Splenic stiffness measurement to predict decompensation in compensated advanced chronic liver disease

Abstract

BACKGROUND

The onset of decompensation is an important event in patients with compensated advanced chronic liver diseases (c-ACLD). Splenic stiffness measurement (SSM) is an emerging non-invasive modality for stratifying patients with c-ACLD.

AIM

To determine the utility of SSM to predict one-year decompensation in patients with c-ACLD.

METHODS

In this prospective study, patients with c-ACLD (Baveno VII) were recruited from 2022 to 2024 at a tertiary care center in north India. Subsequently, they underwent baseline SSM using the vibration-controlled transient elastography method with a 100 Hz spleen-dedicated probe. They were followed up for a year to document decompensation and death.

RESULTS

A total of 294 patients were recruited, 63 were excluded and 231 patients were included in the final analysis. The mean age was 45.18 ± 9.73 years, and the majority were women (58.4%). The most common etiology of c-ACLD was chronic hepatitis C infection [75 (32.5%)]. One year-decompensation rate was 8.26% (19 patients). The patients who decompensated had higher SSM and liver stiffness values (69.61 ± 16.27 kPa vs 43.99 ± 14.81 kPa, P < 0.001; and 30.77 ± 11.26 kPa vs 22.68 ± 7.02 kPa, P < 0.001, respectively). In the Cox proportional hazards model, SSM predicted one-year decompensation [hazard ratio: 1.073 (1.049-1.097), P < 0.001]. With a cut-off 57.5 kPa, SSM has the highest sensitivity (82%), specificity (87.5%), positive predictive value (98.26%), negative predictive value (36.21%) and diagnostic accuracy (82.61%) for predicting decompensation in patients with c-ACLD.

CONCLUSION

A SSM cut-off of 57.5 kPa has a sensitivity and specificity of 82% and 87.5%, respectively, for predicting one-year decompensation in patients with c-ACLD.

Key Words: Splenic stiffness measurement; Compensated advanced chronic liver disease; Liver stiffness measurement; Decompensation; Ascites

Core Tip: The onset of decompensation is an important event in patients with compensated advanced chronic liver diseases (c-ACLD). The presence of clinically significant portal hypertension is associated with a higher risk of decompensation in patients with c-ACLD. Hepatic venous pressure gradient measurement is used to risk-stratify such patients. Hepatic venous pressure gradient measurement is an invasive investigation; hence, liver and splenic stiffness measurements are emerging as useful non-invasive tests to predict clinically significant portal hypertension and risk of decompensation. Herein, we have explored the role of splenic stiffness measurement in predicting decompensation in patients with c-ACLD.

INTRODUCTION

Compensated advanced chronic liver diseases (c-ACLD) encompass a spectrum of patients with advanced liver fibrosis who do not have any history of liver-related decompensation[1]. Decompensation alters the natural history of patients with c-ACLD and is associated with a higher rate of liver-related death without liver transplantation[1,2]. The risk of decompensation can be stratified based on the presence of clinically significant portal hypertension (CSPH)[1]; however, various non-invasive tools have been used to risk-stratify patients with c-ACLD and predict decompensation[2]. Splenic stiffness measurement (SSM) is one of the novel tools to predict decompensation in patients with c-ACLD[3,4]. It has shown promising results in stratifying patients with c-ACLD[4,5]. Herein, we aim to determine the utility of SSM to predict decompensation in patients with c-ACLD.

