Nicotine metabolism ratio increases in HIV-positive smokers on effective antiretroviral therapy: a cohort study

Abstract

Background:

People with HIV (PWH) smoke tobacco at much higher rates than the general population. Prior research has shown that PWH have faster nicotine metabolism than HIV-uninfected individuals which may underlie this disparity, but the cause is unknown. We investigated whether higher nicotine metabolite ratio (NMR; 3-hydroxycotinine:cotinine), a validated biomarker of nicotine metabolism via CYP2A6, was associated with antiretroviral use among HIV-infected smokers.

Methods:

We conducted a retrospective cohort study of HIV-positive smokers in the University of Pennsylvania Center for AIDS Research cohort. We compared the NMR before viral suppression (>10,000 copies/ml) and after viral suppression on ART (<200 copies/ml). We used mixed effects linear regression to analyze the change in NMR after viral suppression and assessed for effect modification by efavirenz use.

Results:

Eighty-nine individuals were included in the study. We observed effect modification by efavirenz use (interaction term for efavirenz use, P<0.001). Among those on non-efavirenz regimens, the mean NMR increased by 0.14 (95% CI 0.05–0.23, P=0.002). Among those on efavirenz-containing regimens, the mean NMR increased by 0.53 (95% CI 0.39–0.66, P<0.001).

Conclusion:

We observed a clinically and statistically significant increase in NMR after viral suppression among smokers with HIV, which more than doubled among those on efavirenz-based regimens. Higher NMR among HIV-positive smokers on ART may help explain the higher rates of tobacco use and lower quit rates among PWH in care. These findings suggest that regimen choice and other modifiable factors may be targets for future attempts to increase success rates for tobacco cessation among PWH.

Introduction

People with HIV (PWH) on effective antiretroviral therapy (ART) in high-income settings may now lose more life years due to smoking than from HIV itself.^1,2 At the same time, PWH have a much higher prevalence of tobacco use than the general population globally.^3,4 Despite progress in reducing overall smoking cessation rates, major disparities in smoking rates among PWH compared with the general population persist.^5,6

Differences in nicotine metabolism in smokers with HIV compared with smokers without HIV may contribute to higher rates of smoking and lower rates of smoking cessation in PWH.⁷ Nicotine is metabolically inactivated by the liver enzyme cytochrome P450 2A6 (CYP2A6) to cotinine, which is further metabolized to 3-hydroxycotinine exclusively by CYP2A6. The nicotine metabolite ratio (NMR; 3-hydroxycotinine:cotinine) is a validated biomarker of nicotine clearance via the CYP2A6 pathway which is independent of time-since-last cigarette.^8,9 Blood NMR has strong test-retest reliability, and also maintains stability over time, freeze-thaw processing, and across laboratories.^10–12 Greater NMR is associated with more tobacco use, nicotine dependence, withdrawal symptoms, and reduced cessation.⁸ It is also important to the success of pharmacologic treatment for nicotine addiction since NMR predicts tobacco cessation rates for both varenicline and nicotine replacement therapy.¹³ In addition, a randomized controlled trial found that NMR could inform clinical decision making in prescribing varenicline vs. nicotine patch for smoking cessation.¹⁴ While we have previously shown that smokers with HIV have significantly higher NMR compared with smokers without HIV,⁷ the cause of this difference is unknown. One potential mechanism is that ART increases NMR. For instance, use of the antiretroviral drug efavirenz, a CYP2A6 substrate, has been associated with greater NMR.¹⁵

We investigated whether viral suppression on effective ART resulted in increased NMR compared with periods of uncontrolled viremia off ART in a cohort of smokers with HIV. In addition, we hypothesized that efavirenz-containing regimens would be associated with the greatest increases in NMR compared with other ART regimens.

Methods

Study Design and Participants

We conducted a retrospective cohort study of smokers with HIV in the University of Pennsylvania Center for AIDS Research (CFAR) cohort between November 1999 and August 2019. This cohort contains longitudinal questionnaire data and specimens for over 3,100 PWH who obtain clinical care in the University of Pennsylvania health system and the Philadelphia Veterans Affairs Medical Center, and was approved by the University of Pennsylvania Institutional Review Board. Signed informed consent was obtained from all study participants.

