We report on the first descriptive follow-up study conducted among a group of patients utilizing a detailed neuropsychological battery to establish the effects of HAART in South Africa where clade C HIV is predominant. We found significant improvement in individuals commencing HAART at one year follow-up. There was improvement in most, but not all neuropsychological domains, with high rates of impairment persisting. Improvement in neuropsychological function was most marked for those with a lower level of education and a greater degree of impairment at baseline. Unadjusted associations with GDS change were noted for male gender, a lower level of education, lower baseline CD4 count and a greater baseline GDS severity. Individuals most impaired at baseline were more likely to improve at one year than those less impaired.
The use of HAART has frequently been reported as improving neuropsychological outcomes in prospective cohorts [3, 13, 24, 25]. Predictors of improvement previously reported include higher CD4 count nadir, high CPE, stability of HAART regimen, lower levels of baseline impairment and good medication adherence [26, 27]. All participants in this study had CD4 counts less than 200 cells/ml due to existing South African HAART guidelines; this might account for high rates of impairment at baseline, and to some extent, to persistence at follow-up. It is possible that we did not detect an effect for low CD4 count as the value range was limited to 119 to 215 in this cohort. We did not gather data on nadir CD4 count specifically, although in most instances, the value obtained at this pre-HAART visit is the nadir. We did not detect an effect of different CPE ranks, as first-line regimens, used by the majority of participants in our study have moderate to high CPE ranks (73 of 82 participants had a CPE rank ≥ 1.5). Also, almost all participants achieved peripheral viral load suppression during the first year of treatment, and therefore were adequately immune reconstituted. This suggests that they were also adherent to treatment.
There are few data enumerating changes in neuropsychological performance by domain. Some have shown improvements in psychomotor speed [12, 28], while others have reported either improvement in global scores or across multiple domains [29–31]. While we report on improvement across most domains of function, we did not observe this for learning or psychomotor processing. The domain of learning and psychomotor processing has been held to be a core feature of HIV-associated neurocognitive impairment, due to its predilection for the deep grey nuclei . The absence of significant improvement suggests that HIV-associated disease is either established earlier and therefore less amenable to reversal by HAART in late stage/low CD4 nadir disease; or that neuro-inflammation is ongoing in these brain regions; or lastly, that HAART is not effective in these areas due to deficient penetration or a direct drug toxicity.
In the neuropsychology literature, improvement is usually associated with less baseline impairment . We noted the converse, that more severe baseline performance was associated with most improvement. A greater degree of GDS change was significantly associated with male gender, a lower level of education, baseline CD4 cell count, baseline GDS severity and neared significance with follow-up CD4 cell count (p = 0.05). Only baseline GDS severity remained significant in adjusted models, while a lower level neared significance (p = 0.05). The strong trend to improvement in those most severely affected likely is explained by the range of possible improvement being wider; those with less impairment probably experienced a ceiling effect on their ability to improve. The improvement shown by individuals with less education in the adjusted model suggests that this effect was present together with the degree of impairment. This effect may be explained by participant characteristics such as test experience, whereby individuals with lower education benefitted more from the experience of being tested. In South Africa, the effects of quality of education, as opposed to levels of education, have been shown to exert effects in test settings .
It is less well understood whether improvement in cohort studies is associated only with a HAART effect, or is also associated with practice effect, test familiarity or other participant characteristics, such as level and quality of education [32, 4]. We re-assessed participants after one year to reduce practice effect, although some degree of familiarity with test environment and administration may have had an impact on improvement. When we investigated the associations of baseline GDS, we noted only lower education to be a significant predictor. In this cohort, the medians across the GDS severity groups ranged from 9-12 years which might have had a substantial effect on performance.