MATERIALS AND METHODS

Patient population

This prospective study was conducted at a North Indian tertiary medical university from 2022 to 2024. Patients with c-ACLD, as defined by the Baveno VII consensus [liver stiffness measurement (LSM) > 10 kPa without any prior history of decompensation], with various etiologies, including chronic hepatitis B virus (HBV) and hepatitis C virus infection (HCV), metabolic dysfunction-associated steatotic liver disease (MASLD), and alcohol-related liver disease (ArLD)[1] were screened and recruited for this study. Patients with c-ACLD underwent a thorough physical examination and laboratory investigations monthly for the next 12 months. Patients with advanced fibrosis underwent liver ultrasound 6-monthly for hepatocellular carcinoma screening, irrespective of etiology[6]. The patients were subjected to esophagogastroduodenoscopy based on guidelines laid down by Baveno VII consensus. Any evidence of decompensation was documented. Patients who developed acute-on-chronic liver failure during the follow-up, who received beta-blockers, had portal vein thrombosis, and hepatocellular carcinoma at baseline and/or with active alcohol intake following the enrollment were also excluded from this study. Additionally, patients with a history of splenectomy, hematological disorders affecting spleen architecture (such as myeloproliferative neoplasms, hemolytic anemia, or infiltrative spleen disease), and congestive splenomegaly from cardiac causes were not enrolled, as these conditions are associated with falsely elevated SSM value independent of portal hypertension.

Written informed consent was obtained from each participant. The institutional ethics board had approved the study, approval No. 1444/Ethics/2021. The study was conducted in strict accordance with the guidelines of the Declaration of Helsinki.

Measurement of SSM

Vibration-controlled transient elastography (VCTE) was used to measure liver and splenic stiffness. The quality control standards applied in our study were as follows: (1) A dedicated 100 Hz splenic probe was used to measure SSM in all individuals; (2) SSM was performed after an overnight fast to standardize measurement conditions and minimize post-prandial splanchnic blood flow effects[6]; (3) An ultrasound of the spleen was performed for better positioning of the VCTE probe[6,7]. The M probe was used for thin-to-average-built patients (body mass index < 30 kg/m²), and the XL probe was used for patients with abdominal obesity (body mass index ≥ 30 kg/m² or significant increase in skin-to-liver capsule distance). This probe selection approach applies to both LSM and SSM during VCTE; (4) A minimum of 10 valid measurements were obtained for each patient, and the median value in kPa was recorded; (5) An interquartile range (IQR)/median ratio of < 30% was applied as the reliability criterion, consistent with established standards for LSM and used for SSM in our protocol[7,8]; (6) All measurements were performed by a single trained operator blinded to the patient’s clinical characteristics and follow-up outcomes, thereby eliminating inter-operator variability[7]. The VCTE machine was calibrated at regular intervals per the manufacturer’s specifications.

Definition and management of decompensation

Decompensation was defined according to the Baveno VII consensus statement[1]. The onset of overt hepatic encephalopathy (HE), moderate to grade 3 ascites and acute variceal bleeding (AVB) was defined as decompensation[1]. Any decompensation was recorded during monthly visit and managed accordingly[9]. Patients with moderate to severe ascites were managed with strict salt restriction and a titrated dose of diuretics. HE was managed with careful identification, treatment of the precipitant, lactulose and rifaximin. Patients who presented with AVB were admitted to the intensive care unit. After initial hemodynamic stability, patients were managed with endoscopic variceal ligation. Along with that, all patients received intravenous terlipressin and antibiotics[9,10]. In case of unavailability or contraindication to terlipressin, intravenous octreotide was used. Patients with CSPH and following AVB received beta-blockers.

Sample size calculation

Sample size calculation based on the expected one-year decompensation rate in patients with c-ACLD. Previous studies have reported a one-year decompensation rate of approximately 8%-12% in this population[1]. Assuming a decompensation rate of 10%, a hazard ratio (HR) of at least 1.8 for high SSM, with 80% power and a two-sided alpha error of 0.05, the minimum required sample size was estimated to be 220 patients considering an anticipated loss to follow-up rate of 15%, a total of at least 260 patients were planned for enrolment.

