Chronic Human Immunodeficiency Virus Infection Is Associated with Accelerated Decline of Forced Expiratory Volume in 1 Second among Women but Not among Men: A Longitudinal Cohort Study in Uganda

To the Editor:

People living with human immunodeficiency virus (PLWH) are at increased risk of chronic lung disease and mortality (1–3). Most studies of human immunodeficiency virus (HIV)-associated lung dysfunction are focused on PLWH in North America and Europe (1), yet two-thirds of the estimated 38 million PLWH globally reside in sub-Saharan Africa. In addition, HIV is more prevalent among men in the United States (4, 5) and Europe (6), resulting in predominantly male research cohorts, yet women comprise more than half of the global population living with HIV and incur nearly two-thirds of new HIV diagnoses in sub-Saharan Africa (7). Sex-specific differences in disease risk, severity, and outcomes have been demonstrated for many chronic lung diseases (8–10) but are not characterized for HIV-associated lung dysfunction. Therefore, we estimated the association between HIV serostatus and lung function trajectory in a longitudinal cohort of men and women with and without HIV in Uganda.

Methods

The Uganda Non-Communicable Diseases and Aging Cohort is a longitudinal, observational cohort of PLWH and age- (by decade) and sex-matched HIV-uninfected adults in southwestern Uganda. PLWH were at least 40 years of age, on antiretroviral therapy for at least 3 years, and in care at Mbarara Regional Referral Hospital’s HIV clinic. HIV-uninfected participants were recruited from their homes from a census-based study in a nearby cluster of villages based on their age and sex. Sex was identified based on clinical documentation in participants’ records or census tract data. Participants completed annual spirometry from 2015 through 2018. Post-bronchodilator forced expiratory volume in one second (FEV₁) and forced vital capacity (FVC) were quantified (nDD Medical Technologies) per American Thoracic Society guidelines (11). Approval was obtained from Ugandan and U.S. research ethics committees, and participants provided informed consent.

We compared participant demographics by HIV serostatus and fit generalized mixed-effects linear regression models with random intercepts and slopes, and an HIV × time product term, to characterize the association between HIV serostatus and FEV₁ and FVC trajectories. We adjusted models for age, sex, height, smoking, socioeconomic status, and prior tuberculosis/pneumonia. We did not adjust models for biomass smoke exposure because nearly all study participants live in homes where biomass fuels are burned for cooking and/or heating, so there is no unexposed comparator population. We conducted preplanned analyses stratified by sex (Stata 16; StataCorp).

Results

A total of 278 (97%) participants completed 613 spirometry tests over a median of two visits (interquartile range [IQR], 1–2 visits) and 1.5 years of follow-up (IQR, 1.0–2.0 yr). Most spirometry (91%) met American Thoracic Society criteria. Among participants, 52% were PLWH, 47% were women, and median age was 52 years (IQR, 48–55 yr) (Table 1). Few (6%) had prior pulmonary tuberculosis, all of whom were PLWH and most of whom (16/17) were men. There were no substantive demographic differences between women with versus without HIV. Among PLWH, most (82%) had a CD4 (cluster of differentiation 4) ⩾ 350 cells/mm³, and 94% had undetectable viral loads. Baseline lung function did not differ by HIV serostatus, either among the entire cohort or among women specifically.

Table 1.

Baseline characteristics of study participants

	Total Cohort (N = 278)	HIV+ (n = 145)	HIV− (n = 133)
Demographics
Age	52 (48–55)	52 (49–55)	52 (48–55)
Women	130 (47)	68 (47)	62 (47)
Smoking history
Current	41 (15)	12 (8)	29 (22)
Former	95 (34)	50 (34)	45 (34)
Never	142 (51)	83 (57)	59 (44)
Prior tuberculosis	17 (6)	17 (12)	0 (0)
Prior pneumonia	24 (9)	17 (12)	7 (5)
Biomass exposure
Charcoal	40 (14)	39 (27)	1 (1)
Firewood	236 (86)	104 (73)	132 (99)
Subsistence farming	184 (66)	76 (52)	108 (81)
Socioeconomic status
Poorest	64 (23)	28 (19)	36 (27)
Poorer	74 (27)	27 (19)	47 (35)
Richer	72 (24)	36 (25)	32 (24)
Richest	72 (26)	54 (37)	18 (14)
Lung function*
FEV1, L	2.52 (0.58)	2.51 (0.61)	2.52 (0.55)
FVC, L	3.19 (0.71)	3.20 (0.73)	3.19 (0.68)
FEV1/FVC	0.79 (0.06)	0.78 (0.07)	0.79 (0.06)
FEV1/FVC < LLN	11 (4)	9 (6)	2 (2)
HIV characteristics
Viral load, copies/μl
Undetectable	—	135 (94)	—
Detectable, ⩽10,000	—	6 (4)	—
Detectable, >10,000	—	2 (1)	—
CD4 count, cells/mm3
100–349	—	26 (18)	—
350–499	—	55 (38)	—
⩾500	—	64 (44)	—
ART duration	—	9 (8–10)	—
ART regimen
AZT/3TC/NVP or EFV	—	113 (78)	—
TDF/3TC/NVP or EFV	—	19 (13)	—
TDF/3TC/LPV/r	—	12 (8)	—
AZT/3TC/ABC	—	1 (1)	—

