Regimen durability and toxicity in 36-month follow up on ART in Khayelitsha, South Africa

Polly Clayden, HIV i-Base

Andrew Boulle from the University of Cape Town, presented findings from 36-months of follow up of the Medecins sans Frontiers and Provincial Government programme in Khayelitsha, South Africa.

Khayelitsha is a township outside Cape Town with a population of 4-500,000, and an antenatal seroprevalence of 33%. ARV provision began in May 2001; about half of those in need of ART are receiving it and treatment and care is provided free. This study aimed to “describe the rate of within regimen substitutions due to adverse events, and regimen switches due to virological failure in a routine ARV roll out setting in South Africa.” All treatment naive adults started on ART from May 2001 until the end of 2004, were followed up until June 2005.

The initial drug regimens comprised an AZT/3TC NRTI backbone combined with either nevirapine (NVP) or efavirenz (EFV). Later AZT was changed to d4T. The analyses included time to treatment failure, time to substituting drugs by exposure to each drug and cause of substitution. Failure was time to second consecutive viral load result above 5000 copies/mL (in line with local protocols for virological failure).

There were 1729 patients in the cohort, of which 70% were women. The median follow up was 12.3 months (IQR 8.1-20.3 months) and 1062 patients had begun treatment in 2004).

The investigators reported that at 36 months, 70% of patients remained on their initial regimen (see Table 1) in a cross-sectional analysis.

Table 1: Proportion of patients remaining on initial regimen

12 months 24 months 36 months
n 812 321 130
Switch/failure 0.7% 7.2% 10.8%
Other reasons 3.0% 3.7% 0.0%
Contraindications 5.5% 8.1% 12.3%
Intolerance 10.7% 8.1% 6.9%
Original regimen 80% 72% 70%

The investigators found that cumulatively 11.2% (95% CI: 7.9 to 15.8) of patients had been switched to second-line by 36 months on ART. They noted that this was fewer than would be anticipated based on the viral load results.

In an analysis of rate of substitutions due to toxicity by drug, by 36 months: 16.5% (95% CI: 12.0-22.6), 8.3% (95% CI: 6.3-10.9), 7.4 (95% CI: 5.4-10.1), and 3.1 (95% CI: 1.8-5.5) of patients had switched from d4T, AZT, NVP and EFV respectively.

The overall rate of substitutions of d4T due to symptomatic hyperlactaemia or lactic acidosis was 30 per 1000 patient years (PY). Analysis looking at time, weight and gender found 2 substitutions per 1000 PY at <6 months vs. 60 at >/= 6-months duration on ART; 17 per 1000 PY in men vs. 21 in women at >/= 6-months and 44 per 1000 PY for women <75kg at >/= 6-months, which rose dramatically to an incidence of 316 per 100 PY for women >/= 75kg at >/= 6-months.

Looking at outcome for patients who substituted d4T due to symptomatic hyperlactaemia/lactic acidosis, the investigators found that 34 patients had substituted or stopped. Of these patients 31 were women; 17 were women >/= 75kg, 4 were also receiving ddI, 13 also had peripheral neuropathy at presentation and 11/16 of hospitalised patients were acidotic.

Multivariate analysis of predictors of toxicity induced substitutions from d4T found being female carried an adjusted hazard ratio (AHR) of 4.5 (95% CI: 1.1-19.1), p=0.03; being female and weighing 60-75kg had an AHR of 5.6 (95% CI: 1.5-20.8), p=0.01 and being female and weighing >75kg had an AHR of 25.0 (95% CI: 7.1-87.6), p=0.001. Additionally three month weight gain of >/= 5kg had an AHR of 2.5 (95% CI: 1.1-5.4), p=0.022.

Predictors of d4T related peripheral neuropathy were low baseline CD4 < 50 cells/mm3 AHR 1.9 (95% CI: 1.0-3.3), p=0.67 and older age >/= 50 years AHR 4.1 (95% CI: 1.5-10.8), p=0.004.

For AZT they found 47 substitutions: 37 due to anaemia, 5 due to anaemia and neutropaenia, and 4 due to neutropaenia alone. Lower CD4 <50 cells/mm3 carried a hazard ratio of substitution due to AZT toxicity of 2.3 (95% CI: 1.3-4.3), p=0.007.

