Liver Stiffness (Fibroscan®) Is a Predictor of All-cause Mortality in People With Non-alcoholic Fatty Liver Disease

Michael Braude; Stuart Roberts; Ammar Majeed; John Lubel; Jirayut Prompen; Anouk Dev; William Sievert; Stephen Bloom; Paul Gow; William Kemp

Disclosures

Liver International. 2023;43(1):90-99.

Abstract and Introduction

Abstract

Background and Aims: Progressive liver fibrosis related to non-alcoholic fatty liver disease (NAFLD) is associated with all-cause and liver-related mortality. We assessed vibration-controlled transient elastography (VCTE) as a predictor of mortality.

Method: Data from patients who underwent VCTE for NAFLD at four large health services in Victoria, Australia between the years 2008 and 2019 were linked to state-wide data registries. Cause of death (COD) and predictors of all-cause mortality were subsequently analysed using descriptive statistics and Cox-proportional regression analysis.

Results: Of 7079 VCTE records submitted for data linkage, 6341 were matched via data registry linkage. There were 217 deaths over a 22 653 person-year follow-up. COD included malignancies other than hepatocellular carcinoma (HCC) (18.0%, n = 39), sepsis (16.1%, n = 35), decompensated liver disease (15.2%, n = 33), cardiac disease (15.2%, n = 33) and HCC 6.0% (n = 13). Controlled attenuation parameter (CAP) was not associated with mortality in univariable analysis (HR = 1.00, CI 1.0–1.0, p = .488). Increased liver stiffness measurement (LSM) (HR 1.02 per kiloPascal, CI 1.01–1.03, p < .001), Charlson comorbidity index (CCI) (HR 1.32 for each point, CI 1.27–1.38, p < .001) and age (HR 1.05 per annum, CI 1.03–1.07, p < .001) were each associated with higher rates of all-cause mortality in multivariable analysis. LSM ≥10 kPa suggestive of compensated advanced chronic liver disease (cACLD) was associated with mortality in multivariable analysis (HR 2.31, CI 1.73–3.09, p < .001).

Conclusion: VCTE LSM, in addition to age and CCI, is independently associated with increased all-cause mortality in a large cohort with NAFLD.

Introduction

Non-alcoholic fatty liver disease (NAFLD) is a highly prevalent disorder that is estimated to affect 25.4% of the global population.^[1] NAFLD is associated with a swathe of extra-hepatic conditions including cardiovascular disease (CVD), renal disease and gastrointestinal neoplasia including hepatocellular carcinoma (HCC) which are thought to be driven by heightened systemic inflammatory and pro-fibrotic pathways.^[2,3]

Progressive liver fibrosis stage is associated with increased liver-related and overall mortality in NAFLD.^[4–8] Progressive fibrosis is frequently asymptomatic until such time that complications of advanced disease manifest. It is hence of paramount importance to identify and stage NAFLD at a population level in order to determine those who are most at-risk for adverse outcomes and to introduce timely interventions that prevent liver and extrahepatic disease complications.^[9]

There is, in addition, an unmet need for an efficient and cost-effective method of risk stratification of NAFLD.^[10] The current gold standard for NAFLD staging is via liver biopsy histological assessment. However, liver biopsy is not practical as a screening test due to cost and potential serious adverse events.^[11] An emerging suite of non-invasive tests, some of which are more scalable than others, have therefore been developed as surrogates for liver histology. These include predictive serum-based algorithms, ultrasound-based elastography, magnetic resonance spectroscopy and magnetic resonance elastography.

Vibration-controlled transient elastography (VCTE) is a widely used non-invasive point-of-care test that provides a liver stiffness measurement (LSM) and also a measure of hepatic steatosis via the controlled attenuation parameter (CAP).^[12] VCTE has excellent diagnostic accuracy and inter-user reliability for detection of fibrosis, and furthermore delineates mild-to-severe fibrosis from cirrhosis, with a sensitivity of 75% and a specificity of 81.2%.^[13,14] Like other non-invasive tests (NITs), VCTE is most powerful in its negative predictive value and can rule out advanced fibrosis in 94.7% of cases.^[13] VCTE has been recommended by current clinical guidelines as a tool for risk stratification of people with suspected NAFLD and concurrent metabolic risk factors, such as obesity or T2DM,^[10,15] and has an increasing role in stratifying compensated advanced chronic liver disease (cACLD) and predicting clinically significant portal hypertension (CSPH).^[16]

Several studies have assessed VCTE as a predictor of mortality, and more recently, as a predictor of cardiovascular, liver and oncological morbidity.^[17–19] However, there is no guidelines-based recommendation for use of VCTE as a prognostication tool for mortality. This reflects in part the paucity of large cohort, real-world data evaluating VCTE as a prognostic tool. We, therefore, evaluated the capability of single time-point VCTE to predict mortality in a large cohort of Australian patients with NAFLD. We used linkage analysis to evaluate outcomes and comorbidity data, and in doing so, were able to adjust for the Charlson Comorbidity Index (CCI) which has been widely validated as a predictor of mortality following hospitalisation.^[20,21]

