When should HAART be initiated in pregnancy to achieve an undetectable viral load?

Polly Clayden, HIV i-Base

Women who do not need treatment for their own HIV in resource-rich countries generally receive a short course of HAART in pregnancy to prevent mother to child transmission (MTCT).

It is considered a safe option for women with low or undetectable viral loads (VL) to choose vaginal delivery. BHIVA guidelines recommend a cut off of <50 copies/mL and the US guidelines <1000 copies/mL.

However the optimum timing to initiate short course HAART and achieve an undetectable viral load is uncertain.

A poster authored by Phillip Read and colleagues showed findings from a retrospective UK study across five centres in London and the South East, conducted to provide data for clinicians for the timing of short course HAART in pregnancy.

All available data were included for women commencing boosted PI, NNRTI or triple NRTI based HAART from 2000 onwards In this study demographics, gestation, drug class, CD4 count, and VL results were collated. VL data were right-censored at delivery date and drug regimen

analyses were intent-to-treat. Survival curves for reaching a viral load <50 copies/mL were stratified by VL at initiation of HAART. Cox’s proportional hazards regression model adjusted for demographics and immuno-virological parameters.

In this study, 439 pregnancies met the inclusion criteria and 378 had sufficient data for analysis.

Of those evaluated, 70% of women were of black African origin and their mean age at conception was 29.9 (range 14.7–49.8) years. Median pre-treatment CD4 and viral load was 330 cells/mm3 (IQR 195–470) and 8243 copies/mL (IQR 2341-32,640).

For their regimen, 246 women (65%) started PI-, 129 (34%) NNRTI-, and 3 (1%) NRTI-based HAART, initiated at a median of 23.2 weeks gestation (IQR 20.4 to 26.3).

By their delivery date (mean 38 weeks), 292 (77%) women achieved VL <50 copies/mL after a median of 58 days. Pre-treatment VL was associated with both the time taken and the proportion achieving a VL  <50 copies/mL at delivery, p<0.001.

A baseline viral load of <10,000, 10,000 to 50,000, 50,001 to 100,000, and >100,000 resulted in 91%, 73%, 54%, and 37% of women achieving <50 copies/mL at delivery, respectively.

In multivariate analysis, the hazard ratio (HR) for a NNRTI regimen achieving a viral load <50 copies/mL compared to a PI was 0.7 (95% CI 0.52 -0.94). If VL was >10,000, 58% of PI and 66% of NNRTI regimens achieved <50 copies/mL.

When baseline VL was <10,000 copies/mL, gestation at initiation of HAART did not significantly change the probability of a VL <50 copies/mL at delivery. With a baseline VL of 10,000–50,000 copies/mL, the HR for a VL <50 copies/mL declined to 0.51 if HAART was initiated after 23.3 weeks (p<0.01) while if viral load was >100,000, starting HAART before 20.4 weeks gave a HR of 0.2 (p <0.01) compared with 0.1 if started after 20.4 weeks (p <0.01).

The investigators concluded with four key messages:

  • Women with a VL >10,000 copies/mL should commence HAART by 20.4 weeks
  • Women with a VL >100,000 copies/mL should commence HAART without delay
  • If the VL is <10,000 copies/mL, HAART may be deferred to 26 weeks
  • If the VL is >10,000 copies/mL NNRTI-based HAART, where appropriate, may be more successful

And they noted: “Final decisions on mode of delivery often depend on the VL at 36 weeks gestation and this needs to be taken into account when starting HAART based on these data”.


This analysis provides useful data to guide when to commence short course HAART in pregnancy.

Presumably, a larger proportion of women would commence long term treatment if they were starting according to current guidelines, as the median baseline CD4 count in this data set was 330 cells/mm3 (IQR 195–470), and previous guidance used a 200 CD4 cells/mm3 threshold at which to start HAART.

Ref: Read P et al. When Should HAART be initiated in pregnancy to achieve an undetectable viral load? 17th CROI, 16-19 February 2010, San Francisco. Poster abstract 896.

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