Liver fibrosis in severe psoriasis
: prevalence, risk factors, and investigations

Student thesis: Doctoral ThesisDoctor of Philosophy


Prevalence of liver disease is rising in the UK. Public health strategies suggested to reduce the burden of liver-related morbidity and mortality include management of obesity and type 2 diabetes in addition to screening for liver fibrosis in at-risk groups. Increased prevalence of liver-related mortality has been reported in people with severe psoriasis since the late 1980s. Historically, this increased risk was attributed to methotrexate use. In the last decade, studies have shown that people with severe psoriasis are also at increased risk for obesity, metabolic syndrome, non-alcoholic fatty liver disease and alcohol excess which are all risks for the development of liver fibrosis in the general population. Despite this, there have been no studies to investigate the natural history of liver fibrosis in people with severe psoriasis. Therefore, who within the severe psoriasis population is most at risk for liver fibrosis is not understood. Liver fibrosis is usually asymptomatic until cirrhosis develops. Liver biopsy is the gold standard test for diagnosis of fibrosis; however, the biopsy procedure is associated with a significant morbidity and mortality, which has limited its use as a research tool. The advent of transient elastography (TE) has enabled liver fibrosis estimation without any associated risks to the patient. Consequently, the risks associated with fibrosis have been studied in other populations such as non-alcoholic fatty liver disease (NAFLD) and viral hepatitis, although not yet in people with severe psoriasis. European dermatology guidelines recommend serial procollagen-3 levels (P3NP) to screen for liver fibrosis in people with psoriasis taking methotrexate, based on studies performed in the 1990s. This is in contrast to guidelines for NAFLD which recommend screening for liver fibrosis using TE and/or panels of serological tests developed as surrogate markers for fibrosis. These composite tests have not been prospectively evaluated in a severe psoriasis population. Earlier detection of liver fibrosis would provide an opportunity to reduce liver-related morbidity and mortality in people with severe psoriasis. Therefore, the work in this thesis aimed to address three research questions. Firstly, to understand how common liver fibrosis is in people with severe psoriasis. Secondly to determine who within the severe psoriasis cohort is at increased risk of liver fibrosis by identification of risk factors associated with liver fibrosis in this group. Thirdly to evaluate the diagnostic accuracy of TE and serological tests to detect liver fibrosis in people with severe psoriasis. In view of studies demonstrating that people with severe psoriasis have multiple risks for the development of liver fibrosis in addition to methotrexate exposure, the initial focus was to evaluate the published literature by systematic review to determine whether methotrexate increases the risk of liver fibrosis in severe psoriasis or whether this is just an association. The primary outcome was histological evidence of advanced liver fibrosis. Studies were only included if patients had at least 2 biopsies (1 pre and 1 post- methotrexate). Eight studies (n=429) were included in the review. The risk difference for advanced liver fibrosis after methotrexate was 0.09 (95% CI 0.03 to 0.20). For any grade of liver fibrosis, the risk difference after methotrexate was 0.22 (95% CI.04-0.41). Although this review demonstrated that people with psoriasis taking methotrexate are at risk for development or progression of liver fibrosis over time, the modest increased risk suggested that the risk associated with methotrexate in this population may not apply to all participants. Secondly, we did not identify a dose-dependent relationship with methotrexate and liver fibrosis which signified that other factors were probably important in fibrosis pathogenesis. The included studies were of generally poor quality and there was a lack of consistency in reporting which meant that results couldn’t be pooled. Notably only three of seven studies evaluated obesity as a risk factor for liver fibrosis. Four studies evaluated diabetes as a potential confounder; two studies demonstrated that diabetes influenced progression of liver fibrosis. As the risk for fibrosis only showed a modest increase after starting methotrexate, this systematic review highlighted the need to evaluate all people with severe psoriasis in the subsequent observational study and not just people taking methotrexate. In view of our current use of P3NP to assess for liver fibrosis in people with psoriasis which is based on a small number of old studies, the second systematic review evaluated published observational studies to assess the accuracy of non-invasive tests to detect fibrosis compared to liver biopsy in people with psoriasis either being considered for or taking methotrexate. Studies were included if there was enough data to calculate sensitivity and specificity. Seventeen studies (n= 843 participants) were included. We demonstrated that standard liver function tests had low diagnostic accuracy for the detection of any fibrosis at standard test thresholds (38% pooled sensitivity) compared to biopsy. Likelihood ratios for P3NP were also sub-optimal for it to be considered a good test (pooled positive LR 5.32). Only two small studies (n=34) evaluated TE in people with psoriasis. We were unable to evaluate liver fibrosis prevalence in the systematic reviews due to significant selection bias in the included studies. Therefore, to address all three research aims, the second phase of research was to perform an observational study to evaluate prevalence of liver fibrosis, clinical factors associated with liver fibrosis and evaluate the diagnostic