Noninvasive prediction of insufficient biochemical response after ursodeoxycholic acid treatment in patients with primary biliary cholangitis based on pretreatment nonenhanced MRI

Objectives To explore the feasibility of pretreatment nonenhanced magnetic resonance imaging (MRI) in predicting insufficient biochemical response to ursodeoxycholic acid (UDCA) in patients with primary biliary cholangitis (PBC). Methods From January 2009 to April 2022, consecutive PBC patients who were treated with UDCA and underwent nonenhanced MRI within 30 days before treatment were retrospectively enrolled. All MR images were independently evaluated by two blinded radiologists. Uni- and multivariable logistic regression analyses were performed to develop a predictive model for 12-month insufficient biochemical response. Model performances were evaluated by computing the area under the receiver operating characteristic curve (AUC), sensitivity, and specificity. Results A total of 74 patients (50.6 ± 11.9 years; 62 females) were included. Three pretreatment MRI features, including hepatomegaly (odds ratio [OR]: 4.580; p = 0.011), periportal hyperintensity on T2-weighted imaging (T2WI) (OR: 4.795, p = 0.008), and narrowing of the bile ducts (OR: 3.491; p = 0.027) were associated with 12-month insufficient biochemical response in the multivariable analysis. A predictive model based on the above indicators had an AUC of 0.781, sensitivity of 85.4%, and specificity of 61.5% for predicting insufficient biochemical response. Conclusions A noninvasive model based on three pretreatment MRI features could accurately predict 12-month insufficient biochemical response to UDCA in patients with PBC. Early identification of PBC patients at increased risk for insufficient response can facilitate the timely initiation of additional treatment. Clinical relevance statement A noninvasive predictive model constructed by incorporating three pretreatment MRI features may help identify patients with primary biliary cholangitis at high risk of insufficient biochemical response to ursodeoxycholic acid and facilitate the timely initiation of additional treatment. Key Points • Noninvasive imaging features based on nonenhanced pretreatment MRI may predict an insufficient biochemical response to UDCA in PBC patients. • A combined model based on three MRI features (hepatomegaly, periportal hyperintensity on T2-weighted imaging, and narrowing of the bile ducts) further improved the predictive efficacy for an insufficient biochemical response to UDCA in PBC patients, with high sensitivity and specificity. • The nomogram of the combined model showed good calibration and predictive efficacy for an insufficient biochemical response to UDCA in PBC patients. In particular, the calibration curve visualised the clinical applicability of the prediction model. Supplementary Information The online version contains supplementary material available at 10.1007/s00330-023-10080-w.


Introduction
Primary biliary cholangitis (PBC) is an autoimmune-mediated intrahepatic cholestatic disease characterised by progressive, nonsuppurative, destructive intrahepatic cholangitis, leading to hepatic fibrosis and eventually cirrhosis and a range of complications [1].According to the latest statistics from the 2022 Asia-Pacific Society for the Study of the Liver (APASL) guidelines [2], the global incidence and prevalence of PBC (17.6 per million people/year and 146 per million people/year, respectively) have exceeded previous estimates and are increasing at an incalculable rate, particularly notable in the Asia-Pacific region (8.4 and 98.2~118.8per million, respectively) [3][4][5][6].PBC has become a serious chronic liver disease threatening human health.
Ursodeoxycholic acid (UDCA) is the only first-line drug recommended by international guidelines and most countries for PBC treatment [7][8][9].However, nearly 30~40% of PBC patients fail to achieve a sufficient therapeutic response and are at high risk of progressing to liver failure or liver transplantation [8,10].Previous studies have shown that PBC patients with an insufficient biochemical response have a significantly lower 10-year transplant-free survival rate than those with a sufficient biochemical response (51% vs. 90%) and that insufficient biochemical response is an

Inclusion and exclusion criteria
From January 2009 to April 2022, consecutive patients who fulfilled the following inclusion criteria were included: (1) established diagnosis of PBC, (2) 13-15 mg/kg/d UDCA prescription, and (3) abdominal MRI examination (a 2D or 3D MRCP sequence was included in the scanning protocol) within 30 days before UDCA treatment.
The exclusion criteria were as follows: (1) comorbid with tumours or other causes of chronic liver disease such as viral hepatitis and alcoholic cirrhosis, or immune system diseases such as systemic lupus erythematosus; (2) combined with decompensated cirrhosis; (3) incomplete baseline clinical or laboratory indicators; (4) absence of follow-up information or follow-up duration of less than 12 months; (5) death during the follow-up period; and (6) poor image quality or incomplete abdominal MRI scans that could not be evaluated.