Statistical analysis

Data were maintained on an Excel sheet, and data analysis was conducted using IBM SPSS version 23 (IBM Corp., Armonk, NY, United States). Results are expressed as mean and standard deviation. Normality of variables was assessed using histogram analysis and the Shapiro-Wilk test. Continuous variables were compared parametrically by using the Student t-test. And a P value < 0.05 was considered statistically significant. The aim and primary objective of the study were to assess the one-year decompensation rate and its predictors in patients with c-ACLD. The cut-off was estimated using the Youden index, and the area under the curve (AUC) was constructed. Cox proportional hazards ratio was used to assess predictors of decompensation and death. Adjusted hazards ratio was calculated as patients with CSPH received beta-blockers according to the Baveno VII criteria[1]. We excluded patients with ongoing alcohol intake, so the impact of concomitant alcohol intake following enrolment was not estimated. The Kaplan-Meier survival analysis (Log rank test) was used to calculate the overall survival and decompensation-free survival using the cut-off SSM. Furthermore, the AUC of SSM and LSM were compared using the DeLong test. Based on the predictive power of parameters on Cox Proportional HR model, we have also devised a score to predict decompensation in patients with c-ACLD. The composite score was used to risk-stratify patients with c-ACLD to predict decompensation rate.

RESULTS

Baseline characteristics of the study population

A total of 294 patients were included in the study. Sixty three patients were excluded from the final analysis (Figure 1). Of the 231 patients, [age of 45.18 ± 9.73 years, male: 96 (41.5%)], the most common etiology of c-ACLD was chronic HCV infection [75 (32.4%) patients], followed by chronic HBV [60 (26%)], MASLD [52 (23%)], and ArLD [43 (18.6%)]. The mean SSM, LSM, aspartate aminotransferase to platelet ratio index (APRI), and fibrosis-4 index (FIB-4) were 46.4 ± 16.8 kPa, 23.5 ± 7.9 kPa, 1.73 ± 1.14, and 4.45 ± 2.5, respectively. Their baseline characteristics are presented in Table 1. The median follow-up period was 15 months (IQR 13-15 months).

Figure 1 Study flowchart. c-ACLD: Compensated advanced chronic liver disease; HCV: Hepatitis C virus; HBV: Hepatitis B virus; MASLD: Metabolic dysfunction-associated steatotic liver disease; ArLD: Alcohol related liver disease; ACLF: Acute on chronic liver failure; ALD: Alcohol related steatotic liver disease; SSM: Splenic stiffness measurement.

Table 1 Baseline parameters of patients with compensated advanced chronic liver disease of different etiology, n (%)/mean ± SD.

Parameters	Overall (n = 231)	MASLD (n = 52)	ArLD (n = 43)	HCV (n = 75)	HBV (n = 60)
Age (years)	45.18 ± 9.73	47 ± 10.7	39.09 ± 7.17	47 ± 8.88	44 ± 9.33
Male:female	96:135	22:31	42:1	14: 61	18:42
Hemoglobin	11.66 ± 1.87	11.53 ± 1.75	11.82 ± 1.76	11.92 ± 1.95	11.37 ± 1.94
Total leukocyte count	5077.0 ± 1622	4857 ± 1436	5154 ± 1454	5322 ±1837	4905 ± 1582
Platelet	1.1 ± 0.43	1.49 ± 0.49	1.14 ± 0.47	1.43 ± 0.49	1.08 ±0.41
Total bilirubin	1.01 ± 0.54	1.6 ± 0.56	0.92 ± 0.67	1.0 ± 0.46	1.11 ± 0.55
Conjugated bilirubin	0.61 ± 0.33	0.7 ± 0.31	0.50 ± 0.33	0.62 ± 0.21	0.64 ± .13
AST	64 ± 30	63 ± 23	71 ± 37	70 ± 33	54.20 ± 21
ALT	49.18 ± 26	48 ± 20	56 ± 37	52 ± 29.5	40.8 ± 16
ALP	117 ± 43	121 ± 56	127 ± 36	121.6 ± 38	127 ± 34
Albumin	4.10 ± 2.6	3.87 ± 0.41	3.84 ± 0.50	3.82 ± 1.38	4.4 ± 2.76
Protein	8.8 ± 5.3	8.7 ± 3.1	8.45 ± 2.47	7.74 ± 0.61	8.1 ± 3.23
INR	1.12 ± 0.19	1.14 ± 0.20	1.09 ± 0.21	1.11 ± 0.16	1.15 ± 0.20
Urea	28.8 ± 9.27	31 ± 10.3	24.5 ± 8.75	28.42 ± 7.96	29.8 ± 9.21
Creatinine	0.88 ± 0.19	0.89 ± 0.19	0.50 ± 0.33	0.86 ± 0.19	0.89 ± 0.18
LSM	23.5 ± 7.9	23 ± 7.3	21.9 ± 8.17	24.5 ± 8.06	23.7 ± 8.12
SSM	46.4 ± 16.8	47.26 ± 16.47	47.26 ± 17.9	45.6 ± 15.9	49.9 ± 17.42
FIB-4	4.45 ± 2.5	4.78 ± 2.63	3.88 ± 2.34	4.90 ± 2.80	4.03 ± 2.11
APRI	1.73 ± 1.14	1.76 ± 1.02	1.89 ± 1.29	1.49 ± 0.94	1.83 ± 1.26
Decompensation	19 (8.22)	7 (13.4)	5 (11.62)	4 (5.3)	3 (5)
Death	5 (2.17)	2 (3.8)	2 (4.6)	0	1 (1.6)