We used the following inclusion criteria for our cohort: at least 18 years of age, confirmed HIV positive, self-reported current cigarette smoking while viremic off ART (>10,000 copies/ml; “viremic stage”), and self-reported current cigarette smoking during periods after virologic suppression on anti-retroviral therapy (<200 copies/ml; “aviremic stage”). We included ART-naïve and ART-experienced individuals in the cohort. We excluded individuals if they did not report current smoking during both viremic and aviremic stages, or if plasma or serum specimens were unavailable for either the viremic or aviremic stages.

Measures

Demographic information (age, race/ethnicity, sex), HIV history, smoking status, alcohol, and drug use data were obtained from baseline and follow-up questionnaires in the CFAR database. All viral load data were determined from the CFAR laboratory database and confirmed in the electronic medical record. We also collected laboratory data including CD4+ count, hepatic function panels, and viral hepatitis from the medical record, in addition to common medications affecting the CYP2A6 pathway, and diagnoses of cirrhosis. We calculated FIB-4 scores, a validated marker of hepatic fibrosis, and used a cutoff of 3.25, which has >95% specificity to predict advanced fibrosis.¹⁶

The primary outcome of interest was the change in NMR between viremic and aviremic stages. Initiation of ART had to precede the aviremic stage by at least 90 days, as determined by chart review. We collected blood samples at the defined viremic and aviremic stages and measured the NMR at each of the two stages. The NMR was treated as a continuous measure. Blood samples were collected and frozen at the patient’s clinic visits and stored in the CFAR Central Repository. Plasma samples (and if unavailable, serum samples) were assessed for cotinine and 3-hydroxycotinine (3-HC), to calculate NMR, using liquid chromatography-tandem mass spectrometry.¹⁴ Samples were sent to the University of Toronto for analysis. Study investigators analyzing NMR were blinded to all study participant data.

The primary exposure of interest was ART use. We ascertained ART regimens from the medical record, and the start date was considered the first date of prescription. Because almost all ART regimens use nucleoside reverse transcriptase inhibitors (NRTIs) as backbones, we categorized ART regimens based on the non-backbone drug: integrase inhibitors (INSTIs), protease inhibitors (PIs), non-nucleoside reverse transcriptase inhibitors (NNRTIs), and entry inhibitors. Based on our previous work,¹⁵ we assessed efavirenz as a potential effect modifier on NMR changes, so we also categorized regimens as either efavirenz or non-efavirenz based regimens.

Statistical analysis

We used mixed effects linear regression to evaluate the change in NMR after viral suppression. We included both a main effect for efavirenz-containing regimens and an interaction with pre-post viral suppression status. We used Akaike and Bayesian information criteria to select the covariance structure used in our model. Covariates in Table 1 were evaluated as potential confounders and retained in adjusted models if the association between viral suppression and NMR changed by greater than 15%. In our primary analysis, we excluded individuals with cotinine values below 10 ng/ml, which likely reflect occasional or light smokers where NMR may be unstable.¹⁷

Table 1:

Baseline characteristics of HIV-positive smokers

Characteristic	Cotinine > 10 ng/ml	Total (All cotinine levels)
	N=89	N=120
Median age at viremic specimen (IQR)	40 (35–46)	40 (31–45)
Gender
Male	56 (63%)	82 (68%)
Female	33 (37%)	38 (32%)
Race/Ethnicity
Non-Hispanic Black	67 (75%)	85 (71%)
Non-Hispanic White	11 (12%)	21 (18%)
Hispanic or Latino	6 (7%)	8 (7%)
Other	5 (6%)	6 (5%)
BMI1 (IQR)	24 (21–28)	24 (21–28)
Baseline NMR, mean (SD)	0.4 (0.3)	0.4 (0.2)
Baseline year of viremic specimen
2000–2005	31 (35%)	35 (29%)
2006–2010	45 (51%)	65 (54%)
2011–2014	13 (15%)	20 (17%)
Median plasma HIV RNA at baseline, log10 copies/ml (IQR)	5 (4–5)	5 (4–5)
Median CD4 count at baseline, cells/mm3 (IQR)	275 (130–424)	274 (120–432)
Median duration of ART before aviremic specimen (months, IQR)	19 (10–33)	18 (11–31)
ART regimen
Efavirenz-based	24 (27%)	32 (27%)
Non-efavirenz NNRTI-based2	8 (9%)	9 (8%)
Protease inhibitor-based	42 (47%)	53 (44%)
Integrase inhibitor-based	10 (11%)	19 (16%)
Maraviroc + other	2 (2%)	4 (3%)
NRTI only3	3 (3%)	3 (2%)
On CYP2A6 inducer or inhibitor other than efavirenz4	4 (5%)	4 (4%)
ART naïve at time of initiation	37 (42%)	59 (49%)
HCV infection5	12 (13%)	13 (11%)
HBV infection6	1 (1%)	2 (2%)
Weekly alcohol use
0 to less than 10 drinks/week	64 (89%)	86 (89%)
10–15 drinks/week	5 (7%)	8 (8%)
At least 15 drinks/week	3 (4%)	3 (3%)
Binge drinking, at least 5 drinks/sitting at least once a month	7 (9%)	9 (9%)
FIB-4 score>3.25 7	5 (6%)	7 (6%)