1 2 3 4 5

Next Section

Table 1. Demographics of population ( n = 7079)
Parameter	n (%)
Age, median (IQR)	60 (IQR 49–69)
Age at death (n = 217), median (IQR)	72.3 (64.4–79.0)
Gender, n = 6001 (%)
Male	2813 (46.9)
Female	3188 (53.1)
Country/region of birth (n = 4655)
Australia	2650 (56.9)
Europe	913 (19.6)
Asia (Southeast/east/central)	512 (11.0)
Polynesia (New Zealand, Fiji, Cook Islands, Samoa)	111 (2.4)
Indonesia, Malaysia, Philippines	135 (2.9)
Middle East	140 (3.0)
Africa	91 (2.0)
South and central America	71 (1.5)
North American	30 (.6)
Charlson comorbidities VAED (n = 5858)^a	Pre-VCTE²⁷
T2DM	1121 (19.1)
Liver disease	660 (11.3)
Cancer	326 (5.6)
CKD	218 (3.7)
Peripheral vascular disease	169 (2.9)
Congestive cardiac failure	159 (2.7)
Chronic ischaemic heart disease	146 (2.5)
Pulmonary disease	134 (2.3)
Stroke	98 (1.7)
Peptic ulcer disease	96 (1.6)
Connective tissue/rheumatological	52 (.9)
Hemiplegia	44 (.8)
HIV/AIDS	10 (.1)
Dementia	1 (.0)
BMI, median (IQR) (n = 2, 126)	30.6 (27.0–35.1)
Smoker (n = 1095)
Yes (current/former)	482 (44.0)
Never	613 (56.0)
ALT, median (IQR) (n = 1047)	41 (23–65)
CAP (n = 3786), median (IQR)^b,25	293 (240–339)
<302 db/m	2063 (54.5)
302–336 db/m	715 (18.9)
CAP ≥337 db/m	1008 (26.6)
LSM, median (IQR)	6.1 (4.7–9.4)
LSM category, n = 7079¹⁶
<5 kpa	2183 (30.8)
5–10 kpa	3272 (46.2)
10–15 kpa	739 (10.4)
15–20 kpa	275 (3.9)
20–25 kpa	206 (2.9)
>25 kPa	404 (5.7)

^aComorbidity data censored within 6-months of VCTE.
^bCAP ≥274 db/m, which has a 90% sensitivity for detecting steatosis, was present in 59.8% (2265/3786).

Table 2. Primary COD based on consolidated death (VDI) and admissions (VAED) datasets
Cause of death (COD)	n (%)	LSM (IQR)
Infection^a	35 (16.1)	12.0 (5.8–21.7)
Decompensated liver disease^b	33 (15.2)	32.2 (18.8–49.1)
HCC	13 (6.0)	22.3 (12.0–26.3)
Non-HCC malignancy
Non-GI malignancy	25 (11.5)	13.9 (6.2–23.9)
Non-HCC GI malignancy	14 (6.5)	7.5 (5.9–14.0)
Cardiac complications
Decompensated cardiac failure	19 (8.8)	9.6 (7.0–21.2)
Myocardial infarction	14 (6.5)	13.0 (7.2–24.5)
Stroke	12 (5.5)	10.8 (8.1–22.6)
Respiratory	9 (4.1)	14.7 (10.0–23.5)
Kidney injury (acute/chronic)	10 (4.6)	8.8 (5.0–24.1)
Vascular	10 (4.6)	13.8 (9.8–21.1)
Other^c	17 (7.8)	6.9 (4.9–21.6)
Nil documented COD	6 (2.8)	12.7 (7.1–30.9)

^aConsidered causal in cases where infection could not be clearly delineated as either caused by or resulting in comorbid system failure.
^bCo-morbid HCC in two of these cases.
^cIncludes post-operative complication, MVA, self-harm, neuromuscular/neurodegenerative disorders, GI bleeding.

Table 3. Cox-regression univariable and multivariable analysis of mortality risk ( n = 5857)
Parameter	Univariable analysis			Multivariable analysis^a
Parameter	HR	CI	p-value	HR	CI	p-value
Age	1.09	1.08–1.10	<.001	1.05	1.03–1.07	<.001
Gender (M)	1.35	1.03–1.77	.028	1.19	.91–1.56	.211
COB (Australia)	.80	.59–1.08	.144	—	—	—
CAP (n = 3840)	1.00	1.00–1.00	.488	—	—	—
LSM (kPa)^b	1.04	1.04–1.05	<.001	1.02	1.01–1.03	<.001
Ischaemic heart disease	2.57	1.43–4.60	.002	—	—	—
Congestive cardiac failure	8.69	5.93–12.75	<.001	—	—	—
Peripheral vascular disease	6.68	4.53–9.82	<.001	—	—	—
Stroke	2.73	1.36–5.54	.005	—	—	—
Pulmonary disease	5.63	3.55–8.93	<.001	—	—	—
Connective tissue disease	2.61	1.07–6.33	.034	—	—	—
Hemiplegia	7.45	3.51–15.96	<.001	—	—	—
CKD	8.34	6.03–11.52	.001	—	—	—
T2DM—uncomplicated	1.71	1.21–2.42	.002	—	—	—
T2DM—complicated	4.67	3.19–6.84	<.001	—	—	—
Cancer—nil metastatic	2.30	1.38–3.84	.002	—	—	—
Cancer—metastatic	19.45	12.31–30.72	<.001	—	—	—
Mild liver disease	1.30	.78–2.17	.30	—	—	—
Severe liver disease	5.62	4.13–7.64	<.001	—	—	—
Charlson comorbidity index (CCI)^c	1.44	1.41–1.49	<.001	1.32	1.27–1.38	<.001
CCI 0–2	.10	.07–.15	<.001	—	—	—
CCI 3–5	1.15	.87–1.52	.324	—	—	—
CCI 5–8	6.00	4.48–8.04	<.001	—	—	—
CCI >8	17.29	12.40–24.09	<.001	—	—	—

^aAdjusted for age, gender, liver stiffness and Charlson comorbidity index.
^bCCI subtracted for liver resulted in similar findings on multivariable ➔ HR 1.03 (CI 1.02–1.04).
^cAssessed as a continuous variable.