New-Onset Psoriasis With Tumor Necrosis Factor-α Inhibitor Treatment in Inflammatory Diseases

• Updated on: 2024-04-06
• Read original article here

Question  What is the extent of new-onset psoriasis associated with tumor necrosis factor-α inhibitor (TNFi) treatment compared with nonbiologic conventional treatment?

Findings  In this cohort study including 109 085 patients, of which 108 024 patients received conventional therapy and 20 910 patients received TNFi treatment, the incidence rates for any type of psoriasis per 1000 patient-years were 3.0 for conventional therapy and 7.8 for TNFi with an estimated absolute risk difference around 5 per 1000 patient-years. The time in TNFi exposure needed for 1 additional event of TNFi-associated psoriasis was 241 patient-years for any type of TNFi-associated psoriasis, and the hazard ratio for any type of new-onset psoriasis for patients receiving TNFi was 2.38 compared with conventional therapy.

Meaning  Although TNFi treatment in patients with immune-mediated inflammatory diseases is associated with increased risk of new-onset psoriasis compared with conventional therapy, the absolute risk remains modest; physicians should be aware of this adverse event but ought not change their initial approach to treatment with TNFis.

Importance  Tumor necrosis factor-α inhibitor (TNFi)–associated psoriasis is a rare adverse event following TNFi treatment. Data on the risk of developing TNFi-associated psoriasis when treated with TNFi are sparse.

Objective  To investigate the associated risk between new-onset psoriasis and TNFi treatment compared with nonbiologic conventional treatment.

Design, Setting, and Participants  Using Danish national registries (1995-2018), this cohort study included patients with inflammatory bowel disease (IBD) and/or rheumatoid arthritis (RA) who received either conventional therapy or TNFi treatment. Patients may not have been diagnosed with psoriasis prior to initiation of treatment. Patients were followed up for up to 5 years. Cox regression models with robust variance were used to compare the risk of developing any type of psoriasis, nonpustular psoriasis, and pustular psoriasis. Patients receiving conventional therapy were used as reference. Data analysis was performed from January 1995 to December 2018.

Exposures  For the present study, the term conventional therapy was used for the nonbiological therapy. For biological therapy, a distinction was made between TNFi treatment and non-TNFi biological therapy.

Main Outcomes and Measures  The outcome of psoriasis was defined as a registered International Statistical Classification of Diseases and Related Health Problems, Tenth Revision code of psoriasis and/or having 2 consecutive prescriptions of topical vitamin D analogues.

Results  The study included 109 085 patients, of which 62% were female. Median (IQR) age was 50 (34-64) years. Of the included patients, 108 024 received conventional therapy and 20 910 received TNFi treatment. During follow-up, 1471 (1.4%) patients developed any type of psoriasis, of which 1332 developed nonpustular psoriasis, 127 patients developed palmoplantar pustulosis, and 12 patients developed generalized pustulosis. The incidence rates for developing any type of psoriasis per 1000 patient-years were 3.0 (95% CI, 2.9-3.2) for conventional therapy and 7.8 (95% CI, 7.5-8.9) for TNFi. During treatment with TNFi, the hazard ratio was 2.12 (95% CI, 1.87-2.40; P < .001) for developing nonpustular psoriasis and 6.50 (95% CI, 4.60-9.23; P < .001) for pustular psoriasis compared with conventional treatment. Exposure needed for 1 additional patient to be harmed was 241 patient-years for any type of TNFi-associated psoriasis, 342 patient-years for nonpustular psoriasis, and 909 patient-years for pustular psoriasis.

Conclusions and Relevance  In a Danish nationwide cohort of patients with immune-mediated inflammatory diseases treated with TFNi or conventional treatment and no history of psoriasis, in TFNi-treated patients, nonpustular types of psoriasis constituted the most events, whereas pustular types of psoriasis had the highest relative risk. Although the risk of new-onset psoriasis increased for both nonpustular and pustular types of psoriasis in TFNi-treated patients, the absolute risk remained modest at 241 patient-years of exposure need for 1 additional event and an estimated absolute risk difference around 5 per 1000 patient-years, indicating that the approach to treatment of patients in need of TNFi treatment should not change.

