems reasonable to assume that the indications for discontinuation are resistance to treatment, mild localized lesions, generalized lesions, and some diagnoses, such as adverse mucocutaneous reactions (injection site reactions, infusion reactions, paradoxical reactions, eczema, psoriasiform reactions, and life‑threatening disorders), cutaneous malignancies, lupus‑like symptoms, or vasculitis.^7,8,31 A proposed regimen for the management of dermal lesions can be found in Supplementary material, Figures S6–S10.