Higher ELF Scores Linked to Cognitive Decline Over 11 Years
Key Takeaways
- Among Finnish adults in nationally representative population surveys, high ELF was associated with lower baseline global cognition and poorer wordlist learning after full adjustment.
- High ELF was also associated with slower simple and multiple-choice reaction times at baseline.
- Higher baseline ELF was associated with greater declines in global cognition and wordlist learning over follow-up after full adjustment.
- Continuous baseline ELF was associated with change in wordlist learning but not change in global cognition.
The baseline analysis included 5,115 adults; the longitudinal analysis included 2,490 across approximately 11 years. Exclusions included alcohol intake above sex-specific limits, missing examination or cognitive assessments or enhanced liver fibrosis (ELF) measurements, and insufficient fasting. ELF combines serum hyaluronic acid, procollagen type III amino terminal propeptide, and tissue inhibitor of metalloproteinases 1; high and low scores were defined as ELF ≥9.8 and <9.8. Investigators separately assessed continuous baseline ELF and mean ELF across baseline and follow-up. Verbal fluency, wordlist learning, delayed wordlist recall, and a global cognitive composite were measured at both visits; simple and multiple-choice reaction times were measured only at baseline. Progressively adjusted regressions accounted for demographics, apolipoprotein E ε4 status, metabolic and cardiovascular conditions, body mass index, smoking, depression, and physical activity. Change models also accounted for the corresponding baseline cognitive score.
In fully adjusted baseline comparisons, adults with high ELF had lower global cognitive scores and poorer wordlist learning than those with low ELF. Simple and multiple-choice reaction times were slower in the high-ELF group, while verbal fluency and delayed recall did not differ significantly. Continuous baseline ELF was not associated with global cognition or other baseline outcomes after full adjustment; its initial association with wordlist learning did not persist.
Over follow-up, high baseline ELF was associated with greater declines in global cognition and wordlist learning after full adjustment, including adjustment for the corresponding baseline test score. An association with delayed-recall decline did not persist after full adjustment. Continuous baseline ELF remained associated with wordlist-learning change across adjustment models, but not global-score change. Mean ELF across the two visits was separately associated with poorer wordlist-learning performance at follow-up and greater wordlist-learning decline; its association with global-score change diminished with fuller adjustment.
Serum ELF is a nondiagnostic indicator of possible fibrosis; fibrosis was not confirmed by elastography or biopsy, and false positives are possible. Residual confounding and selection from follow-up nonparticipation or missing cognitive assessments may affect these Finnish findings. The investigators also noted possible overfitting: only 148 participants had high ELF in the fully adjusted longitudinal analysis. Reaction times were not repeated, so change in reaction time could not be assessed. The associations establish neither a causal effect of fibrosis on cognition nor a validated cognitive-screening cutoff.
Vataja and colleagues concluded that an ELF-derived fibrosis indicator was longitudinally associated with performance in domains involving verbal learning, attention, and working memory in Finnish adults.
Clinician Questions
What alcohol intake excluded participants from the Finnish ELF and cognition analyses?
In the Finnish Health 2000 baseline population examined for enhanced liver fibrosis (ELF) and cognition, investigators excluded women drinking more than 20 g of alcohol per day and men drinking more than 30 g per day. These were participation exclusions, not ELF or cognitive thresholds.
How was the global cognitive score calculated for Finnish adults assessed for ELF and cognition?
Investigators standardized verbal fluency, wordlist learning, and delayed wordlist recall against the baseline population’s mean and standard deviation. The global cognitive z-score was the mean of those standardized scores. Cognitive change was calculated as follow-up score minus baseline score, so a lower change indicated greater decline.