Severe anaemia or a body mass index below 18 were good predictors of a very high risk of developing AIDS or dying within the following three months among Asian patients receiving antiretroviral therapy, and could be used in the absence of CD4 cell counts to identify patients in need of close attention after starting treatment, researchers from South-East Asia and China report in the journal Clinical Infectious Diseases.
Male patients aged 40 or below, and those with mild anaemia were at high risk of AIDS or death.
The risk equations were devised by looking at the risk of disease progression in HIV-positive patients receiving antiretroviral therapy in clinics that form part of the TREAT Asia network.
Risk equations identifying people infected with HIV at high risk for disease progression are in use in the developed world allowing clinicians to intervene.
Based primarily on populations in the developed world and dependent upon diagnostic tests not readily available in resource-limited countries their usefulness in this setting is questionable.
Patients were included from the TREAT Asia HIV Observational Database (TAHOD). The model is intended for extensive use in busy clinics in Asia and the Pacific region as well as other resource-limited settings.
The researchers evaluated the ability of three risk equations to predict progression to AIDS or death: a clinical model, a CD4-based model and a model using viral load and CD4 counts.
- The clinical model used clinical variables only;
- The CD4-based model used CD4 counts and the clinical criteria contained in the first model,
- The third model used CD4 cell counts, viral load measurements and clinical criteria.
The inclusion criteria for the analyses of the 200 patients recruited from each of the 17 sites of TAHOD were:
- Clinical model: following initiation of ≥ 3 antiretroviral agents, individuals with available data on demographics BMI, haemoglobin levels (development of anaemia) and alanine aminotransferase (ALT) (liver function)
- CD4 model: the preceding variables and CD4 cell counts and
- Viral load and CD4 model: the preceding two sets of variables and viral load measurements.
Follow-up began with the start of ART and ended with a diagnosis of a new AIDS defining-illness, death or last clinical visit if there was no progression.
Explanatory variables were included in univariate analyses and used to describe risk over the short-term.
Calculation of risk scores determined the risk categories of low, medium and high. The difference in patient risk was determined by the observed incidence rate of AIDS-defining illness or death within each category.
Of 3516 patients in TAHOD, 1679, 1663 and 1231 were eligible for inclusion in the clinical, CD4 cell count and CD4 cell count and viral load level models, respectively. In all three evaluation groups more than 80% of patients were over the age of 30 with approximately 70% male and over 70% having acquired HIV heterosexually.
The authors chose to develop simple patient risk-factor groups where calculation for, and identification of, those at short-term high risk was easy.
Calculation of full risk scores was neither practical in a busy clinical setting, nor would it allow for easy identification of those at medium or low risk.
Severe anaemia and very low BMI in all three models, as in other studies (for developing and developed world populations), were common predictive factors.
Contrary to other studies being of younger age and being male were associated with higher risk. The authors believe this is explained by better adherence to ART by female and older patients. For patients at the highest risk, that risk was greatest in the first three months following assessment.
In an accompanying editorial Amanda Mocroft and Jens Lundgren note that apart from the World Health Organization staging system for HIV disease, most staging systems in the developing world have been used for surveillance, whereas the TAHOD study represents one of the first prognostic staging systems for patients already on ART to be developed and applied in a resource-limited setting.
They note that while prognostic markers for short-term disease progression are reported, the mean frequencies of the clinical and laboratory assessments are not reported. This information would have been helpful in determining over which short-term period the predictions would be likely to apply.
Citing the difficulties inherent in simplifying the information of scoring systems and prognostic staging to find a model that is both statistically sound and easy to use by clinicians they acknowledge the need for a trade-off.
They also note the arbitrary cut-off values used. For example, a low BMI in the TAHOD study is
They conclude that the risk equations need to be tested in various populations before being used in clinical practice.
Srasuebkul P et al. Short-term clinical disease progression in HIV type-1-infected patients receiving combination antiretroviral therapy: results from the TREAT Asia HIV Observational Database. Clin Infect Dis 48: 940-950, 2009.
Mocroft A and Lundgren JD Use of risk equations for predicting disease progression in HIV infection. Editorial commentary. Clin Infect Dis 48: 951-953, 2009.