This Figure provides information on the adjusted hazard ratio (aHR) (95% CI) for any cancer except nonmelanoma skin cancer (NMSC) as well as individual cancers of interest in the overall, mild, and moderate-to-severe psoriasis groups. a Adjusted for age, BMI, drinking and smoking status, b adjusted for age and sex, c adjusted for age, sex, and smoking status, d adjusted for age only, e includes only men, f includes only women. CTCL indicates cutaneous T-cell lymphoma.
Customize your JAMA Network experience by selecting one or more topics from the list below.
Chiesa Fuxench ZC, Shin DB, Ogdie Beatty A, Gelfand JM. The Risk of Cancer in Patients With Psoriasis: A Population-Based Cohort Study in the Health Improvement Network. JAMA Dermatol. 2016;152(3):282–290. doi:10.1001/jamadermatol.2015.4847
Copyright 2016 American Medical Association. All Rights Reserved. Applicable FARS/DFARS Restrictions Apply to Government Use.
The risk of cancer in patients with psoriasis remains a cause of special concern due to the chronic inflammatory nature of the disease, the use of immune-suppressive treatments and UV therapies, and the increased prevalence of comorbid, well-established risk factors for cancer, such as smoking and obesity, all of which may increase the risk of carcinogenesis.
To compare the overall risk of cancer, and specific cancers of interest, in patients with psoriasis compared with patients without psoriasis.
Design, Setting, and Participants
Population-based cohort study of patients ages 18 to 89 years with no medical history of human immunodeficiency virus, cancer, organ transplants, or hereditary disease (albinism and xeroderma pigmentosum), prior to the start date, conducted using The Health Improvement Network, a primary care medical records database in the United Kingdom. The data analyzed had been collected prospectively from 2002 through January 2014. The analysis was completed in August 2015.
Exposures of Interest
Patients with at least 1 diagnostic code for psoriasis were classified as having moderate-to-severe disease if they had been prescribed psoralen, methotrexate, cyclosporine, acitretin, adalimumab, etanercept, infliximab, or ustekinumab or phototherapy for psoriasis. Patients were classified as having mild disease if they never received treatment with any of these agents.
Main Outcomes and Measures
Incident cancer diagnosis.
A total of 937 716 control group patients without psoriasis, matched on date and practice visit, and 198 366 patients with psoriasis (186 076 with mild psoriasis and 12 290 with moderate-to-severe disease) were included in the analysis. The adjusted hazard ratios (aHRs) with 95% CIs for any incident cancer excluding nonmelanoma skin cancer (NMSC) were 1.06 (95% CI, 1.02-1.09), 1.06 (95% CI, 1.02-1.09), and 1.08 (95% CI, 0.96-1.22) in the overall, mild, and severe psoriasis group. The aHRs for incident lymphoma were 1.34 (95% CI, 1.18-1.51), 1.31 (95% CI, 1.15-1.49), and 1.89 (95% CI, 1.25-2.86); for NMSC, 1.12 (95% CI, 1.07-1.16), 1.09 (95% CI, 1.05-1.13), and 1.61 (95% CI, 1.42-1.84); and for lung cancer, 1.15 (95% CI, 1.03-1.27), 1.12 (95% CI, 1.01-1.25), and 1.62 (95% CI, 1.16-2.28) in the overall, mild, and severe psoriasis groups, respectively. No significant association was seen with cancer of the breast, colon, prostate, or leukemia.
Conclusions and Relevance
The association between psoriasis and cancer, albeit small, was present in our cohort of patients with psoriasis. This association was primarily driven by NMSC, lymphoma, and lung cancer.
The risk of cancer in patients with psoriasis has been of special concern owing to the chronic inflammatory nature of the disease, use of immune suppressive treatments and UV therapies, and the increased prevalence of comorbid, well-established risk factors for cancer, such as smoking, all of which may increase the risk of carcinogenesis.1 Indeed, studies2-8 have previously demonstrated that patients with more severe psoriasis are at an increased risk of cancer-related mortality, and psoriasis has been associated with an increased risk of cancer, including lymphoma. However, these studies have generally not controlled for important confounders, and have often failed to examine the rates of specific cancers or the impact of disease severity on cancer risk. Furthermore, there have often been conflicting data on the risk of specific cancers including lymphoma.1,3,4,6,9 Additional studies are necessary to further define the risk of cancer in patients with this disease.
We performed a population-based cohort study using a nationwide, electronic medical records database from the United Kingdom to evaluate the association between psoriasis and cancer. In addition, our study also evaluated whether the risk of cancer in patients with psoriasis could be explained by a specific group of cancers and if this risk was higher in patients with more severe disease.
