Predicting risk in primary biliary cholangitis

Potential conflict of interest: Nothing to report. See Article On Page 930. Primary biliary cholangitis (PBC) is a chronic immune‐mediated liver disease that ultimately results into destruction of the small bile ducts, which leads to cholestasis, progressive fibrosis, and, in some individuals, end‐stage liver disease (ESLD). Incidence of PBC varies significantly from 0.33 to 5.8 per hundred thousand individuals per year with a prevalence rate ranging from 1.91 to 40.2 per hundred thousand people.1 Despite its relative rarity, it contributes significantly to mortality and morbidity from chronic liver disease and is a major indication for liver transplantation (LT). The ability to accurately diagnose PBC by use of antimitochondrial antibodies directed to the pyruvate dehydrogenase complex (PDC‐E2) epitopes, which are positive in over 95% of individuals, has largely replaced the need for a liver biopsy to make the diagnosis. The majority of historical PBC cases are presented at a stage of advanced liver disease and cirrhosis and are accurately captured by short‐term indices to predict mortality as well as selection for LT, such as MELD, Na‐MELD, and United Kingdom Model for End‐Stage Liver Disease. However, this is no longer the case with the vast majority of individuals being diagnosed at very early stages of disease, with possibly the majority never progressing to ESLD. A great challenge for clinicians is to be able to perform a risk assessment at diagnosis to determine which individuals are at high or low risk for progression so as to prioritize who should receive therapy and be appropriately counseled. The question of adequate risk stratification extends further to clinical trial design as the select high‐risk individuals that may benefit most from investigated medicinal products, and to provide dynamic surrogate endpoints that predict progression and/or complications or death from liver‐related causes.2 Liver fibrosis, as measured by liver biopsy or by noninvasive means such as ultrasound‐based transient elastography or magnetic resonance imaging‐based elastography, accurately captures late‐stage liver fibrosis and correlates with complications from ESLD or liver‐related death. Similarly, the use of hepatic portal‐venous gradient is an accurate predictor of complications in ESLD, such as development of varices.3 Unfortunately, the progression between different stages of fibrosis in PBC is highly variable and thus requires more dynamic models to predict the medium‐ and long‐term prognosis of particularly individuals with early liver disease. Multiple clinical and biochemical parameters have been used to predict progression in PBC, with particularly mild patients and individuals younger than 50 years of age being associated with worse prognosis.4 Similarly, a raised bilirubin and jaundice at diagnosis as well as pruritus and fatigue are associated with worse outcomes, but these observations are most often only helpful in advanced stages of disease. Possibly, the best surrogate marker available is the use of serum alkaline phosphatase (ALP), which, in a multivariate analysis of 4,845 individuals, showed a near relationship with the risk of subsequent LT or death across multiple time points.2 The prognostic value of ALP lies in its prognostic ability in the early stages of PBC and was independent of presentation, age, sex, disease stage, or treatment. The prognostic ability of ALP has been extended to the use of ALP changes (delta ALP) in response to therapy in individuals treated with ursodeoxycholic acid (UDCA) at a dose of 10‐15 mg/kg/day. UDCA is the only licensed therapy for PBC and has a demonstrated efficacy in reducing liver‐related death or transplantation. Treatment with UCDA has created the paradigm of individuals that will normalize or reduce their ALP on treatment (termed responders) and those with an ALP that remains high despite therapy (nonresponders).5 This has led to multiple treatment‐response‐based models, such as the Paris‐1, Toronto, and Paris‐2 criteria, to predict long‐term outcomes in patients on UDCA treatment.2 These not only predict progression to ESLD or death, but also an increased risk of hepatocellular carcinoma in nonresponders. The difficulty with most of these response algorithms is that response to UDCA therapy is evaluated at 1 year after initiation, which potentially disadvantages nonresponders that may have been suitable for other therapeutic options, such as second‐line agents or clinical trials during that period. A