Diagnostic criteria for PBC
The diagnosis of PBC was based on the criteria proposed in the 2018 Practice Guidance from the American Association for the Study of Liver Diseases (AASLD) [7].A definitive diagnosis was established when two or more of the following three criteria were met: (1) cholestasis as indicated by elevated alkaline phosphatase (ALP); (2) the presence of anti-mitochondrial antibodies (AMAs) or PBC-specific autoantibodies such as sp100 or gp210, if the AMA test was negative; and (3) the presence of histological evidence supporting nonsuppurative destructive cholangitis and interlobular bile duct destruction.

Clinical and laboratory data collection
All laboratory parameters were examined in the Department of Laboratory Medicine of West China Hospital, which was certified by the College of American Pathologists (CAP).The diagnosis of cirrhosis was made through imaging tests, including ultrasonography (US), computed tomography (CT), and/or MRI.
The clinical and laboratory data were retrieved from electronic medical records.L.Y.F. and G.Y.X.reviewed the data under the supervision of a specialist in gastroenterology/hepatology (F.X.L.) with > 8 years of experience in the field of autoimmune liver disease.The investigators who reviewed the clinical data were blinded to the imaging data.The normal values of biochemical results were based on the standards for common clinical biochemical tests proposed by the Health Care Commission of the People's Republic of China (http:// www.nhc.gov.cn/).Biochemical results were normalised by the upper limit of normal (ULN) of each laboratory to homogenise data interpretation.

UDCA treatment and evaluation criteria for biochemical response
All patients received standard UDCA treatment for more than 12 months (a daily dose of 13-15 mg/kg of body weight).The response to UDCA treatment was evaluated according to the Paris I criteria (ALP level ≤ 3 × ULN, together with AST level ≤ 2 × ULN and a normal bilirubin level at 12 months) for cirrhotic PBC patients and the Paris II criteria (ALP and AST ≤ 1.5 × ULN with normal bilirubin at 12 months) for noncirrhotic patients [24].

MRI examination
All MRI examinations were performed by using a 3.0-T (MAGNETOM Skyra, Siemens Healthineers; SIGNA™ Premier, GE) or 1.5-T (uMR588, Shanghai United Imaging Healthcare) MRI system equipped with a body coil.The MRI sequences and parameters are listed in Supplementary Table 1.Diffusion-weighted images were obtained by using a navigator-triggered technique at b values of 0, 50, and 800 s/mm 2 , and the corresponding apparent diffusion coefficient maps based on b values of 0 and 800 s/mm 2 were reconstructed.
Two radiologists, one with 8 years of experience in abdominal imaging (Z.Y.) and one with 10 years of experience in abdominal imaging (C.Y.D.), blinded to all clinical and laboratory data independently analysed the nonenhanced MRI scans on a picture archiving and communication system (PACS).To resolve discrepancies between the two reviewers, all MR images were reassessed together with a third, more experienced radiologist (H.Z.X.) with more than 15 years of experience in abdominal MRI until a consensus was reached.

Statistical analysis
Continuous variables, reported as the mean ± standard deviation or median (range), were compared by Student's t test or the Mann-Whitney U test, where applicable.Categorical variables, expressed as counts and percentages, were compared by the chi-square test or Fisher's exact test, where applicable.
The kappa statistic was computed to assess interobserver agreement.Univariable and multivariable logistic regression analyses were used to identify clinical and imaging predictors for 12-month insufficient biochemical response and to develop the predictive model.All variables with p < 0.1 at the univariable analyses were included in the multivariable analysis, which was performed using a forwards stepwise method.A nomogram was constructed to improve the model visualisation and utility based on the regression coefficients of the multivariable logistic regression analysis.The area under the receiver operating characteristic curve (AUC), sensitivities, and specificities were calculated to assess discriminative performances of each indicator and a combined model constructed by three indicators.Calibration curves were plotted to assess model calibration.
All statistical analyses were performed using SPSS Statistics (IBM, Version 25) and R software (R Foundation for Statistical Computing, version 3.2.5).A two-tailed p < 0.05 was considered statistically significant.

