School of Precision Instruments and Opto-Electronics Engineering, Tianjin University, Tianjin, China.
Tianjin Institute of Heptology, Tianjin Second People's Hospital, Tianjin, China.
J Med Virol. 2024 Aug;96(8):e29863. doi: 10.1002/jmv.29863.
This study aimed to establish a novel noninvasive model based on the serum N-glycan spectrum for providing an objective value for determining the stage of liver necroinflammation related to chronic hepatitis B (CHB) patients. N-glycan profiles of the sera of 295 treatment-naïve CHB patients were analyzed. N-glycan profiles were tested for different liver necroinflammation stages using DNA sequence-assisted fluorophore-assisted carbohydrate electrophoresis. A serum N-glycan model named N-glycan-LI (NGLI) using support vector machine was selected to evaluate the classification of liver necroinflammation (G < 2 and G ≥ 2). The area under the receiver operating characteristic curves (AUROCs) was 0.898 (training set, n = 236) and 0.911 (validation set, n = 59) regardless of the stage of liver fibrosis (AUROC = 0.886 and 0.926, respectively, in S < 2 and S ≥ 2 group). The NGLI correspondingly had the highest specificity (SP) of 90.79% and negative predictive value of 92.00% in an inactive stage (including immune-tolerant [IT] and inactive-carrier [IC] stage), had the highest positive predictive value of 95.18% in stage immune-active, and had the highest SP of 93.94% in grey zone IT + IC. N-glycan profiles appear to correlate well with hepatic necroinflammation in CHB when compared with liver biopsy. The newly developed model appears to reliably predict liver damage in naïve-treatment patients with CHB.
本研究旨在建立一种基于血清 N-糖谱的新型非侵入性模型,为确定与慢性乙型肝炎(CHB)患者相关的肝坏死炎症阶段提供客观价值。分析了 295 例未经治疗的 CHB 患者的血清 N-糖谱。使用 DNA 序列辅助荧光辅助糖电泳检测 N-糖谱在不同肝坏死炎症阶段的表现。使用支持向量机选择了一种名为 N-糖谱-LI(NGLI)的血清 N-糖模型,用于评估肝坏死炎症的分类(G<2 和 G≥2)。无论肝纤维化阶段如何,受试者工作特征曲线(AUROCs)下面积在训练集(n=236)和验证集(n=59)中均为 0.898(AUROC=0.886 和 0.926,分别在 S<2 和 S≥2 组)。在非活动期(包括免疫耐受[IT]和非活动携带[IC]期),NGLI 具有最高的特异性(SP)90.79%和 92.00%的阴性预测值,在免疫活跃期具有最高的阳性预测值 95.18%,在灰色区 IT+IC 中具有最高的 SP 93.94%。与肝活检相比,N-糖谱与 CHB 中的肝坏死炎症相关性良好。新开发的模型似乎可以可靠地预测未经治疗的 CHB 患者的肝损伤。