Definition of abbreviations: 3TC = lamivudine; ABC = abacavir; ART = antiretroviral therapy; AZT = zidovudine; EFV = efavirenz; FEV₁ = forced expiratory volume in 1 second; FVC = forced vital capacity; HIV = human immunodeficiency virus; LLN = lower limit of normal, applying National Health and Nutrition Examination Survey III lung function prediction equations; LPV/r = lopinavir/ritonavir; NVP = nevirapine; TDF = tenofovir.

Data are presented as median (interquartile range) or n (%). Lung function parameters are presented as mean (standard deviation).

^*Baseline lung function summarized for the 266 participants with American Thoracic Society acceptable spirometry.

In adjusted main effects models, FEV₁ and FVC declined by −24.4 (95% confidence interval [CI], −51.1 to 2.3; P = 0.07) and −34.5 (95% CI, −64.2 to −4.7; P = 0.02) ml/yr, respectively, neither of which differed by HIV serostatus. In sex-stratified models, women with HIV (WWH) had more accelerated FEV₁ decline than women without HIV (additional −46.3 ml/yr; 95% CI, −91.3 to −1.4; P for interaction = 0.04) (Figure 1). There was no HIV-associated difference in FEV₁ decline among men or in FVC decline among women or men (Table 2). The three-way interaction term between HIV serostatus, time, and sex was not statistically significant.

Figure 1.

Change in FEV₁ and FVC over the course of the study, by HIV serostatus and sex. FEV₁ = forced expiratory volume in 1 second; FVC = forced vital capacity; HIV = human immunodeficiency virus.

Table 2.

Correlates of lung function among people with and without human immunodeficiency virus in southwestern Uganda

	Total Cohort (N = 266)		Women (n = 126)		Men (n = 140)
	Δ FEV1	95% CI	Δ FEV1	95% CI	Δ FEV1	95% CI
Characteristic
Age, per year	−13.4	−19.4 to −7.4	−9.1	−15.9 to −2.3	−17.7	−27.8 to −7.6
Female sex	−581.3	−702.1 to −460.5
HIV positive	−23.0	−127.1 to 81.2	55.4	−60.2 to 171.0	−89.1	−260.4 to 82.1
Per study year	−24.4	−51.1 to 2.3	−21.6	−53.0 to 9.7	−20.9	−63.8 to 22.0
HIV × year	−25.8	−62.4 to 10.9	−46.3	−91.3 to −1.4	−11.5	−68.5 to 45.5
Height, per cm	15.9	9.2 to 22.6	14.6	6.7 to 22.5	18.5	8.0 to 29.0
Current smoking	41.9	−48.4 to 132.1	−95.1	−245.7 to 55.5	82.7	−38.5 to 203.8
Prior TB or pneumonia	−41.0	−91.8 to 9.8	−14.5	−78.3 to 49.3	−61.5	−140.3 to 17.2
Socioeconomic status
Lowest quartile		REF		REF		REF
Lower quartile	−1.6	−57.8 to 54.5	13.1	−54.6 to 80.7	−30.1	−119.7 to 59.5
Higher quartile	−34.8	−99.8 to 30.3	−42.2	−125.7 to 41.3	−40.0	−135.5 to 55.4
Highest quartile	15.2	−65.7 to 96.1	3.3	−101.0 to 107.7	22.1	−95.7 to 139.9
Characteristic
Age, per year	−11.4	−18.1 to −4.6	−10.6	−18.9 to −2.4	−12.1	−23.2 to −1.0
Female sex	−749.0	−884.4 to −613.7
HIV positive	−15.2	−131.9 to 101.5	44.8	−92.4 to 181.9	−62.8	−251.8 to 126.3
Per study year	−34.5	−64.2 to −4.7	−42.7	−81.9 to −3.6	−22.5	−68.1 to 23.1
HIV × year	−19.3	−60.2 to 21.6	−27.2	−83.4 to 29.0	−14.7	−75.1 to 45.7
Height, per cm	23.6	16.0 to 31.2	20.6	11.1 to 30.1	27.1	15.5 to 38.6
Current smoking	32.8	−69.5 to 135.1	−79.9	−267.2 to 107.5	84.4	−47.8 to 216.7
Prior TB or pneumonia	−32.3	−90.7 to 26.1	−0.9	−80.3 to 78.6	−59.1	−144.9 to 26.7
Socioeconomic status
Lowest quartile		REF		REF		REF
Lower quartile	−25.6	−89.8 to 38.7	1.2	−82.8 to 85.1	−72.2	−169.8 to 25.5
Higher quartile	−88.0	−162.0 to −14.0	−69.2	−172.2 to 33.7	−124.4	−228.3 to −20.6
Highest quartile	−26.3	−118.1 to 65.4	−17.3	−144.5 to 110.0	−47.3	−175.3 to 80.8

