Inter-reader agreement of interpretation of radiological course of bile duct changes between serial follow-up magnetic resonance imaging/3D magnetic resonance cholangiopancreatography of patients with primary sclerosing cholangitis

Interobserver agreement and prognostic value of image-based scoring systems in patients with primary sclerosing cholangitis

BACKGROUND

Primary sclerosing cholangitis (PSC) is a cholestatic liver disease that progresses to cirrhosis and liver failure. The Anali and Amsterdam scores are based upon imaging features on MRI and ERCP, respectively.

AIMS

We aimed to compare the interobserver variability and performances of these scores.

METHODS

Patients with PSC with at least 1 MRCP were included. Images were independently scored by 2 experts. Agreement and prognostic performance with a primary end point of hepatic decompensation was assessed.

RESULTS

Fifty-nine patients were included (67.8% male, 86.4% IBD). Interobserver agreement for the Anali and Amsterdam scores were moderate (k = 0.49; 95% CI 0.35-0.64 and k = 0.43; 95% CI 0.30-0.56, respectively). Among the Anali components, dysmorphy (caudate/right lobe ratio > 0.9) had fair agreement (k = 0.37; 95% CI 0.14-0.60) and portal hypertension (k = 0.64, 95% CI 0.32-0.89) and intrahepatic dilation (k = 0.70; 95% CI 0.53-0.87) had substantial agreement. The Amsterdam extrahepatic and intrahepatic scores had fair agreement (k = 0.38; 95% CI 0.23-0.52) and moderate agreement (k = 0.50; 95% CI 0.34-0.67), respectively. Anali score (HR 5.90, 95% CI 1.64-21.21), total bilirubin (HR = 3.23; 95% Cl 1.06-9.91), and age (HR = 1.05; 95% CI 1.00-1.11) were independent predictors of hepatic decompensation. Mayo risk score and Anali score had good discriminative ability with c-statistics of 0.78 (CI 0.59-0.96) and 0.76 (CI 0.56-0.91). Anali score remained an independent predictor after adjusting for Mayo risk score.

CONCLUSION

Anali score adds additional predictive value for hepatic decompensation in patients with PSC.

Prognostic role of magnetic resonance imaging of the abdomen with intravenous contrast and magnetic resonance cholangiopancreatography in primary sclerosing cholangitis

Objective

To evaluate the usefulness of Anali scores, determined by magnetic resonance imaging, for predicting the prognosis of primary sclerosing cholangitis (PSC) and to analyze interobserver variability, as well as to assess the impact of periportal edema and heterogeneous signal intensity on diffusion-weighted imaging of the liver.

Materials and Methods

This was a retrospective cohort study of 29 patients with PSC and baseline magnetic resonance imaging. Anali scores, without gadolinium (0-5 points) and with gadolinium (0-2 points), were calculated by two radiologists. Clinical end-points included liver transplantation, cirrhotic decompensation, and death. We calculated intraclass correlation coefficients (ICCs) for interobserver agreement on the Anali scores, performed Kaplan-Meier survival analysis comparing event-free survival among the score strata, and calculated the areas under receiver operating characteristic curves to determine sensitivity and specificity.

Results

Among the patients with a clinical event, the median Anali score was 4 (interquartile range [IQR], 2-5) without gadolinium and 2 (IQR, 1-2) with gadolinium, compared with 1 (IQR, 1.0-2.5) and 1 (IQR, 0.25-1.0), respectively, among those without a clinical event. The ICC was 0.79 (95% confidence interval: 0.57-0.91) for the Anali score with gadolinium and 0.99 (95% confidence interval: 0.98-0.99) for the Anali score without gadolinium. Periportal edema and heterogeneous signal intensity in the liver on diffusion-weighted imaging showed no statistical impact on clinical events (p = 0.65 and p = 0.5, respectively).

Conclusion

Anali scores correlate with clinical events in PSC, with a high level of interobserver agreement.

Adult Primary Sclerosing Cholangitis (PSC) subjects have worse biliary disease at diagnosis compared to pediatric PSC subjects

INTRODUCTION

Adult Primary Sclerosing Cholangitis (PSC) subjects have worse outcomes compared to pediatric PSC subjects. The reasons for this observation are not completely understood.

METHODS

In this single-center, retrospective (2005-17) study we compared clinical information, laboratory data, and previously published MRCP-based scores between 25 pediatric (0-18 years at diagnosis) and 45 adult (19 years and above) subjects with large duct PSC at the time of diagnosis. For each subject, radiologists determined MRCP-based parameters and scores after reviewing the MRCP images.

