Leptin, Adiponectin and Cognition in Middle-aged HIV-infected and Uninfected Women. The Brooklyn Women’s Interagency HIV Study

Abstract

Context

Case-control study of women with and without HIV infection.

Objective

To explore the association of cognition and the adipokines, leptin and adiponectin (total; high molecular weight, HMW), in women with (HIV+) and without HIV (HIV−) infection.

Design

Cross-sectional analyses of adipokines and cognition using linear regression models of log-transformed adipokines, and Trails A, Trails B, Stroop interference time, Stroop word recall, Stroop color naming and reading, and Symbol Digit Modalities Test (SDMT) with consideration for age, HIV infection status, education, CD4 count, diabetes, body mass index (BMI), waist circumference (WC) and race/ethnicity.

Setting

Brooklyn, NY.

Participants

354 participants (247 HIV+, 107 HIV−), in the Brooklyn Women’s Interagency HIV Study (WIHS), average age 38.9 years, with measured levels of leptin and adiponectin (total and high molecular weight, HMW).

Main Outcome Measure

Cognition

Results

Higher levels of leptin were positively associated with worse cognition on the basis of Trails A completion time and SDMT score. Among at risk HIV− women, leptin was associated with worse performance on Trails B. No associations were observed for total or HMW adiponectin.

Conclusion

Blood adipokine levels were measured to provide mechanistic insights regarding the association of adipose with cognitive function. These data suggest that higher levels of leptin, consistent with more adipose tissue, are associated with worse cognitive function in middle age. Monitoring leptin over time and with increasing age in relation to cognition and dementia, may lend insights to the role of adipose tissue in successful body and brain aging among women with HIV infection.

Introduction

Human Immunodeficiency Virus (HIV) infection is one of the most prevalent infectious diseases worldwide. With current therapies, HIV infection has evolved from a fatal infection, to a treatable, chronic condition of aging among older adults [1,2]. In 2012, 50% of all Ryan White HIV/AIDS Program clients were age 45–64 years and older; [3] and data suggest that over half of HIV-infected people in the United States (US) today are age 50 years and older [4]. This increase in life expectancy is necessarily accompanied by concerns about the concomitant physiological, social and mental consequences of aging with chronic HIV infection. Some studies show that HIV infection is synergistic with adverse influences of aging on the immune, endocrine, vascular, and central nervous systems, thereby accelerating the aging process [5,6]. Evaluating these systems in middle-aged HIV infected persons is essential to understanding HIV impact on the natural processes of aging.

In many countries, the advent of ART has been accompanied by an increase in body mass index (BMI). Thus, in contrast to the wasting syndrome that initially characterized HIV infection and AIDS, individuals with HIV infection and on antiretroviral therapies (ART) are overweight and obese. Importantly, overweight and obesity are leading causes of disability and death in the United States and around the world, and are associated with risk of cognitive impairment and late-onset dementia in populations without HIV [7].

During adult life, BMI is a gross indicator of amount of adipose tissue in the human body. This is mainly white adipose tissue, which functions as the largest endocrine organ in the human body by secreting many hormones, peptides and cytokines, some of which are referred to as adipokines [8]. These adipokines are not exclusively secreted by adipose tissue, nor is adipose tissue the primary source of these compounds, however their bioactivity is indisputable [9]. Adipokines affect processes in the periphery and the central nervous system (CNS), and may be dysregulated in individuals with HIV infection, whether on or off ART, perhaps due to lipodystrophies [10]. It is important to try to partition out the bioactive aspects of adipose tissue and their relationship to cognition in individuals with HIV infection. Using blood adipokine levels may be one way to do this.

Leptin is an adipose tissue hormone that has been most reported in association with dementia and Alzheimer’s Disease (AD). Leptin is a 16 kDa protein hormone [11,12] that is primarily secreted by adipose tissue and positively correlated with anthropometric measures [13,14]. Correlations of approximately r=0.7 between BMI and blood leptin levels are observed in adults, even among those with obesity syndromes [15,16]. There have been numerous reports from human population studies showing that higher leptin levels are associated with less cognitive impairment and dementia in later life. The Framingham Heart Study, [17] Study of Osteoporotic Fractures, [18] and Health ABC Study, [19] have all shown leptin levels to be protective for dementia and Mild Cognitive Impairment (MCI), or cognitive decline when measured 4–8.3 years before onset of disease. During this prodromal period, leptin may merely be a marker of adipose tissue since higher levels of BMI have also shown protection for dementia during this period; or leptin may have independent, beneficial effects on the brain and/or reflect neurodegenerative processes. However, it has also been observed that leptin levels 24 years prior to dementia onset are not predictive of dementia risk [15].