Study population
A total of 1072 consecutive patients who were diagnosed with PBC from January 2009 to April 2022 in our centre constituted the population database for this study.A total of 74 patients with PBC treated with UDCA who had a baseline abdominal MRI examination were finally included according to the inclusion and exclusion criteria (Fig. 1).

Patient characteristics
Seventy-four patients (62 females; 83.8%) were included in this study, with an average age of 50.6 ± 11.6 years.Overall, 8 of the 74 (10.8%) patients had cirrhosis; regarding PBCspecific autoantibodies, 8 (10.8%) patients had AMA positivity, and 73 (98.6%) had ANA positivity.There were no significant differences between the sufficient responders and insufficient responders in terms of the presence of autoantibodies (p = 1.000) or cirrhosis (p = 1.000).However, male patients had a higher rate of sufficient biochemical response to UDCA than female patients (p = 0.047).Regarding laboratory parameters, insufficient responders showed higher levels of serological parameters at diagnosis, including TBIL (p = 0.012), ALT (p = 0.015), AST (p = 0.005), ALP Fig. 1 Flow chart of the study population.PBC, primary biliary cholangitis; MRI, magnetic resonance imaging (p < 0.001), and GGT (p = 0.011).The baseline clinical and laboratory characteristics of the patients with PBC are summarised in Table 1.

Baseline MRI features of patients
All patients underwent specialised imaging evaluations of five aspects: liver morphology, characteristics of the liver parenchyma, portal hypertension, characteristics of the bile ducts, and lymphadenopathy.The imaging characteristics of the bile ducts were evaluated on 2D or 3D MRCP images.The baseline MRI features of the patients are summarised in Table 2.

Prediction of biochemical response to UDCA
After 12 months of follow-up, 48 of 74 patients (64.9%) with PBC showed a sufficient biochemical response to UDCA treatment, and 26 (35.1%) patients with PBC showed an insufficient biochemical response.Sex, hepatomegaly, periportal halo sign, periportal hyperintensity on T2WI, splenomegaly, minimal perihepatic effusion, and narrowing of the bile ducts were identified as significant predictors of an insufficient biochemical response to UDCA in univariate analysis.However, only hepatomegaly (odds ratio [OR]: 4.580; p = 0.011), periportal hyperintensity on T2WI (OR: 4.795, p = 0.008), and narrowing of the bile ducts (OR: 3.491; p = 0.027) were significant predictors of insufficient biochemical response in multivariate analysis (Table 3).
ROC curves were utilised to compare the predictive efficacies of the individual predictors and the combined prediction model (the concomitant three MRI characteristics were present) (Fig. 5).The results showed that the combined prediction model had a good prediction efficiency.A predictive model based on the above indicators showed an AUC of 0.781, sensitivity of 85.4%, and specificity of 61.5% for predicting insufficient biochemical response (shown in Table 4).
A nomogram was constructed based on the combined model (Fig. 6).The calibration curve of the model, as a visual nomogram, for predicting the probability of an insufficient biochemical response demonstrated near-to-perfect alignment with the 45° line, indicating a good agreement between the prediction and observation.The H-L test showed a satisfactory goodness of fit (p = 0.6933) (Supplemental Figure 1).