This study is a population-based cohort study analyzing data that had been collected prospectively since 2002 through January 2014 by general practitioners (GPs) in the United Kingdom who were unaware of the hypothesis being tested.
Our source of data was The Health Improvement Network (THIN), an electronic medical records (EMR) database that contains anonymized medical record information from nearly 11.9 million patients across the United Kingdom. THIN has been extensively used for epidemiological research, and prior studies10-13 have validated the diagnostic accuracy of psoriasis and malignant disease codes used in THIN. Participants within THIN are considered to be representative of the general population of the United Kingdom in terms of their age, sex, geography, and medical diagnoses.14 Information recorded includes patient demographics, medical diagnosis, and treatments prescribed, as well as referrals to specialists, hospitalizations, and laboratory results. THIN has the advantage over administrative claims databases (ie, Medicare) because it contains information on important risk factors for cancer, such as smoking, alcohol intake, and body mass index (BMI).15
Our initial cohort within THIN consisted of all patients with a diagnosis of psoriasis and unexposed controls (ie, those with no recorded diagnosis of psoriasis). All patients with psoriasis who were ages 18 to 89 years at the start date and had observation time within THIN, after implementation of the EMR, were included. Patients were excluded if they died or transferred out of the practice prior to the implementation of EMR. Each patient with psoriasis was matched to up to 5 unexposed controls that were also 18 to 89 years at the start date and were from the same practice. Unexposed controls were assigned a “diagnosis” date based on having a recorded visit with the GP within 6 months (ie, ±180 days) of the corresponding patient’s visit date to the GP. This algorithm was designed to minimize bias by ensuring that exposed patients and unexposed controls were followed by similar clinicians during similar time periods. Patients were initially excluded from the entire cohort if they had a personal history of any cancer, organ transplant, or human immunodeficiency virus (HIV) and/or AIDS, or had a diagnosis of hereditary diseases that have an increased susceptibility for developing malignant abnormalities, including albinism and xeroderma pigmentosum.
This study was developed in accordance with the STROBE guidelines and was approved by the institutional review board of the University of Pennsylvania and by the Scientific Review Committee of CSD Medical Research, United Kingdom. This study was a population-based study using data already collected for these purposes. Data were collected by a third party and deidentified before they were made available for research purposes. Because our study analyzed data already collected for these purposes, patient consent was not required.
Patients with psoriasis were identified by the presence of at least one diagnostic Read code for psoriasis in the patient’s medical file. Read codes are a comprehensive numerical system developed in the United Kingdom to record diagnosis, symptoms, and tests, and is similar to the International Classification of Disease (ICD) codes.16 The use of Read codes for modeling disease outcomes in patients with psoriasis has been previously validated in THIN.13 Severity of psoriasis was classified as mild or moderate-to-severe depending on whether the patient was prescribed systemic therapy or phototherapy for psoriasis. Patients were classified as having moderate-to-severe psoriasis if they had ever received any prescription for psoralen, phototherapy, methotrexate, cyclosporine, acitretin, adalimumab, etanercept, infliximab, and/or ustekinumab. Patients were classified as having mild disease if they had never received treatment with any of these agents. This approach has been used in previous studies as a proxy for estimating disease severity.4,17
For exposed patients (ie, those who received at least 1 diagnostic code for psoriasis) the start of follow-up time occurred at the latest of the following: date of first diagnostic code for psoriasis in THIN; 180 days after patient registration with the practice; or when the practice began recording information using Vision software. For unexposed patients (ie, those with no diagnosis of psoriasis) the start time occurred at the latest of 3 dates: 180 days after patient registration, when the practice began recording information using Vision software, or the closest corresponding visit to the exposed patient’s visit date with the GP (±180 days), whichever was latest. Patient follow-up continued for both groups until the earliest of the following: first diagnosis of cancer, death, transfer out of the practice, or end of the study period.
The outcome of interest was incident cancer diagnosis. Cancer cases were identified using a series of Read codes selected a priori by the primary investigator (Z.C.C.F.). Patients were classified as having an incident cancer diagnosis if they received a code for any cancer (except carcinoma in situ or metastatic cancer) after the start date and on or before the end date of the study period. We also selected a priori specific cancer outcomes that were of interest to us, as part of secondary outcome analyses, either because prior studies have shown a possible association between psoriasis and risk of developing these malignant diseases or because they are considered important owing to public health implications.