possible alternative to UDCA‐based treatment response algorithms as prognostic indices is the use of the AST/platelet ratio index (APRI), which has gained popularity as a mechanism to assist in determining prognosis.7 It has been used to predict prognosis at baseline and at 1 year as an independent predictor of transplant‐free survival in four cohorts.7 Of note, the 1‐year APRI score highlighted a population of PBC sufferers that experienced disease progression and earlier mortality despite responses to UDCA, indicating a possible additive effect of the APRI score to the treatment‐based algorithms. These observations formed the basis of the current study by Carbone et al., in which elements of the APRI score as well as biochemical parameters in response to UDCA treatment were combined into complex computation algorithms to generate a more accurate prediction model for PBC to predict transplant‐free survival. The study is based on the UK PBC research cohort, which aims to capture every PBC sufferer within the United Kingdom.8 A total of 1,916 UDCA‐treated participants were selected as a derivation cohort from which they performed Cox proportional hazards regression analysis of diverse variables and derived a multivariable fractional polynominal model to predict LT or liver‐related mortality within 5, 10, and 15 years. This model was subsequently validated in a further independent cohort of 1,249 UDCA‐treated participants. Individuals who never received UDCA or had other forms of coexisting liver disease, such as autoimmune hepatitis, were excluded from the analysis. Because their data were derived from multiple centers across the UK with different analysis platforms for clinical liver biochemistry, they normalized individual values in relation to the upper limit of normal values to create normalized ratios. They assessed these ratios at baseline and after 12 months of UDCA therapy. The five parameters demonstrated statistical significance in multivariate analysis, which were baseline ratios of albumin and platelet count, and the ratios at 12 months of therapy of bilirubin, transaminase levels, and ALP. This was then used to generate a complex mathematical model that was validated in a total of 1,109 participants in the validation cohort that had recorded values for responses to treatment at 12 months. During the follow‐up period, 9.1% of individuals suffered liver‐related events. The new UK PBC risk score proved superior to the existing predictive models in that the area under the curve was 0.96 (95% confidence interval: 0.93‐0.99) for the 5‐year risk score, 0.95 (0.93‐0.98) for the 10‐year risk score, and 0.94 (0.91‐0.97) for the 15‐year risk score. This outperformed the existing scoring systems from Barcelona, Paris‐1, Toronto, and Paris‐2. Overall, the current study by Carbone et al. incorporates elements of the APRI score with biochemical markers of treatment response to UDCA to generate the most up‐to‐date and accurate score to facilitate risk stratification in PBC. The utility of the UK‐PBC risk score is exemplified by three hypothetical cases in the article and are adapted in Table 1. The UK‐PBC risk score currently provides the most complete tool in assigning risk of progression in the clinical counseling of patients and has potential utility as a surrogate marker for clinical trials. A limitation of the UK‐PBC risk score is that parts of the score require at least 12 months of UDCA therapy and hence does not provide thresholds at baseline to prioritize patients for therapy.Figure 1: Adapted from Carbone et al. Model cases of how the risk score might impact patient care. UK‐PBC Risk Scores = 1‐baseline survival function∧exp(.0287854 *(ALP12×ULN‐1.722136304) −.0422873* (((ALTAST12×ULN/10)∧‐1) −8.675729006) +1.4199* (ln(Bil12×ULN/10) +2.709607778) −1.960303*(ALB×lln‐1.17673001)−.4161954* (PLT×lln‐1.873564875))With Baseline survivor function = 0.982 (5 years); 0.941 (10 years); 0.893 (15 years).One of the current difficulties in managing individuals with PBC is to select those who are at risk of progression for therapy or possibly avoiding therapy in those in whom progression is unlikely. As it stands, the current recommendations are for all individuals with newly diagnosed PBC to receive UDCA therapy, and to that extent, the scoring system is extremely timely and will be valuable in the risk assessment of these individuals. It will also provide a new, interesting surrogate marker to be used in clinical trial design of new investigated products.

Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.