RESULTS

The median age at diagnosis for pediatric subjects was 14 years, while that of adult subjects was 39 years. At the time of diagnosis, adult subjects had a higher incidence of biliary complications like cholangitis and high-grade biliary stricture (27% vs. 6%, p = 0.003) and higher serum bilirubin (0.8 vs. 0.4 mg/dl, p = 0.01). MRCP analysis showed that adult subjects had a higher incidence of hilar lymph node enlargement (24.4% vs. 4%, p = 0.03) at diagnosis. Adult subjects had worse sum-IHD score (p = 0.003) and average-IHD score (p = 0.03). Age at diagnosis correlated with higher average-IHD (p = 0.002) and sum-IHD (p = 0.002) scores. Adult subjects had worse Anali score without contrast (p = 0.01) at diagnosis. MRCP-based extrahepatic duct parameters and scores were similar between groups.

DISCUSSION

Adult PSC subjects may have higher severity of disease at diagnosis compared to pediatric subjects. Future prospective cohort studies are required to confirm this hypothesis.

DeePSC: A Deep Learning Model for Automated Diagnosis of Primary Sclerosing Cholangitis at Two-dimensional MR Cholangiopancreatography

Purpose

To develop, train, and validate a multiview deep convolutional neural network (DeePSC) for the automated diagnosis of primary sclerosing cholangitis (PSC) on two-dimensional MR cholangiopancreatography (MRCP) images.

Materials and Methods

This retrospective study included two-dimensional MRCP datasets of 342 patients (45 years ± 14 [SD]; 207 male patients) with confirmed diagnosis of PSC and 264 controls (51 years ± 16; 150 male patients). MRCP images were separated into 3-T (n = 361) and 1.5-T (n = 398) datasets, of which 39 samples each were randomly chosen as unseen test sets. Additionally, 37 MRCP images obtained with a 3-T MRI scanner from a different manufacturer were included for external testing. A multiview convolutional neural network was developed, specialized in simultaneously processing the seven images taken at different rotational angles per MRCP examination. The final model, DeePSC, derived its classification per patient from the instance expressing the highest confidence in an ensemble of 20 individually trained multiview convolutional neural networks. Predictive performance on both test sets was compared with that of four licensed radiologists using the Welch t test.

Results

DeePSC achieved an accuracy of 80.5% ± 1.3 (sensitivity, 80.0% ± 1.9; specificity, 81.1% ± 2.7) on the 3-T and 82.6% ± 3.0 (sensitivity, 83.6% ± 1.8; specificity, 80.0% ± 8.9) on the 1.5-T test set and scored even higher on the external test set (accuracy, 92.4% ± 1.1; sensitivity, 100.0% ± 0.0; specificity, 83.5% ± 2.4). DeePSC outperformed radiologists in average prediction accuracy by 5.5 (P = .34, 3 T) and 10.1 (P = .13, 1.5 T) percentage points.

Conclusion

Automated classification of PSC-compatible findings based on two-dimensional MRCP was achievable and demonstrated high accuracy on internal and external test sets.Keywords: Neural Networks, Deep Learning, Liver Disease, MRI, Primary Sclerosing Cholangitis, MR Cholangiopancreatography Supplemental material is available for this article. © RSNA, 2023.

Comparative Performance of Quantitative and Qualitative Magnetic Resonance Imaging Metrics in Primary Sclerosing Cholangitis

Background and Aims

Several quantitative and qualitative magnetic resonance imaging (MRI) metrics have been reported to predict outcomes among those with primary sclerosing cholangitis (PSC). We aimed to compare the reproducibility and prognostic performances of MRI biomarkers and examine if combining these measurements adds value.

Methods

We performed a retrospective review of 388 patients with PSC who underwent a magnetic resonance elastography and magnetic resonance cholangiopancreatography. Liver stiffness (LS) was determined by validated automated software, whereas spleen volume was calculated by semiautomated software, and radiologists manually determined the ANALI scores. The primary endpoint was hepatic decompensation.

Results

LS and spleen volume values had perfect and near-perfect agreement (intraclass correlation coefficient of 1.00 and 0.9996, respectively), whereas ANALI with and without gadolinium had a moderate inter-rater agreement between 3 radiologists (kappa = 0.42-0.54 and 0.46-0.57, respectively). As a continuous variable, LS alone was the best predictor of hepatic decompensation (concordance score = 0.90; 95% confidence interval, 0.87-0.93). A quantitative-only MRI model [LS (>4.70 kPa = 2 or ≤4.70 kPa = 0) + spleen volume (>600 mm3 = 1 or ≤600 mm3 = 0)] had the optimal reproducibility and performance (concordance score = 0.85; 95% confidence interval = 0.80-0.89) and enabled patient risk stratification by estimating the 5-year incidence of hepatic decompensation: 7.49%, 44.50%, 70.00%, and 91.30% (score 0-3).