The introduction of biologics has improved treatment and management of immune-mediated inflammatory diseases (IMIDs) such as psoriasis, inflammatory bowel disease (IBD), and rheumatoid arthritis (RA). Biologics are in most cases effective, and the majority of adverse effects are mild and easy to manage. Paradoxically, psoriasis develops in a number of patients during tumor necrosis factor-α inhibitor (TNFi) treatment—henceforth referred to as TNFi-associated psoriasis. This psoriasis appears to be a result of a prolonged production of type I interferon caused by the inhibition of the TNF-dependent maturation process of plasmacytoid dendritic cells. The clinical presentation is heterogeneous and includes both nonpustular (eg, guttate psoriasis and psoriasis vulgaris) and pustular types of psoriasis (eg, palmoplantar pustulosis). It most often occurs within the first 1.5 years of treatment with TNFi but can occur within a week and up to several years after the initiation of treatment. Even though TNFi agents are widely used, studies on the epidemiology of TNFi-associated psoriasis are scarce, limited in the number of included patients, and often include patients with established disease, making it difficult to interpret data clearly because the development of psoriasis may be a result of treatment failure or an adverse reaction. Furthermore, IBD is associated with development of psoriasis, and as studies on TNFi-associated psoriasis seldom use control groups, the extent of TNFi-associated psoriasis remains unknown. The objective of this Danish nationwide register-based cohort study was to investigate the risk of new-onset psoriasis in patients with IBD or RA during TNFi treatment compared with conventional nonbiologic treatment.

The Danish health care system is based on a tax-financed system that offers free and equal access to health services throughout the primary, secondary, and tertiary sector. Treatment with biologics has mostly been managed at hospital departments throughout follow-up. In Denmark, ethical approval is not required for register-based studies. The study was registered in the Capital Regions inventory (No. VD-2018-286), which constitutes the necessary legal approvals.

At the date of birth or immigration, all Danish residents are assigned a unique personal identification number allowing for accurate linkage of patient data. The following data sources were used: (1) the Danish National Patient Registry, which holds information on all hospital admissions and outpatient visits (from 1995 based on International Classification of Diseases [ICD]; ICD-8 from 1977-1993 and ICD-10 1994-recent), and information on all medical procedures (NOMECO Classification of Surgical Procedures since 1996); (2) the Danish Register of Medicinal Product Statistics, which contains information on all dispensed prescriptions since 1995 using the international Anatomical Therapeutical Chemical system; and (3) the National Population Registry, which contains information on vital status and sex.

All adult patients in Denmark with IBD or RA were included from the year 1995 through 2018. Patients were excluded if they had a history of any type of psoriasis prior to treatment or if a patient did not receive relevant pharmacotherapy after age 18 years.

Disease status in the form of IBD, RA, and baseline comorbidities was defined by the presence of a relevant ICD-10 registration (eTable 1 in the Supplement). Most ICD-10 diagnoses in the registries were validated. For ulcerative colitis, the sensitivity was 81% and specificity was 86%; for Crohn disease, the sensitivity was 95% and the specificity was 93%; and for RA, the positive predictive value was 79%.

The active pharmacologic therapy defined the exposure in the present study design. The individual drugs were grouped as biological and nonbiological treatment. The nonbiological treatment consisted mostly of conventional synthetic disease-modifying antirheumatic drugs. The term conventional therapy was used for the nonbiological therapy. For the biological therapy, a distinction was made between TNFi treatment and non-TNFi biological therapy. The active pharmacological therapy was determined by a combination of drugs dispensed at the pharmacy using Anatomical Therapeutical Chemical codes and drugs administered at the hospital using Sundhedsvæsenets Klassifikations System codes (eTable 2 in the Supplement). During the time frame of 1995 to 2018, patients could be exposed to more than 1 therapy from the prespecified pharmacologic groups. For these patients, each treatment option was handled independently.

The date of inclusion was defined as the first date by which a relevant pharmacotherapy was dispensed for IBD or RA after age 18 years. For each relevant therapy, patients were followed up from the date of inclusion to 6 months after the cessation of that therapy, the development of psoriasis, or for a maximum of 5 years, whichever occurred first. We introduced the grace period of 6 months following the last administration or filled prescription of a drug to minimize the risk of missing relevant psoriasis events (eFigure 1 in the Supplement). This was implemented because patients who developed new-onset psoriasis—drug associated or not—might make changes to their treatment as a result of the new diagnosis. The time in treatment was calculated as the sum of each period in that treatment. Patients were stratified according to the first diagnosis of IBD or RA. Patients could have multiple IMIDs.

