Tumor necrosis factor inhibitors are associated with a decreased risk of COVID‐19‐associated hospitalization in patients with psoriasis—A population‐based cohort study

Abstract The risk of coronavirus disease 2019 (COVID‐19) and its complications among patients with psoriasis treated by tumor necrosis factor inhibitors (TNFis) remains to be decisively delineated. We aimed to assess the risk of COVID‐19 infection, COVID‐19‐associated hospitalization, and mortality among Israeli patients with psoriasis treated by TNFi relative to other systemic agents. A population‐based cohort study was conducted to compare psoriasis patients treated by TNFi (n = 1943), with those treated by methotrexate (n = 1929), ustekinumab (n = 348), and acitretin (n = 1892) regarding COVID‐19 outcomes. Risk of investigated outcomes was assessed using uni‐ and multi‐variate Cox regression analyses. The incidence rate of COVID‐19, COVID‐19‐associated hospitalization, and mortality in the TNFi group was 35.8 (95% CI, 26.1‐47.9), 0.8 (95% CI, 0.0‐4.2), and 0.0 per 1000 person‐years, respectively. Exposure to TNFi was associated with a comparable risk of COVID‐19 infection [adjusted hazard ration (HR) for TNFi vs methotrexate: 1.07 (95% CI, 0.67‐1.71); TNFi vs ustekinumab: 1.07 (95% CI, 0.48‐2.40); TNFi vs acitretin: 0.98 (95% CI, 0.61‐1.57)]. TNFi was associated with a decreased risk of COVID‐19‐associated hospitalization relative to methotrexate (adjusted HR, 0.10; 95% CI, 0.01‐0.82) and ustekinumab (adjusted HR, 0.04; 95% CI, 0.00‐0.64), but not to acitretin (adjusted HR, 1.00; 95% CI, 0.16‐6.16). No significant difference in COVID‐19‐associated mortality was found between the four different groups. TNFi was associated with a decreased risk of admissions due to COVID‐19. Our findings substantiate the continuation of TNFi treatment during the pandemic. TNFi may be positively considered in patients with moderate‐to‐severe psoriasis warranting systemic treatment during the pandemic.


| INTRODUCTION
Given that severe coronavirus disease 2019 (COVID-19) is associated with a hyperinflammatory state, 1,2 it is highly pertinent to investigate whether the presence of preexisting immune-mediated diseases or the previous exposure to immunomodulatory drugs affect the manifestations of  While several studies displayed an increased risk and more aggressive course of SARS-CoV-2 infection in patients with certain immune-mediated and inflammatory diseases, [3][4][5] others disproved these observations. [6][7][8][9][10] Tumor necrosis factor inhibitor (TNFi), the largest drug class worldwide, 11 was proved efficacious in a wide array of autoimmune and inflammatory diseases. 12 While TNFi has dramatically improved outcomes of patients with moderate-to-severe plaque psoriasis, it is implicated with different infections. 13 While the use of this immunomodulatory drug during the current pandemic raised many concerns, recent observational studies ascribed a favorable protective role for TNFi among patients with inflammatory bowel disease (IBD) and rheumatic diseases who developed 14 In psoriasis, however, the risk of COVID and its complications among patients undergoing TNFi treatment remains to be delineated. The current knowledge about the safety of immunomodulatory drugs in psoriasis stems from observational studies of short follow-up durations and lacks head-to-head comparison between different agents, as most former studies pooled patients under ''biologics'' or ''immunomodulatory drugs'' together. [15][16][17][18][19] In the current study, we sought to investigate the risk of COVID- 19 infection, COVID-19-associated hospitalization, and mortality among patients with psoriasis treated by TNFi (adalimumab, etanercept, and infliximab). To precisely assess the safety of this treatment during the pandemic, patients on TNFi were compared with three reference groups: (i) psoriasis patients treated by methotrexate, (ii) ustekinumab, and (iii) acitretin. A sensitivity analysis was held to dissect the outcomes under each one of the three investigated TNF agents.

| Study design and dataset
We performed a historical retrospective cohort study that followed patients with psoriasis to assess the incidence of COVID-19, COVIDassociated hospitalization, and mortality. The study was approved by the institutional review board (IRB) in accordance with the declaration of Helsinki (approval code: 0212-17-COM).
The current study was based on the computerized database of Clalit Health Services (CHS). CHS is the largest healthcare maintenance organization in Israel, covering a wide variety of private and public healthcare services for 4 540 768 enrollees as of October 2018. CHS database is typified by an inclusive nature owing to its ability to retrieve data from numerous sources originating both from ambulatory and hospitalized care settings. CHS is additionally eminent in negligible loss to follow-up and free access to healthcare services.
All these features render the CHS database highly compatible with the conduction of reliable and robust epidemiological analyses. 20

| Study population
The computerized database of CHS was systematically screened for all prevalent cases with a diagnostic code of psoriasis as registered by board-certified dermatologists. Eligible patients had to be alive and active members of CHS at the onset of the pandemic, defined as the date of the first confirmed case of COVID-19 in Israel (February 27th, 2020).