Conclusion

Quantitative MRI markers of fibrosis and portal hypertension generated by automated and semiautomated software are highly reproducible. LS is the single best imaging predictor of hepatic decompensation. However, a quantitative MRI score using LS and spleen volume is well suited to risk stratify those with PSC.

A Quantitative Magnetic Resonance Cholangiopancreatography Metric of Intrahepatic Biliary Dilatation Severity Detects High-Risk Primary Sclerosing Cholangitis

Magnetic resonance imaging with magnetic resonance cholangiopancreatography (MRI‐MRCP) in primary sclerosing cholangitis (PSC) is currently based on qualitative assessment and has high interobserver variability. We investigated the utility and performance of quantitative metrics derived from a three‐dimensional biliary analysis tool in adult patients with PSC. MRI‐MRCP, blood‐based biomarkers, and FibroScan were prospectively performed in 80 participants with large‐duct PSC and 20 healthy participants. Quantitative analysis was performed using MRCP+ (Perspectum Ltd., United Kingdom), and qualitative reads were performed by radiologists. Inter‐reader agreements were compared. Patients were classified into high risk or low risk for disease progression, using Mayo risk score (MRS), Amsterdam‐Oxford model (AOM), upper limit of normal (ULN) alkaline phosphatase (ALP), disease distribution, and presence of dominant stricture. Performance of noninvasive tools was assessed using binomial logistic regressions and receiver operating characteristic curve analyses. Quantitative biliary metrics performed well to distinguish abnormal from normal bile ducts (P < 0.0001). Interobserver agreements for MRCP+ dilatation metrics (intraclass correlation coefficient, 0.90‐0.96) were superior to modified Amsterdam intrahepatic stricture severity score (κ = 0.74) and Anali score (κ = 0.38). MRCP+ intrahepatic dilatation severity showed excellent performance to classify patients into high‐risk and low‐risk groups, using predictors of disease severity as the reference (MRS, P < 0.0001; AOM, P = 0.0017; 2.2 × ULN ALP, P = 0.0007; 1.5 × ULN ALP, P = 0.0225; extrahepatic disease, P = 0.0331; dominant stricture, P = 0.0019). MRCP+ intrahepatic dilatation severity was an independent predictor of MRS >0 (odds ratio, 31.3; P = 0.035) in the multivariate analysis. Conclusion: Intrahepatic biliary dilatation severity calculated using MRCP+ is elevated in patients with high‐risk PSC and may be used as an adjunct for risk stratification in PSC. This exploratory study has provided the groundwork for examining the utility of novel quantitative biliary metrics in multicenter studies.

Assessment of prognostic value and interreader agreement of ANALI scores in patients with primary sclerosing cholangitis

PURPOSE

ANALI-scores are two prognostic magnetic resonance imaging (MRI)-based scores developed for patients with primary sclerosing cholangitis (PSC). Our study aims to assess the interreader agreement between expert radiologists of the two ANALI-scores and of the radiological parameters they utilize, and to test the prognostic performance of the scores in our population.

METHOD

Three radiologists evaluated MRIs of 98 PSC-patients from a prospectively collected cohort with median follow-up of 6.7 years. Each parameter of ANALI-scores was assessed, and the scores were calculated. Interreader agreement was assessed with intraclass correlation coefficient (ICC). After consensus reading was reached, the prognostic value of ANALI-scores was assessed with Cox regression, and outcome-free survival rates were evaluated with Kaplan-Meier estimates.

RESULTS

The ANALI-score without gadolinium had poor to moderate (ICC = 0.56, 95 %CI: 0.42-0.68) and with gadolinium poor (ICC = 0.30, 95 %CI: 0.16-0.44) agreement. Liver deformity (ICC = 0.28, 95 %CI: 0.13-0.44) and parenchymal enhancement heterogeneity (ICC = 0.24, 95 %CI: 0.12-0.38) had poor agreement. Portal hypertension had poor to moderate (ICC = 0.48, 95 %CI: 0.36-0.59) and dilatation of the intrahepatic ducts had moderate (ICC = 0.64, 95 %CI: 0.54-0.73) agreement. Hazard ratios for liver-related death, transplantation or cirrhosis decompensation of the ANALI-scores with and without gadolinium were 3.53 (95 %CI: 1.40-8.93) and 2.25 (95 %CI: 1.56-3.24), respectively. Outcome-free survival was better for patients with low ANALI-scores.

