 |
|
|
| |
FIB-4 Alone Proves Best Predictor of ESLD in People With HCV/HIV
|
| |
| |
CROI 2014, March 3-6, 2014, Boston
Mark Mascolini
In people with treated HIV infection but untreated HCV, FIB-4 alone proved a better predictor of end-stage liver disease (ESLD) than various predictive models incorporating other baseline variables [1]. Researchers analyzing the Veterans Aging Cohort Study Virtual Cohort (VACS-VC) suggested that FIB-4 may be useful in deciding on treatment for HCV/HIV-infected people.
Noting the lack of a model to predict ESLD in coinfected people, the VACS-VC investigators proposed that a reliable model could help clinicians decide which patients should begin therapy now or defer therapy until appropriate direct-acting antivirals (DAAs) become available. In places where DAAs might not be readily available for some time, a reliable model could help clinicians judge the risks and benefits of starting interferon-based therapy.
With those possibilities in mind, the researchers conducted a retrospective cohort study using 1997-2010 data from the VACS-VC. Included patients had detectable HCV RNA and HIV infection treated for at least 12 months in the VACS-VC. The researchers excluded people who already had ESLD or who received HCV therapy before follow-up, which began after 12 months of antiretroviral therapy in the VA system and continued until ESLD, death, HCV treatment, or December 31, 2010.
The VACS-VC team defined ESLD as hepatic decompensation, hepatocellular carcinoma, or liver-related death, whichever came first. They considered the following baseline predictors: FIB-4 (a hepatic fibrosis index calculated by age, platelet count, and alanine and aspartate aminotransferase levels*), anemia (hemoglobin below 10 g/dL), CD4 count below 200, HIV load above 400 copies, diabetes, hepatitis B, nonblack race, and obesity.
The researchers used Cox regression to evaluate factors associated with ESLD through 1 to 5 years of follow-up. They constructed predictive models with FIB-4 alone and FIB-4 plus other variables, either starting HCV therapy in all patients or treating none. For each model, the investigators plotted (1) net benefit (relative harm of false-negative versus false-positive classifications) on the y axis and (2) threshold probability (predicted risk of ESLD that prompts HCV therapy) on the x axis.
The analysis included 4280 coinfected people on ART with a median age of 48. One third (34.9%) were nonblack, 7.4% had diabetes, 11.1% were obese, 2.5% had anemia, 4.4% were hepatitis B surface antigen-positive, 27.1% had a CD4 count below 200, and 63.4% had an HIV load above 400 copies. Of the 4280 participants, 646 (15.1%) had a FIB-4 score above 3.5. Through a median follow-up of 6.8 years, 373 people (8.7%) reached ESLD, including hepatic decompensation in 72.7%, liver-related death in 29.5%, and hepatocellular carcinoma in 19.8%.
A predictive model based on FIB-4 alone had the highest discriminatory ability for ESLD (c statistic 0.73). The FIB-4-only model performed better across decision-making thresholds than other prognostic models including FIB-4 and other variables. Three FIB-4 brackets could predict ESLD over periods of 1, 3, 5, 7, or 9 years: FIB-4 below 1.45, 1.45 to 3.25, and above 3.25. For those three brackets, ESLD risks were 0.1%, 0.6%, and 4% at 1 year; 2%, 5%, and 17% at 5 years; and 6%, 12%, and 29% at 9 years.
The researchers concluded that baseline "FIB-4 has prognostic value and might be useful for HCV treatment decisions" in antiretroviral-treated people with HCV and HIV.
*FIB-4 = (age x AST)/(platelets x ALT[1/2])
Reference
1. Lo Re V, Tate JP, Butt AA, et al. Predicting risk of ESLD in HIV/HCV patients for individualized HCV therapy decisions. CROI 2014. Conference on Retroviruses and Opportunistic Infections. March 3-6, 2014. Boston. Abstract 650.
|
| |
|
|
|
|
|