Covariates assessed in multivariable models as potential confounders included age, sex, BMI, smoking, and alcohol intake. Age and BMI were treated as continuous variables whereas smoking and alcohol intake were treated as categorical variables.
The sample size was fixed as we included all patients with a diagnosis of psoriasis in THIN who met our inclusion criteria. Using data on the incidence rates for all cancers, we would expect to detect a hazard ratio (HR) of 1.06 to 1.98 for patients with mild psoriasis and 2.02 to 7.79 for patients with severe disease for specific cancers assuming a 2-sided .05 significance level with a median follow-up period of 5 years and 80% power.
Descriptive statistics were used to examine age, sex, disease severity, person-time, and covariate distribution among the cohort with exposed psoriasis and the control group. χ2 Test was used to examine dichotomous variables. The nonparametric Wilcoxon-Mann-Whitney rank sum test was used to examine the continuous variables age and BMI. The rates of incident cancer diagnosis were descriptively reported as the cumulative incidence with 95% CIs for all cancers overall and stratified by cancer type in controls and patients with mild and severe psoriasis, separately. An unadjusted Cox proportional hazards model was then used to determine the HRs with 95% CIs for incident cancer diagnosis. The overall models were then adjusted for age and sex and were also adjusted for additional covariates using a purposeful selection modeling approach.18 Covariates with biological plausible relationship to the exposure and outcome and those that resulted in a change in the point estimates of the main effects by more than 10% were kept in the final model. Multivariate models were examined for proportionality using diagnostic log-log plots. Finally, we performed multiple sensitivity analyses to test the underlying assumptions of our primary analysis. Statistical analyses were performed using STATA statistical software (version 13; Stata Corp).
We identified 937 716 controls and 198 366 patients with psoriasis; 186 076 were classified as having mild psoriasis, and 12 290 were classified as having moderate-to-severe disease based on having received treatment with either phototherapy or systemic therapies (Table 1). Overall, patients with psoriasis were younger, had higher BMI, and a greater proportion were male and current smokers compared with controls. In patients with moderate-to-severe disease, the most commonly prescribed systemic therapy was methotrexate (69.48%) (Table 2).
The mean follow-up time across all outcomes varied from 6.39 to 6.57 years in the control group and 6.06 to 6.22 in the overall psoriasis group. The mean follow-up time in the mild and moderate-to-severe psoriasis group varied from 6.11 to 6.28 years, and 5.20 to 5.35 years, respectively. For all cancers, excluding NMSC, the number of new cancer cases identified was 32 241 in the control group and 6289 in the overall psoriasis group, of which 5490 were in the mild psoriasis group and 2186 in the moderate-to-severe psoriasis group. Unadjusted HRs with 95% CI showed that patients with moderate-to-severe psoriasis had a statistically significant higher risk of any lymphoma (HR, 1.69; 95% CI, 1.12-2.55); cutaneous T-cell lymphoma (CTCL) (HR, 8.86; 95% CI, 3.53-22.20); lung cancer (HR, 1.73; 95% CI, 1.26-2.39); and NMSC (HR, 1.40; 95% CI, 1.23-1.60) compared with patients without psoriasis (Table 3).
The adjusted HR with 95% CI for specific cancer types are summarized in the Figure. Fully adjusted models showed a statistically significant increased risk for any cancer excluding NMSC, lymphoma, CTCL, lung cancer, and NMSC in patients with any psoriasis compared with controls, with this risk being higher in patients classified as having moderate-to-severe disease. A similar association was seen for any lymphoma, excluding CTCL, melanoma, and pancreatic cancer, where an increased risk was seen across all 3 groups; however, results were not statistically significant in the moderate-to-severe psoriasis group. In fully adjusted models, no statistically significant increased risk was observed across any of the groups for leukemia, breast, prostate, or colon cancer (Table 3).
Results were robust to a number of sensitivity analyses (Table 4). Because smoking is an especially strong risk factor for lung cancer, we performed additional sensitivity analyses with the fully adjusted model limited to patients who were classified as never smokers. In this model, the association between psoriasis and lung cancer was lost in the overall (HR, 0.98; 95% CI, 0.76-1.25) and mild (HR, 0.97; 95% CI, 0.75-1.24) group and was decreased in the severe psoriasis group (HR, 1.18; 95% CI, 0.44-3.17). In contrast, an age- and sex-adjusted model for lung cancer including only patients classified as current smokers showed a positive association between psoriasis and risk of lung cancer across all study groups.