CONCLUSIONS

The ANALI-scores show poor to moderate agreement, which challenges their usefulness in clinical practice. They are associated with clinical outcomes, confirming the value of imaging in prognosis of PSC, but need further multicenter evaluation.

Evaluation of quantitative MRCP (MRCP+) for risk stratification of primary sclerosing cholangitis: comparison with morphological MRCP, MR elastography, and biochemical risk scores

OBJECTIVES

To study the association of MRCP+ parameters with biochemical scoring systems and MR elastography (MRE) in primary sclerosing cholangitis (PSC). To evaluate the incremental value of combining MRCP+ with morphological scores in associating with biochemical scores.

METHODS AND MATERIALS

MRI images, liver stiffness measurements by MRE, and biochemical testing of 65 patients with PSC that were retrospectively enrolled between January 2014 and December 2015 were obtained. MRCP+ was used to post-process MRCP images to obtain quantitative measurements of the bile ducts and biliary tree. Linear regression analysis was used to test the associations. Bootstrapping was used as a validation method.

RESULTS

The total number of segmental strictures had the strongest association with Mayo Risk Score (R² = 0.14), minimum stricture diameter had the highest association with Amsterdam Oxford Prognostic Index (R² = 0.12), and the percentage of duct nodes with width 0-3 mm had the strongest association with PSC Risk Estimate Tool (R² = 0.09). The presence of Ducts with medians > 9 mm had the highest association with MRE (R²= 0.21). The strength of association of MRCP+ to Mayo Risk Score was similar to ANALI2 and weaker than MRE (R² = 0.23, 0.24, 0.38 respectively). MRCP+ enhanced the association of ANALI 2 and MRE with the Mayo Risk Score.

CONCLUSIONS

MRCP+ demonstrated a significant association with biochemical scores and MRE. The association of MRCP+ with the biochemical scores was generally comparable to ANALI scores. MRCP+ enhanced the association of ANALI2 and MRE with the Mayo Risk Score.

KEY POINTS

• MRCP+ has the potential to act as a risk stratfier in PSC. • MRE outperformed MRCP+ for risk stratifcation. • Combination of MRCP+ with MRE and ANALI scores improved overall performace as risk stratifiers.

Prevalence and long-term outcome of sub-clinical primary sclerosing cholangitis in patients with ulcerative colitis

BACKGROUND

Primary sclerosing cholangitis (PSC) is closely associated with inflammatory bowel disease, particularly ulcerative colitis (UC), with an increased risk of biliary and colorectal malignancy. We sought to clarify the prevalence, characteristics and long-term outcome of sub-clinical PSC diagnosed by magnetic resonance cholangiogram (MRC) in patients with UC and normal liver biochemistry, with or without colorectal dysplasia (CRD).

METHODS

In this prospective case-control study, 70 patients with UC and normal liver function (51 extensive UC, 19 CRD), 28 healthy volunteers (negative controls) and 28 patients with PSC and cholestasis (positive controls) underwent MRC and blood evaluation. MRC scans were interpreted blindly by two radiologists who graded individually, the scans as definitive for PSC, possible for PSC or normal. Clinical outcome was assessed by blood monitoring, abdominal imaging and endoscopic surveillance.

RESULTS

7/51 (14%) with extensive UC and 4/19 (21%) with CRD had biliary abnormalities on MRC consistent with PSC. 7/11 (64%) with sub-clinical PSC had isolated intrahepatic duct involvement. Sub-clinical PSC was associated with advanced age (P = .04), non-smoking (P = .03), pANCA (P = .04), quiescent colitis (P = .02), absence of azathioprine (P = .04) and high-grade CRD (P = .03). Inter-observer (kappa = 0.88) and intra-observer (kappa = 0.96) agreement for MRC interpretation was high. No negative controls were assessed as definite PSC, 4/28 were considered on blinding as possible PSC. During follow-up of sub-clinical PSC (median 10.1(3.1-11.9) years), four patients developed abnormal liver biochemistry, two had radiological progression of PSC and seven developed malignancy, including two biliary and one colorectal carcinoma.

CONCLUSIONS

Prevalence of sub-clinical PSC appears high in patients with extensive UC and normal liver biochemistry, with or without CRD. Disease progression and malignancy were identified on long-term follow-up. MRC should be considered for all patients with extensive UC or CRD to stratify surveillance.