accuracy of non-invasive tests in people with severe psoriasis. Patients were recruited to this single centre, observational cohort study (The Co-morbidities in Severe Psoriasis (CORE) study ) performed at St John’s Institute of Dermatology within Guys and St Thomas’ Hospital, London from 2012 to 2015 . TE, fasting bloods and study questionnaire were collected during a single visit. Lack of consistency in reporting in systematic reviews meant that there were no clear ‘best’ parameters to measure obesity and diabetes and this influenced by how we captured data. In the Comorbidities in Severe Psoriasis (CORE) study we recorded obesity by weight in kg, BMI and waist circumference, fasting glucose and insulin, calculated Homeostatic Model Assessment (a method for assessing insulin resistance) and recorded a diagnosis of diabetes according to medical records, medication history or fasting glucose result. Of the 400 patients recruited (108 women and 289 men) 135 patients (37%) were overweight, 135 (37%) were obese 247 (79%) had a raised waist circumference. 184 (53%) were insulin resistant, 79 (22%) had type 2 diabetes, and 163 (50%) had NAFLD. Low levels of alcohol consumption were reported. 251 (67%) were taking a biologic drug. 85 patients (21%) were taking mtx, and according to patients’ history, 340 (85%) had been exposed to methotrexate at some stage in the past. Median duration of methotrexate exposure was 0.6 years (interquartile range, 0.0-2.4 years). 333 participants an evaluable TE and were included in analysis. 47 of 333 participants (14.1%; 95% CI, 10.4%-17.9%) had advanced fibrosis by TE. Central obesity (measured by waist circumference) and insulin resistance (by HOMA-IR) were associated with the outcomes of both any and advanced fibrosis on multivariate modelling. The best multivariate model for advanced fibrosis (R2 54%) was a combination of increased central obesity, insulin resistance, psoriasis disease severity (PASI), AST, platelet count and reduced reported alcohol use. Central obesity, insulin resistance and psoriasis disease severity are all potentially reversible factors in this population and not currently screened for in routine care. Demographic data from our participants was compared with a larger severe psoriasis cohort (n=3,722) recruited for the Biomarkers of Systemic Treatment Outcomes in Psoriasis (BSTOP) study. The CORE population had equivalent waist circumference, body mass index, alcohol intake, and age as the wider UK and Ireland population with severe psoriasis, suggesting our results are likely to be generalisable to other secondary care centres in the UK. Evaluation of diagnostic tests revealed that all tests showed greater diagnostic accuracy for the detection of advanced fibrosis compared to any fibrosis. Receiver operating characteristic curve analysis determined that most tests evaluated had a similar diagnostic accuracy for advanced fibrosis compared to TE; AST, ELF, Fibrotest, P3NP, NAFLD score all had AUCs 0.69-0.75. Waist and HOMA-IR were the most accurate tests evaluated. Best thresholds to detect advanced fibrosis were female waist measurement of 109 cm or above (AUC 0.83), male waist measurement of 111 cm or above (AUC, 0.86), HOMA-IR (threshold, 3.65; AUC, 0.81). Waist and HOMA-IR were superior to first- line tests recommended in current guidance (LFTs in USA, P3NP in Europe) and more recently developed composite tests such as ELF and Fibrotest. AST was the most accurate of all LFTs compared to TE. Strengths of the CORE study include accurate phenotyping of psoriasis, use of TE to accurately assess for liver fibrosis and low levels of missing data (92% were assessed by TE). The main limitation is single-centre cross-sectional analysis. In the absence of longitudinal data, we could not describe incident advanced fibrosis and estimate degree of risk for individual variables. The CORE study has demonstrated a significant burden of liver fibrosis in the severe psoriasis population. Our data suggest that people with central obesity, insulin resistance, and active psoriasis are the most at risk for liver fibrosis. These parameters are not currently evaluated as part of routine care. In our study, most individuals with advanced fibrosis were not taking methotrexate and would not have been identified using current guidelines. Whilst we cannot exclude the fact that methotrexate may contribute towards liver fibrosis in people with psoriasis, we can be confident that it is only one risk factor of several and people with severe psoriasis taking other disease-modifying drugs should not be excluded from fibrosis screening. We identified that waist measurement may be a useful screening tool for liver fibrosis. People who have severe psoriasis should be counselled to make appropriate lifestyle interventions if they have central obesity and/or insulin resistance to prevent onset of diabetes and reduce likelihood of developing liver fibrosis. Systematic screening for liver fibrosis should be considered in people with severe psoriasis who have central obesity and insulin resistance, irrespective of therapeutic modality. Future work should involve investigation of the CORE study findings in other cohorts. Comparison of liver fibrosis in a severe psoriasis cohort to a matched NAFLD cohort may enable an estimation of the disease burden associated with psoriasiform inflammation in the severe psoriasis population. Importantly, longitudinal analysis of the associated clinical factors identified in the CORE study and assessment of all prescribed medications (including biologics and anti-diabetic medication), may delineate causality and factors influencing progression to advanced liver fibrosis in severe psoriasis.
Date of Award1 May 2020
Original languageEnglish
Awarding Institution
  • King's College London
SupervisorJonathan Barker (Supervisor) & Catherine Smith (Supervisor)

Cite this