Non-invasive and clinical aspects of individuals at risk for steatotic liver disease

Introduction: Non-alcoholic fatty liver disease (NAFLD) represents an increasingly massive burden on our healthcare system. The disease is a spectrum of varying degrees of severity, and people who have risk factors (e.g., severe obesity or type 2 diabetes) face an enhanced risk of the disease progressing into non-alcoholic steatohepatitis (NASH) and even cirrhosis. In cirrhosis, the liver gradually becomes severely impaired with scar tissue and steadily loses its vital functions. Today, NASH can only be diagnosed with a liver biopsy, which can result in adverse complications. Therefore, there is a need for methods to detect significant fatty liver disease in patients with risk factors without performing a biopsy, and to be able to monitor disease activity.


Aim: This thesis aimed to explore non-invasive biomarkers to diagnose and monitor NASH. In addition, we wanted to investigate whether people with different degrees of fatty liver disease show impaired cognition. We have focused on investigating “The triggering receptor expressed on myeloid cell 2” (TREM-2), a marker on the surface of liver macrophages that has biological functions related to inflammation.


Methods: Several prospective clinical trials (Odense University Hospital and Esbjerg Hospital, 2016–2022) have contributed data to the studies and manuscripts included in this thesis. All participants gave consent to have several examinations carried out, including liver elastography, blood samples, and liver biopsy at the first visit (Papers 1 and 2 are cross-sectional studies). At a follow-up and final visits, the same examinations and tests were repeated (Paper 3, follow-up study). All participants included had metabolic risk factors. People with severe obesity, recruited from Esbjerg Hospital, had several cognitive tests performed during the baseline visit.


In Paper 1, we investigated the diagnostic ability of TREM2 to distinguish between NAS and no NASH. The Fibro-STARD guidelines was used.

In Paper 2, we investigated the prevalence of impaired cognition and the predictors associated with impaired cognitive function. All included participants had a round of basic tests (CRT, PSE, Stroop test), and a representative sub-group had an additional extensive RBANS test. The STROBE guidelines was used.

In Paper 3, we investigated the ability of several non-invasive tests (NITs) to reflect a change of histologic NAFLD Activity Score (NAS). Measurements: sTREM2, collagen markers PRO-C3, PRO-C4, PRO-C6, PRO-C8, and PRO-C18L, liver stiffness using Fibroscan, FAST-score, and HOMA-IR.


Results: In Paper 1, we included 48 + 170 people (derivation and validation study). All had an elevated liver stiffness ≥8 kPa (measured by elastography), a liver biopsy, and no excessive alcohol consumption. The level of TREM2 in plasma was 2.1 times higher in subjects with NAS≥4 and showed high diagnostic accuracy in the validation group with an AUROC of 0.83 (95% CI: 0.77–0.89, p <0.0001).

In Paper 2, we examine 180 people with severe obesity (72% women, 54% fatty liver (NAFL), and 24% NASH). Cognition was impaired in 20%, but not more frequently in people with NAFL or NASH. In the sub-group, the RBANS test (a far more comprehensive test) found that 40% of tested persons showed impaired cognition. Factors associated with impaired cognition were: male sex (OR 3.67, 95% CI 1.32– 10.27); low LDL (0.59, 95% CI 0.37–0.96); and taking at least two psychoactive drugs (5.24, 95% CI 1.34–20.4), and only male sex and LDL was associated in the RBANS sub-group. TREM2 levels were lower in subjects with an abnormal RBANS test (p=0.015) but were not associated with cognitive impairment in multivariate logistic regression analysis.

In Paper 3, 173 participants were included, mean age of 52 years (±12), 38% males, 70% had low fibrosis (F0-F1), 29% had a NAS above 4 (n=51), and 23% had NASH (n=39) at baseline. A stepwise correlation of NAS improvement was seen for TREM2, PRO-C3, and FAST scores (p=0.0001). A regression model with sTREM2 and PRO-C3 could predict NAS improvement (AUROC 0.75) with an OR 1.13 for every unit decrease (p=0.001, 95% CI 1.04–1.21). HOMA-IR performed well (AUROC 0.76, OR 1.22 p<0.001), while FIB-4 and NFS did not (AUROC < 0.60, OR <1.05 p>0.5).


Conclusion: This combined work has found that TREM2 is a reliable diagnostic biomarker for NASH and has good accuracy. Cognitive impairment appears to be frequent in people with severe obesity, but it is not associated with either NAFLD severity or NASH. sTREM2 was not elevated in patients with cognitive impairment, but was lower, however did not predict cognitive impairment in regression analysis. In patients with metabolic comorbidity and moderate inflammation and fibrosis, a combination of sTREM2 and PRO-C3 reflects NAS improvement and is a potential surrogate marker of NAS improvement. FIB4 and NFS showed low accuracy of histologic response in our cohort. 

TREM2 needs validation in external cohorts. Whether this cognitive impairment affects persons in their daily life should be investigated, preferably with patients as research partners.