The outcome of psoriasis was defined as a registered ICD-10 code of psoriasis and/or having 2 consecutive prescriptions of topical vitamin D analogues (eTable 2 in the Supplement). Both are established and validated methods with a positive predictive value for adults of 98% and 83%, respectively. Combining both methods increases the sensitivity and minimizes the risk of missing events of psoriasis. Patients were additionally divided into pustular and nonpustular types of psoriasis according to ICD-10 (eTable 1 in the Supplement). The registries do not differentiate between drug-associated psoriasis and conventional psoriasis but allow us to establish new-onset psoriasis. The statistical comparison between the development of psoriasis in patients treated with conventional therapy and TNFi thus served as a proxy for the development of TNFi-associated psoriasis. For patients receiving 2 or more overlapping treatments as they developed psoriasis, the event could not be clearly assigned to a single treatment. To avoid unnecessary complexity in the interpretation of results, overlapping events were not considered in the main analysis.

For baseline characteristics, differences between groups were compared by 1-way t test or Kruskal-Wallis test for continuous variables, and Pearson χ for categorical variables. Incidence rates were calculated per 1000 person-years, accounting for the competing risk of death. Cumulative incidence curves were reported for the type of psoriasis developed (all events, nonpustular events, and pustular events). Exposure needed for 1 additional patient to be harmed was estimated based on 1/[unexposed incidence rate × (HR − 1)]. This measure was used to investigate discrepancy in new-onset psoriasis associated with TNFi compared with conventional therapy and can thus approximate the increase in TNFi-associated psoriasis. Because patients might be represented in more than 1 group and the proportionality assumption was not met for all variables (age and sex), Cox-proportional hazard regression with a robust variance adjusted for age, sex, and IMID was used. We tested for interaction between outcome and sex and outcome and age, respectively, by a likelihood ratio test and found no significant interactions unless stated otherwise. For all analyses, a P value of <.05 was considered statistically significant. The following sensitivity analyses were performed: (1) Outcomes occurring in patients with 2 or more overlapping treatments were assigned to each of the individual treatments (overlapping events). (2) We included overlapping events, but because most of the events were in the grace period, we revoked the 6-month grace period. (3) With similar setup as the main analysis, patients treated with non-TNFi biologics served as the comparison group. (4) Analysis was performed only including diagnosis made by a dermatologist. All statistical analyses were performed in R statistical software, version 4.0.2 (R Foundation for Statistical Computing). This study followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline.

A total of 199 398 patients diagnosed with either IBD (99 565 [49%]) or RA (102 939 [52%]) were identified. Of these, 90 313 patients were excluded owing to a history of psoriasis, overlapping events, and a lack of relevant pharmacotherapy, resulting in 109 085 unique patients (62% female; median [IQR] age, 50 [34-64] years), of which 106 765 (98%) had received conventional therapy and 20 387 (19%) had received TNFi (Figure 1). Baseline characteristics according to disease can be found in eTable 3 in the Supplement. An additional 4909 (5%) patients received non-TNFi biologics. Baseline characteristics on patients receiving non-TNFi biologics can be found in eTable 4 in the Supplement. From the total number of included patients, 18 167 (19%) patients received more than 1 type of therapy during the follow-up period, and 2756 (3%) patients received concomitant therapy—conventional and TNFi treatment at the same time. For all treatment options, the mean (SD) age of patients receiving TNFi treatment (40.9 [16.2] years) was lower than the mean age of patients receiving conventional pharmacotherapy (50.9 [18.4] years) (Table 1).

During follow-up, a total of 1471 (1.4%) patients developed any type of new-onset psoriasis. Of these patients, 1332 (91%) developed nonpustular psoriasis, 139 (9%) developed pustular psoriasis of which 127 (91%) developed palmoplantar pustulosis, and 12 (9%) developed generalized pustulosis. The median (IQR) follow-up time in therapy was 3.1 (1.1-5.0) years for patients who did not develop psoriasis, 1.3 (0.6-2.6) years for patients who developed nonpustular psoriasis, and 1.2 (0.5-3.0) years for patients who developed pustular psoriasis.

