A non-invasive diagnostic nomogram for CHB-related early cirrhosis: a prospective study

Abstract

This study aimed to construct a non-invasive diagnostic nomogram based on high-frequency ultrasound and magnetic resonance imaging results for early liver cirrhosis patients with chronic hepatitis B (CHB) which cannot be detected by conventional non-invasive examination methods but can only be diagnosed through invasive liver puncture for pathological examination. 72 patients with CHB were enrolled in this prospective study, and divided into S4 stage of liver cirrhosis and S0-S3 stage of non-liver cirrhosis according to pathological findings. Binary logistic regression analysis was performed to identify independent predictors, and a diagnostic nomogram was constructed for CHB-related early cirrhosis. It was validated and calibrated by bootstrap self-extraction. Binary logistic regression analysis showed that age (OR 1.14, 95% CI (1.04-1.27)), right hepatic vein diameter (OR 0.43, 95% CI 0.23-0.82), presence or absence of nodules (OR 31.98, 95% CI 3.84-266.08), and hepatic parenchymal echogenicity grading (OR 12.82, 95% CI 2.12-77.51) were identified as independent predictive indicators. The nomogram based on the 4 factors above showed good performance, with a sensitivity and specificity of 90.70% and 89.66%, respectively. The area under the curve (AUC) of the prediction model was 0.96, and the predictive model showed better predictive performance than APRI score (AUC 0.57), FIB-4 score (AUC 0.64), INPR score (AUC 0.63), and LSM score (AUC 0.67). The calibration curve of the prediction model fit well with the ideal curve, and the decision curve analysis showed that the net benefit of the model was significant. The nomogram in this study can detect liver cirrhosis in most CHB patients without liver biopsy, providing a direct, fast, and accurate practical diagnostic tool for clinical doctors.

Keywords: Chronic hepatitis B; Early liver cirrhosis; High-frequency ultrasound; Nomogram; Non-invasive diagnostic.

Figure 1

Flow chart for patient selection. CHB = chronic hepatitis B.

Figure 2

High-frequency ultrasound examination parameters and grading. (A) No nodule in liver (grade 0), fine granular echo in liver parenchyma (grade 0); (B) Scattered small hyperechoic nodules were seen in the liver, and the echo of the liver parenchyma showed coarse granules (grade 0); (C, D) The hyperechoic structure of the liver parenchyma was widened and lengthened, with an irregular shape and cord-like shape. The hypoechoic structure between the cords was irregular and widened, with strong and hypoechoic structures arranged in a staggered manner, showing a large cribriform or reticular shape. In the reticular shape, multiple hypoechoic nodules (grade 1) were scattered in the liver parenchyma.

Figure 3

Measurement of magnetic resonance examination parameters. (A) hepatic fissure distance: measurement of the thickness of the fat around the hepatic hilum. In the axial view, the distance between the anterior wall of the right portal vein and the posterior margin of the medial segment of the left hepatic lobe was measured at the plane where the right portal vein appeared, with the measurement line perpendicular to the middle of the right portal venous system; (B) portal vein-caudate lobe distance (C), right lobe-portal distance (RL): (C) In the axial plane, draw a line parallel to the mid-sagittal plane through the right lateral wall of the first bifurcation of the right portal vein. Measure the distance from this drawn line to the innermost edge of the caudate lobe; RL: from the drawn line to the outer edge of the right lobe, situated between the main portal vein and the inferior vena cava; (C) hepatic veins on the axial image, measured at the end near the inferior vena cava.

Figure 4

Diagnostic nomogram for CHB-related early cirrhosis.

Figure 5

Application of the diagnostic nomogram for CHB-related early cirrhosis.

Figure 6

Comparison of the ROC curve between the prediction model and the APRI, FIB-4, INPR scores, and LSM.

Figure 7

Calibration curve of prediction model.

Figure 8

Comparison diagram of the DCA curves of our prediction model and APRI, FIB-4, INPR, LSM.