A cost-effectiveness analysis of the OCTANE trial
Polly Clayden, HIV i-Base
The Optimum Combination Therapy After Nevirapine Exposure (OCTANE)/ACTG5208 trial found superior outcomes among women exposed to single dose nevirapine (NVP) receiving lopinavir/ritonavir (LPV/r)-based HAART compared to those receiving NVP-based regimens. [1, 2]
In OCTANE, women with CD4 <200 cells/mm3 exposed to NVP as PMTCT prophylaxis a median of 17 months prior to HAART initiation were randomised to receive either LPV/r- or NVP-based HAART. Over a median of 73 weeks follow up, women in the LPV/r arm had significantly lower risks of virological failure or death compared to those in the NVP arm. We covered these findings in previous issues of HTB.
However, the difference in cost between the two drugs is quite considerable: LPV/r is12 times more expensive than NVP.
Andrea Ciaranello presented an analysis that examined the cost-effectiveness of first line LPV/r based HAART, compared to first line NVP-based HAART, for women with prior single dose NVP exposure in South Africa. 
This analysis utilised the Cost-effectiveness of Preventing AIDS Complications (or CEPAC) computer simulation model. The model included incidence, prophylaxis, and treatment of opportunistic infections in South Africa, CD4 count, viral load and HAART efficacy. The investigators used data for the modeling from OCTANE and from published South African cohorts, including natural history data from the Cape Town AIDS Cohort.
They projected two year and lifetime outcomes associated with first line HAART strategies for women with prior single dose NVP exposure. These projections included risk of opportunistic infections, deaths, and per-person HIV-related costs.
Dr Ciaranello explained that cost-effectiveness analysis a method of comparing alternative health care strategies. This method uses an incremental cost-effectiveness ratio (ICER). To calculate the ICER, the number of additional health care resources needed for one strategy compared to another is determined. This is divided by the additional health benefits gained by this strategy compared to the other.
Currency per year of life saved (YLS) is the most common unit for an ICER. A lower ratio, when less money is needed to produce a health benefit, shows a more cost-effective intervention.
To determine whether an intervention is cost effective, WHO compare ICER to per-capita GDP. ICER <1xGDP/YLS is considered to be very cost effective and ICER <3x GDP/YLS is cost effective. South African GDP (2006) is $5400.
Three strategies were evaluated in this analysis:
- No HAART (comparitor)
- First line NVP/TDF/FTC (Second-line LPV/r/ddI/AZT)
- First line LPV/r/TDF/FTC (Second-line NVP/ddI/AZT).
Following second line failure, women were assumed to receive a maintenance regimen of LPV/r and 3TC, which was common practice at OCTANE sites.
Baseline data for the women were used from the OCTANE cohort and included median: 31 years of age; CD4139 cells/mm3; viral load, 5.15 log copies/mL and 17 months since exposure to single dose NVP.
The efficacy of each regimen was modeled on that observed in the OCTANE trial. Efficacy was defined as achieving viral load suppression <400 copies/mL at 24 weeks after initiation of HAART.
With a median time from single dose NVP exposure of 17 months, efficacy of the first line NVP regimen was 84.6%, and the efficacy of the first-line LPV/r regimen was 96.7%.
The investigators noted that these data are slightly different from those previously presented, as a composite endpoint of virologic failure or death was shown.
There are very few data on the efficacy of NNRTI-based second-line regimens. For the model data were extrapolated from two studies to give an estimate of 43% virologic suppression at 24 weeks for second-line NVP. Data for second-line PI-based HAART are more common and, using multiple sources, an estimate of 72% suppression was used.
The investigators assumed that routine viral load tests were not available and that HAART was switched for severe opportunistic infection, 50% CD4 decline or toxicity.
They used HIV-related healthcare costs derived from the South African Health Systems Trust. These included the cost of a day in hospital of $221, the cost of an outpatient clinic visit of $11 and the cost of a CD4 test of $9.
For drug costs, they used prices from the Clinton Foundation HIV/AIDS Initiative. Annual drug costs were $38 for NVP and $444 for LPV/r. For the nucleosides, the figures were $142 for TDF/FTC and $238 for ddI/AZT.