ArLD: Alcohol related liver disease; HCV: Hepatitis C virus; HBV: Hepatitis B virus; MASLD: Metabolic dysfunction-associated steatotic liver disease; AST: Aspartate aminotransferase; ALT: Alanine aminotransferase; ALP: Alkaline phosphatase; INR: International normalized ratio; SSM: Splenic stiffness measurement; LSM: Liver stiffness measurement; FIB-4: Fibrosis-4 index; APRI: Aspartate aminotransferase to platelet ratio index.

One-year frequency of decompensation and predictors of decompensation

A one-year decompensation was observed in 19 patients (8.26%). The median time to develop decompensation was 10 (IQR: 7-11) months. The most common decompensation was ascites [16 (84.2%)], followed by HE [11 (5.7%)] and AVB [8 (4.2%)] (Table 1). In laboratory investigations, the patient who decompensated had a lower platelet count (0.74 ± 0.16/mm³vs 1.15 ± 0.44/mm³; P < 0.001) and a higher international normalized ratio (INR) (1.24 ± 0.17 vs 1.11 ± 0.19; P = 0.002) (Table 2). The baseline mean SSM, LSM, APRI, and FIB-4 were higher in decompensated group (69.61 ± 16.27 kPa vs 43.99 ± 14.81 kPa, P < 0.001, 30.77 ± 11.26 kPa vs 22.68 ± 7.02 kPa, P < 0.001; 2.24 ± 0.95 vs 1.68 ± 1.16, P = 0.022; 5.97 ± 2.31 vs 4.28 ± 2.52, P = 0.020 respectively) (Table 3).

Table 2 Comparison of baseline laboratory parameters in patients with decompensation as compared to patients who did not have any decompensation within the follow-up period, mean ± SD.

Parameters	No decompensation (n = 212)	Decompensation (n = 19)	P value
Haemoglobin	11.73 ± 1.84	11.25 ± 2.17	0.245
Total leukocyte count	5117.77 ± 1606.44	4727.5 ± 1748.51	0.266
Platelet	1.15 ± 0.44	0.74 ± 0.16	< 0.001
Albumin	4.12 ± 2.81	3.41 ± 0.22	0.218
Protein	8.39 ± 6.03	7.68 ± 5.36	0.132
INR	1.11 ± 0.19	1.24 ± 0.17	0.002
Total bilirubin	0.99 ± 0.54	1.22 ± 0.59	0.054
Conjugated bilirubin	0.6 ± 0.34	0.69 ± 0.35	0.226
AST	64.73 ± 31.15	62.75 ± 19.48	0.761
ALT	49.1 ± 26.18	49.92 ± 25.43	0.884
ALP	124.3 ± 32.7	127 ± 37.4	0.971
Urea	28.66 ± 9.24	30.08 ± 9.67	0.477
Creatinine	0.88 ± 0.19	0.95 ± 0.22	0.086

INR: International normalized ratio; AST: Aspartate aminotransferase; ALT: Alanine aminotransferase; ALP: Alkaline phosphatase.

