Scientific Papers

A novel imaging index for predicting adverse progression in acute-on-chronic liver failure related to hepatitis B virus: the low erector spine index | BMC Gastroenterology


General patient characteristics

In this study, 118 patients with HBV-ACLF were included. Of these patients, 104 (88.1%) were men. The average age was 41.3 ± 10.7 years. The average level of HBV DNA was 3.2 ± 2.2 log10 IU/ml. The average model for end-stage liver disease (MELD) score and MELD-sodium (MELD-Na) score were 27.4 ± 6.8 and 30.4 ± 9.4, respectively. The potential precipitating event included discontinuance of nucleos(t)ide analogs for 36 patients who had taken entecavir or adefovir combined with lamivudine, the resistance of nucleos(t)ide analogs for 9 patients who had taken lamivudine or telbivudine, and severe acute attack for 34 patients who had not previously received nucleos(t)ide analogs. There was no record of identified precipitating for 39 patients. All patients received entecavir (92.4%) or a combination of entecavir and adefovir (7.6%) after onset. The presence of ascites, SBP, kidney dysfunction, hyponatremia, respiratory infections, and hepatic encephalopathy at baseline was 89.8%, 21.2%, 13.5%, 42.4%, 14.4%, and 11.0%, respectively. Besides the infections located in the abdominal and pulmonary, infections involved in the urinary system, gastrointestinal tract, and unknown locations were 16.1%. The average ESI was 12.6 ± 3.8 cm2/m2. Laboratory parameters, such as serum albumin, bilirubin, creatinine, INR, and white blood cell count, were shown in detail in Table 1.

Table 1 General characteristics at baseline

All patients were followed up to 90 days after enrollment. The free-transplant mortality was 39.0%, and the rates of new-onset kidney dysfunction, hepatic encephalopathy, and in-hospital infection were 32.4%, 25.7%, and 68.8%, respectively (Table 2). The in-hospital infections primarily included SBP, respiratory infections, and other types of infections, and the details were shown (Supplementary Table 1). Other types of infections in this study referred to conditions beyond those of abdominal and pulmonary and involved the urinary and gastrointestinal systems, as well as unlocated infections.

Table 2 Outcomes for 90-day follow up

Correlation of ESI with baseline characteristics

According to the median of ESI (12.05 cm2/m2), the patients were classified into two groups to evaluate the factors associated with erector spine mass in HBV-ACLF patients. The cut-off value of ESI calculated through ROC was 12.06 cm2/m2, which closely approximated the median. The baseline characteristics between the two groups were compared in Table 1. Patients with ESI<12.05 cm2/m2 (n = 58) showed a higher level of serum ammonia (P = 0.030) than those patients with ESI ≥ 12.05 cm2/m2 (n = 60). There was no significant difference in age, MELD score, MELD-Na score, or other laboratory indicators. Further binary logistic regression analysis was used to determine factors independently associated with ESI (Table 3). The presence of hepatic encephalopathy (the odds ratio (OR) = 2.26, 95% confidence interval (CI): 2.05–3.18; P = 0.006), SBP (OR = 3.95, 95% CI: 1.01–5.46; P = 0.037), and kidney dysfunction (OR = 4.47, 95% CI: 1.02–9.64; P = 0.032) were independently associated with low ESI in HBV-ACLF patients.

Table 3 Risk factors associated with ESI in HBV-ACLF patients

Impact of ESI on the free-transplant mortality of HBV-ACLF patients

During the 90-day follow-up period, 30 of 58 (51.7%) patients with a low ESI and 16 of 60 (26.7%) patients with a high ESI died(P = 0.005) (Table 2). Kaplan-Meier survival analysis was conducted, and the cumulative survival rates were 71.0%±4.6 and 85.8%±3.9, respectively (log-rank P = 0.003) (Fig. 2A). The hazard ratio (HR) calculated using univariable analysis was 2.23(95% CI: 1.25–4.21; P = 0.005) and further multivariable analyses showed that a low ESI increased the risk for death dependently (adjusted HR = 2.52; 95% CI: 1.34–9.24; P = 0.011) (Table 4). The details about the comparison between the survival patients with the deceased patients and the multiple analysis were provided in Supplementary Table 2 and Table 1, respectively.

Fig. 2
figure 2

Kaplan–Meier curves of patients with a low ESI and patients with a high ESI level. Statistical significance was calculated by the log-rank test for (A) cumulative survival, (B) cumulative incidence rate of kidney dysfunction, (C) cumulative incidence rate of hepatic encephalopathy, and (D) cumulative incidence rate of in-hospital infection

Table 4 Cox Regression Evaluation of ESI Associated with Outcomes in HBV-ACLF patients

Impact of ESI on the development of kidney dysfunction, hepatic encephalopathy, and in-hospital infection

In the study, there were 102 patients with the absence of kidney dysfunction at baseline. Among these patients, 20 of 46 (43.5%) patients with a low ESI and 13 of 56 (23.2%) patients with a high ESI developed kidney dysfunction during the 90-day follow-up period(P = 0.029) (Table 2). There was a significant difference in the cumulative incidence rate of kidney dysfunction between groups (log-rank P = 0.017) (Fig. 2B). Univariable and multivariable analyses are shown in Table 4, indicating that a low ESI was an independent risk factor for kidney dysfunction (adjusted HR = 1.36; 95% CI: 1.05, 2.93; P = 0.043) (Table 4, and Supplementary Table 4 for details).

Furthermore, among 105 patients without the presence of hepatic encephalopathy at baseline, 49 patients with a low ESI had a higher cumulative incidence rate of hepatic encephalopathy compared with 56 patients with a high ESI as shown by Kaplan-Meier analysis (39.6% vs. 14.0%, P = 0.003; log-rank P = 0. 010) (Table 2; Fig. 2C), and a low ESI was an independent risk factor for the development of hepatic encephalopathy through multivariable analyses (adjusted HR = 2.26; 95% CI: 2.05, 3.13; P = 0.036) (Table 4, and Supplementary Table 5 for details). However, there was no difference in the cumulative incidence rate of in-hospital infection between the two groups (75.0% vs. 63.4%, P = 0.273; log-rank P = 0. 079; adjusted HR = 1.62; 95% CI: 0.60, 3.08; P = 0.138) (Fig. 2D; Table 4, and Supplementary Table 6 for details).

Correlation of ESI with SMI

In this study, there were 23 patients with whole abdominal CT images. The ESI and SMI at L3 of these patients were calculated. The average ESI and SMI were 11.7 ± 3.8 and 39.4 ± 12.3, respectively (P = 0.231). There was a significant positive correlation between ESI and SMI (r = 0.714, P < 0.001). We also assessed the predictive ability of ESI and SMI for mortality outcomes. The areas under the ROC of ESI and SMI were 0.755 (0.536–0.973) and 0.773 (0.563–0.983), respectively (Fig. 3). The ROC analysis yielded no significant results (Z=-1.449, P = 0.147).

Fig. 3
figure 3

ROC for evaluating mortality in patients with HBV-ACLF. The areas under the ROC (95% CI) of the ESI and SMI were 0.755 (0.536–0.973) and 0.773 (0.563–0.983), respectively. ESI: erector spine index. SMI: skeletal muscle index



Source link