Collagen proportionate area predicts long-term, but not short-term mortality in alcoholic hepatitis

04 Oct 2021
Collagen proportionate area predicts long-term, but not short-term mortality in alcoholic hepatitis
Alcoholism in Malaysia: A serious public health issue

Collagen proportionate area (CPA) is a good predictor of long-term mortality in alcoholic hepatitis (AH) patients regardless of their abstinence from alcohol, a recent study has found. However, CPA seems to have no clear interaction with short-term mortality.

Researchers enrolled 140 patients who had biopsy-verified AH, retrospectively collecting relevant data such as clinical, laboratory, and outcome measures. CPA was measured using standard magnification on the paraffin-embedded liver biopsies. Five AH prognostic scores were used: Maddrey’s Discriminant Function (DF), model of end-stage liver disease (MELD), Glasgow AH score (GAHS), age-bilirubin-INR-creatinine score (ABIC), and Lille Model.

Seventy-five percent of participants had a Maddrey’s DF score ≥32, while 72 percent had MELD-score ≥21; both prognostic tools showed high concordance with each other (r, 0.67). On the other hand, GAHS score ≥9 and ABIC score >9 was reported in 38 percent and 21 percent, respectively, while seventh day Lille score >0.45 was observed in 27 percent.

CPA showed good correlation with the Maddrey’s DF (r, 0.32) and MELD (r, 0.26) scores. CPA was also significantly elevated in patients with poor prognosis as defined by either prognostic tool. In contrast, GAHS, ABIC, and Lille Model were all unrelated to CPA.

Sixty-seven patients (48 percent) died after a median follow-up of 66 months. Multivariable Cox regression analysis accounting for all prognostic risk scores found that higher CPA was a significant and independent predictor of long-term mortality (hazard ratio [HR], 95 percent confidence interval [CI], 1.01–1.05; p=0.003). Alcohol abstinence, on the other hand, exerted a strong protective effect (HR, 0.28, 95 percent CI, 0.13–0.58; p=0.001).

However, CPA could not predict 90-day mortality risk (HR, 1.01, 95 percent CI, 0.98–1.05; p=0.506).

Editor's Recommendations