Discussion
Although UDCA has been recognised as the first-line treatment for PBC, up to 40% of patients show insufficient biochemical response to UDCA and are still at risk of progressing to advanced disease stages, such as liver fibrosis and cirrhosis.For this group of patients, the use of second-line therapies such as OCA, fibrate, or budesonide is expected to improve the prognosis and survival of patients [2].However, assessing the biochemical response of PBC patients to UDCA using existing criteria requires a wait of 12 months or more after UDCA treatment.Thus, there is an urgent need for predictive indicators to screen those patients who are more likely to have an insufficient biochemical response to UDCA to facilitate early access to additional treatment.In this preliminary study, we found that the presence of three nonenhanced MRI features (hepatomegaly, periportal hyperintensity on T2WI, and narrowing of the bile ducts) was suggestive of an insufficient biochemical response to UDCA in PBC.The combined model based on three indicators showed a good predictive efficacy with high sensitivity and specificity.
The nomogram based on this noninvasive model showed good calibration and predictive ability for predicting the risk of insufficient response to UDCA in PBC.
In our study, the feature of periportal hyperintensity on T2WI was significantly associated with an insufficient biochemical response after UDCA treatment in patients with PBC, and its OR value (OR: 4.795, p = 0.008) was the highest among the three risk factors.Periportal hyperintensity on T2WI is not unique to PBC and may be observed in all processes leading to periportal oedema, ductal hyperplasia, lymphatic vessel dilation, and inflammatory cell infiltration [20].In PBC, this sign may be associated with active inflammation of the portal venous branches and interfacial hepatitis [32].Previous studies have reported that patients with interfacial hepatitis are more likely to have an insufficient biochemical response [11,15].Our results may validate their findings.In addition, a controlled imagingpathology study showed that periportal hyperintensity on T2WI was present in 85.7%, 66.7%, 66.7%, and 66.7% of PBC patients with stage 1, 2, 3, and 4 disease, respectively [19].The decreased probability of this feature in advanced stages may be due to the replacement of the inflammatory process with the accumulation of fibrous tissue [33].The presence of this feature may suggest that we need to closely monitor the follow-up images (e.g., disappearance of the feature) to determine whether the disease is progressing to an advanced stage.
Hepatomegaly is defined as an abnormal diffuse enlargement of the liver, but there is still a lack of consensus on the appropriate thresholds [34].Imaging diagnosis of hepatomegaly is often based on the automatic acquisition on software of a workstation, and virtual value of liver volume can also be obtained based on calculation formulas [27,[35][36][37].Nevertheless, to facilitate consideration of the volume of  phy (e) shows irregularity and segmental narrowing of the medium to small left intrahepatic bile ducts (white arrows), and the dilatation of the common hepatic bile (red arrow).Haematoxylin-eosin (HE) staining at ×200 (f) magnification illustrates some infiltrating lymphocytes, monocytes, plasma cells, and neutrophils in the portal tract as well as bile duct loss.The patient was evaluated as 12-month insufficient biochemical response to UDCA by the Paris criteria I the liver in clinical practice, recent studies have commonly used unidimensional measurement as a diagnostic criterion for hepatomegaly in PBC [18,19].According to the results of previous studies, approximately 11 to 50% of patients with PBC present the feature of hepatomegaly, which varies depending on the stage of the disease [18,27].In the early stages of PBC, the liver tends to be normal in volume or mildly enlarged [27], whereas in patients with advanced stage or very severe portal hypertension, liver atrophy or segmental lobar hypertrophy or atrophy is more common [38,39].Our study found that hepatomegaly was a significant predictor of an insufficient response to UDCA in PBC.We speculate that this might be associated with damaged liver parenchymal cells, inflammatory cell infiltration, and liver stasis.The presence of hepatomegaly might suggest that the disease has started to progress from the earliest stages (normal liver volume) to the intermediate stages, and in patients treated with UDCA, an insufficient biochemical response may occur at this moment.
In addition, the presence of narrowing of the bile ducts was also associated with an insufficient biochemical response to UDCA in PBC.As reported in previous studies, in the early stages of PBC, when cirrhosis or liver fibrosis has not yet developed, the bile duct may appear normal on imaging in the vast majority of patients, while as the disease progresses towards cirrhosis, the intrahepatic bile ducts become destroyed and damaged, leading to fewer intrahepatic bile ducts and an irregular morphology [20].Bergasa et al [40] reported that patients with fatigue or pruritus at diagnosis were at significantly higher risk of an insufficient UDCA response and were more likely to develop cirrhosis and its complications.Moreover, patients with advanced histology and bile duct deficiency were also reported to be more likely to show an insufficient biochemical response [11,15].Given these results, we hypothesise that patients with PBC who presented with narrowing of bile ducts are more likely to be in an advanced stage or have symptoms of liver function impairment.Therefore, treatment with UDCA at this stage may not be effective in improving patient prognosis.
We further developed a nomogram to visualise and validate the predictive power of the noninvasive model based on three MRI features.Compared with the findings of Tian et al [41] who developed a predictive model based on baseline clinical and laboratory indicators, the noninvasive imaging-based prediction model created in our work achieved a similar AUC (0.781 vs. 0.790) but with a higher sensitivity (85.4% vs. 80.4%).This may indicate that imaging can reflect pathological changes associated with the disease process earlier than laboratory markers.
Despite the relatively small sample size of this study, the results are exciting.Previous studies have mainly focused on the biochemical response in PBC patients by clinical symptoms, pathological staging, and biochemical indicators [11,15,40,42].To our knowledge, this study was the first to use pretreatment MRI features to predict the insufficient biochemical response to UDCA in PBC patients.Our study has good generalizability and application prospects.On the one hand, the selection of pretreatment indicators can help in the screening of high-risk patients with a poor response so that early interventions can be implemented.On the other hand, the predictive indicators proposed in our study are based on nonenhanced MRI examinations, which are simple, easy to master, economical to use in practice, and free from risks such as contrast allergy.
This study had several limitations.First, substantial selection bias could have been introduced due to the retrospective nature of the study.Second, due to the limited sample size (most patients underwent only ultrasound without MRI examination before UDCA treatment), in future studies, we would like to prospectively include a large sample size to find more stable noninvasive indicators that can effectively predict the biochemical response to UDCA therapy in PBC patients before changes in serological indicators occur, such as 6 months or earlier.Finally, this study was a retrospective case-control study, which might not accurately reflect real clinical conditions.Further large-scale multicentre studies are warranted to validate our model.
In conclusion, a noninvasive model based on pretreatment nonenhanced MRI features could predict an insufficient  biochemical response in PBC patients after UDCA treatment.A nomogram based on potential risk factors may be used as a tool for clinicians to screen out patients who are not responding well to UDCA in advance so that early interventions can be implemented.

Fig. 2
Fig. 2 A 55-year-old female with AMA+ and ANA+ primary biliary cholangitis.T1-weighted imaging (a) shows hepatomegaly, splenomegaly, and liver surface nodularity; T2-weighted imaging (b) shows heterogeneity of the liver parenchyma and periportal hyperintensity (arrows); diffusion-weighted imaging (b value = 800 s/mm 2 ) (c) shows lymphadenopathy in the hilar and hepatogastric ligament regions (arrows); the apparent diffusion coefficient map (d) shows an absence of diffusion restriction lesion; and magnetic resonance cholangiopancreatography (e) shows poor visualisation of small to

Fig. 4 A
Fig. 4 A 53-year-old female with AMA− and ANA+ primary biliary cholangitis.T1-weighted imaging (a) shows the liver with regular contour, T2-weighted imaging (b) shows heterogeneity of the liver parenchyma and periportal hyperintensity (arrow), diffusion-weighted imaging obtained at b values of 800 s/mm 2 (c) and the corresponding apparent diffusion coefficient map (d) show no obvious abnormal diffusion restriction, and magnetic resonance cholangiopancreatogra-

Table 1
Baseline clinical and laboratory characteristics of patients with PBC PBC, primary biliary cholangitis; AMA, anti-mitochondrial antibody; ANA, anti-nuclear antibodies; TBIL, total bilirubin; ALT, alanine aminotransferase; AST, aspartate aminotransferase; ALP, alkaline phosphatase; GGT , gamma-glutamyl transferase; ALB, albumin; GLB, globulin *Unless otherwise indicated, data are number of patients, with percentage in parentheses.Categorical variables were compared by using the chisquare test, Fisher exact test, or Kruskal-Wallis H test # Data are continuous variables, reported as mean ± standard deviation or median (range), and were compared by using the two-sample t test or Mann-Whitney U test

Table 2
Baseline MRI features of patients with PBC

Table 4
Diagnostic performance of significant indicators in prediction of insufficient biochemical response after UDCA treatment in patients with PBC UDCA, ursodeoxycholic acid; PBC, primary biliary cholangitis; T2WI, T2-weighted imaging; AUC , area under the receiver operating characteristic curve