MELD score predictive of pre-transplant mortality in HCV coinfected patients
1 April 2008. Related: Conference reports, Hepatitis coinfection, CROI 15 (Retrovirus) 2008.
Simon Collins, HIV i-Base
Aruna Subramanian from Johns Hopkins University looked at determining incidence, cause, and time to pre-transplant mortality in transplant candidates compared to HIV-negative patients in a prospective cohort study at 20 US sites, with particular reference to the MELD score.
The MELD score (Model for End Stage Liver Disease) incorporates creatinine, bilirubin and INR checked at the same visit. MELD is validated as predictor of mortality in HIV-negative patients. It is used as a basis for organ allocation, so that sick patients get earlier access to transplant.
Patient in this study needed to fulfil local criteria to be included on a transplant list, with CD4 count of >100 cells/mm3 within 16 weeks of transplant (>200 if a recent OI), and to have undetectable viral load (except in cases when ART was discontinued due to hepatotoxicity, and a resistance profile indicated that HIV suppression post transplant would be likely. Clinical follow-up was at least every three months from joining the list until transplant.
Each case was matched (by age, gender, race, time of listing and HCV coinfection) with up to five controls, and compared by time to death, transplant and reaching MELD >25.
During follow-up the cohort included 167 HIV-positive patients (51% were not transplanted, 14% died and 35% received a transplant) and 792 controls (41% not transplanted, 11% died and 48% transplanted).
Median baseline CD4 was lower in patients who died compared to those who received a transplant (median 237 vs 315, p=0.01). There was no difference in the percentage with undetectable viral load, use of PI-based treatment or percent with HCV coinfection.
Cause of death pre-transplant were broadly similar in the HIV-positive vs control group, including sepsis (25% vs 20%), multi-organ failure (17% vs 26%), GI haemorrhage (13% vs 6%), other causes (29% vs 27%), and unknown (17% vs 20%).
Comparative time to death was similar in cases and controls, as was time to transplantation and to elevated MELD >25.
However, in multivariate model baseline MELD score showed the strongest risk (HR=21.8 95% CI 6.3, 75.7, p<0.0001). CD4 count <200 had only borderline significance (HR 2.6, 95%CI 0.98, 6.9, p=0.05), and viral load was not predictive.
The researchers concluded that low CD4 count at time of listing may be predictive of greater risk of death, but that after controlling for CD4 and viral load. MELD had excellent predictive value for pre-transplant mortality, and should be used routinely for patients with cirrhosis to help guide decisions for early transplant referral.
The group plans to develop a scoring method that incorporates CD4 count and MELD to predict mortality that could be validated for all patients, not just at transplant listing, and to determine optimum CD4 count for transplantation, and to determine any relationship between MELD score and post-transplant outcomes (which limited data indicate may be poorer in coinfected patients).
An online MELD calculator is available:
http://www.unos.org/resources/MeldPeldCalculator.asp?index=98
Transplant study for people with HIV
Comment
These results should not be a surprise as MELD is well validated for assessing liver failure. MELD is used by European and US transplant centres, The post-transplant data are very important, because MELD score at the time of transplant may be an accurate indicator of post-transplant survival.
There are clearly concerns that standard criteria for listing urgency may not apply for HIV-positive patients due to faster risk of progression and re-thinking listing priorities in this group of patients may be important. It is re-assuring that standard MELD criteria still apply.
Ref Subramanian A et al MELD is the best predictor of pre-transplant mortality in HIV-infected liver transplant candidates. Oral abstract 64.
http://www.retroconference.org/2008/Abstracts/31927.htm
This oral presentation is available to view online from the conference website (Monday 4 February).