Since 2001, Médecins Sans Frontières (MSF) has provided antiretroviral therapy (ART) to more than 100 000 people in resource-limited settings (RLS), using World Health Organization (WHO)-recommended first-line regimens involving two nucleoside reverse transcriptase inhibitor (NRTI) and one non-nucleoside reverse transcriptase inhibitor (NNRTI) antiretroviral drugs. Satisfactory short-term outcomes in such settings have been described by MSF and others [1–4], and this contributed to the rapid scaling up of ART. As access to and time on ART increases, the need for second-line regimens in RLS becomes a priority due to the development of drug resistance [5,6]. Clinical programme data are urgently needed to help design longer-term treatment strategies, and to allow programme managers, governments, donors, and pharmaceutical manufacturers to more accurately forecast antiretroviral drug first-line and second-line requirements .
Detection of first-line ART failure in most RLS relies on immunological or clinical criteria, as routine viral load monitoring is frequently not available due to financial and technical constraints. Thus, diagnosis of failure is often delayed , and might favour not only clinical disease progression , but also development of antiretroviral resistance due to concurrent antiretroviral drug exposure and high viral replication. The appearance of resistance mutations [e.g. thymidine analogue mutations (TAMs)] may thus jeopardize the effectiveness of second-line ART regimens in settings where few treatment options are available . Additional constraints regarding second-line ART in RLS include availability of less robust second-line ART regimens (e.g., nonboosted protease inhibitors); difficulty in ensuring long-term adherence due to increased pill burden or meal restrictions; and high price of drugs (up to ten times more expensive than first-line regimens) . Therefore, concerns about the effectiveness of second-line ART in RLS have been raised, but few data on this exist [12,13].
We report here the rate of switch from first-line to second-line ART in MSF programmes; survival and clinico-immunological outcomes of patients on second-line ART; and factors contributing to death when on second-line ART.
We used routinely collected individual patient data (FUCHIA software, Epicentre, Paris) from 62 MSF-supported HIV programmes in 26 countries between October 2001 and December 2006. Data collected included sex; age; treatment history; ART prescription date and regimen; dates of visit, appointment, or death; WHO clinical stage; and CD4 cell count and viral load, when available.
All programmes provided free care, including antiretroviral drugs and laboratory investigations, with clinical consultation performed by doctors, clinical officers, or nurses. Eligibility criteria for ART were based on the 2003 WHO recommendations  and generic antiretroviral drugs used, mainly in the form of NNRTI-containing fixed-dose combinations. Adherence counseling was provided to all patients both prior to and during ART. Programmes routinely provided prophylaxis and treatment of opportunistic infections and often nutritional support for malnourished individuals. CD4 cell counts were measured using either automated or manual methods (Partec, Dynabeads, Beckton Dickinson). Although virological monitoring was not routinely performed, viral load was occasionally determined when clinicians suspected treatment failure and adequate laboratory facilities were available.
We analysed information from all ART-naive adults (> 15 years) at MSF programme inclusion and who received an NNRTI first-line regimen for more than 6 months. Missing age data was the only reason for exclusion from this study. To exclude antiretroviral drug substitutions for toxicity, we defined second-line therapy as a concomitant initiation of a protease inhibitor-containing regimen and a change in at least one NRTI drug in patients who had received NNRTI first-line therapy for more than 6 months. Women who had received prevention of mother-to-child transmission (PMTCT) prophylaxis were considered naive. Reasons for switch to second line were not prospectively collected. Treatment failure was defined as CD4 at switch less than CD4 cell count at ART initiation; CD4 cell count at switch less than 100 cells; new WHO stage 3 or 4 event within 3 months before switch; and/or viral load more than 1000 copies/ml, when available .