¹BMI = Body mass index.

²NNRTI = Non-nucleoside reverse transcriptase inhibitor.

³NRTI = Nucleoside reverse transcriptase inhibitor.

⁴CYP2A6 inducers and inhibitors included isoniazid, valproic acid, letrozole, dexamethasone, griseofulvin, phenobarbital, rifampin.

⁵HCV infection: Defined as current hepatitis C viremia at time of HIV viremia.

⁶HBV infection: defined as Hepatitis B surface antigen positivity at time of HIV viremia.

⁷FIB-4 score: Fibrosis-4 Index for Liver Fibrosis.

Based on prior research on NMR variability,¹⁴ we initially targeted a sample size of 73 to have 80% power to detect a 0.1 mean difference in NMR after viral suppression, assuming a standard deviation at the high end of 0.30, with a standard two-sided alpha of 0.05. We inflated our sample size by 1.5, assuming that some individuals may have low cotinine levels that may make NMR estimates less stable. In our primary analysis, we only included individuals with cotinine levels >10 ng/ml, consistent with regular smoking. We performed three sensitivity analyses: 1) including those with cotinine levels < 10 ng/ml, 2) excluding those on medications that affect the CYP2A6 pathway, 3) excluding those who reported recent illicit drug use, and 4) excluding those with FIB4 scores > 3.25 (those with significant hepatic fibrosis).

Results

One hundred twenty individuals had detectable cotinine levels, of whom 89 had cotinine levels >10 ng/ml. Patient characteristics are displayed in Table 1. The median age was 40 (IQR 35–46), 63% identified as male gender, and 75% identified as non-Hispanic Black. The mean NMR during HIV viremia was 0.42 (SD 0.30). Viremic specimens were obtained between 2001 and 2014 (median 2008, IQR 2005–2010). The median duration of ART before the aviremic specimen was 19 months (IQR 10–33); 47% were on PIs, 27% were on efavirenz, 11% were on INSTIs, and 9% were on NNRTI regimens excluding efavirenz.

We observed increases in NMR after viral suppression on ART in HIV-positive smokers regardless of efavirenz use, and effect modification of NMR by efavirenz use (P-value for interaction term for efavirenz use<0.001). Results are therefore stratified by efavirenz use. The mean NMR after viral suppression 0.54 (95% CI 0.44–0.63, P<0.001) on non-efavirenz regimens, and 0.95 (95% CI 0.80–1.10, P<0.001) on efavirenz.

In unadjusted analyses, the mean NMR increased by 0.10 (95% CI 0.01–0.19) among those on non-efavirenz regimens, and by 0.57 (95% CI 0.42–0.71) among those on efavirenz regimens (Table 2). Self-reported alcohol use was retained in adjusted models as a confounder. In adjusted models, the mean NMR increased by 0.14 (95% CI 0.05–0.23) among those on non-efavirenz regimens, NMR increased by 0.53 (95% CI 0.39–0.66) among those on efavirenz. Sensitivity analyses including individuals with cotinine levels<10 ng/ml, excluding those on medications known to affect the CYP2A6 pathway, excluding those who reported recent illicit drug use, and excluding those with significant hepatic fibrosis did not alter our findings (data not shown).

