Highly active antiretroviral therapy in resource-poor settings: the experience of Medecins Sans Frontieres

We describe the short-term results of highly active antiretroviral therapy (HAART) in seven projects in low and middle income countries. A total of 743 adults were included, and clinical, immunological and virological responses were analysed. At 6 months, outcomes were similar to those observed in western countries, and the probability of remaining on treatment was 94%. The challenge now is to extend access to HAART to the millions in urgent need.

Médecins Sans Frontières has initiated highly active antiretroviral therapy (HAART) programmes in various settings in resource-poor countries over the past 2 years. After adherence consultations to explain the treatment and its constraints, HAART was proposed to severely immunocompromised patients following the World Health Organization (WHO) recommendations [¹]. Only triple-therapy regimens were proposed, mainly including a non-nucleoside reverse-transcriptase inhibitor (NNRTI) as preference was given to the easiest scheme to follow, with the lowest pill burden. The CD4 cell count was measured on admission and every 6 months. The HIV viral load was systematically measured when possible (three projects). Antiretroviral supply relied on local market competition, including the use of quality-assured generics. Treatment was provided free to patients.

As of May 2002, 881 patients had started HAART in seven projects (Chiradzulu District Hospital, Malawi; Homa Bay District Hospital, Kenya; Khayelitsha Health Centers, South Africa; Institut de Recherche et Développement, Military Hospital, Yaoundé, Cameroon; PBN Sihanouk Hospital, Phnom Penh, Cambodia; Surin Provincial Hospital, Thailand; and Roosevelt Hospital, Guatemala City, Guatemala). This first analysis aimed at describing the inclusions and short-term outcomes after compiling data from all the projects.

All patients aged 14 years or more, followed under HAART, were included in the analysis. Individual data systematically collected were sex, age, treatment background, dates of starting HAART, last visit or death, WHO clinical stage and regimen prescribed at baseline, CD4 cell count and HIV viral load at baseline and at 6 months if available. Kaplan-Meier methods were used to estimate the probability of survival and of remaining on treatment. At 6 months, any change in weight and CD4 cell counts were analysed, as well as the proportion with undetectable viral loads. Pooled results are presented and the extreme ranges among projects.

A total of 743 adults were included (median age 33 years, 50.3% women); 61 children and 77 adults just beginning HAART were excluded. Patients were at an advanced stage at the start of therapy (Table 1). The median CD4 cell count on inclusion (n = 684) was 48 cells/mm³ (25-75th percentiles: 11-120). The median baseline viral load, available for 231 patients, was 132 000 HIV copies/ml (5.12 log₁₀ copies/ml; 25-75th percentiles: 4.56-5.52). Among the 432 patients for whom the information was available, 339 (78.5%) were antiretroviral naive. Overall, 620 patients (83.4%) initiated NNRTI-containing regimens and 123 (16.6%) protease inhibitor-containing regimens. The median period of observation was 4.0 months (25-75th percentiles: 1.7-6.9). During this period, 61 patients died (of whom 26 died within the first 30 days; 42.6%), 25 stopped treatment (for adherence problems or patient request), and 18 were lost to follow-up for 60 days or more (Table 1). Overall, 240 patients (32.3%) were on treatment for 6 months or more. The probability of survival at 6 months was estimated at 89.5% [95% confidence interval (CI) 86.8-92.1]. Among those surviving, the probability of remaining on treatment at 6 months was estimated at 94.0% (95% CI 91.8-96.1). Among 200 patients controlled at 6 months, the median increase in the CD4 cell count was 104 cells/mm³ (25-75th percentiles: 47-163) (Table 1). Among 118 patients controlled for viral load at 6 months, 106 (89.8%) were undetectable (below 500 copies/ml). A total of 110 patients (14.8%) changed regimen. The reasons were systematically assessed in four projects (56 patients switching): 39 (69.6%) for intolerance [25 zidovudine (24 anaemia, one myopathy), six indinavir/ritonavir (two urinary lithiasis, two gastrointestinal disorders, four hyperlipidemia), five nevirapine (three rash, two hepatitis), two efavirenz (dizziness), one didanosine and one stavudine (both for neuropathy)]; 17 (30.4%) for other reasons (tuberculosis treatment in 15 cases).

Table 1

These preliminary results show the major benefits brought by HAART in resource-poor settings, even for patients at an advanced stage of AIDS. Patients show regular attendance, and immunological and virological responses indirectly indicate a high level of individual drug adherence. This first pooled description also highlights differences between projects in baseline characteristics and, consequently, outcomes (Table 1). A comparative analysis of prognosis factors is limited by the short follow-up period. Opportunistic infections occurring under HAART were not described. The main problems encountered in our experience are the occurrence of tuberculosis in Africa and Asia, of severe fungal infections in Guatemala and the persistence of Kaposi's sarcoma in Africa.

Currently, 6 million people in developing countries are in need of treatment. Regardless of the need for operational research, our results support the belief that treating severely immunocompromised AIDS patients is feasible in various settings, including in peripheral health facilities, as has also recently been reported by Ugandan and Senegalese national initiatives [^2,3]. Technical issues need to be addressed urgently in order to scale up access to HAART. Greater access to HAART will be facilitated by the ongoing price reduction of antiretroviral drugs. Triple therapy is currently available at US$361 per year from quality generic suppliers [⁴]. This price should come down to as low as US$50 per year. New fixed-drug combinations (in one pill) are also needed to improve patients' adherence and the management of antiretroviral therapy in peripheral health structures. Besides treatment protocols, monitoring should be simplified and adapted to each context [^5,6]. Pilot projects using clinical and biological proxy markers instead of CD4 cell counts and viral loads should be developed as a priority, together with the reinforcement of health structures to ensure access to testing and the availability of trained staff.

Funding HAART programmes requires a huge international effort to meet the growing political commitment in affected countries. As of July 2002, the Global Fund to Fight AIDS, Tuberculosis and Malaria effectively received only US$616 million for 2 years compared with the estimated US$10 billion per year needed to fight AIDS alone. Above all, increasing access to treatment is a challenge requiring strong political and financial commitments from both developed and developing countries.