The incidence rates for any new-onset psoriasis per 1000 patient-years were 3.0 (95% CI, 2.9-3.2) for conventional therapy and 7.8 (95% CI, 7.5-8.9) for TNFi treatment (Table 2). The corresponding incidence rates for nonpustular psoriasis were 2.8 (95% CI, 2.7-3.0) for conventional therapy and 6.5 (95% CI, 5.8-7.2) for TNFi, while the event rates for pustular psoriasis were 0.2 (95% CI, 0.15-0.25) for conventional therapy and 1.3 (95% CI, 1.0-1.6) for TNFi treatment. Cumulative incidence curves are shown in Figure 2. The cumulative incidence plots for each disease (IBD and RA) are shown in eFigures 2 and 3, respectively, in the Supplement.

The hazard ratio (HR) of developing any type of new-onset psoriasis for patients receiving TNFi treatment was 2.38 (95% CI, 2.13-2.68; P < .001) compared with conventional therapy (Figure 3). The HR of developing a nonpustular type of psoriasis was 2.12 (95% CI, 1.87-2.40; P < .001) for TNFi treatment, while the HR of developing a pustular type of psoriasis was 6.50 (95% CI, 4.60-9.23; P < .001) for TNFi treatment. No difference in risk was found between the sexes. The overall risk of developing any type of new-onset psoriasis was lower for IBD compared with RA (HR, 0.63; 95% CI, 0.54-0.72; P < .001). Similar results were found for nonpustular psoriasis but not pustular psoriasis. However, for all types of psoriasis, we found an interaction between TNFi treatment and IBD. Treatment with TNFi in patients with IBD was thus associated with a higher risk of new-onset psoriasis compared with TNFi treatment in patients with RA (HR, 2.07; 95% CI, 1.65-2.60; P < .001).

The increased risk of psoriasis in patients receiving TNFi compared with conventional treatment was considered a result of TNFi-associated psoriasis. For patients with IBD or RA, the time in TNFi exposure needed for 1 additional event (1 event of TNFi-associated psoriasis) was 241 patient-years for any type of TNFi-associated psoriasis, 342 patient-years for nonpustular psoriasis, and 909 patient-years for pustular psoriasis.

A total of 229 events occurred in patients with more than 1 treatment at the time of event (overlapping events), eg, treatment with both conventional therapy and TNFi. Inclusion of overlapping events resulted in increased risk of developing any type of psoriasis (HR, 2.78; 95% CI, 2.53-3.06; P < .001) and nonpustular psoriasis (HR, 2.49; 95% CI, 2.25-2.77; P < .001) and lowered the risk of pustular psoriasis (HR, 6.25; 95% CI, 4.83-8.08; P < .001) in patients treated with TNFi compared with conventional therapy. In most cases, the overlapping events were observed after treatment cessation during the grace period (eFigure 4 in the Supplement).

In an additional sensitivity analysis, the grace period was revoked while overlapping events were allowed for. Here, 1236 events were included. In this sensitivity analysis, the HR of developing any type of new-onset psoriasis was 2.69 (95% CI, 2.40-3.02; P < .001), 2.55 (95% CI, 2.26-2.87; P < .001) for nonpustular psoriasis, and 4.15 (95% CI, 0.70-5.53; P = .20) for pustular psoriasis in patients treated with TNFi compared with conventional therapy. Cumulative incidence plots are shown in eFigure 5 in the Supplement.

Aware that the increased risk of psoriasis may be associated with the increased disease burden found in patients receiving biological therapy, we compared patients who received TNFi treatment with patients receiving non-TNFi biological therapies. The incidence rates per 1000 person-years were 3.9 (95% CI, 2.8-5.4) for any types of new-onset psoriasis, 3.5 (95% CI, 2.5-4.9) for nonpustular psoriasis, and 0.4 (95% CI, 0.1-1.0) for pustular psoriasis among patients treated with non-TNFi biologics. Cumulative incidence plots are shown in eFigure 6 in the Supplement. The HR of developing any type of psoriasis when comparing TNFi with non-TNFi biological therapy was 1.98 (95% CI, 1.43-2.73; P < .001) for any type of psoriasis, 1.85 (95% CI, 1.31-2.61; P < .001) for nonpustular psoriasis, and 3.11 (95% CI, 1.15-8.46; P = .03) for nonpustular psoriasis. Results on the sensitivity analysis using diagnosis only made by a dermatologist demonstrated an increase in exposure needed for 1 additional event (eTable 5 in the Supplement).