Table 3 Differences in baseline non-invasive parameters in patients who developed decompensation, mean ± SD.

Parameters	No decompensation (n = 212)	Decompensation (n = 19)	P value
LSM (kPa)	22.68 ± 7.02	30.77 ± 11.26	< 0.001
SSM (kPa)	43.99 ± 14.81	69.61 ± 16.27	< 0.001
FIB-4	4.28 ± 2.52	5.97 ± 2.31	0.002
APRI	1.68 ± 1.16	2.24 ± 0.95	0.022
Spleen size (cm)	13.41 ± 1.37	14.59 ± 1.02	< 0.001
Liver size (cm)	12.31 ± 1.14	12.2 ± 1.57	0.676
PV diameter (mm)	12.83 ± 0.82	13.11 ± 0.86	0.123

LSM: Liver stiffness measurement; SSM: Splenic stiffness measurement; FIB-4: Fibrosis-4 index; APRI: Aspartate aminotransferase to platelet ratio index; PV: Portal vein.

On Cox proportional HR model, SSM [HR: 1.073 (1.049-1.097), P < 0.001], LSM [HR: 1.084 (1.047-1.122), P < 0.001], lower platelet count [HR: 0.056 (0.056-0.014) P < 0.001], INR [1.045 (1.34-1.57); P < 0.001], FIB-4 [HR: 1.203 (1.065-1.359), P = 0.003] and APRI [HR: 1.357 (1.035-1.778), P = 0.027] predicted one-year decompensation (Table 4).

Table 4 Predictors of decompensation based on the Cox proportional hazard ratio model and adjusted hazard ratio based on beta-blocker therapy.

Parameters	HR (95%CI)	Adjusted HR	P value
SSM	1.073 (1.049-1.097)	0.54 (0.28-1.62)	< 0.001
LSM	1.084 (1.047-1.122)	0.97 (0.14-1.67)	< 0.001
Platelet	0.056 (0.056-0.014)	-	< 0.001
INR	1.045 (1.34-1.57)	-	< 0.001
FIB-4	1.203 (1.065-1.359)	-	0.003
APRI	1.357 (1.035-1.778)	-	0.027

HR: Hazard ratio; CI: Confidence interval; LSM: Liver stiffness measurement; SSM: Splenic stiffness measurement; INR: International normalized ratio; FIB-4: Fibrosis-4 index; APRI: Aspartate aminotransferase to platelet ratio index.

Sensitivity and specificity of SSM and other parameters to predict decompensation

With a cut-off 57.5 kPa, when compared to LSM, APRI, and FIB-4, SSM has the highest sensitivity (82%), specificity (87.5%), positive predictive value (PPV) (98.26%), negative predictive value (NPV) (36.21%) and diagnostic accuracy (82.61%) for predicting decompensation in patients with c-ACLD as compared to LSM (cut-off 27 kPa) (78.16%, 62.5%, 94.7%, 25% and 76.5%) respectively and other non-invasive markers (Table 5; Figure 2). APRI and FIB-4 had an excellent PPV in predicting decompensation (96.06% and 97%, respectively). On Kaplan-Meier survival analysis, SSM > 57.5 kPa was associated with lower decompensation-free survival (P < 0.001) (Figure 3A).

Figure 2 Receiver operating curve comparing the diagnostic utility of splenic stiffness with other parameters for predicting one-year decompensation. SIP: SSM-INR-Platelet ratio; SSM: Splenic stiffness measurement; FIB-4: Fibrosis-4 index; APRI: AST to Platelet ratio index; LSM: Liver stiffness measurement.