We estimated the probabilities of remaining on first-line ART, rate of switch to second-line ART, and probabilities of remaining alive and in care after first-line and second-line therapy initiation using Kaplan–Meier and censoring-naive methods. For patients on ART for more than 19.8 months (the median time of follow-up on ART before the switch to second line), we compared the probabilities of remaining alive and in care in patients started or not on second-line therapy with the log-rank test. We then described patient characteristics at the start of ART and at switch, and BMI and CD4 gains at 6, 12, and 24 months after switch, using medians, interquartile ranges (IQR), and percentages, as appropriate. Finally, we investigated factors associated with death or lost to follow-up (LFU; a missed appointment for > 2 months) using Poisson regression. To control for the heterogeneity of the data, as information from several projects was included (and access to CD4 testing and/or to diagnostic facilities for opportunistic infections vary in different contexts), all the models were adjusted for geography (sub-Saharan Africa, Asia, and Latin America) and for factors significantly associated with the outcomes (P value from likelihood ratio tests < 0.05).
Description of the global Médecins Sans Frontières cohort
After excluding 856 (1.7%) patients with unknown age, we analysed data from 48 338 naive adults on ART for more than 6 months, 78% treated in MSF-supported projects in Africa, 18% in Asia, 4% in Latin America, and 0.2% in Eastern Europe (Table 1). At the start of ART, patient median age was 35 years (IQR 30–42), and 62% were women. The median CD4 cell count was 110 cells/μl (IQR 46–178; n = 34 799), and 84% were classified as WHO clinical stage 3 or 4. The first-line ART regimen most frequently prescribed was stavudine/lamivudine/nelfinavir (d4T/3TC/NVP) (86%), and median duration of ART was 18 months (IQR 11–25), with 13 871 (29%) patients on treatment for more than 2 years. For patients on regular follow-up, the probability of remaining on first-line ART after 36 months was 0.98 (95% CI 0.97–0.98). The probability of remaining alive and in care at 24 months was 0.87.
Description of patients on second-line antiretroviral therapy
A total of 370 (0.8%) patients began a second-line regimen after a median of 20 months (IQR 14–27), corresponding to a switch rate from first to second line of 4.8/1000 person-years (95% CI 4.3–5.3) (Fig. 1) (Table 2). Switch rates ranged from 4.2/1000 person-years in sub-Saharan projects to 6.5/1000 in Asia, 7.4/1000 in Latin America, and 14.5/1000 in Eastern Europe. At the start of first-line ART, 334 (90%) patients were in cumulative WHO clinical stage 3 or 4, and had a median BMI of 20 kg/m2 (IQR 18–22; n = 321) and median CD4 cell count of 52 cells/μl (IQR 18–131; n = 304). Median CD4 nadir was 39 cells/μl (IQR 13–86; n = 368).
At the time of switch, median CD4 cell count was 99 cells/μl (IQR 39–200; n = 244), 32% having less than 50 cells/μl, 50% had less than 100 cells/μl, and 75% had less than 200 cells/μl. Median viral load was 43 188 copies/ml (IQR 16 406–166 197; n = 75), and median BMI was 21 kg/m2 (IQR 19–24; n = 320) with 45 (14%) patients having less than 17 kg/m2.
During the 3 months before the switch, a new WHO clinical stage 3 or 4 condition had been recorded for 111 (30%) patients and a CD4 value of less than or equal to CD4 cell count at the start of ART for 69 (34%). Among patients who were switched to second line, 230 (62%) had at least one of the criteria of treatment failure according to the 2006 WHO criteria. A total of 139 patients had no CD4 data collected and thus could not be evaluated with regard to immunological treatment failure criteria.
The protease inhibitor component of the second-line regimen was lopinavir/ritonavir (LPV/r) for 188 (51%) patients and nelfinavir (NFV) for 160 (43%); 56% received a boosted protease inhibitor. The most frequently administered NRTI combinations were zidovudine-didanosine (ZDV-ddI; 34%) and abacavir (ABC)-ddI (22%). Only one NRTI drug had been changed instead of two for 115 (31%) patients, and only 12% received ZDV-lamivudine (ZDV-3TC) in combination with a protease inhibitor.