Definition of abbreviations: CI = confidence interval; FEV₁ = forced expiratory volume in 1 second (in milliliters); HIV = human immunodeficiency virus; REF = reference group; TB = tuberculosis.

Models adjusted for age, sex, HIV serostatus, height, smoking status, a history of TB or pneumonia, and socioeconomic status (asset index, a principal components analysis based on asset ownership, in which higher quartiles correlate with higher wealth).

Discussion

In this longitudinal cohort of mostly nonsmoking Ugandan adults, FEV₁ decline was accelerated among WWH compared with women without HIV. By contrast, there was no difference in FEV₁ decline among men by HIV serostatus or in FVC decline by HIV serostatus among either sex. To our knowledge, this is the first demonstration of a sex-specific difference in HIV-associated FEV₁ decline.

Prior data characterizing the relationship between HIV serostatus and lung function trajectory describes mixed results. FEV₁ decline among PLWH on antiretroviral therapy ranges from 23–62 ml/yr (12–15), consistent with our findings. By contrast, Li and colleagues described a more rapid FVC decline (12) than was observed in our cohort, which may be due to a higher smoking prevalence. None of these studies included an HIV-uninfected comparator group with which to evaluate HIV-associated differences in lung function decline, nor did they explore sex-specific differences in these predominantly male cohorts. HIV serostatus was not associated with accelerated FEV₁ or FVC decline in two additional cohorts of people with and without HIV (16, 17), though sex-specific estimates were not reported.

We offer three potential hypotheses to explain our findings. Women have a greater immune response to HIV infection, which acutely results in better HIV control (18, 19), but chronic inflammation among PLWH is associated with lung dysfunction (2). We have shown that WWH in this cohort have higher systemic inflammation and that systemic inflammation is associated with lower FEV₁ among PLWH (20, 21). Air pollution exposure differences may also underlie our findings. Women in sub-Saharan Africa are exposed to more household air pollution because of gendered meal preparation roles (22), and air pollution causes acute and chronic lung dysfunction (23). Indeed, air pollution exposure is higher among women and PLWH in this cohort (24). Extrapolating the known synergistic effects of HIV and smoking (25), WWH exposed to higher amounts of air pollution may have elevated risks of lung dysfunction. Last, sex hormone–based differences may also explain our findings. WWH may experience premature ovarian failure and early menopause (26), which has been associated with lung dysfunction and accelerated lung function decline (27, 28), potentially via sex steroid–associated lung inflammation.

The main strengths of this analysis are that it is based within a well-phenotyped longitudinal cohort that includes an even balance of women, an HIV-uninfected comparator group, and objective measures of lung function. However, the 2 years of follow-up may not have been sufficient to identify clinically meaningful changes in lung function (although we would expect this limit to bias our results toward a finding of no sex-based differences). Data confirming sex assigned at birth were not available, so there may be a small degree of misclassification by sex, which we expect to be random in nature. Although WWH had accelerated lung function decline, baseline lung function was not different between women with versus without HIV, which may be related to the as-yet-undefined duration of time required for the local and systemic effects of chronic HIV infection to result in lung dysfunction. The wide CI around our point estimate may suggest that the accelerated FEV₁ decline is clinically insignificant. However, on a population level, a slightly accelerated FEV₁ decline could significantly increase the proportion of WWH who develop lung dysfunction (29). Although the interaction term between HIV serostatus, sex, and time was not statistically significant, we are likely underpowered for three-way interaction terms in this cohort. Our findings would not achieve statistical significance if multiple testing was considered (P value threshold, 0.025). Finally, participants did not undergo total lung capacity or diffusing capacity of the lung for carbon monoxide measurements, which may have resulted in underestimation of HIV-associated lung dysfunction (30).

In conclusion, WWH in Uganda may be particularly susceptible to chronic lung disease. Future work is needed to explore the potentially causative role of immune regulation, air pollution, and/or sex hormones on lung function decline among PLWH.