Adiponectin is produced solely by adipose tissue and exists as complex multimeric isoforms comprised of High Molecular Weight (HMW), hexamers and trimmers [20]. It is an effective insulin sensitizer, and circulating levels are inversely correlated with insulin resistance, metabolic syndrome, obesity, type 2 diabetes, and cardiovascular diseases. HMW adiponectin or the ratio of HMW adiponectin to total adiponectin may be better indicators of insulin sensitivity than total adiponectin in obesity, diabetes, and cardiovascular disease [20]. Adiponectin modulates inflammatory responses, energy expenditure (CNS and periphery), food intake (CNS), and a number of metabolic processes, including glucose regulation and fatty acid catabolism in the periphery [21]. Adiponectin has also been implicated in cognition and dementia, but data are mixed and available for total adiponectin only. The Framingham Heart Study [22] and the Mayo Clinic Study on Aging (MCSA) [23] have shown mixed associations between total adiponectin and dementia. Framingham reported that higher total adiponectin levels were associated with all-cause dementia and AD risk over a 13 year follow-up; whereas there was no cross-sectional difference by Mild Cognitive Impairment (MCI) status in the MCSA.

In the present study, we determined whether leptin and adiponectin levels are differentially associated with concurrent measures of cognitive performance in women who are HIV+ or HIV− and are participants at the Brooklyn site of the Women’s Interagency HIV Study (WIHS). Demographic, cardiovascular, and HIV-related factors were considered as confounders.

Materials and Methods

WIHS is an ongoing prospective study of HIV in women [24]. WIHS began in 1994 and enrolled 3766 women across six sites in San Francisco, Los Angeles, Chicago, Washington, DC, Brooklyn and the Bronx. Across all sites, WIHS initially recruited 2054 HIV infected (HIV+) and 569 HIV uninfected (who in WIHS, are their ‘at risk’ counterparts, HIV−) women in 1994–95, and an additional 737 HIV-infected and 406-HIV uninfected women in 2001–2002. The Brooklyn WIHS site has participated since the WIHS’ inception. Three hundred and fifty-four of the participants (247 HIV+, 107 HIV−) enrolled in Brooklyn WIHS had available cognitive test scores.

Demographic measures

All demographic measures were self-reported. Race [24,25] was reported as white, Hispanic, African-American, or ‘other’ (self-reported as Native American/Alaskan, Asian/Pacific Islander or other). Participants were also asked to report their current smoking status, and use of marijuana, ‘crack’, cocaine, and heroin.

Clinical measures

Anthropometric measures were conducted according to the U.S. National Health and Nutrition Examination Survey (NHANES) III protocol and included body weight (pounds), body height (inches), waist (WC) and hip circumferences (cm), and BMI (kg/m²) [26]. Anthropometric measurements were conducted with participants wearing only undergarments. Those who conduct the measurements are recertified every two years. Body weight was recorded to the nearest 1.0 pound, and body height was measured to the nearest 1.00 inch. After conversion of body weight and height to metric units, BMI was calculated as kilograms per meter squared (kg/m²). Categories of BMI included: underweight, <18.5; ‘normal’ or healthy weight, 18.5–24.9; overweight, 25.0–29.9; and obese, ≥30 kg/m² [27]. WC and hip circumference were measured to the nearest 0.5 cm. WHR was calculated as the ratio of WC to hip circumference. Central obesity was defined as WHR >0.85 [28].

Eight hour fasted blood samples were collected and total cholesterol levels were determined as previously described [29]. Systolic (SBP) and diastolic blood pressures (DBP) were recorded using a standardized protocol [30]. Hypertension was defined as either average measured SBP >140 mm Hg, or DBP >90 mm Hg, or a self-reported hypertension with use of antihypertensive medications. Previous myocardial infarction (MI) and diabetes mellitus (DM) were self-reported [24,25].

Adipokine measures

Plasma samples for adipokine analyses were drawn within one visit of cognitive test administration. Standards and controls were tested in duplicate using High Molecular Weight (HMW) adiponectin, total adiponectin, and leptin ELISAs (Millipore, Billerica, MA) [20,31]. For leptin, undiluted samples were tested and plates were prepared according to protocol. The 7-point standard curve ranged from 0.5–100 ng/mL. Plates were read using a Molecular Devices Plate reader and Softmax Pro data analysis software (Molecular Devices, Sunnyvale, CA). A 4-point logistic (PL) curve fit was used. For HMW adiponectin, samples were digested as directed in the assay protocol using reagents provided in the kit, the 7-point standard curve ranged from 1.5 to 200 ng/mL and samples were tested at a final dilution of 1:200. For total adiponectin, the diluted samples were assayed and plates were prepared according to protocol. The 7-point standard curve range was 1.56–100 ng/mL and tested at a final dilution of 1:500.