Second-line treatment outcomes
Median follow-up on second-line ART was 8 months (IQR 2–18), with 138 (37%) patients on treatment for more than 12 months. Twenty-eight (8%) patients died after a median of 5 months (IQR 3–8), and 18 (5%) were LFU after a median of 9 months (IQR 3–14). Recorded causes of death were Kaposi sarcoma (n = 7), tuberculosis (n = 4), wasting syndrome (n = 3), and one suspicion of cerebral mass. The probabilities of remaining alive and in care at 12 and 24 months were 0.86 (95% CI: 0.81–0.90) and 0.77 (95% CI: 0.69–0.83), respectively (Fig. 2a)(Table 3). These figures did not differ by the number of NRTI drugs changed (P = 0.99) and were slightly, but not significantly, higher for patients on LPV/r-second-line therapy (unadjusted P = 0.06, compared with NFV-based therapy). The probability of remaining alive and in-care for patients with a follow-up of at least 19.8 months was similar in patients switched and not switched to second-line therapy at 20 months of ART, but it was lower for patients on second line after that (log-rank test P value 0.03; Fig. 2b and Table 4). However, latter estimates were based on data from few patients.
Median CD4 cell count was 184 cells/μl (IQR 128–306; n = 106) and 247 cells/μl (IQR 132–302; n = 78) at 6 and 12 months, respectively. Median CD4 increase was 90 (37–141; n = 73) and 135 (50–198; n = 55) at 6 and 12 months, respectively (Fig. 3). One year after switch, six (10.9%) patients had a CD4 cell count of less than 50 cells/μl, 14 (25.5%) had less than 100 cells/μl, and 30 (54.5%) had less than 200 cells/μl. Furthermore, median weight gains at 6 and 12 months after switch were 0.5 kg (IQR −2 to 3; n = 205) and 1 kg (IQR −2 to 4; n = 139), respectively. Only eight (6%; n = 127) patients had a BMI of less than 17 kg/m2 1 year after the switch. After 6 months of treatment, 46 (12.4%) patients developed a WHO condition stage 3 or 4 (14 at stage 4 condition), and the CD4 cell count was equal or below the value recorded at switch for 11% (8/75).
Antiretroviral drug-related toxicity leading to cessation or change of the antiretroviral drug regimen was recorded for only three (1%) patients: two on ZDV-ddI-NFV (neuropathy WHO grades 1 to 2 and lactic acidosis) and one on ZDV-ddI-IDV/r (hepatoxicity grade 4).
Factors associated with death and lost to follow-up
In multivariable analyses, death and LFU rates were higher in patients classified as WHO stage 4 at first-line ART initiation [incidence rate ratio (IRR) 2.35, 95% CI 1.29–4.31; P = 0.006], and in those with CD4 cell count nadir less than 50 cells/μl (IRR 1.73, 95% CI 0.91–3.29; P = 0.09) (Table 5). Interestingly, the number of NRTI drugs changed (P = 0.39), level of CD4 cell count at switch (P = 0.26), and type of protease inhibitor (boosted versus nonboosted; P = 0.31) were not significant predictors of survival in this analysis.
In this first published study of second-line ART in RLS, we have shown encouraging early treatment outcomes, with clinical and immunological outcomes similar to those published for first-line regimens in RLS [1,3,16]. Nine out of ten patients were still alive after 12 months of treatment, and few patients were diagnosed with new severe AIDS-related illnesses. However, over half of the patients were still at a significant risk of life-threatening opportunistic infections (CD4 < 200 cells/μl) after 12 months of treatment, showing that room for improvement exists even for second-line therapy.
We found that overall a relatively small proportion of patients, after at least 6 months on ART in MSF programmes, switched to a second-line regimen for treatment failure (switch rate 4.8/1000 person-years; 6% of cohort at 48 months). Switch rates to second-line therapy were lowest in sub-Saharan patients and highest in Eastern Europeans, probably reflecting differences in access to viral load and CD4 testing in those contexts (many MSF projects in Africa are based in rural areas). Our observed switch rate probably reflects only the most obvious cases of treatment failure, due to our inability to accurately diagnose when a first-line regimen should be changed to second line. This is also suggested by our finding of lower survival in patients switched to second-line therapy compared with those who did not, suggesting that late diagnosis of failure would increase the risk of death in patients started on second-line ART. In the absence of routine virological monitoring in our programmes, diagnosis of failure is usually based on either the occurrence of a clinical event, or on immunological criteria as per WHO guidelines .
