The results reported here are consistent with no or little protective effect from LPV/r versus NVP-based ART regimen on the incidence of clinical malaria. Subjects randomized to LPV/r-based ART were as, or more, likely to develop clinically diagnosed malaria as subjects randomized to receive a NVP-based ART regimen. The results were similar across both trials of ACTG A5208 and when the analysis was restricted to only confirmed cases of malaria.
These results are inconsistent with laboratory evidence supporting an antimalarial effect of HIV PIs. Skinner-Adams et al. demonstrated the inhibitory effects of ritonavir, saquinavir, and indinavir on the in vitro
growth of P. falciparum
, whereas the NNRTI NVP had no such inhibitory effect (33
). Additional studies using murine models of malaria have also demonstrated the antimalarial effect of certain HIV PIs. After infection with P. chabaudi
, for example, mice exposed to LPV/r had a delayed onset of parasitemia by 2 days and a decrease from 20 to 4% in median parasitemia (3
). Evidence from a different murine model infected with P. yoelii
has suggested that HIV PIs, including lopinavir (LPV), can also inhibit the growth of pre-erythrocytic-stage parasites at concentrations effective to inhibit HIV-1 replication (17
The absence of a protective effect of LPV/r on the incidence of clinical malaria seen in the present study needs to be considered in the context of the study design of the trials. Women who experienced virologic failure in either trial, as well as those with adverse reactions to their assigned medications, were eligible to switch their antiretroviral regimen and receive the treatment available in the other study arm. Among the 445 women included in this secondary analysis, 50 (11%) did so; almost all of these subjects switched from a NVP-based to a LPV/r-based ART regimen. Although this could have potentially influenced the results of the study, the small percentage of subjects who actually did switch treatment may have limited the effect that switching could have had on our results.
There are several potential explanations for why LPV/r-based therapy was not associated with a decrease in the incidence of clinical malaria in the present study. First, Nathoo et al. (24
) have proposed that the decrease in CD36 expression associated with exposure to HIV PIs, while perhaps being beneficial to the patient in controlling P. falciparum
infection, may also make it more difficult for a patient's innate immune system to produce an effective immune response against P. falciparum
infection through its inhibitory effect on the nonopsonic phagocytosis of parasitized red cells by macrophages. As a result, this reduction in the innate immune response to malaria may have offset any potential antiparasitic activity of LPV/r, resulting in little or no net protective benefit.
In addition, the concentration of LPV in the patients' sera might not have achieved a level sufficient enough to exert an antimalarial activity. In vitro
activities of the HIV PIs against P. falciparum
are modest compared to standard antimalarial medications, including chloroquine (3
); they are also much less active against malaria in vitro
than HIV (26
). In addition, LPV is highly protein bound, and free drug concentrations in plasma are likely to be 100-fold less than measured total concentrations (9
). Although the antimalarial activity of sera taken from patients treated with HIV PIs has been evaluated ex vivo
to explore the effect of clinically relevant peak serum concentrations (29
), it is unknown whether the parasites' drug exposures in these laboratory studies are truly equivalent to the drug concentrations one would anticipate in a patient on ART and, because there was no systematic sampling of LPV levels in subjects who initiated LPV/r therapy, we could not evaluate that in the present study.
Limitations to the present study include the imperfect sensitivity and specificity of relying on a clinical syndrome (6
) and microscopy (37
) for the diagnoses of malaria. This reduced our ability to determine the true incidence of malaria; however, the results from the sensitivity analyses indicate that the results were robust to moderate levels of outcome misclassification. Also, since the study sites are generally in areas of high malaria transmission and high preexisting immunity, they may not be generalizable to areas with low malaria transmission.
These results should be considered cautiously given the reliance on clinically diagnosed malaria as the primary outcome in the present study. Studies on the effects of PI's on laboratory-confirmed malaria in this study population are ongoing and will provide additional data. In addition to studies in adults that are designed to directly address whether PIs have a beneficial effect on malaria in regions where malaria is endemic and the prevalence of HIV-1 is high, further research is warranted to determine the effects of HIV PIs on malaria in children and pregnant women, two groups that are at greater risk for clinical malaria. In addition, since laboratory evidence suggests that the coadministration of HIV PIs with additional antimalarial agents, including chloroquine or mefloquine, may enhance the antimalarial activity of these drugs (16
), the utility of such combinations also warrants further research. In conclusion, the results from our analyses, which were not consistent with large a decreased hazard of clinical malaria in subjects receiving LPV/r-based ART, provide evidence, albeit preliminary and imperfect, that supports the possibility of no or little effect of LPV/r-based therapy on clinical malaria.