HIV-related variables

Methods for determining HIV status, AIDS diagnosis, CD4 count, viral load, and duration of ART use were described previously [24,25,32].

Cognitive tests

Cognitive tests were administered to all English-speaking WIHS participants during visits 21 to 24 (October 2004 to September 2006) as part of the WIHS core assessment; the Comalli-Kaplan Stroop was administered to a subgroup during visits 25–28, October 2006 to September 2008 (Table 1). These tests have been previously described [29]. Among participants who completed testing on multiple visits, and therefore have more than one score, only the first score was used. Times greater than 240 seconds were coded as 240 seconds. Errors were recorded, but were not used to adjust interference times. For all cognitive tests, we used raw scores rather than normalized data.

Table 1.

Cognitive tests administered in WIHS and corresponding cognitive domains measured [29].

Cognitive Domain	Test
Executive Function	Trails A, Trails B Stroop Interference[63,64]
Speed of Information Processing	Symbol Digit Modalities Test (SDMT)* [65,66] Stroop Color Naming and Reading
Learning and Memory	Stroop Word Recall

^*The SDMT score is the number of correct items in 90 seconds, all other test scores are times with lesser time indicating better performance.

Inclusion criteria

We include all data collected by visit 28, concluding in September 2008 on 354 participants (247 HIV+, 107 HIV−) with data available on both anthropometric and cognitive measures.

Statistical analysis

All adipokines were log-transformed and considered as continuous variables. We estimated leptin resistance via the leptin: BMI [33]. Linear regression analyses were used to examine associations between continuous or categorical log adipokines and cognitive test scores (or time to completion) of Trails A, Trails B, SDMT score, Stroop interference, Stroop Color Naming, and Stroop Word Recall. Regression models were run including all participants, and separately for HIV+ and HIV− individuals.

Several covariates were considered, including: age, race, highest educational level attained, HIV status, ART, CD4 count, CD4 nadir, prevalent DM, SBP, DBP, use of anti-hypertensive medications, blood cholesterol level, current smoking status, and use of marijuana, ‘crack’, cocaine, and/or heroin. Potential covariates were included if they were significantly associated with cognitive test scores and/or adipokines in age-adjusted models at a level of p<0.05. The final models included the following covariates: age, education, race, DM, and BMI. In analyses of women who were HIV+, we also adjusted for current CD4 count. STATA 12 was used for all statistical analyses. Results were considered statistically significant at p<0.05.

Results

Both adipokines and cognitive measures were available for 354 Brooklyn WIHS participants (213 HIV+ and 97 HIV− women). Demographic, anthropometric, and health characteristics of the participants are presented in Table 2. HIV+ women were approximately 3.5 years older than HIV− women, however educational attainment, a key influencer of cognitive performance, did not differ between HIV+ and HIV− women. Most (65%) women were overweight or obese (≥25.0 kg/m²) and the prevalence of central obesity was 60%. Less than 2% of women had a BMI <18.5, probably in concordance with the low prevalence of AIDS among those who were HIV+. HIV+ women had a higher WHR (P<0.05) and there was a trend for them (p<0.10) to have a lower average BMI, compared to HIV− women. Among published vascular risk factors for cognitive impairment, there were no differences.

Table 2.

Demographic and health characteristics of Brooklyn WIHS participants with adipokine and cognitive measures.