The epidemiologic characteristics of TNFi-associated psoriasis have been reported with conflicting results. Incidence rates of new-onset psoriasis during TNFi have been recorded as high as 50 per 1000 person-years in patients with IBD (based on 42 cases) and as low as 1 per 1000 patient-years in RA (based on 25 cases). A larger study of 125 cases of new-onset psoriasis during TNFi treatment in patients with IBD found an incidence rate of 5.3 per 1000 person-years. However, these studies often include patients with established psoriasis and may thus report incidence of new-onset psoriasis combined with psoriasis flares without being able to distinguish between the two and take the background risk of psoriasis or treatment failure into account. To our knowledge, no other study has been able to include this high number of patients and cases with both IBD and RA, exclude patients with established psoriasis, compare patients treated with a TNFi to a relevant group of patients receiving conventional treatment (control group), and complete sensitivity analyses including patients receiving non-TNFi biological therapy. This thus allowed the current study to estimate the relative risk and absolute risk of new-onset psoriasis in patients with IBD or RA receiving TNFi treatment with greater certainty than previous attempts as well as assessing the exposure needed for 1 additional event.

In the current study, we found an incidence rate of any type of new-onset psoriasis of 7.8 per 1000 person-years in patients treated with a TNFi and an incidence rate of 3.0 per 1000 person-years in patients treated with conventional nonbiologic treatment. This could potentially indicate that the incidence rate of TNFi-associated psoriasis to be approximated to 5 per 1000 person years of treatment. Because there is a possible association between disease severity and the risk of developing an additional IMID, the increased risk of new-onset psoriasis during treatment with TNFi could be a result hereof. However, when including non-TNFi biologics as a reference group, similar results were observed. This finding indicates that new-onset psoriasis in patients receiving TNFi treatment is not associated with disease severity and supports the notion that risk is likely associated with TNFi. However, we cannot refute that the interaction between TNFi and IBD may potentially be associated with the existing association between IBD and psoriasis or the increased use of infliximab in this patient group.

We observed that psoriasis during treatment with TNFi occurred throughout the entire treatment period and that the median time until development of new-onset psoriasis was estimated to be a little more than a year, in line with previous studies. Additionally, we found nonpustular TNFi-associated psoriasis to be the most prevalent type of TNFi-associated psoriasis, while pustular psoriasis constituted the highest risk in relative terms. Although the finding that a majority of events were of the nonpustular type is in line with recent studies on TNFi-associated psoriasis, these studies have found that 30% to 40% of TNFi-associated psoriasis cases are palmoplantar pustulosis, which is a significantly higher proportion of patients when compared with the study. This discrepancy could potentially be a consequence of patient selection and the definitions of new-onset psoriasis, as we cannot refute that cases of misdiagnosis may have occurred. This would most likely affect patients diagnosed with RA, which shares some symptoms found in psoriatic arthritis. However, the risk of misdiagnosis would be evident across all treatment options and thus if relevant would only be expected to affect the results marginally. Additionally, palmoplantar pustulosis may potentially be underreported relative to nonpustular types of psoriasis because this disease requires specialist’s treatment available at hospitals, whereas we used both hospital diagnosis and prescription medicine to increase the sensitivity.