Projecting outcomes for the entire cohort at two years revealed that with no HAART, 41.7% of women would survive at a per-person cost of $2650. Using a NVP-based first-line regimen survival was 96.1% at a per-person cost of $2450, and providing LPV/r-based HAART first-line, survival was 97.1% at a per-person cost of $2780. Compared to first-line NVP, LPV/r gave a 26% risk reduction in mortality at an additional per-person cost of $330.
Projecting long-term outcomes showed live expectancy of 1.8 years at a per-person cost of $3540 if the cohort were untreated. With NVP-based first line survival increased by 13.6 years to 15.4 years at a per-person cost of $14,040 for an ICER of $770/YLS. Using LPV/r-based first-line gained a further 1.1 years survival (16.5 years) at a per-person cost of $16,180 for an ICER of $1970/YLS. So using LPV/r first-line would be very cost effective compared to NVP according to WHO criteria for South Africa.
Additionally, the investigators then conducted a sensitivity analysis. They evaluated many model inputs parameters and assumptions. Dr Ciaranello presented estimates for parameters of particular importance.
Importantly, they evaluated the efficacy of second-line NVP, for which there are few data (in the range of 16-45%). The investigators looked at 0-100% efficacy and found that LPV/r remains very cost effective by WHO criteria unless the efficacy of second-line NVP is less than 15%.
They also looked at the influence of NNRTI resistance at the time of initiation of HAART. Among the OCTANE cohort, 86% of women had no detectable NNRTI resistance using standard genotype assay at the time of starting treatment. For this group of women the efficacy of first line NVP was greater than for the cohort overall, 89% vs 85% (97% for LPV/r first line). For women with no resistance the ICER of first line LPV/r compared to NVP was $10,990/YLS, and no longer a very cost-effective intervention in South Africa.
OCTANE also included stratification by time from NVP exposure to initiation of HAART. Looking at cost effectiveness according to these strata, the investigators found LPV/r was very cost effective with 6-24 months between NVP exposure and treatment, at an ICER of $2000/YLS. However, >24 months the ICER reached the WHO threshold for South Africa at $5400. They noted that LPV/r first-line became less cost effective as time from NVP exposure increased.
Dr Ciaranello acknowledged that the limitations to these estimates include some input data from cohorts of both men and women; costs and cost effectiveness thresholds that are specific to South Africa and results are sensitive to data that are not yet available from the OCTANE trial.
She stressed that reducing the impact of NNRTI resistance related to single dose NVP is particularly important in order for women to benefit from available classes of antiretrovirals in the treatment of their own HIV.
She also emphasised the importance of HIV and CD4 testing of pregnant women in order to initiate HAART in eligible women before delivery, both improving maternal health and reducing mother-to-child transmission.
However she added: Despite these efforts, many of the single dose NVP-exposed women living in resource-limited settings are likely to need to initiate HAART soon. The choice of optimal first-line HAART for these women will require important data about long-term outcomes, particularly outcomes of second-line HAART. Such outcomes can only be observed after the conclusion of most clinical trials. OCTANE, cohort studies, and HAART programme monitoring and evaluation efforts will be crucial sources of these data.
This cost-effectiveness analysis is useful to demonstrate that for NVP-exposed women, particularly those with an interval since exposure of <24 months, using LPV/r-based HAART first-line is very cost-effective.
However, it would be better to avoid risk of resistance in the first place by early identification and treatment for women indicated for their own health (with CD4 <350 cells/mm3), and more complex PMTCT regimens for healthier women (short course PI based HAART, AZT plus single dose NVP plus tail coverage).
As the investigators suggest, the future options for NVP-exposed women for second-line treatment are unclear in resource-limited settings.
- Ciaranello A et al. Lopinavir/ritonavir (LPV/r)- compared to nevirapine (NVP)-based ART following receipt of single-dose nevirapine (sdNVP) for prevention of mother-to-child HIV transmission in South Africa: a cost-effectiveness analysis of the OCTANE (ACTG A5208) trial. 5th IAS Conference, Cape Town.19-22 July 2009. Poster abstract TUAD103.