Prior studies evaluating the risk of cancer in patients with psoriasis have often shown conflicting results. Our results are consistent with those of a meta-analysis1 that showed a small increased risk of cancer overall excluding NMSC (HR, 1.16; 95% CI, 1.07-1.25). As expected, we observed an increased risk for NMSC and lymphoma. In both instances, this risk was highest in patients with psoriasis who received systemic treatment or phototherapy. In patients defined as having moderate-to-severe disease, the HR for incident NMSC was 1.61 (95% CI, 1.42-1.84). The risk for NMSC in patients with psoriasis has been previously described in the literature and has been primarily attributed to phototherapy exposure, particularly psoralen and ultraviolet A (PUVA).1,9,19,20 Chen et al6 postulated that although systemic medications, including PUVA, increased the risk of cancer in patients with psoriasis, they are not the sole cause of malignant neoplasms seen in this population. Indeed, multiple studies have suggested that in the case of lymphoma, including CTCL, prolonged immune stimulation could potentially lead to the development of a dominant clone which in turn promotes lymphomagenesis.4,21,22 Our results showed that patients with psoriasis had an increased risk of any lymphoma across all study groups (overall: HR, 1.34 [95% CI, 1.18-1.51]; mild disease: HR, 1.31 [95% CI, 1.15-1.49]; and moderate-to-severe disease: HR, 1.89 [95% CI, 1.25-2.86]). These results replicate those of Gelfand et al3 who, using a similar study population and case definition, found that patients with mild psoriasis had increased risk for any lymphoma (HR, 1.34; 95% CI, 1.16-1.54) and patients with moderate-to-severe disease had increased risk (HR, 1.59; 95% CI, 0.88-2.89) compared with patients without psoriasis.
The risk for common solid organ malignant neoplasms, breast, colon, and prostate cancer was not higher than the general population. This finding suggests that positive associations seen for other internal malignant neoplasms, lymphoma, and lung cancer are not likely owing to screening bias. In our age-, sex-, and smoking-adjusted model, patients with moderate-to-severe psoriasis had an approximately 60% increased risk of developing lung cancer compared with patients without psoriasis (HR, 1.62; 95% CI, 1.16-2.28). Three prospective cohort studies23-25 showed a positive association between psoriasis and lung cancer; however, models were not adjusted for smoking. In 2 prior studies in which adjustment for smoking was performed, one8 found no positive association between psoriasis and lung cancer (incidence rate ratio [IRR], 0.79; 95% CI, 0.60-1.06), and the other26 resulted in loss of the association when adjusting for this risk factor (HR, 1.3; 95% CI, 0.8-2.0). In additional sensitivity analyses, we showed a loss of this association when only including patients who were classified as never smokers. However, in models including only patients classified as former or current smokers, this association persisted. This observation would suggest that the relationship between smoking and lung cancer in patients with psoriasis is more complex.
We found no clear association between psoriasis and either melanoma or pancreatic cancer. In the case of melanoma, results from prior studies have been mixed. Chen et al6 found an increased risk for melanoma in Taiwanese patients with psoriasis compared with patients without psoriasis (HR, 3.10; 95% CI, 1.24-7.71). A more recent study27 found that patients with psoriasis had a lower probability of developing melanoma when compared with a group of nondermatological patients. With respect to pancreatic cancer, a standardized incidence ratio (SIR) of 1.46 (95% CI, 1.10-1.95) has been previously reported, which is somewhat similar to our results (HR, 1.25; 95% CI, 1.04-1.51).1 Interestingly, Brauchli et al8 also found an increased incidence rate of pancreatic cancer in patients with psoriasis overall (HR, 2.20; 95% CI, 1.18-4.09). Our results showed a positive association across all study groups, yet we found no evidence of a dose-response effect with psoriasis severity. Chronic systemic inflammation, such as that seen in patients with severe psoriasis, has been postulated as a potential risk factor for pancreatic cancer.28 Chronic alcohol consumption, which is reportedly higher in this population and often results in chronic pancreatitis, could be a contributing factor.29,30