Characteristic	ALL (n=354)		HIV+ (n=247)		HIV− (n=107)
	N	Mean (SD)/n(%)	N	Mean (SD)/n(%)	N	Mean (SD)/n(%)	p-value
Age	354	38.9 (9.1)	247	40.0 (8.6)	107	36.4 (9.8)	<0.001
Race	354		247		107		0.465
White		31 (8.7%)		24 (9.7%)		7 (6.5%)
African American (AA)		280 (79.0%)		193 (78.2%)		87 (81.5%)
Non-white, non-AA Hispanic		34 (9.6%)		23 (9.3%)		11 (10.2%)
Other		9 (2.7%)		7 (2.8%)		2 (1.8%)
Highest education	354		247		107		0.285
Grades 7–11		124 (35.1%)		91 (36.8%)		33 (31.1%)
Completed HS		127 (36.0%)		88 (35.6%)		39 (26.8%)
Some college		86 (24.4%)		58 (23.5%)		28 (26.4%)
4-yr degree		14 (3.9%)		10 (4.1%)		4 (3.8%)
attend/complete grad school		2 (0.6%)		0		2 (1.9%)
CD4 count, need log			245	518.3 (323.4)
Viral load			244	27,619 (140,879)
Leptin (ng/ml)	353	30.8 (29.7)	246	29.0 (28.7)	107	34.8 (31.7)	0.091
HMW Adiponectin (ng/ml)	348	5074.8 (5009.7)	244	5283.7 (5488.1)	104	4584.5 (3626.0)	0.234
Total Adiponectin (ng/ml)	354	6998.5 (3873.6)	247	7035.6 (4124.2)	107	6912.9 (3238.1)	0.785
HMW/Total Adiponectin	348	0.68 (0.36)	244	0.70 (0.39)	104	0.63 (0.30)	0.115
BMI (kg/m2)	349	29.2 (7.9)	245	28.8 (7.3)	104	30.3 (9.1)	0.091
<18.5		6 (1.7%)		5 (2.0%)		1 (1.0%)
18.5–24.9		115 (33.0%)		83 (33.9%)		32 (30.8%)
25.0–29.9		104 (29.8%)		75 (30.6%)		29 (27.9%)
≥30.0		124 (35.5%)		82 (33.5%)		42 (40.4%)
WHR	264	0.88 (0.08)	180	0.89 (0.08)	84	0.85 (0.07)	<0.001
>0.85		157 (59.5%)		119 (66.1%)		38 (45.2%)	0.002
Waist Circumference, WC (cm)	266	91.1 (16.2)	181	90.9 (15.3)	85	91.3 (17.8)	0.821
Marijuana Use since last visit	354	64 (18.1%)	247	41 (16.6%)	107	23 (21.5%)	0.294
Any indicator of hypertension*	354	127 (35.9%)	247	94 (38.1%)	107	33 (30.8%)	0.119
Total Cholesterol (mg/dl)	354	176.8 (36.5)	247	175.4 (37.4)	107	180.1 (34.1)	0.265
Diabetes mellitus (yes)	312	15 (4.8%)	217	9 (4.2%)	95	6 (6.3%)	0.401

^*Either SBP ≥ 140, DBP ≥90, self-reported hypertension, or taking anti-hypertensive medication.

Correlations between anthropometric measures and adipokines indicated strong associations in the directions expected (Tables 3 and 4). Leptin was positively, and adiponectin inversely, correlated with anthropometric measures. Notably the correlation coefficients were quite strong for each hormone in line with previous reports in the literature in non-HIV population samples [15]. An exception to this was the correlation between leptin and WHR among HIV+ women. In addition, viral load was correlated with log leptin only after age adjustment; and CD4 count was correlated with all adipokine measures except for the ratio of HMW/total adiponectin.

Table 3.

Correlations between anthropometric measures and adipokines: Brooklyn WIHS.

	BMI					Waist Circumference					WHR
		crude		age-adjusted			crude		age-adjusted			crude		age-adjusted
	n	r	p-value	r	p-value	n	r	p-value	r	p-value	n	r	p-value	r	p-value
Log Leptin	348	0.706	<0.0001	0.707	<0.0001	265	0.699	<0.0001	0.709	<0.0001	263	0.156	0.011	0.182	0.003
HMW adiponectin	343	−0.294	<0.0001	−0.298	<0.0001	262	−0.334	<0.0001	−0.345	<0.0001	260	−0.265	<0.0001	−0.303	<0.0001
Total adiponectin	349	−0.260	<0.0001	−0.266	<0.0001	266	−0.314	<0.0001	−0.325	<0.0001	264	−0.282	<0.0001	−0.321	<0.0001
HMW/total adiponectin ratio	343	−0.292	<0.001	−0.292	<0.0001	262	−0.263	<0.0001	−0.263	<0.0001	260	−0.206	<0.001	−0.213	<0.001

Table 4.

Correlations between anthropometric measures and adipokines by HIV infection status: Brooklyn WIHS.

	Log Leptin			Log Adiponectin			Log Total Adiponectin			Log Total/HMW Adiponectin Ratio
	n	r	p-value	n	r	p-value	n	r	p-value	n	r	p-value
BMI - HIV+
crude	244	0.708	<0.0001	242	−0.316	<0.0001	245	−0.271	<0.0001	242	−0.235	0.0002
age-adjusted	244	0.706	<0.0001	242	−0.314	<0.0001	245	−0.264	<0.0001	242	−0.241	0.0002
BMI – HIV−
crude	104	0.707	<0.0001	101	−0.484	<0.0001	104	−0.277	0.004	101	−0.452	<0.0001
age-adjusted	104	0.697	<0.0001	101	−0.478	<0.0001	104	−0.281	0.004	101	−0.446	<0.0001
WC - HIV+
crude	180	0.695	<0.0001	179	−0.315	<0.0001	181	−0.312	<0.0001	179	−0.196	0.009
age-adjusted	180	0.699	<0.0001	179	−0.316	<0.0001	181	−0.313	<0.0001	179	−0.196	0.009
WC– HIV−
crude	85	0.720	<0.0001	83	−0.489	<0.0001	85	−0.351	0.001	83	−0.412	0.0001
age-adjusted	85	0.721	<0.0001	83	−0.473	<0.0001	85	−0.355	0.001	83	−0.384	0.0004
WHR - HIV+
crude	179	0.085	0.259	178	−0.335	<0.0001	180	−0.332	<0.0001	178	−0.205	0.006
age-adjusted	179	0.112	0.136	178	−0.349	<0.0001	180	−0.360	<0.0001	178	−0.204	0.007
WHR – HIV−
crude	84	0.419	0.0001	82	−0.497	<0.0001	84	−0.398	0.0002	82	−0.388	0.0003
age-adjusted	84	0.402	0.0002	82	−0.485	<0.0001	84	−0.413	0.0001	82	−0.355	0.001
Viral load
crude	243	−0.125	0.052	241	0.079	0.220	244	0.089	0.168	241	0.042	0.518
age-adjusted	243	−0.128	0.046	241	0.081	0.212	244	0.096	0.137	241	0.040	0.542
CD4
crude	272	0.146	0.016	268	−0.159	0.009	273	−0.169	0.005	268	−0.099	0.107
age-adjusted	272	0.144	0.018	268	−0.159	0.009	273	−0.167	0.006	268	−0.101	0.101
Nadir CD4
crude	221	0.062	0.357	219	−0.058	0.394	222	−0.047	0.486	219	−0.053	0.432
age-adjusted	221	0.040	0.551	219	−0.054	0.427	222	−0.010	0.885	219	−0.078	0.250

Associations with cognitive test scores (Table 5), revealed that higher leptin levels were associated with longer time to completion of Trails A in both HIV+ and HIV− women after multivariate adjustment (Model 2). Similar associations were also observed for Trails B performance, but only among those without HIV infection. Also observed was an inverse association of leptin with SDMT among women without HIV infection. There were no linear associations observed for total or HMW adiponectin. Calculated leptin resistance was not informative (data not shown).

Table 5.

Leptin and adiponectin in association with cognitive test score, by HIV infection status: Brooklyn WIHS.

		All					HIV+			HIV−
Log Adipokine		n			b(95% CI)	p-value	n	b(95% CI)	p-value	n	b(95% CI)	p-value
Trails A
Leptin
Model 1		353			0.90 (−0.63, 2.42)	0.248	246	0.77 (−1.20, 2.74)	0.444	107	2.26 (0.05, 4.46)	0.045
Model 2		306			3.16 (0.97, 5.35)	0.005	215	2.96 (0.12, 5.81)	0.041	91	4.84 (1.58, 8.11)	0.004
HMW adiponectin
Model 1		348			−1.35 (−3.81, 1.10)	0.279	244	−0.91 (−4.01, 2.18)	0.561	104	−2.91 (−6.59, 0.78)	0.121
Model 2		302			−2.13 (−4.84, 0.58)	0.123	213	−1.98 (−5.32, 1.37)	0.245	89	−1.49 (−6.32, 3.34)	0.541
Total adiponectin
Model 1		354			2.30 (−0.91, 5.51)	0.160	247	2.32 (−1.66, 6.30)	0.251	107	1.68 (−3.53, 6.89)	0.523
Model 2		307			1.05 (−2.44, 4.55)	0.554	216	2.77 (−1.73, 7.26)	0.227	91	−2.71 (−8.61, 3.18)	0.363
HMW adiponectin/total adiponectin ratio
Model 1		348			−1.35 (−3.81, 1.10)	0.279	244	−0.91 (−4.01, 2.18)	0.561	104	−2.91 (−6.59, 0.78)	0.121
Model 2		302			−3.55 (−8.48, 1.39)	0.159	213	−3.77 (−9.71, 2.18)	0.213	89	−2.85 (−12.13, 6.43)	0.542
Trails B
Leptin
Model 1		352			1.64 (−3.26, 6.53)	0.512	245	−0.94 (−7.39, 5.52)	0.776	107	8.29 (1.50, 15.08)	0.017
Model 2		305			7.67 (0.75, 14.58)	0.030	214	4.86 (−4.01, 13.74)	0.281	91	15.48 (4.50, 26.44)	0.006
HMW adiponectin
Model 1		347			0.85 (−7.05, 8.76)	0.832	243	2.49 (−7.62, 12.61)	0.628	104	−4.48 (−16.10, 7.13)	0.446
Model 2		301			−2.11 (−10.58, 6.36)	0.624	212	−2.58 (−12.81, 7.65)	0.619	89	−3.48 (−19.77, 12.80)	0.671
Total adiponectin
Model 1		353			6.94 (−3.37, 17.26)	0.186	246	12.59 (−0.37, 25.55)	0.057	107	−8.69 (−24.80, 7.42)	0.287
Model 2		306			4.51 (−6.40, 15.42)	0.416	215	12.42 (−1.31, 26.16)	0.076	91	−18.22 (−37.64, 1.21)	0.066
HMW adiponectin/total adiponectin ratio
Model 1		347			0.85 (−7.05, 8.76)	0.832	243	2.49 (−7.62, 12.61)	0.628	104	−4.48 (−16.10, 7.13)	0.446
Model 2		301			−3.14 (−18.57, 12.29)	0.689	212	−2.48 (−20.65, 15.69)	0.788	89	−16.21 (−47.31, 14.90)	0.303
Symbol Digit Modalities Test (SDMT)
Leptin
Model 1		348			0.37 (−0.75, 1.49)	0.515	242	0.73 (−0.66, 2.13)	0.303	106	−0.80 (−2.71, 1.10)	0.405
Model 2		302			−1.01 (−2.53, 0.50)	0.190	212	−0.33 (−2.18, 1.52)	0.727	90	−3.22 (−5.96, −0.48)	0.022
HMW adiponectin
Model 1		343			−0.33 (−2.12, 1.46)	0.716	240	−0.26 (−2.43, 1.90)	0.810	103	−0.28 (−3.50, 2.93)	0.862
Model 2		298			0.38 (−1.48, 2.25)	0.686	210	0.52 (−1.63, 2.68)	0.633	88	−1.42 (−5.44, 2.60)	0.485
Total adiponectin
Model 1		349			−0.83 (−3.17, 1.51)	0.486	243	−1.28 (−4.08, 1.52)	0.370	106	0.47 (−3.97, 4.90)	0.835
Model 2		303			0.46 (−1.93, 2.85)	0.706	213	−0.67 (−3.58, 2.23)	0.649	90	3.94 (−0.87, 8.75)	0.107
HMW adiponectin/total adiponectin ratio
Model 1		343			−0.33 (−2.12, 1.46)	0.716	240	−0.26 (−2.43, 1.90)	0.810	103	−0.28 (−3.50, 2.93)	0.862
Model 2		298			0.16 (−3.23, 3.56)	0.925	210	0.35 (−3.50, 4.21)	0.856	88	−0.80 (−8.56, 6.96)	0.837
Stroop – Color
Leptin
Model 1		313			1.64 (−0.23, 3.51)	0.086	215	2.12 (−0.30, 4.55)	0.086	98	1.07 (−1.81, 3.96)	0.463
Model 2		274			1.58 (−1.14, 4.30)	0.254	189	1.51 (−2.00, 5.03)	0.397	85	0.61 (−3.67, 4.89)	0.777
HMW adiponectin
Model 1		308			0.55 (−2.54, 3.64)	0.726	213	−0.14 (−3.93, 3.65)	0.944	95	2.13 (−3.20, 7.46)	0.430
Model 2		270			−0.08 (−3.40, 3.24)	0.963	187	−1.48 (−5.55, 2.60)	0.475	83	2.13 (−3.95, 8.21)	0.487
Total adiponectin
Model 1		314			−2.36 (−6.37, 1.65)	0.248	216	−2.47 (−7.40, 2.45)	0.324	98	−1.75 (−8.78, 5.28)	0.623
Model 2		275			−3.86 (−8.24, 0.50)	0.083	190	−4.04 (−9.58, 1.50)	0.152	85	−2.00 (−9.54, 5.55)	0.600
HMW adiponectin/total adiponectin ratio
Model 1	308		−0.06 (−3.25, 3.14)			0.973	213	−0.14 (−3.93, 3.65)	0.944	95	2.13 (−3.20, 7.46)	0.430
Model 2	270		1.99 (−4.27, 8.25)			0.532	187	−0.59 (−8.14, 6.97)	0.879	83	7.58 (−4.03, 19.19)	0.197
Stroop – Word
Leptin
Model 1	314		1.12 (−0.44, 2.69)			0.158	216	1.91 (−0.05, 3.87)	0.056	98	−0.07 (−2.75, 2.60)	0.956
Model 2	275		0.85 (−1.44, 3.13)			0.467	190	0.96 (−1.86, 3.78)	0.503	85	−0.07 (−4.29, 4.16)	0.975
HMW adiponectin
Model 1	309		0.60 (−1.98, 3.19)			0.647	214	0.70 (−2.37, 3.77)	0.653	95	0.09 (−4.85, 5.03)	0.972
Model 2	271		0.66 (−2.13, 3.46)			0.640	188	0.23 (−3.05, 3.50)	0.891	83	0.79 (−5.23, 6.80)	0.796
Total adiponectin
Model 1	315		−1.41 (−4.76, 1.94)			0.408	217	−0.71 (−4.70, 3.28)	0.726	98	−3.77 (−10.23, 2.69)	0.250
Model 2	276		−2.24 (−5.93, 1.44)			0.231	191	−1.52 (−5.98, 2.95)	0.503	85	−4.12 (−11.52, 3.27)	0.270
HMW adiponectin/total adiponectin ratio
Model 1	309			0.25 (−2.37, 2.87)		0.850	214	0.70 (−2.37, 3.77)	0.653	95	0.09 (−4.85, 5.03)	0.972
Model 2	271			1.12 (−4.15, 6.40)		0.676	188	0.72 (−5.34, 6.78)	0.815	83	0.33 (−11.25, 11.91)	0.955
Stroop – Interference
Leptin
Model 1	308			4.67 (1.54, 7.79)		0.004	2 11	4.90 (0.96, 8.84)	0.015	97	3.93 (−1.46, 9.31)	0.151
Model 2	271			4.12 (−0.27, 8.51)		0.066	187	4.53 (−0.91, 9.98)	0.102	84	3.66 (−4.66, 11.97)	0.384
HMW adiponectin
Model 1	303			−1.02 (−6.28, 4.24)		0.703	209	−1.37 (−7.65, 4.90)	0.667	94	−0.28 (−10.26, 9.69)	0.955
Model 2	267			−0.41 (−5.80, 4.98)		0.882	185	−3.05 (−9.44, 3.35)	0.348	82	7.33 (−4.25, 18.91)	0.211
Total adiponectin
Model 1	309			−0.27 (−7.08, 6.54)		0.938	212	2.63 (−5.46, 10.73)	0.522	97	−7.71 (−20.90, 5.48)	0.249
Model 2	272			0.29 (−6.84, 7.43)		0.936	188	2.26 (−6.44, 10.96)	0.609	84	−6.70 (021.40, 8.01)	0.367
HMW adiponectin/total adiponectin ratio
Model 1	303			−1.85 (−7.21, 3.51)		0.498	209	−1.37 (−7.66, 4.90)	0.667	94	−0.28 (−10.26, 9.69)	0.955
Model 2	267			−1.24 (−11.37, 8.89)		0.810	185	−4.83 (−16.58, 6.91)	0.418	82	10.62 (−12.04, 33.27)	0.353

Model 1 adjusts for age

Model 2 adjusts for age, education, race, BMI, diabetes. Models of HIV+ also included current CD4 count.

Conclusion

Among women at mid-life with HIV infection for at least 10 years, cross-sectional associations between cognition and adipokines (leptin, total adiponectin and HMW adiponectin), were observed for tests of executive function for leptin only, in which case higher leptin levels were associated with worse performance. This is in contrast to some of our findings concerning anthropometric measures – BMI, WC, or WHR – and cognition in WIHS, where there appeared to be an association between higher levels of adiposity (suggesting higher leptin levels) and better performance on certain cognitive tests, as well as lower adiposity (BMI <18.5 kg/m²) and worse performance [16]. This is also in contrast to published observations for leptin in non-HIV populations suggesting protection via leptin, albeit at older ages. Interestingly, and as a type of validation, similar correlations between BMI and adipokine levels were observed here in mid-life for women with or at risk for HIV infection in Brooklyn WIHS as have been observed in other epidemiologic studies for non-HIV populations [15]. Given potential aberrations in fat metabolism, such as lipodystrophies, that are observed in HIV, this was deemed somewhat surprising. Also of note, a BMI of 40–59 kg/m² (9.7% of women; max BMI, 59 kg/m²) brought substantial heterogeneity in leptin levels in the sample. The association of leptin with BMI was no longer linear (data not shown); the highest isolated leptin levels were observed; and there was vague evidence of leptin resistance at BMI >50 kg/m², since leptin levels plateaued as BMI continued to increase. Attempts to estimate this phenomenon via calculation of leptin:BMI did not lend further insight into this theory.

To our knowledge, published data on leptin and/or adiponectin levels in association with cognition among those with HIV infection have been reported from the Charter Study [34] and the Multicenter AIDS Cohort Study (MACS) [35]. The Charter Study conducted a metabolic substudy that included 52 cognitively impaired and 78 unimpaired HIV+ women and men, at the average age of 46 years. Average leptin levels did not differ between cognitively impaired versus unimpaired. The MACS, which includes men only, conducted an analysis of leptin and adiponectin in 509 HIV+ (average age: 55 years) and 271 HIV− (average age: 53 years). There was, once again, no association between cognition and leptin levels. However, among HIV+ men, higher adiponectin levels were associated with worse cognitive test scores. Thus, observations spanning the average ages of 39–55 years in women and men from Charter, the WIHS, and the MACS, do not support a protective role of leptin for cognition among adults with HIV infection. However, as adiponectin is a marker of inflammation, [36–41] perhaps higher levels of this adipokine, especially among older adults (i.e., MACS participants) and/or men specifically, may be informative.