Figure 3 Kaplan-Meier survival graph comparing 1-year decompensation-free and overall survival with a splenic stiffness measurement cut-off > 57.5 kPa. Censored observations are indicated by “+”. Log-rank test was used to compare survival distributions. A: Decompensation-free survival; B: Overall survival. SSM: Spleen stiffness measurement.

Table 5 Sensitivity, specificity, positive predictive value, negative predictive value, and diagnostic accuracy of splenic stiffness measurement compared to other parameters.

Parameters	SSM > 57.5 kPa	LSM > 27 kPa	APRI > 1.6	FIB-4 > 4.01	SIP score
Sensitivity	82.04	78.16	59.22	61	87.5%
Specificity	87.50	62.50	79.17	82	82.5%
Area under the curve	0.848	0.703	0.70	0.76	0.89
Positive predictive value	98.26	94.70	96.06	97	36.8%
Negative predictive value	36.21	25.00	18.45	21.5	98.3%
Diagnostic accuracy	82.61	76.50	61.30	64	83.0

SSM: Splenic stiffness measurement; LSM: Liver stiffness measurement; APRI: Aspartate aminotransferase to platelet ratio index; FIB-4: Fibrosis-4 index; SIP: Splenic stiffness measurement-international normalized ratio-platelet score.

Composite score to predict decompensation: SSM-INR-platelet score

Based on the predictive value of SSM, platelets, and INR, we have devised an SSM-based decompensation predictive model, SSM-INR-platelet (SIP) score. Total score ranges from 0-8. A score of 0-2, 3-5 and 6-8 signifies mild, moderate and severe risk of decompensation (Table 6). It demonstrated excellent discrimination for predicting decompensation [AUC: 0.89; 95% confidence interval (CI): 0.819-0.950] with sensitivity, and specificity of 87.5%, and 82.5% respectively. It performed better than LSM, SSM, APRI and FIB-4 alone (Table 5 and Figure 2). The estimated risk of decompensation in the low (0-2), moderate (3-5) and high (6-8) risk group was < 10%, 10%-30% and 30%, respectively.

Table 6 Splenic stiffness measurement-based decompensation predictive model: Splenic stiffness measurement-international normalized ratio-platelet score and risk of decompensation.

Variable	Cut-off	Points
SSM	< 40 kPa	0
	40-59.9 kPa	2
	≥ 60 kPa	4
Platelets	≥ 150 × 109/L	0
	100-149 × 109/L	1
	< 100 × 109/L	2
INR	< 1.20	0
	1.20-1.49	1
	≥ 1.50	2
SIP score	Total score	0-8
Score	Risk stratification	Estimated risk
0-2	Low	< 10%
3-5	Moderate	10%-30%
6-8	High	> 30%

SSM: Splenic stiffness measurement; INR: International normalized ratio; SIP: Splenic stiffness measurement-international normalized ratio-platelet.

SSM as a predictor of death among patients with c-ACLD

Five (2.17%) of 231 patients died during the follow-up period. Three patients died of liver-related decompensation and sepsis. Two patients had cardiac death with suspected acute coronary syndrome. SSM and other non-invasive markers failed to predict death in our cohort with c-ACLD (Table 7). On Kaplan-Meier survival analysis, survival did not differ between patients with SSM > 57.5 kPa vs SSM < 57.5 kPa (Figure 3B).

Table 7 Predictors of death among patients with compensated advanced chronic liver diseases.

Parameters	HR (95%CI)	P value
SSM	1.021 (0.977- 1.067)	0.347
LSM	1.049 (0.964-1.142)	0.263
Platelet	0.361(0.043-3.026)	0.347
FIB-4	1.088 (0.569-2.081)	0.799
APRI	0.96 (0.69-1.334)	0.806

HR: Hazard ratio; CI: Confidence interval; SSM: Splenic stiffness measurement; LSM: Liver stiffness measurement; FIB-4: Fibrosis-4 index; APRI: Aspartate aminotransferase to platelet ratio index.