These data from a large number of RLS highlight the increased need for second-line therapy in similar settings. Evidence of the efficacy of rescue regimens has been shown in studies conducted in resource-rich settings, where median CD4 cell count at second-line initiation is higher, viral load monitoring routinely performed, and treatment options readily available. In a European study  conducted before the widespread availability of genotyping, patients initiating a second protease inhibitor regimen at lower viral load with higher CD4 cell counts, or receiving additional nucleosides, were more likely to achieve undetectable viral loads. Studies in South Africa [18,19], where viral load is routinely measured and treatment failure is defined as two consecutive viral load measurements more than 5000 copies/ml, reported that after 36 months of NNRTI-based first-line therapy, 5.6 to 11.9% of patients switched to a second-line regimen.
Although the number of initially naive patients switched to second line in our study was small, in RLS there is an increasing need for second-line ART regimens as ART cohorts mature and access to virological monitoring increases . Detection of early treatment failure will ensure that patients are able to switch regimens before the occurrence of severe clinical events and will prevent unnecessary early switches. One major strategy for improving the diagnosis of first-line treatment failure in RLS is to increase viral load monitoring. In the absence of regular viral load monitoring, diagnosis of treatment failure might be delayed due to reliance on less sensitive immunological or clinical methods . Although not a RLS, our ART programme in Khayelitsha, South Africa, routinely measures patient viral load (and CD4 cell count) at baseline and 3 and 6 months after the start of treatment, as well as every 6 months thereafter . As costs and technological limitations decrease for viral load testing , its use in RLS could beneficially increase switching to second-line therapy while optimizing the duration of first-line regimens. Ideally, treatment-failure algorithms  could be designed based on diagnostic parameters, such as viral load, CD4 cell count, or haemoglobin levels, and incorporated into HIV treatment guidelines in the field.
In RLS, constraints at the treatment level also negatively affect therapy outcomes. Even in the presence of virological failure, clinicians working in RLS may be reluctant to change to second-line regimens, as shown in one of the before-mentioned South African studies . Evidence shows that clinical and immunological benefits can be obtained on a virologically failing regimen, but this effect has been demonstrated only in patients on a protease inhibitor-based regimen . Prices of second-line drugs have been reported to be about ten times higher than first-line agents . Also, concerns exist about the limited efficacy of available second-line regimens involving boosted and nonboosted protease inhibitors or the addition of a single new NRTI. Clinicians often also doubt the immediate benefit of second-line therapy because of the difficulties in ensuring patient treatment adherence due to the high pill burden of protease inhibitor-containing regimens, absence of fixed-dose drug combinations, need for refrigeration, and necessary meal restrictions. Finally, the fear that no further treatment options will be available if subsequent failure on second-line regimens occurs might also delay therapy switching .
In our study , more than half of the patients were put on NFV-based regimens, due to a lack of refrigeration systems and heat-stable boosted protease inhibitors, which are less effective than a regimen containing a ritonavir-boosted protease inhibitor. Also, the choices for replacing the NRTI drugs were limited due to a restricted formulary of drugs. Despite these constraints, our results showed that rescue following the failure of WHO-recommended first-line treatment (including the regimen d4T/3TC/NVP) is feasible and efficient in RLS, at least in MSF programmes. Therefore, in addition to improving the diagnosis of treatment failure, the obstacles to second-line drug access and usage must be addressed by reducing costs, increasing the availability of newer, more potent molecules (including heat-stable formulations of boosted protease inhibitors), and facilitating adherence through fixed-dose formulations that do not require food restrictions.
