This cross-sectional study described the variability of NMR and its association with sociodemographic, tobacco use, and clinical characteristics in a population of PWH who smoke in South Africa. NMR was dichotomized into fast and regular metabolizers using a cut-off of the fourth quartile; few associations were seen between NMR and variables of interest. NMR was minimally associated with tobacco use characteristics, and the NMR had weak to no association with sociodemographic or clinical characteristics. Our study sample, which was reflective of the burden of smoking among PWH in South Africa, was a racially and ethnically homogenous population of Black South Africans, primarily of low socioeconomic status; this likely contributed to the observed relative lack of variability in NMR values.

We reported little variability in values of NMR in this population, and we found a minimal association between fast NMR and more intense smoking characteristics, which was inconsistent with evidence from smokers from both the general population and among PWH [11,18,27]. We found no evidence that those with a fast NMR tended to smoke more cigarettes per day; our observed association was inconsistent with existing literature, which showed a positive association between high NMR and a higher number of cigarettes per day [11,18,27]. Although we also found a negative association between NMR and quit attempts in the last year, there was limited to no association between NMR and nicotine dependence, as measured by the Heaviness of Smoking Index. We also found that those reporting alcohol misuse were less likely to be fast metabolizers of nicotine. This was inconsistent with recent studies in the general population [27,28], which reported positive associations between alcohol consumption and NMR. Similarly, in a trial of men receiving treatment for alcohol use disorder, a 50% reduction in NMR was reported among those who successfully abstained from alcohol, compared to the baseline, despite little change in tobacco use, suggesting that chronic alcohol exposure induces CYP2A6 activity, which is responsible for nicotine metabolism [29].

Overall, the values of NMR found in our study sample were lower than what was reported in other populations of PWH in high-resource settings; however, the existing evidence came largely from more diverse Western settings, with a large proportion of participants of Caucasian descent. Mean values of NMR in literature among PWH range from 0.35 to 0.47 [11,25,27,30], whereas the mean in this study was 0.32. Given our sample was entirely Black South African, and given African Americans on average have a lower NMR than other racial/ethnic groups, the homogeneity of our population, with a genetic makeup closer to those with typically lower NMRs, aligned our findings with these other studies. The variability in the NMR observed in this study may be too minimal to allow the use of NMR as a tool for clinical decision-making at the individual level in this population. Existing studies among PWH reported greater variability in NMR, with values ranging from 0.054 to 1.42 [11], while our minimum and maximum values were 0.29 to 0.57, respectively. In addition, participants of African descent in other studies also reported a higher average number of cigarettes per day than our study population, with an average number of cigarettes per day of 16.6 among individuals of African descent [18], compared to 11.1 in this present study. Existing studies largely found no association between NMR and sociodemographic characteristics other than sex, race/ethnicity, and age; associations in our study between NMR and sex, age, and other sociodemographic characteristics were weak and inconsistent in direction. A handful of studies in the general population demonstrated a relationship between high NMR and older age [24,26,31]. However, studies in HIV populations did not report these associations [11,30], which was consistent with our findings. We also found no association with sex, which was inconsistent with existing evidence showing that women have a higher NMR than men [18,27].

Our findings are not without limitations. NMR was measured in urine in this study, which tends to correlate imperfectly with NMR measured in plasma [32,33]. NMR values obtained in our study may not accurately represent the sample or the population but is likely to be a reasonable approximation of plasma NMR. Recent evidence also suggests that some antiretroviral medications may impact nicotine metabolism. Efavirenz was found to raise NMR, given it is metabolized by the same pathway as nicotine [13]; this implies that people with HIV who smoke and are on efavirenz may have more intense smoking behaviors and more difficulty quitting smoking compared to those on other ARTs who are given an elevated NMR. However, our sample was drawn from a population with limited access to antiretroviral drugs at the time of recruitment, given policies at the time limiting the initiation of ART to individuals with a CD4+ count under a defined level, compared to current universal test and treat policies. Additionally, other predictors of smoking behavior, such as extreme poverty, may have influenced smoking behavior in our sample to a degree that would have reduced the impact of NMR on smoking behavior [26]. Given that included subjects were recruited for participation in a clinical trial for smoking cessation, our sample may have included a disproportionate number of individuals with a slower metabolism. Participants in the primary trial must have been willing to set a quit date, and those with faster metabolisms may have been less willing to attempt to quit and therefore less willing to participate in the primary cessation study; this may have limited the generalizability of our findings. While a more racially/ethnically diverse sample may result in the greater variability of NMR, our study sample is reflective of the burden of HIV in this setting. The results of this study are most generalizable to ethnically African populations with HIV in the region; however, evaluations of those without HIV in this population would be informative.