Predictor of decompensation across the different groups of c-ACLD

Of 19 patients, seven in MASLD (36.8%), five in ArLD (26.3%), four (21%) in HBV and three (15.7%) in the HCV group developed decompensation. SSM was found to be the best modality to predict decompensation in the HCV-c-ACLD group (AUC: 0.96). With a cut-off of 57.5 kPa, SSM performed better than LSM, APRI and FIB-4 in MASLD (AUC: 0.87), ArLD (AUC: 0.86), and HBV (AUC: 0.80) groups for the prediction of decompensation (Figure 4).

Figure 4 Receiving operating curve for splenic stiffness measurement > 57.5 kPa for predicting decompensation in different etiology groups. A: Hepatitis B virus-related compensated advanced chronic liver diseases (c-ACLD); B: Alcohol-related liver disease-related c-ACLD; C: Metabolic dysfunction-associated steatotic liver disease-related c-ACLD; D: Hepatitis C virus-related c-ACLD. AUC: Area under the curve; LSM: Liver stiffness measurement; SSM: Splenic stiffness measurement; FIB-4: Fibrosis-4 index; APRI: AST to Platelet ratio index; CI: Confidence interval.

DISCUSSION

Decompensation changes the trajectory of the natural history of patients with cirrhosis[11]. It is associated with higher mortality than compensated cirrhosis[12]. Our study highlighted the utility of SSM to predict decompensation in the natural history of c-ACLD. Furthermore, it has established the usefulness of SSM in various etiologies of c-ACLD. We have also devised the SIP score, which outperformed pre-existing modalities in predicting decompensation. Various cross-sectional studies have established the role of SSM in predicting varices in patients with c-ACLD, thus stratifying them[13,14]. Although LSM, APRI, and FIB-4 performed well in predicting decompensation, SSM outperformed all these parameters with excellent sensitivity, specificity, and diagnostic accuracy.

Hepatic venous pressure gradient (HVPG) measurement is the gold standard for assessing portal hypertension and predicting decompensation risk. However, HVPG requires specialized hepatic venous catheterization expertise and infrastructure not routinely available at all tertiary care centers in India and in many resource-limited settings globally. The primary objective of our study was to evaluate SSM as a non-invasive surrogate for HVPG-based risk stratification of patients with c-ACLD. The correlation between SSM and HVPG is well-established and endorsed by the Baveno VII consensus, which employs SSM > 50 kPa as one of the non-invasive criteria for diagnosing CSPH. Our SSM cut-off of 57.5 kPa is consistent with this threshold and identifies patients with high CSPH probability and elevated decompensation risk. Moreover, HPVG is an invasive test and not feasible in patients with c-ACLD. Hence, this study adds to the data on non-invasive tests for evaluation of c-ACLD.

Few studies have been published on use of SSM in this regard. A recent study has revealed similar results, establishing SSM as one of the promising tests to predict decompensation[15]. Of 242 patients with c-ACLD, 11.6% patients developed decompensation during a median follow-up period of 501.5 days. SSM > 50 Kpa had an AUC 0.823 (95%CI: 0.74-0.90). Another study by Rigamonti et al[16] recruited 114 patients with PBC. The probability of liver-related decompensation was higher (41% at 24 months) in patients with SSM > 40 kPa[16]. Similar results were obtained in a study by Karagiannakis et al[17]. In that study, SSM was independently associated with the probability of liver-related decompensation (HR: 1.063, 95%CI: 1.009-1.120; P = 0.021), with an AUROC of 0.710 (P = 0.003) for predicting one-year liver decompensation (the cut-off point of 37 kPa). Our AUC of 0.848 is superior to the AUROC of 0.710 reported by Karagiannakis et al[17] and comparable to other published cohorts, with better combined sensitivity (82%) and specificity (87.5%). Etiology-specific subgroup analyses confirm SSM as a robust predictor across all four c-ACLD etiologies, with the best performance in HCV-c-ACLD (AUC: 0.96). Recent data from India utilized SSM to predict high-risk oesophageal varices. A cut-off of 35 kPa has 95.6% sensitivity to predict high-risk oesophageal varices in that study[18]. However, the probability of decompensation in such patients was not evaluated. Our study is one of the 1^st-study from India to evaluate the role of SSM to predict decompensation in a large cohort of c-ACLD of different aetiology. SSM failed to predict mortality in our cohort. This can be explained by higher cardiovascular death in patients with c-ACLD[19,20]. The remaining 3 liver-related deaths occurred in the context of acute decompensation events, where outcome is driven by multifactorial organ failure rather than baseline stiffness values.