Less than 2% of patients eligible for the study were excluded from the analyses because of unknown age, and this percentage was similar across continents. We recognize, however, several limitations in our study. First, it was based on monitoring data from a multicentric observational cohort, and the reason for switch to second line was not recorded prospectively. Therefore, we cannot completely exclude that some of the switches in therapy were in fact antiretroviral drug replacements due to reasons other than treatment failure, such as drug toxicity. We are, however, confident that the majority of patients eligible for this analysis were true treatment failures, as 62% had at least one of the WHO criteria for treatment failure recorded, and our definition of second-line therapy (changes of both antiretroviral drug class and NRTI drug after more than 6 months of NNRTI therapy) is likely to have excluded most of the patients with antiretroviral drug replacements due to toxicity in the MSF context. Data on CD4 cell counts were not collected for the remaining patients. Second, clinico-immunological failure can occur in the presence of virological control , and, although clinical and immunological failure was confirmed by a viral load level in about 20% of our patients, we cannot exclude that some of our patients were not. Third, the length of follow-up was relatively short, and long-term monitoring of these patients is warranted. Despite these favourable early clinical and immunological outcomes, the absence of viral load measurements did not allow exclusion of suboptimal virological suppression that would lead to less satisfactory long-term outcomes. Finally, as we have included ART-naive patients in the analysis, the outcomes cannot be extrapolated to patients who might have received first-line regimens prior to entry in the MSF cohort.
Unsurprisingly, we showed that severe immunosuppression at baseline for first-line ART, and the history of a severe clinical event (WHO stage 3 or 4), increased the risk of mortality on second-line treatment. Our findings thus stress the need to enable access to first-line ART before severe immunosuppression has developed, reinforcing the need to scale up early access to HIV testing and treatment for those in RLS.
In summary, we report a relatively low rate of switch to second-line HIV treatment in ART-naive adults in MSF programmes in RLS, but good early outcomes on second-line therapy. Severe immunosuppression at first-line ART initiation increased mortality on second-line treatment. Considering the success of patients put on a second-line regimen, improving the tools to efficiently diagnose first-line treatment failure, and clearing the hurdles of access and adherence to more effective drug regimens, are critical actions that should be taken to allow more patients in RLS to benefit from second-line HIV therapy.
We thank the ministries of health of all the countries, the MSF field teams, and the patients, for providing the data for this analysis. Our thanks to the MSF AIDS Working Group for their comments and Oliver Yun for his editorial support on the manuscript.
1. Calmy A, Pinoges L, Szumilin E, Zachariah R, Ford N, Ferradini L. Medecins Sans Frontieres. Generic fixed-dose combination antiretroviral treatment in resource-poor settings: multicentric observational cohort. AIDS 2006; 20:1163–1169.
2. Laurent C, Kouanfack C, Koulla-Shiro S, Nkoué N, Bourgeois A, Calmy A, et al
. Effectiveness and safety of a generic fixed-dose combination of nevirapine, stavudine, and lamivudine in HIV-1-infected adults in Cameroon: open-label multicentre trial. Lancet 2004; 364:29–34.
3. Ivers LC, Kendrick D, Doucette K. Efficacy of antiretroviral therapy programs in resource-poor settings: a meta-analysis of the published literature. Clin Infect Dis 2005; 41:217–224.
4. Ferradini L, Jeannin A, Pinoges L, Izopet J, Odhiambo D, Mankhambo L, et al
. Scaling up of highly active antiretroviral therapy in a rural district of Malawi: an effectiveness assessment. Lancet 2006; 367:1335–1342.
5. Galárraga O, O'Brien ME, Gutiérrez JP, Renaud-Théry F, Nguimfack BD, Beusenberg M, et al
. Forecast of demand for antiretroviral drugs in low and middle-income countries: 2007–2008. AIDS 2007; 21(Suppl 4):S97–S103.
6. Boyd MA, Cooper DA. Second-line combination antiretroviral therapy in resource-limited settings: facing the challenges through clinical research. AIDS 2007; 21(Suppl 4):S55–S63.
7. Renaud-Théry F, Nguimfack BD, Vitoria M, Lee E, Graaff P, Samb B, Perriëns J. Use of antiretroviral therapy in resource-limited countries in 2006: distribution and uptake of first- and second-line regimens. AIDS 2007; 21(Suppl 4):S89–S95.
8. Chaiwarith R, Wachirakaphan C, Kotarathititum W, Praparatanaphan J, Sirisanthana T, Supparatpinyo K. Sensitivity and specificity of using CD4+ measurement and clinical evaluation to determine antiretroviral treatment failure in Thailand. Int J Infect Dis 2007; 11:413–416.