Our study has certain strengths. We used a large, population-based database that has been used in similar epidemiological studies and contains information on well-established risk factors for cancer. We provide further evidence that the risk of cancer in patients with psoriasis, albeit small, applies to a specific group of cancers, including NMSC, lymphoma, and lung cancer. Nonetheless, our study has certain limitations. First, because THIN does not capture information on disease severity using physician-assessed measures, we used treatment exposure as a proxy for disease severity. This limited our ability to separate the effect of disease severity from the effect of treatment exposure on cancer risk. Second, the effect of biologic therapies on cancer risk cannot be properly assessed because the use of these medications is more restricted in the UK and therefore may not be readily captured within THIN. While data on the effect of treatment on cancer risk in patients with psoriasis are limited, studies31-33 in patients with other systemic inflammatory diseases, such as rheumatoid arthritis, who are exposed to similar immune suppressive medications, have shown that the risk of malignant neoplasms cannot be entirely explained by exposure to these types of systemic medications. Third, while data on well-established risk factors for cancer (ie, smoking), are available within THIN, greater efforts need to be made to improve the capture of these risk factors in large population-based databases if we are to further clarify the impact of psoriasis on cancer risk. Finally, although the risk of NMSC associated with PUVA treatment has been previously demonstrated, there are limited data to support that exposure to UV-B therapy, including narrowband UV-B, also increases this risk.20,34 In addition, we could not adjust for exposure to environmental UV radiation, a strong risk factor for NMSC,35 because these data are not collected in EMR databases such as THIN. Misclassification could also have resulted in an increased estimate for cancer risk. For example, it is well known that patients with CTCL are sometimes incorrectly classified as having psoriasis, resulting in a biased positive association. Yet, it is unlikely that the misdiagnosis of psoriasis would completely explain this association. In the United Kingdom, patients who require initiation of systemic treatment would have been seen by a dermatologist prior to the start of therapy; therefore, we would expect most of these patients to be correctly classified as having psoriasis and not CTCL.4
Finally, surveillance bias could have affected our results because patients with severe psoriasis might be followed more closely than their counterparts who do not have psoriasis or only have mild disease. We performed additional analyses limiting our entire cohort to patients seen at least once a year by their GP and observed similar results. In addition, if there was surveillance bias, we would have expected to see an increased risk of common cancers, such as breast, prostate, and colon cancer, and this was not observed. Care must be taken when interpreting our findings; in some cases our effects were small and could be related to chance because multiple outcomes were evaluated.
In conclusion, our data suggest that there is a small increased risk of cancer overall in patients with psoriasis. Patients with psoriasis who received treatment with systemic medications or phototherapy were shown to be at a higher risk for malignant neoplasms compared with controls. Future work should be aimed at developing a better understanding of the effect of disease severity and treatment exposure separately on cancer risk in this population. Dermatologists who care for patients with psoriasis should consider incorporating current cancer screening guidelines and counseling, such as smoking cessation, into their daily practice.
Corresponding Author: Zelma C. Chiesa Fuxench, MD, Department of Dermatology, University of Pennsylvania, Perelman School of Medicine, 3400 Spruce St, 1107 Dulles, Philadelphia, PA 19104 (firstname.lastname@example.org).
Accepted for Publication: October 13, 2015.
Published Online: December 16, 2015. doi:10.1001/jamadermatol.2015.4847.
Author Contributions: Dr Chiesa Fuxench had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: Chiesa Fuxench, Gelfand.
Acquisition, analysis, or interpretation of data: All authors.
Drafting of the manuscript: Chiesa Fuxench.
Critical revision of the manuscript for important intellectual content: All authors.
Statistical analysis: All authors.
Obtained funding: Gelfand.
Administrative, technical, or material support: Ogdie Beatty.
Study supervision: Chiesa Fuxench, Gelfand.
Conflict of Interest/Financial Disclosure: Dr Gelfand served as a consultant for AbbVie, Amgen Inc, Coherus, Celgene Corp, Eli Lilly, Merck, Janssen Biologics (formerly Centocor), Regeneron Novartis Corp, Leo Pharmaceuticals, Endo, and Pfizer Inc, receiving honoraria; had grants or has pending grants from AbbVie, Janssen, Regeneron, Novartis Corp, and Pfizer Corp; and received payment for continuing medical education work related to psoriasis. No other disclosures were reported.
Funding/Support: This study was supported by an unrestricted grant from Pfizer Pharmaceuticals to the trustees of the University of Pennsylvania, and grants from the National Psoriasis Foundation Fellowship Program (Dr Chiesa Fuxench), Training Grant 5T32GM075766-09 from the National Institutes of Health (Dr Chiesa Fuxench), and grant K24-AR064310 36 from the National Institute of Arthritis and Musculoskeletal and Skin Diseases (Dr Gelfand).
Role of the Funder/Sponsor: The sponsors had no role in the design and conduct of the study; in the collection, management, analysis, and interpretation of the data; in the preparation, review, or approval of the manuscript; or in the decision to submit the manuscript for publication.
Create a personal account or sign in to: