Imaging features of porto-sinusoidal vascular disorder in a case-control study: diagnostic value for differentiation from liver cirrhosis

Introduction

The term porto-sinusoidal vascular disorder (PSVD) was recently introduced to describe a group of vascular liver diseases characterized by lesions involving portal venules and sinusoids, irrespective of portal hypertension (PH) status, and marked by the absence of cirrhotic liver parenchyma changes (1). The term was proposed to replace and extend the previous term idiopathic non-cirrhotic PH (2,3), addressing the diagnostic gap for patients with histological features of PSVD who do not necessarily have PH (1).

Clinically, PSVD with PH is misdiagnosed as cirrhosis in 70–80% cases (4). Both conditions share complications of PH, including variceal bleeding, ascites and hepatic encephalopathy related to PH (5); however, their etiologies differ. Approximately 50% of PSVD cases are associated with rare conditions such as specific drug exposure, immune disorders, coagulation disorders, infectious diseases, and congenital or hereditary diseases (1). Conversely, cirrhosis commonly results from viral hepatitis, alcohol liver disease, non-alcoholic fatty liver disease, cholestatic diseases, and other causes (6).

The accurate differentiation of PSVD and cirrhosis is critical for patient management. PSVD typically preserves liver function and has a better prognosis than cirrhosis (7,8). Importantly, selected patients with PSVD may benefit from anticoagulation, do not require hepatocellular carcinoma surveillance, and often present with benign focal lesions (3,9-11). In contrast, cirrhosis management focuses on etiologic treatment, decompensation prevention, and malignancy screening. Thus, the differentiation of the two diseases is quite important.

Liver biopsy must be performed to diagnose PSVD (12), as it can exclude cirrhosis and identify specific or non-specific histological features. However, in the absence of clear indications, many clinicians may fail to consider this diagnosis and refrain from performing a biopsy, resulting in delayed diagnosis. Moreover, histological changes in liver biopsies may be subtle, making diagnosis challenging in the absence of strong clinical suspicion (13). Thus, non-invasive imaging modalities such as computed tomography (CT) and magnetic resonance imaging (MRI) are essential in the diagnosis of PSVD. However, the imaging characteristics of PSVD remain poorly understood, particularly those that can be used to distinguish PSVD from cirrhosis.

Based on these considerations, we hypothesized that differences in CT and/or MRI imaging features could be used to distinguish PSVD from cirrhosis. This study aimed to analyze the CT and MRI features of PSVD compared to cirrhosis, and to develop and validate a predictive model to distinguish PSVD from cirrhosis based on CT and MRI features. We present this article in accordance with the TRIPOD reporting checklist (available at https://qims.amegroups.com/article/view/10.21037/qims-2025-1775/rc).

Methods

Patient selection

This retrospective study was approved by the institutional review boards of all the participating hospitals (Shanghai Public Health Clinical Center, Fudan University, No. 2024-S095-01; Shuguang Hospital Affiliated to Shanghai University of Traditional Chinese Medicine, No. 2015-445-73; Huashan Hospital, Fudan University, No. 2023-788; 900TH Hospital of Joint Logistics Support Force, No. 2025-087; the First Affiliated Hospital of Soochow University, No. 2025-1037; Renji hospital, Shanghai Jiao Tong University, No. RA-2023-564). The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. The requirement for informed consent was waived due to the retrospective nature of the study. All the datasets were independent of each other.

The data of 269 adult patients (118 with PSVD and 151 with cirrhosis) from six hospitals, diagnosed between January 2018 and December 2023, were collected for imaging analysis and model development. For independent model validation, the data of 106 patients (40 with PSVD and 66 with cirrhosis), diagnosed between January 2018 and April 2025, were collected. All patients underwent contrast-enhanced liver imaging (CT or MRI) within one month of definitive diagnosis, and the images were storied Digital Imaging and Communications in Medicine format.

The diagnosis of PSVD was based on the Vascular Liver Disease Group criteria (3), which requires the exclusion of cirrhosis via adequate liver biopsy and the fulfillment of one of the following: (I) at least one specific clinical manifestation of PH; (II) at least one specific histological feature of PSVD; or (III) the presence of both at least one unspecific clinical feature of PH and one unspecific histological feature of PSVD. Patients with a history of bone marrow transplantation, hepatic schistosomiasis, cardiac failure, Fontan surgery, Abernethy syndrome, hereditary hemorrhagic telangiectasia, chronic cholestatic diseases, tumor cell infiltration of the liver, sarcoidosis, or congenital hepatic fibrosis were excluded from the study, as were those with Budd-Chiari syndrome or hepatic venous outflow obstruction. The current PSVD criteria do not exclude portal vein thrombosis (PVT); however, patients with PVT were only included in the study if their clinical records clearly indicated that PH preceded PVT development or if liver biopsy revealed specific histological features of PSVD.

To ensure diagnostic clarity and highlight the imaging differences between PSVD and cirrhosis, all patients with cirrhosis had biopsy-confirmed stage 4 fibrosis. While this selection may enhance differentiation, it also reflects real-world diagnostic challenges, especially in China, where the prevalence of hepatitis B is high.

The imaging exclusion criteria were as follows: (I) a history of hepatectomy, or transjugular intrahepatic portosystemic shunt (TIPS); (II) a history of splenectomy; and/or (III) poor CT or MRI image quality due to respiratory motion or metal artifacts. The study design and inclusion and exclusion criteria are summarized in Figure 1A,1B.

Figure 1 Flow diagram of eligibility criteria and selection of the (A) dataset for imaging analysis and model construction, and the (B) dataset for model external testing. LASSO, least absolute shrinkage and selection operator; PSVD, porto-sinusoidal vascular disorder; TIPS, transjugular intrahepatic portosystemic shunt.

Clinical features

Medical records were retrospectively reviewed to extract the parameters needed for Child-Pugh score (CPS) assessment, including serum bilirubin, albumin, the international normalized ratio, the presence and severity of ascites, and hepatic encephalopathy. Etiology, specific or unspecific clinical signs of PH, specific or unspecific histological signs of PSVD, the preliminary clinical diagnosis, and the duration of final diagnosis were also recorded. Laboratory results were obtained within 3 months of pathological confirmation, and clinical findings were recorded based on the most recent physician documentation in that same time frame.

Imaging techniques

All patients underwent abdominal contrast-enhanced CT or MRI. Detailed information of the machines is provided in Tables S1-S5. Three-phase CT images were obtained, including pre-contrast, arterial (scanned 30–35 seconds after contrast injection), and portal venous (scanned 70–75 seconds after contrast injection). Contrast agents [e.g., iopromide (Ultravist 370, Bayer Vital GmbH, Leverkusen, Germany); iohexol (Omnipaque, GE Healthcare, Oslo, Norway) and iohexol injection (Beijing Beilu Pharmaceutical Co., Ltd., Beijing, China); etc.] were administered at a rate of 3–4 mL/s using a power injector. The minimal required MRI sequences included coronal and axial T2-weighted images, T1-weighted pre-contrast images, and contrast sequences in arterial and portal venous phases. Gadopentetic acid was administered via a bolus injection of 0.1 mmol/kg at a rate of 1.0 mL/s, followed by a 20-mL saline flush, delivered using a power injector.

Imaging analysis

Two experienced abdominal radiologists (Fengxiang Song and Y.S., with 15 and 25 years of experience, respectively) independently reviewed the images. Discrepancies were resolved through consensus, and both radiologists were blinded to the final diagnosis.

For the CT images, hepatic parenchyma, portal/venous systems, spleen size, and ascites were analyzed. The hepatic parenchymal analysis included: liver surface nodularity (LSN), liver morphologic changes, liver enhancement patterns, and widening of the Glisson sheath with low-density shadows around the portal vein in the pre-contrast and enhanced images. LSN was determined based on its presence or absence in portal venous phase (14,15). Liver morphologic changes were evaluated based on hypertrophy or atrophy of hepatic segments, including the atrophy-hypertrophy complex, which was defined as peripheral parenchymal atrophy with compensatory hypertrophy of the central segments and segment I (16). Hepatic parenchymal enhancement patterns were assessed as homogeneous or heterogeneous in the arterial or portal venous phases. Abnormalities of the portal and venous systems included cavernous transformation of the portal vein (CTPV), intrahepatic or extrahepatic PVT, splenic and/or mesenteric vein thrombosis, collateral vessel grading, and increased small branches of intrahepatic blood vessels. Collateral vessels were graded on a 0–3 scale (0, no visible varices; 1, one site involved; 2, two sites involved; 3, three or more sites involved) in five locations (gastroesophageal, paraesophageal, splenorenal, paraumbilical, and other) (17). Increased small branches of intrahepatic blood vessels, considered intrahepatic shunts, were defined as porto-venous or veno-venous shunts that developed due to underlying liver damage (2), and manifested as an increased number, thickening, and distortion of small intrahepatic blood vessels on CT or MRI. Ascites was graded on a 0–3 scale (0, no ascites; 1, minimal perihepatic and perisplenic fluid; 2, intraperitoneal fluid with no significant abdominal distention; 3, fluid causing marked abdominal distention) (18). Spleen size was measured based on the craniocaudal length in axial or coronal images, and splenomegaly was graded on a 0–3 scale (0, <13 cm; 1, 13–15 cm; 2, 16–20 cm; 3, >20 cm) (18).

For MRI, the features analyzed were similar to those on CT, with additional parameters including regenerative nodules (RNs), reticular delayed enhancement, and abnormal signal intensity on T2-weighted imaging (T2WI). RNs were liver nodules showing iso- or hyperintensity on T1-weighted imaging (T1WI), iso- or hypointensity on T2WI, iso-intensity on diffusion-weighted imaging (DWI), iso-enhancement in the hepatic arterial phase, and no delayed washout (19). Reticular delayed enhancement also showed reticular abnormal signal on T2WI. Widening of the Glisson sheath showed hypointensity on T2WI and hyperintensity on pre-contrast and post-contrast T1WI. Abnormal signal intensity on T2WI included hyperintensity around the portal area, increased peripheral parenchymal intensity, and decreased central parenchymal intensity.

Statistical analysis

Differences between groups were analyzed using Stata software (version 10.0; StataCorp, College Station, TX, USA). The continuous variables are presented as the mean ± standard deviation, or median. Pearson χ² test was used to compare the discrete variables, while the Wilcoxon test was used to compare the continuous variables. A p value <0.05 was considered statistically significant. Inter-reader agreement for imaging features was assessed using the kappa coefficient.

Predictive model construction was performed on an integrated platform-uAI Research Portal (uRP) (20,21). Variance threshold and least absolute shrinkage and selection operator (LASSO) regression were used for feature selection. Five-fold cross-validation and logistic regression were used for model development. CT, MRI, and combined predictive models were developed to select the optimal model. For five-fold cross-validation, the entire dataset was first randomly divided into five subsets. Four subsets were used in turn as the training set to train the model, while the remaining subset was used as the validation set to evaluate model performance. This process was repeated five times, with a different subset selected as the validation set each time. Finally, the model was tested on an independent dataset. The DeLong test was used to compare the performance of the MRI model and combined model. A p value <0.05 was considered statistically significant.

Results

Patient characteristics

After screening, 142 patients with PSVD and 155 patients with cirrhosis were included in the study. The internal dataset, which was used to analyze the imaging features and develop the model, comprised 106 patients with PSVD (58 with CT scans and 48 with MRI scans) and 104 patients with cirrhosis (51 with CT scans and 53 with MRI scans). The test set, which was used to test the model, comprised 36 patients with PSVD and 51 patients with cirrhosis, all of whom had MRI scans.

In the internal dataset, the etiologies of the 106 patients with PSVD included exposure to toxins or medications (36/106, 34.0%), immunodeficiency (25/106, 23.6%), and hematological and prothrombotic diseases (12/106, 11.3%). At first diagnosis, 50.9% (54/106) of the patients were misdiagnosed with cirrhosis, 34% (36/106) had no definite diagnosis, and a suspicion of PSVD or non-cirrhotic PH was only considered for 15.1% (16/106). The interval to final diagnosis ranged from 6 to 283 months (median 18.5 months, interquartile range, 9.5, 90 months). A CPS of A was significantly more common in the patients with PSVD than those with cirrhosis (79/106, 74.5% vs. 45/104, 43.3%; P<0.005). Consistent with previous PSVD studies (2,22,23), although the CPS is primarily validated for cirrhotic patients, it was applied to both groups in this study to reflect liver functional reserve. For further details, see Table 1.

Comparison of imaging characteristics

Good inter-reader agreement was observed between the two readers, with kappa coefficients between 0.740 and 0.981 (Table S6). Comparisons of the imaging characteristics between the PSVD and cirrhosis groups are presented in Table 2. PSVD exhibited pronounced PH imaging features of similar severity to cirrhosis, including grade 1–3 splenomegaly, grade 1–3 collateral vessels, and grade 1–3 ascites. In addition, PSVD showed a higher frequency of small branches of intrahepatic blood vessels than cirrhosis. Conversely, PSVD exhibited fewer cirrhosis-specific imaging features, such as reticular delayed enhancement of the hepatic parenchyma, RNs, and LSN (Figures 2A-2F,3A-3D,4A-4D). The four imaging features that showed the greatest difference between PSVD and cirrhosis were reticular delayed enhancement of the hepatic parenchyma, RNs, LSN, and increased small branches of intrahepatic blood vessels (Figure 4A-4D). Thus, PSVD showed distinct imaging dissociation-severe PH without cirrhotic features; this was accompanied by a marked increase in small intrahepatic vascular branches.

Figure 2 Representative MRI features of PSVD versus liver cirrhosis. (A-C) MRI scans of a 34-year-old male patient with confirmed PSVD showed homogeneous hepatic parenchyma (A), multiple gastro collateral vessels (B,C, white arrows), and grade 2 splenomegaly, without LSN, RNs, or reticular delayed enhancement of the hepatic parenchyma. (D-F) MRI scans of a 68-year-old female patient with confirmed liver cirrhosis by hepatitis B showed reticular abnormal signal on T2WI (D), RNs (E, liver nodules with high signal intensity on T1WI, black arrows), LSN (E, white arrows), reticular delayed enhancement (F), gastroesophageal and splenic collateral vessels (F, white arrows). To clarify, reticular delayed enhancement (F) presented as a fine, net-like pattern of delayed contrast uptake in the hepatic parenchyma, consistent with fibrotic septa. LSN, liver surface nodularity; MRI, magnetic resonance imaging; PSVD, porto-sinusoidal vascular disorder; RN, regenerative nodule; T1WI, T1-weighted imaging; T2WI, T2-weighted imaging.

Figure 3 Imaging and histopathological features of a 49-year-old female patient with confirmed PSVD. MRI scans showed prominent portal hypertension signs, including grade 2 splenomegaly (A) and large collateral vessels in the splenorenal area (B, white arrows), with an absence of LSN, RNs, and reticular delayed enhancement of the hepatic parenchyma (A-C). Microscopic findings revealed herniation of the portal vein branches into the surrounding hepatic parenchyma (D, yellow arrow; hematoxylin and eosin staining, ×100). LSN, liver surface nodularity; MRI, magnetic resonance imaging; PSVD, porto-sinusoidal vascular disorder; RN, regenerative nodule.

Figure 4 Representative MRI features of a 56-year-old male patient with confirmed hepatitis B-related cirrhosis. MRI scans showed LSN (A-D), multiple RNs (A, multiple liver nodules with low signal intensity on T2WI; B, multiple liver nodules with high signal intensity on T1WI, white arrows), reticular delayed enhancement of the hepatic parenchyma (C,D), grade 1 splenomegaly (A), and large collateral vessels in the gastroesophageal area (C,D, white arrows). This patient also showed grade 2 ascites. LSN, liver surface nodularity; MRI, magnetic resonance imaging; RN, regenerative nodule; T1WI, T1-weighted imaging; T2WI, T2-weighted imaging.

Other imaging features more common in PSVD than in cirrhosis included hypertrophy of the right liver lobe, the atrophy-hypertrophy complex (Figure 5A-5F), abnormal liver signal intensity on T2WI (Figure 5C), and heterogeneous parenchymal enhancement. Segment IV atrophy and segment II and III hypertrophy were less common in PSVD.

Figure 5 Comprehensive imaging and histopathological features of a 55-year-old female patient with confirmed PSVD. (A-C) MRI scans showed peripheral parenchymal atrophy and compensatory hypertrophy of the central segments and segment I. On contrast-enhanced images, numerous small vessels in the liver parenchyma were observed (A,B). Severe stenosis of the right portal vein due to PVT (A, black arrows), collateral vessels in the hepatic hilum and gastric varices (A,B), splenomegaly (B), and grade 1 ascites were also present (C). T2WI showed high signal intensity of the liver parenchyma around the portal vein area (short white arrows, C). (D-F) The microscopic findings showed thickening of the vessel wall of the portal vein (D, yellow arrow; hematoxylin and eosin staining, ×200), occlusion of the lumen of the portal vein (E, yellow arrow; hematoxylin and eosin staining, ×200), and incomplete septal fibrosis (F, yellow arrow; hematoxylin and eosin staining, ×100). Although hematoxylin and eosin staining was used, Masson’s trichrome or reticulin staining was performed in the original pathological assessment to confirm the presence of fibrosis but not included here due to figure limitations. MRI, magnetic resonance imaging; PSVD, porto-sinusoidal vascular disorder; PVT, portal vein thrombosis; T2WI, T2-weighted imaging.

Widening of the Glisson sheath, PVT, splenic and/or mesenteric vein involvement, and CTPV did not differ significantly between the groups.

Development of a predictive model using imaging features to diagnose PSVD

After feature selection by variance threshold and LASSO regression, different features were selected for model development. The selected features and performance of the CT model, MRI model and combined model (which included features of both the CT and MRI models) are shown in Table 3.

In the validation set, the MRI model [area under the curve (AUC): 0.970, 95% confidence interval (CI): 0.912–1.0] demonstrated superior discriminatory performance compared to the CT model (AUC: 0.825, 95% CI: 0.646–1.0), with a sensitivity of 0.818 and a specificity of 0.889. The narrow CI of the MRI model provides a precise and robust estimate of its high performance. The combined model (AUC: 0.97, 95% CI: 0.912–1.0) showed slightly improved performance compared to the MRI model alone, with a sensitivity of 0.821 and a specificity of 0.889; however, we still recommend the MRI model as the preferred modality for diagnosing PSVD.

For the MRI model, the four features showing the greatest difference between the two diseases were used for model construction (i.e., reticular delayed enhancement of the hepatic parenchyma, RNs, LSN, and increased small branches of intrahepatic blood vessels). Another 87 patients, including 36 patients with PSVD and 51 patients with cirrhosis, served as the independent test set for evaluating model performance. The AUC of the model on the independent test set was 0.988 (95% CI: 0.967–1), with a sensitivity of 0.972, and a specificity of 0.826. Model performance on the independent test set was comparable to that on the internal validation set. The receiver operating characteristic (ROC) curves, and the decision curves in the internal validation set and independent test set are shown in Figure 6A-6D. The nomogram is shown in Figure 7. The feature score was calculated as follows:

Figure 6 Validation of model efficacy. ROC and decision curve analysis showed stable and generalizable performance, with high discriminative ability and a clinical net benefit maintained across the internal validation set (A,B) and the independent test set (C,D). ROC, receiver operating characteristic.

Figure 7 The nomogram of the predictive model. 0, absence of the feature; 1, presence of the feature.

Radscore = 0.7116092 * Reticular_delayed_enhancement_of__hepatic_parenchyma + 0.129915655 * Regenerative_nodules + 0.0244622566 * Liver_surface_nodularity + –0.0275296513 * Increased_small_branches_of_intrahepatic_blood_vessels + 0.06379032.

Discussion

This multi-center, large-scale study systematically characterized the imaging features distinguishing PSVD from cirrhosis. A key and novel insight is that PSVD presented with a distinct imaging dissociation—severe PH without cirrhotic features; this was accompanied by a marked increase in small intrahepatic vascular branches. We also developed and independently tested an MRI-based predictive model, which demonstrated robust and generalizable diagnostic performance.

Consistent with the findings of Lampichler et al., the patients with PSVD in our cohort exhibited severe PH imaging features comparable to those seen in cirrhosis (2). However, Kang et al. (23) reported more severe PH imaging manifestations (including varices, ascites, and splenomegaly), and Valainathan et al. (22) observed that spleen height was higher in patients with PSVD than in patients with cirrhosis. Notably, Valainathan et al.’s study only included 43–50 patients with PSVD, and the small sample size may limit the generalizability of the results. A critical methodological distinction should be noted: while current PSVD criteria do not exclude PVT, our study specifically included PVT patients only if clinical records documented pre-existing PH prior to thrombosis or if liver biopsy revealed characteristic PSVD histology. Thus, cases where PVT (and consequent PH signs) arose from distinct causes such as hematological disorders like polycythemia vera (thrombosis due to coagulopathy) or after biliary pancreatitis (thrombosis due to regional inflammation) were excluded to prevent their misclassification as PSVD. The two aforementioned studies did not address this specific inclusion/exclusion criterion regarding the etiology of PVT.

Unlike PH in cirrhosis, which primarily stems from parenchymal liver damage, PH in PSVD results more directly from vascular lesions. These lesions include obliterative portal venopathy/portal vein stenosis (characterized by thickened vessel walls, luminal occlusion, or complete portal vein disappearance), and abnormal portal areas (characterized by increased and dilated arteries, periportal vascular shunts, and aberrant vasculature) (3). Consequently, patients with PSVD typically lack the classic imaging hallmarks of cirrhosis, such as LSN, RNs, and reticular delayed parenchymal enhancement.

This characteristic imaging pattern—severe PH in the absence of cirrhosis imaging features—serves as a strong indicator for PSVD. Recognition of this imaging pattern is crucial, as it prompts further diagnostic evaluation. Although liver biopsy is essential for diagnosing PSVD, it is not routinely performed due to its invasiveness, sampling limitations, and the limited awareness of the disease among clinicians. Moreover, the early diagnosis of PSVD is crucial, as it significantly influences patient management, particularly by preventing unnecessary hepatocellular carcinoma surveillance or cirrhosis-based treatments. In this context, cross-sectional imaging plays a critical role in raising clinical suspicion and guiding the decision to biopsy. Mironova et al. (4) also highlighted the importance of imaging in suggesting PSVD. Our study specifically highlighted this characteristic imaging pattern—severe PH in the absence of cirrhotic features—as an important radiologic indicator of PSVD.

In our study, increased small branches of intrahepatic blood vessels were observed much more frequently in PSVD than in cirrhosis (74.5% vs. 39.4%), which may be due to porto-venous or veno-venous shunts developing secondary to underlying liver damage (2,22) or increased portal pressure. Lampichler et al. (2) also reported that intrahepatic shunts were more common in PSVD than in cirrhosis, but the proportions reported were much lower than those in our study (19% vs. 75.5%). This inconsistency may be due to their analysis of contrast CT images; MRI images provide better visualization of small intrahepatic vessels due to higher spatial resolution. Conversely, Valainathan et al. (22) reported fewer arterioportal or arteriovenous shunts in PSVD than in cirrhosis (12% vs. 33%). This may be attributed to their small sample size of 50 patients with PSVD and their reliance on contrast CT images. Additionally, the severity of PSVD may vary among different studies.

Many studies (24-26) have linked LSN, resulting from the formation of RNs and/or the contraction of fibrotic tissue in cirrhosis, to the severity of chronic liver disease. A quantitative LSN score has been proposed as a potential biomarker for predicting cirrhosis stage (27-29). Consistent with our research, Valainathan et al. (22) and Lampichler et al. (2) also reported that LSN was more common in cirrhosis than in PSVD. Reticular delayed enhancement of the hepatic parenchyma, resulting from fibrotic septa and bridges, and RNs are common MRI features of cirrhosis (30). In our study, 94.3% of the cirrhosis group exhibited reticular delayed enhancement of the hepatic parenchyma, and 54.7% showed RNs, compared to only 20.0% and 6.3% in the PSVD group, respectively. However, previous studies did not highlight the importance of these indices in differential diagnosis. Kang et al. (23) included 129 patients with pathologically proven cirrhosis as the control group, 79.1% of whom had hepatitis B; however, they did not report on reticular delayed enhancement of the hepatic parenchyma or RNs. Lampichler et al. (2) did not report on these features either. This may because these authors focused on focal nodular hyperplasia-like lesions and “periportal hyperintensity” in the hepatobiliary phase of gadolinium-ethoxybenzyl-diethylenetriamine pentaacetic acid-enhanced MRI.

In addition, 90.4% of the cirrhosis patients in our study had hepatitis B, which is associated with macro-nodular cirrhosis (5) and more frequent RNs, making the typical cirrhosis imaging features more apparent. In our study, three patients with PSVD had RNs, which are considered indicative of cirrhosis. This may be due to PSVD causing nodular regenerative hyperplasia (1,31), which can mimic RNs on MRI. Moreover, unlike complete fibrotic septa in cirrhosis, PSVD typically forms incomplete fibrotic septa (1,8,31), which may explain the lack of reticular delayed enhancement of the hepatic parenchyma in PSVD.

Two morphological changes in the hepatic parenchyma should be noted: right lobe hypertrophy (34.0% vs. 17.3%) and the atrophy-hypertrophy complex (17.9% vs. 0.96%) were more frequently observed in PSVD than in cirrhosis, consistent with findings reported by Lampichler et al. (2). Abnormal signal intensity of the hepatic parenchyma on T2WI was significantly more common in PSVD than in cirrhosis (22.2% vs. 1.9%) and has not been reported in previous studies. The presence of these imaging features may serve as a valuable indicator for the diagnosis of PSVD.

The predictive model based on MRI demonstrated excellent performance in both the internal validation and independent test sets. Valainathan et al. (22) also developed a model using a median LSN <2.5 (with LSN quantification software) and a normal-sized or enlarged segment IV on CT images. Their model had a specificity of 94% and an accuracy of 87% in an independent multi-center validation cohort. However, it was based on only 50 patients with PSVD, and LSN quantification software is currently not widely available.

Given the superior ability of MRI to visualize LSN, RNs, reticular delayed enhancement, increased small intrahepatic vessels, and hepatic parenchymal signal changes, along with the optimal performance of our predictive model, we recommend contrast-enhanced MRI as the preferred method for differentiating PSVD from cirrhosis. Additionally, enhanced MRI with hepatobiliary-specific contrast agents is recommended, as focal nodular hyperplasia-like lesions and periportal hyperintensity in the hepatobiliary phase have also proven valuable for differential diagnosis (2,23).

Our study had several limitations. First, variations in the scanning instruments, protocols, and contrast agents may have influenced the imaging features. Second, incomplete baseline characteristics, such as clinical symptoms and laboratory findings, resulted from the long enrollment period. Third, most of the cirrhosis patients had hepatitis B. Future studies should explore PSVD differentiation from cirrhosis caused by hepatitis C, autoimmune hepatitis, alcoholic liver disease, and cholestatic liver disease. Fourth, although we enrolled 106 patients for model development, the amount of MRI data is relatively small. Finally, the imaging features of PSVD without PH were not analyzed, due to the small simple size of five cases. The early diagnosis of PSVD before PH onset could enable timely intervention, potentially delaying disease progression and improving prognosis.

Conclusions

Various imaging features can be used to differentiate PSVD from cirrhosis. When severe PH is present without typical imaging signs of cirrhosis, a diagnosis of PSVD should be considered. Our MRI-based predictive model, incorporating reticular delayed enhancement of the hepatic parenchyma, RNs, LSN, and increased small branches of intrahepatic blood vessels, performed well in both the internal validation and independent test sets. This model provides a new non-invasive method to aid in identifying patients with PSVD, which is important in guiding accurate clinical diagnosis and treatment.

Acknowledgments

None.

Footnote

Reporting Checklist: The authors have completed the TRIPOD reporting checklist. Available at https://qims.amegroups.com/article/view/10.21037/qims-2025-1775/rc

Data Sharing Statement: Available at https://qims.amegroups.com/article/view/10.21037/qims-2025-1775/dss

Funding: This work was supported by Shanghai Key Specialty of Traditional Chinese Clinical Medicine (No. shslczdzk01201), IIT study grant of Huashan Hospital (No. 2023-YN012), the Shanghai Municipal Health Commission (No. 202140084), and National Natural Science Foundation of China (Nos. 82302335 and 82172029).

Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://qims.amegroups.com/article/view/10.21037/qims-2025-1775/coif). The authors have no conflicts of interest to declare.

Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. This retrospective study was approved by the institutional review boards of all participating hospitals (Shanghai Public Health Clinical Center, Fudan University, No. 2024-S095-01; Shuguang Hospital Affiliated to Shanghai University of Traditional Chinese Medicine, No. 2015-445-73; Huashan Hospital, Fudan University, No. 2023-788; 900TH Hospital of Joint Logistics Support Force, No. 2025-087; the First Affiliated Hospital of Soochow University, No. 2025-1037; Renji hospital, Shanghai Jiao Tong University, No. RA-2023-564). The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. Informed consent was waived due to the retrospective nature of the study.

Open Access Statement: This is an Open Access article distributed in accordance with the Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International License (CC BY-NC-ND 4.0), which permits the non-commercial replication and distribution of the article with the strict proviso that no changes or edits are made and the original work is properly cited (including links to both the formal publication through the relevant DOI and the license). See: https://creativecommons.org/licenses/by-nc-nd/4.0/.

References

1. De Gottardi A, Sempoux C, Berzigotti A. Porto-sinusoidal vascular disorder. J Hepatol 2022;77:1124-35. [Crossref] [PubMed]
2. Lampichler K, Semmler G, Wöran K, Simbrunner B, Jachs M, Hartl L, Bauer DJM, Balcar L, Burghart L, Trauner M, Tamandl D, Ba-Ssalamah A, Mandorfer M, Reiberger T, Scheiner B, Scharitzer M. Imaging features facilitate diagnosis of porto-sinusoidal vascular disorder. Eur Radiol 2023;33:1422-32. [Crossref] [PubMed]
3. De Gottardi A, Rautou PE, Schouten J, Rubbia-Brandt L, Leebeek F, Trebicka J, Murad SD, Vilgrain V, Hernandez-Gea V, Nery F, Plessier A, Berzigotti A, Bioulac-Sage P, Primignani M, Semela D, Elkrief L, Bedossa P, Valla D, Garcia-Pagan JC. Porto-sinusoidal vascular disease: proposal and description of a novel entity. Lancet Gastroenterol Hepatol 2019;4:399-411. [Crossref] [PubMed]
4. Mironova M, Gopalakrishna H, Koh C, Kleiner DE, Heller T. Portal sinusoidal vascular diseases: Assessment and therapy. Hepatology 2025; Epub ahead of print. [Crossref]
5. Krasinskas AM, Eghtesad B, Kamath PS, Demetris AJ, Abraham SC. Liver transplantation for severe intrahepatic noncirrhotic portal hypertension. Liver Transpl 2005;11:627-34; discussion 610-1. [Crossref] [PubMed]
6. Faria SC, Ganesan K, Mwangi I, Shiehmorteza M, Viamonte B, Mazhar S, Peterson M, Kono Y, Santillan C, Casola G, Sirlin CB. MR imaging of liver fibrosis: current state of the art. Radiographics 2009;29:1615-35. [Crossref] [PubMed]
7. Khanna R, Sarin SK. Non-cirrhotic portal hypertension - diagnosis and management. J Hepatol 2014;60:421-41. [Crossref] [PubMed]
8. Premkumar M, Anand AC. Porto-sinusoidal Vascular Disease: Classification and Clinical Relevance. J Clin Exp Hepatol 2024;14:101396. [Crossref] [PubMed]
9. Montenovo MI, Jalikis FG, Yeh M, Reyes JD. Progression of Hepatic Adenoma to Carcinoma in the Setting of Hepatoportal Sclerosis in HIV Patient: Case Report and Review of the Literature. Case Reports Hepatol 2016;2016:1732069. [Crossref] [PubMed]
10. Sempoux C, Paradis V, Komuta M, Wee A, Calderaro J, Balabaud C, Quaglia A, Bioulac-Sage P. Hepatocellular nodules expressing markers of hepatocellular adenomas in Budd-Chiari syndrome and other rare hepatic vascular disorders. J Hepatol 2015;63:1173-80. [Crossref] [PubMed]
11. Sugimoto K, Kondo F, Furuichi Y, Oshiro H, Nagao T, Saito K, Yoshida H, Imai Y, Fukusato T, Moriyasu F. Focal nodular hyperplasia-like lesion of the liver with focal adenoma features associated with idiopathic portal hypertension. Hepatol Res 2014;44:E309-15. [Crossref] [PubMed]
12. Giri S, Singh A, Roy A, Patel RK, Tripathy T, Angadi S. Noninvasive differentiation of porto-sinusoidal vascular disorder from cirrhosis: a systematic review. Abdom Radiol (NY) 2023;48:2340-8. [Crossref] [PubMed]
13. Isabel Fiel M, Thung SN, Hytiroglou P, Emre S, Schiano TD. Liver failure and need for liver transplantation in patients with advanced hepatoportal sclerosis. Am J Surg Pathol 2007;31:607-14. [Crossref] [PubMed]
14. Glatard AS, Hillaire S, d'Assignies G, Cazals-Hatem D, Plessier A, Valla DC, Vilgrain V. Obliterative portal venopathy: findings at CT imaging. Radiology 2012;263:741-50. [Crossref] [PubMed]
15. Krishnan P, Fiel MI, Rosenkrantz AB, Hajdu CH, Schiano TD, Oyfe I, Taouli B. Hepatoportal sclerosis: CT and MRI appearance with histopathologic correlation. AJR Am J Roentgenol 2012;198:370-6. [Crossref] [PubMed]
16. Vilgrain V, Condat B, Bureau C, Hakimé A, Plessier A, Cazals-Hatem D, Valla DC. Atrophy-hypertrophy complex in patients with cavernous transformation of the portal vein: CT evaluation. Radiology 2006;241:149-55. [Crossref] [PubMed]
17. Ito K, Mitchell DG, Hann HW, Kim Y, Fujita T, Okazaki H, Honjo K, Matsunaga N. Viral-induced cirrhosis: grading of severity using MR imaging. AJR Am J Roentgenol 1999;173:591-6. [Crossref] [PubMed]
18. Saygili OB, Tarhan NC, Yildirim T, Serin E, Ozer B, Agildere AM. Value of computed tomography and magnetic resonance imaging for assessing severity of liver cirrhosis secondary to viral hepatitis. Eur J Radiol 2005;54:400-7. [Crossref] [PubMed]
19. Watanabe A, Ramalho M, AlObaidy M, Kim HJ, Velloni FG, Semelka RC. Magnetic resonance imaging of the cirrhotic liver: An update. World J Hepatol 2015;7:468-87. [Crossref] [PubMed]
20. Zhou L, Yang W, Liu Y, Li J, Zhao M, Liu G, Zhang J. Correlations between cognitive reserve, gray matter, and cerebrospinal fluid volume in healthy elders and mild cognitive impairment patients. Front Neurol 2024;15:1355546. [Crossref] [PubMed]
21. Wang S, Tong X, Cheng Q, Xiao Q, Cui J, Li J, Liu Y, Fang X. Fully automated deep learning system for osteoporosis screening using chest computed tomography images. Quant Imaging Med Surg 2024;14:2816-27. [Crossref] [PubMed]
22. Valainathan SR, Sartoris R, Elkrief L, Magaz M, Betancourt F, Pellegrino S, et al. Contrast-enhanced CT and liver surface nodularity for the diagnosis of porto-sinusoidal vascular disorder: A case-control study. Hepatology 2022;76:418-28. [Crossref] [PubMed]
23. Kang JH, Kim DH, Kim SY, Kang HJ, Lee JB, Kim KW, Lee SS, Choi J, Lim YS. Porto-sinusoidal vascular disease with portal hypertension versus liver cirrhosis: differences in imaging features on CT and hepatobiliary contrast-enhanced MRI. Abdom Radiol (NY) 2021;46:1891-903. [Crossref] [PubMed]
24. Smith AD, Branch CR, Zand K, Subramony C, Zhang H, Thaggard K, Hosch R, Bryan J, Vasanji A, Griswold M, Zhang X. Liver Surface Nodularity Quantification from Routine CT Images as a Biomarker for Detection and Evaluation of Cirrhosis. Radiology 2016;280:771-81. [Crossref] [PubMed]
25. Colli A, Fraquelli M, Andreoletti M, Marino B, Zuccoli E, Conte D. Severe liver fibrosis or cirrhosis: accuracy of US for detection--analysis of 300 cases. Radiology 2003;227:89-94. [Crossref] [PubMed]
26. Venkatesh SK, Yin M, Takahashi N, Glockner JF, Talwalkar JA, Ehman RL. Non-invasive detection of liver fibrosis: MR imaging features vs. MR elastography. Abdom Imaging 2015;40:766-75. [Crossref] [PubMed]
27. He Y, Yan Y, Zhang S. Quantitative liver surface nodularity score based on imaging for assessment of early cirrhosis in patients with chronic liver disease: A protocol for systematic review and meta-analysis. Medicine (Baltimore) 2021;100:e23636. [Crossref] [PubMed]
28. Besa C, Wagner M, Lo G, Gordic S, Chatterji M, Kennedy P, Stueck A, Thung S, Babb J, Smith A, Taouli B. Detection of liver fibrosis using qualitative and quantitative MR elastography compared to liver surface nodularity measurement, gadoxetic acid uptake, and serum markers. J Magn Reson Imaging 2018;47:1552-61. [Crossref] [PubMed]
29. Lo GC, Besa C, King MJ, Kang M, Stueck A, Thung S, Wagner M, Smith AD, Taouli B. Feasibility and reproducibility of liver surface nodularity quantification for the assessment of liver cirrhosis using CT and MRI. Eur J Radiol Open 2017;4:95-100. [Crossref] [PubMed]
30. Ozturk A, Olson MC, Samir AE, Venkatesh SK. Liver fibrosis assessment: MR and US elastography. Abdom Radiol (NY) 2022;47:3037-50. [Crossref] [PubMed]
31. Noble S, Linz M, Correia E, Shalaby A, Bittencourt LK, Sclair SN. Porto-sinusoidal Vascular Disease and Portal Hypertension. Clin Liver Dis 2024;28:455-66. [Crossref] [PubMed]