In relationship to leptin, a null finding with cognition in mid-life may be expected as most published data on the association between leptin and dementia have been collected in late life, [21] when the average age at baseline was 74–83 years, much older than our sample. Another report has shown that over a 24 year follow-up period, midlife (age 38–60 years) leptin levels were not associated with late life dementia [15]. Based on these findings in old age, it has been suggested that leptin may be a cognitive enhancer when given at the time of cognitive impairment or dementia, [42] and, in light of these findings, influence appears to be temporal in relation to dementia onset. It will be interesting to observe whether there is a late life role for leptin in HIV infected women as the WIHS cohort ages.

There is much mechanistic evidence for a role of peripheral leptin and the brain. Leptin is one of the adipokines most studied in association with brain structure and function, and has been found to have numerous effects on brain development [43] and potentially on brain health in cognition and aging. Peripheral leptin enters the central nervous system and interacts with specific areas of the brain such as the hypothalamus and hippocampus [44]. However, besides leptin transport into the cerebrospinal fluid (CSF) via leptin receptor a (LepRa), and a second, not yet characterized, transport mechanism, [45] several studies indicate that leptin is also produced in the brain, for example in the hypothalamus, cortex and cerebellum [46,47].

Leptin regulates food intake and energy expenditure, improves insulin sensitivity, facilitates lipolysis, and inhibits lipogenesis [48,49]. In addition, leptin plays a permissive role in neuroendocrine immune function [48]. Leptin affects hypothalamic function, and learning and memory processes controlled by the hippocampus [43,50–52]. Experimental data show that leptin, and other adipose tissue compounds, interact directly with hypothalamic nuclei, such as the arcuate nucleus, and regulate energy expenditure and food intake through production of orexigenic (NPY, agrp) and anorectic (aMSH) peptides [53,54]. In addition, leptin appears to facilitate pre- and postsynaptic transmitter release and sensitivity, respectively, in hippocampal CA1 neurons [51]. This translates to improved performance related to spatial learning and memory function. Leptin may even shape the hypothalamus in the earliest stages of development and enhance cognition [43]. Since most data suggest leptin to be positive for the brain, it is unclear as to why the findings reported here, in women with HIV, are observed. Either higher leptin and/or adiposity levels in later life occur as a part of the aging process and are truly protective for the brain, or they are a response to earlier brain injury to which these data would attest. There is a dearth of data in HIV-infected adults.

Adiponectin is a visceral adiposity marker and is moderately inversely correlated with BMI (compared to leptin for example), and blood-brain barrier (BBB) transport mechanisms are unclear. Thus blood levels may not provide an adequate indication of potential interactions between adiponectin and the brain [55]. Studies evaluating adiponectin in association with dementia, have only reported on total adiponectin levels. Isolating HMW adiponectin and the smaller adiponectin fragments can present problems in the laboratory and inter- and intra-assay variability can be high [20].

While adiponectin is produced solely by adipose tissue, its receptors are not [56]. The peripheral effects of adiponectin are mediated mainly via 2 receptors, AdipoR1 and AdipoR2. Expression of these receptors is reported in adipose, brain, ovaries, endometrium, and placenta [57]. Both AdipoR1 and AdipoR2 are widely found throughout the CNS in brain microvessels, hippocampus, hypothalamus and brainstem in humans [58–60]. However, because there is a 1000-fold lower cerebrospinal fluid (CSF)/serum adiponectin ratio observed in humans, [59] the origin of brain adiponectin is debated. Trimeric and low molecular weight adiponectins are detectable in the CSF of humans and rodents [61,62]. In combination with a lack of HMW adiponectin observed in CSF, this may imply that only smaller forms of adiponectin cross the BBB [61,62].

This is a large study of adipokines and cognition in women with and without HIV infection. Strengths include the large multi-ethnic participant sample, and a variety of anthropometric measures that are commonly used to estimate total and regional body adiposity. The primary limitations include a limited battery of cognitive tests, no body composition measures, and cross-sectional analyses. Due to multiple comparisons, one must also consider risk for false discoveries; and our analyses were not adjusted for multiple comparisons. Observed p-values were <0.005 for most correlations between adipokines and anthropometric factors and for some of the regression analyses. However, this is one of the first reports of promising biomarkers of adiposity and cognition in women with or at risk for HIV infection. These data suggest the need for continued follow up of these women to determine mid-life and late-life effects of adipose tissue and adipose tissue alterations on cognition and eventually dementia in HIV, measured as BMI and as adipokines, particularly leptin.