Abstract

Nowadays, most preclinical HIV treatment trials use a protocol of administering repeated low-dose challenges (RLCs) of simian immunodeficiency virus (SIV) to macaques. Statistical analyses of treatment efficacy under the RLC protocol need to consider two confounding hypotheses, both pertinent biologically to HIV: (1) the non-infecting challenges may immunize animals against SIV; and (2) the animals may vary in intrinsic infectability ("frailty"). To explore the two hypotheses, a previous study (Regoes 2012) assembled a database from 7 articles with SIV-macaque treatment trials. With two explicable exceptions, Regoes concluded that the control data did not support either confounding hypothesis. Recent SIV-macaque trials present opportunities to evaluate the conclusions robustness. Accordingly, the present article assembles from 24 articles an updated database containing net survival curves from both control and treatment arms in SIV-macaque treatment trials. Broad patterns of statistical significance (at p<0.05, uncorrected for multiple testing) made it difficult to dismiss the confounding hypotheses completely in the controls. Although statistical analysis has focused on defense against variable frailty, only one set of controls showed significant variable frailty, whereas many sets showed significant immunization. As trials progressed, changes in the probability of infection per challenge were significant in 8/28 trials (1/3 trials using oral challenges; 2/4 trials using vaginal challenges; and 5/21 trials using rectal challenges). The results suggest the possibility that vaginal challenges may immunize animals faster than rectal challenges, and they also bear on previous conclusions that repeated exposure to HIV without treatment may have no effect on infectability or may even reduce it. Author SummaryMany preclinical trials of HIV treatments rely on repeatedly administering low-dose SIV challenges to macaques until infection occurs. The repeated low-dose protocol reuses macaques and is more sensitive to subtle therapeutic efficacies than a protocol administering a single large dose to each macaque. The animal reuse raises some pertinent biological questions, notably: (1) do macaques have intrinsically variable infectabilities? and (2) do the repeated SIV challenges immunize macaques against infection? A 2012 study collected a database of eight macaque trials, concluding that variable infectability and immunization were at most sporadic and readily explicable. I expanded the 2012 database to twenty-eight trials, discovering that the conclusions were not robust. Although only 1/28 SIV-macaque trials showed variable infectability, 7/28 showed immunization, with few ready explanations. Statistical analysis of SIV-macaque trials has focused on the confounding effects of variable infectability to the neglect of immunization, so the expanded database provides a rich empirical resource. The trials have general medical importance because they provide a model for analyzing animal trials of infectious disease therapies and other sparse trials, e.g., for breast cancer. My findings also indirectly suggest that repeated human exposure to HIV inconsistently immunizes and can foster either immune priming or tolerance.