The present study was subject to several limitations. The databases are unable to differentiate patients diagnosed with psoriasis and patients diagnosed with psoriasis directly due to treatment. Using conventional treatment to compare the risk of TNFi-associated new-onset psoriasis, we attempted to eliminate that problem to the furthest extent. Additionally, both pustular and nonpustular types of psoriasis are likely underdiagnosed and underreported in the databases, and although the introduction of calcipotriol in addition to ICD-10 diagnoses increased the sensitivity, it may yet be difficult to capture all relevant events. Thus, the incidence rates of psoriasis during treatment were most likely underestimated, and the exposure needed for 1 additional patient to be harmed was probably lower. However, cases not identified are believed to have milder forms of psoriasis and might not be of clinical significance, and though the case definition for psoriasis has been validated by this group, it may fail to capture all patients. Furthermore, the pustular type of TNFi-associated psoriasis has been highlighted in the literature to be associated with TNFi, which could introduce observation bias toward symptoms associated with this type of TNFi-associated psoriasis. We cannot refute the higher risk of pustular psoriasis to be associated with this, but as this contributed to a small part of the overall cases of psoriasis, the interpretation would likely remain unaltered. Additionally, the outcome of psoriasis was assessed by dermatologists, but other medical specialists were able to register relevant psoriasis diagnoses in the databases. This may introduce a possibility of misdiagnosis, but such a risk would most likely be nondifferential and thus not introduce a bias toward either treatment option. Subsequent sensitivity analyses were completed only including diagnosis made by a dermatologist (eTable 5 in the Supplement). We could not assess the individual drug for TNFi, but the risk of TNFi-associated psoriasis appears to be a class effect. Lastly, important clinical parameters, relevant family history, disease severity, and the reason for discontinuation of treatment were unavailable. The strengths of the study include the large number of included patients, the number of events, the inclusion of RA and IBD, the use of non-TNFi biologics to take into account disease severity, and the exclusion of patients with established psoriasis.

In this cohort study, treatment with TNFi in patients with IBD or RA was associated with a 2-fold increased risk of new-onset psoriasis compared with conventional therapy. Nonpustular types of TNFi-associated psoriasis constituted the most events, whereas pustular types of TNFi-associated psoriasis had the highest relative risk with a 6-fold increase. Nevertheless, new-onset psoriasis due to TNFi treatment was a rare adverse event with the lowest estimate of exposure needed for 1 additional event of approximately 250 patient-years of TNFi treatment and an estimated absolute risk difference around 5 per 1000 patient-years. Practitioners and patients should be aware and observant of the potential for TNFi-associated psoriasis during TNFi treatment but keep in mind that the absolute risk appears to be low.

Corresponding Author: David Thein, MB, Department of Dermatology, Bispebjerg Hospital, University of Copenhagen, Bispebjerg Bakke 23, 2400 København NV, Denmark (david.nikolai.thein.aagaard@regionh.dk).

Author Contributions: Mr Thein and Dr Loft had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

Critical revision of the manuscript for important intellectual content: All authors.

Conflict of Interest Disclosures: Mr Thein reported receiving research grants from Herlev and Gentofte Hospital, Mogens Wedell-Wedellsborgs Fond, and Ebba Celinders Legat during the conduct of the study. Dr Egeberg reported having received research funding from Pfizer, Eli Lilly, Novartis, Bristol Myers Squibb, AbbVie, Janssen Pharmaceuticals, the Danish National Psoriasis Foundation, the Simon Spies Foundation, and the Kgl Hofbundtmager Aage Bang Foundation, and honoraria as consultant and/or speaker from AbbVie, Almirall, LEO Pharma, Zuellig Pharma Ltd, Galápagos NV, Sun Pharmaceuticals, Samsung Bioepis Co, Ltd, Pfizer, Eli Lilly, Novartis, Union Therapeutics, Galderma, Dermavant, UCB, Mylan, Bristol Myers Squibb, and Janssen Pharmaceuticals outside the submitted work. Prof Skov reported personal fees (speaker) from AbbVie, Eli Lilly, Novartis, Pfizer, and LEO Pharma; personal fees (advisory board) from AbbVie, Janssen-Cilag, Novartis, Eli Lilly, LEO Pharma, UCB, Almirall, Bristol Myers Squibb, Boehringer Ingelheim, and Sanofi; investigator feed to the hospital from AbbVie, Sanofi, Janssen-Cilag, Boehringer Ingelheim, AstraZeneca, Eli Lilly, Novartis, and Regeneron; and grants to the hospital from Novartis, Sanofi, Bristol Myers Squibb, Janssen-Cilag, UCB, and LEO Pharma outside the submitted work. Dr Loft reported personal fees from Janssen-Cilag and Eli Lilly outside the submitted work.

Additional Information: The data supporting the findings of this study are available from Statistics Denmark, but restrictions apply to the availability of these data, which were used under license for the current study, and thus are not publicly available. Data are available from the authors upon reasonable request and with permission from Statistics Denmark.