9. Nicastri E, Chiesi A, Angeletti C, Sarmati L, Palmisano L, Geraci A, et al
. Clinical outcome after 4 years follow-up of HIV-seropositive subjects with incomplete virologic or immunologic response to HAART. J Med Virol 2005; 76:153–160.
10. Sungkanuparph S, Manosuthi W, Kiertiburanakul S, Piyavong B, Chumpathat N, Chantratita W. Options for a second-line antiretroviral regimen for HIV type 1-infected patients whose initial regimen of a fixed-dose combination of stavudine, lamivudine, and nevirapine fails. Clin Infect Dis 2007; 44:447–452.
11. Campaign for Access to Essential Medicines; Médecins Sans Frontières. Untangling the web of price reductions: a pricing guide for the purchase of ARVs for developing countries. 9th ed. Geneva: Médecins Sans Frontières; 2006.
12. Gallant JE. Drug resistance after failure of initial antiretroviral therapy in resource-limited countries. Clin Infect Dis 2007; 44:453–455.
13. Vekemans M, John L, Colebunders R. When to switch for antiretroviral treatment failure in resource-limited settings? AIDS 2007; 21:1205–1206.
14. World Health Organization. Scaling up antiretroviral therapy in resource-limited settings: treatment guidelines for a public health approach. Geneva: World Health Organization; 2003.
15. World Health Organization. Antiretroviral therapy for HIV infection in adults and adolescents in resource-limited settings: towards universal access. Geneva: World Health Organization; 2006.
16. Braitstein P, Brinkhof MW, Dabis F, Schechter M, Boulle A, Miotti P, et al
. Mortality of HIV-1-infected patients in the first year of antiretroviral therapy: comparison between low-income and high-income countries. Lancet 2006; 367:817–824.
17. Mocroft A, Phillips AN, Miller V, Gatell J, van Lunzen J, Parkin JM, et al
. The use of and response to second-line protease inhibitor regimens: results from the EuroSIDA study. AIDS 2001; 15:201–209.
18. Boulle A, Van Cutsem G, Coetzee D, Hilderbrand K, Goemaere E, Maartens G. Regimen durability and tolerability to 36-month duration on ART in Khayelitsha, South Africa. 13th Conference on Retroviruses and Opportunistic Infections (CROI); 5–8 February 2006; Denver, Colorado, USA.
19. Orrell C, Harling G, Lawn SD, Kaplan R, McNally M, Bekker LG, et al
. Conservation of first-line antiretroviral treatment regimen where therapeutic options are limited. Antivir Ther 2007; 12:83–88.
20. Calmy A, Ford N, Hirschel B, Reynolds SJ, Lynen L, Goemaere E, et al
. HIV viral load monitoring in resource-limited regions: optional or necessary? Clin Infect Dis 2007; 44:128–134.
22. Colebunders R, Moses KR, Laurence J, Shihab HM, Semitala F, Lutwama F, et al
. A new model to monitor the virological efficacy of antiretroviral treatment in resource-poor countries. Lancet Infect Dis 2006; 6:53–59.
23. Ledergerber B, Lundgren JD, Walker AS, Sabin C, Justice A, Reiss P, et al
. Predictors of trend in CD4-positive T-cell count and mortality among HIV-1-infected individuals with virological failure to all three antiretroviral-drug classes. Lancet 2004; 364:51–62.
24. Vasan A, Hoos D, Mukherjee J, Farmer P, Rosenfield A, Perriens J. The pricing and procurement of antiretroviral drugs: an observational study of data from the global fund. Bull World Health Org 2006; 84:393–398.
25. Srasuebkul P, Calmy A, Zhou J, Kumarasamy N, Law M, Lim PL. Impact of drug classes and treatment availability on the rate of antiretroviral treatment change in the TREAT Asia HIV Observational Database (TAHOD). AIDS Res Ther 2007; 4:18.
26. Walmsley S, Bernstein B, King M, Arribas J, Beall G, Ruane P, et al
. Lopinavir-ritonavir versus nelfinavir for the initial treatment of HIV infection. N Engl J Med 2002; 346:2039–2046.
© 2008 Lippincott Williams & Wilkins, Inc.