Table 2:

Mean increase in nicotine metabolite ratio after viral suppression

	Cotinine >10 ng/ml		All cotinine levels
	Efavirenz regimen	Non-efavirenz regimen	Efavirenz regimen	Non-efavirenz regimen
Adjusted estimate	0.53 (95% CI 0.39, 0.66) P<0.001	0.14 (95% CI 0.05, 0.23) P=0.002	0.46 (95% CI 0.26, 0.65) P<0.001	0.13 (95% CI 0.01, 0.24) P=0.03
Unadjusted estimate	0.57 (95% CI 0.42, 0.71) P<0.001	0.10 (95% CI 0.01, 0.19) P=0.03	0.55 (95% CI 0.40, 0.71) P<0.001	0.17 (95% CI 0.08, 0.26) P<0.001

Final adjusted model included weekly alcohol use.

Discussion

In this longitudinal cohort of HIV-positive smokers, we observed an increase in the NMR among individuals on ART, which doubled among those on efavirenz-based regimens. Prior research has observed higher NMR among smokers with HIV compared with smokers without HIV, and numerous studies indicate a strong association between higher NMR values and a lower likelihood of cessation following tobacco use treatment.⁸ However, past studies with HIV-positive smokers that assessed differences in NMR levels compared to HIV-negative smokers used cross-sectional data, and did not evaluate changes in NMR after viral suppression on ART.⁷ Our research is the first to evaluate longitudinal changes in NMR among HIV-positive smokers. In addition, our research finding that efavirenz significantly increased NMR compared with other ART may have significant implications in low- and middle-income countries, where efavirenz use remains widespread.

Our findings offer important insights into understanding the potential mechanisms underlying the links between HIV infection, ART use, and tobacco use. Even among those on ART not containing efavirenz, the observed NMR increase may be of clinically significant magnitude.¹⁸ These findings were robust after evaluating multiple potential confounders, and after performing sensitivity analyses excluding individuals with significant hepatic fibrosis, individuals on common medications affecting the CYP2A6 pathway, and those reporting illicit drug use, all of which could theoretically have been implicated in nicotine metabolism changes. One potential mechanism that may explain higher NMR after viral suppression may be diminution of the inflammatory milieu associated with HIV viremia.¹⁹ Cytochrome P450 enzyme function can increase with decreased inflammation, which may in part explain our finding.²⁰ Moreover, higher NMR has been observed in individuals with greater age and lower BMI, which may be altered after viral suppression on ART.^13,21 Faster nicotine clearance after viral suppression may make smoking cessation more difficult among PWH by increasing tobacco dependence, cravings, and withdrawal symptoms.⁸

In addition, our finding that efavirenz was an effect modifier on NMR changes is important because it suggests that the choice of ART can have a differential impact on nicotine metabolism among HIV-positive smokers. While use of efavirenz has significantly declined in high-income countries, regimens containing efavirenz remain first-line treatment in many low-income and middle-income settings.²² We found that among those on efavirenz-containing regimens, the NMR nearly doubled after viral suppression was achieved. Efavirenz, a CYP2A6 substrate,²³ has been previously implicated in higher NMR among HIV-infected smokers.¹⁵ Our study expands upon this finding by demonstrating, within subject, that the effect of efavirenz on NMR was significantly greater compared with other ART regimens, a mechanism likely mediated through the CYP2A6 metabolic pathway. One potential implication of this finding is to tailor the selection of smoking cessation medications based on ART regimens to support tobacco cessation among HIV-positive smokers. For instance, prior studies have found that slow metabolizers of nicotine (e.g., lower NMR) may have greater success with transdermal nicotine (vs. varenicline) compared with fast metabolizers.^14,24 Among HIV-positive smokers on efavirenz, a change in ART may reduce NMR.

Our study has several limitations. Unmeasured confounders may have biased our estimate of the effect of viral suppression on NMR. However, unmeasured confounders would need to have been time-varying to have a significant effect in light of having within subject paired samples. Secondly, our sample size precluded assessing multiple differential treatment effects of other ART regimens. Thirdly, we were unable to assess the degree to which NMR is influenced by HIV infection itself. Additional studies are needed to evaluate whether other ART regimens and HIV infection can impact NMR and tobacco cessation rates. Finally, we did not assess tobacco cessation rates in this cohort. Future research is warranted to test if higher NMR rates portend lower quit rates among PWH.

Our cohort study observed an association between viral suppression on ART and an increase in NMR among smokers with HIV, an effect that was substantially greater among those on efavirenz-containing regimens. These findings support that choice of ART may impact nicotine metabolism among HIV-infected smokers. Our research suggests there may be a potentially modifiable mechanism that can be targeted to decrease the numbers of deaths from lung cancer, cardiovascular and lung disease among PWH.