For nevirapine, weight <60 kg was associated with increased risk of substitution due to toxicity, AHR 2.3 (95% CI: 1.3-3.90), p= 0.003. The investigators noted that this concurs with previous findings on hepatoxicity but this analysis failed to demonstrate a higher risk in women, AHR 0.6 (95% CI: 0.4-1.0), p=0.062.

Additionally, they found that each regimen interruption resulted in an increased risk of virological failure, AHR 3.2 (95% CI: 2.0-5.1), p<0.001; controlling for baseline CD4 count.

Dr Boulle summarised the findings from this analysis. In the Khayelitsha cohort:

  • 70% patients remained on their 1st line regimen at 36 months. Just over 10% were on 2nd line regimens at the time of analysis.
  • Most toxicity driven switches are associated with d4T (occurring after 6 months from initiation), followed by AZT and nevirapine (usually within the first 6 months)
  • Symptomatic hyperlactaemia and lactic acidosis (SH/LA) occur almost exclusively beyond six months and in this cohort have always been ascribed in total or in part to d4T.
  • Rates of SH/LA in women beyond 6 months of ART, especially in women >75kg are exceptionally high. 70% of patients initiating HAART are women.
  • Haematological adverse events in patients on AZT are fewer in patients initiating therapy with CD4 counts >/= 50 cells/mm3.
  • NNRTI toxicity is higher in patients <60kg.

Additionally, he raised some questions for first-line treatment policy in this setting asking “Should we change the starting regimen for naive patients?” and:

  • Start with AZT rather than d4T? This would require extra safety monitoring early on.
  • Replace d4T with tenofovir? This drug is not yet available in South Africa and even at the access price still ten times that of d4T.
  • Use a reduced dose of d4T? Either for everyone or selectively in heavier patients.
  • Use a stratified approach based on risk factors? So women with higher weight or BMI could be started on AZT instead of d4T.


A number of other studies also reported toxicity data from sub-Saharan African cohorts including two posters from rural Uganda and Kenya (the majority of people in both cohorts starting with d4T, 3TC and nevirapine).

In the Ugandan study of 1037 people, the investigators reported the probability of remaining free from any toxicity at 6, 12, and 18 months was 0.76, 0.59, and 0.47, respectively; while the probability of remaining free from severe toxicities was 0.92, 0.86, and 0.84 respectively. Because of toxicity, 220 patients (21%) had 225 drug changes:  181 d4T to AZT; 30 nevirapine to efavirenz; 9 nevirapine to lopinavir; 4 AZT to tenofovir; and 1 efavirenz to nevirapine.

The probability of remaining on the original regimen, without change for toxicity, at 6, 12, and 18 months was 0.91, 0.78, and 0.68, respectively.

In the Kenyan study there were 12/283 cases of toxicity that were severe enough to lead to a substitution: 4 rash, 4 hepatitis (nevirapine) , 3 neuropathy and 1 lipodystrophy (d4T). The probability of remaining on the original regimen at 6, 12, and 18 months was 0.98, 0.97, and 0.96, respectively.

Neither study went into the same degree of detail as the Khayelitsha group nor was hyperlactaemia and lactic acidosis observed in these cohorts.

Additionally, a substudy of the DART trial from two Ugandan centres, reported lower rates of discontinuation and a trend toward a lower rate of serious adverse reactions, in African patients initiating ARV therapy with abacavir vs nevirapine plus AZT/3TC.


All references are to the Programme and Abstracts of the 13th Conference on Retroviruses and Opportunistic Infections, 5- February 2006, Denver.

  1. Boulle A, Van Cutsem G, Coetzee D et al. Regimen durability and tolerability to 36-month duration on ART in Khayelitsha, South Africa. 13th CROI, Denver, 2006. Abstract 66.
  2. Forna F, Liechty C, Solberg P et al. Early clinical toxicity to nonnucleoside reverse transcriptase inhibitor-based HAART in a home-based AIDS care programme in Rural Uganda. 13th CROI, Denver, 2006. Abstract 142. 
  3. Kim A, Ngan’ga L, Macharia D et al. Adverse events in HIV-infected patients receiving ART in a treatment programme in a Nairobi Slum, Kenya, 2003 to 2005. 13th CROI, Denver, 2006. Abstract 143.
  4. Paula Munderi and the DART Trial Team. Safety of nevirapine compared to abacavir on a background of zidovudine/lamivudine as first-line antiretroviral therapy: A randomised double-blind trial. 13th CROI, Denver, 2006. Abstract 109LB. 

Links to other websites are current at date of posting but not maintained.