| Definition of exposure and different analyses
Exposure to all drugs was defined in case the drug was prescribed for at least 1 month during the pandemic. For the main analysis, patients with psoriasis receiving TNFi were compared to those receiving methotrexate as a reference group. Methotrexate was selected as a referent drug given that it is the most commonly prescribed nonbiologic systemic therapy for psoriasis, 21 and in accordance with other international registry-based studies. 22 Patients managed by TNFi or methotrexate in conjunction with other concomitant systemic immunomo dulatory/immunosuppressive drugs were excluded from the analysis enabling to evaluate the independent influence of the drug of interest.
A sensitivity analysis was performed to assess the risk of COVID-19 outcomes in patients treated by the three widely utilized TNFi agents in Israel; adalimumab, etanercept, and infliximab. Each of these drugs was separately compared to methotrexate. Numerous patients were exposed to more than a single TNFi agent during the pandemic.

| Definition of COVID-19-related outcomes
The medical records of eligible patients were checked for a diagnosis of COVID-19. The diagnosis of COVID-19 was based on confirmation of cases by US FDA-approved molecular tests. COVID-19-associated hospitalization was defined in COVID-19-confirmed patients admitted to intensive care units, internal medicine, or COVID-19-specific respiratory inpatient wards. COVID-19-associated mortality was defined in COVID-19-confirmed patients whose cause of death was ascribed to COVID-19 or its complications.
Study participants date of death was ascertained by linking the study cohort with the Ministry of Interior registry. All study participants were followed up from the onset of the pandemic in Israel or the date of drug initiation, whichever occurs later, until October 2, 2020, drug discontinuation, death, or fulfilling the study outcomes, whichever occurs earlier.

| Covariates
Outcome measures were adjusted for the following comorbid conditions: chronic obstructive pulmonary disease (COPD), chronic renal failure (CRF), diabetes mellitus, ischemic heart disease (IHD), hypertension, hyperlipidemia, and malignancies, all of which were evidenced to project worse prognosis of COVID-19. 23-27 COVID-19 outcomes were additionally adjusted for smoking owing to the association of the latter with worse outcomes of COVID-19. 23,24 The chronic registry of CHS was utilized to identify comorbidities of eligible patients prior to the development of COVID-19.

| Statistical analysis
The comparison between different variables was performed utilizing the chi-square test and t test for categorical and continuous variables, respectively. Incidence rates of outcomes were calculated and expressed as the number of events per 1000 person-years. Hazard ratios (HR)s for the risk of incident outcomes were obtained by the use of the Cox regression model. Two-tailed p-values less than 0.05 were considered as statistically significant. All statistical analyses were performed using the SPSS software, version 25 (SPSS, Armonk, NY:

| RESULTS
The current study included 1943, 1929, 348, and 1892 patients with psoriasis treated by TNFi, methotrexate, ustekinumab, and acitretin during the pandemic, respectively. Relative to patients treated by methotrexate, those treated by TNFi were younger at the onset of the pandemic, had a male predominance, and a lower frequency of COPD, diabetes mellitus, hypertension, hyperlipidemia, ischemic heart disease, and malignancy. The baseline characteristics of study participants are delineated in Table 1. 3.1 | Primary analysis comparing the risk of COVD-19 outcomes associated with TNFi relative to methotrexate   Table 2).
In a sensitivity analysis, we estimated the risk of the aforementioned outcomes in patients under adalimumab (n = 1166; Table S1), etanercept (n = 643; Table S2), and infliximab (n = 176; Table S3) relative to psoriasis patients managed by methotrexate.
While the incidence rate of COVID-19-associated hospitalization and mortality was numerically lower among patients managed by each one of the three drugs, the HRs of the aforementioned outcomes fell out of significance. Of note, 42 patients were treated by more than a single agent during the course of the pandemic.
The second secondary analysis evaluated the differential risk of TNFi (n = 1869) as compared to acitretin (n = 1892). Out of the original subgroup of patients exposed to TNFi, 74 patients were excluded in the current analysis since they were concomitantly treated by acitretin. The risk of the three COVID-19 outcomes of interest was comparable between the two subgroups (Table 4).

DATA AVAILABILITY STATEMENT
The datasets generated during and/or analysed during the current study are available from the corresponding author on reasonable request.