Stratifying patients with c-ACLD is of paramount importance, as treating the underlying aetiology and the addition of a beta-blocker prevents decompensation in such patients[21]. Ascites is the most common decompensating event, and the onset of ascites changes the natural course of the disease[1]. It is often associated with acute kidney injury and spontaneous bacterial peritonitis. Presence of other decompensation like HE and AVB is also associated with higher mortality in such patients[21]. Hence following stratification, strict titration of beta-blockers may prevent decompensation[22]. Baveno VII uses SSM and LSM to predict CSPH in patients with c-ACLD, although the grey zone exists where, despite having a CSPH, patients have lower SSM and LSM and vice versa[1,22-24].

An important aspect to be highlighted is the relatively low NPV of SSM at the 57.5 kPa cut-off (36.21%) in our study which reflects, in part, the low prevalence of decompensation in our cohort (8.26%). In low-prevalence settings, NPV is mathematically constrained even when sensitivity and specificity are high. Importantly, SSM still demonstrates the highest NPV among all evaluated markers (LSM: 25.0%, APRI: 18.45%, FIB-4: 21.5%), confirming its relative superiority as a non-invasive predictor. Clinically, an SSM below 57.5 kPa should not be interpreted as excluding decompensation risk, and patients below this threshold must continue to receive regular clinical surveillance, aetiology-specific treatment, and periodic reassessment of stiffness values. Conversely, the high positive predictive value (PPV: 98.26%) strongly supports the use of SSM > 57.5 kPa to identify high-risk patients warranting intensified surveillance and early preventive intervention (e.g., beta-blocker initiation in confirmed CSPH) (Figure 5).

Figure 5 Graphical abstract showing the final results of our study (splenic stiffness measurement to predict decompensation in patients with compensated advanced chronic liver decease. ArLD: Alcohol related liver disease; MASLD: Metabolic dysfunction-associated steatotic liver disease; AVB: Acute variceal bleeding; HBV: Hepatitis B virus; HCV: Hepatitis C virus; HE: Hepatic encephalopathy; AUC: Area under the curve; SSM: Splenic stiffness measurement; LSM: Liver stiffness measurement; FIB-4: Fibrosis-4 index; APRI: AST to Platelet ratio index.

Our study has a few limitations, as it is a single-centre study with a short follow-up period. No patients underwent HVPG; SSM could not be tested against the gold standard. A formal cost-effectiveness analysis was not within the primary scope of the current study as it was done free of cost. Moreover, SIP score demonstrated a superior predicative modality to predict decompensation, but a reliable multivariable model typically requires approximately 10 events per predictor variable (EPV ≥ 10), which would restrict a robust multivariable model to no more than two predictors in our cohort, thereby limiting its clinical utility. Hence, validation of this new SIP score warrants a large, multicentre study. Also, a formal health economic analysis as an important direction for future research. Despite the limitations, it included many patients with c-ACLD with known aetiology and established a cut-off for Indian patients at risk of decompensation. SSM has the potential to substantially reduce the cost burden of decompensation-related hospitalizations and invasive complication management. As compared to HVPG, SSM measurement offers an excellent alternative to risk-stratify such patients.

CONCLUSION

One-year decompensation rate among Indian patients with c-ACLD is 8.26%. SSM performed better than LSM to predict decompensation in patients with c-ACLD. An SSM cut-off of 57.5 kPa has a sensitivity and specificity of 82% and 87.5%, respectively, for predicting one-year decompensation in these patients.