Pro-Neurotensin/Neuromedin N and Risk of Cognitive Impairment in a Prospective Study

Abstract

Background:

The neuropeptide neurotensin (NT) has been linked to cardiometabolic disease. Cardiovascular risk factors are being recognized as risk factors for cognitive impairment.

Objective:

To examine the association of the stable precursor of NT, pro-neurotensin/neuromedin N (pro-NT/NMN), with incident cognitive impairment (ICI).

Methods:

We conducted a prospective nested case-control study in the REasons for Geographic And Racial Differences in Stroke (REGARDS) cohort. In 2003-2007, REGARDS enrolled 30,239 Black and White adults aged ≥45 years. ICI was identified using a 3-test cognitive battery administered biannually. Baseline pro-NT/NMN was measured by immunoassay in 393 cases of ICI and 490 controls after 3.4 years follow up. Multivariable logistic regression was used to calculate odds ratios (OR) of ICI by pro-NT/NMN quartiles. Race, age, and sex differences were studied with stratified models and interaction testing.

Results:

Pro-NT/NMN was higher in Black participants and those with hypertension and diabetes. Women with a 4th versus 1st-quartile pro-NT/NMN had 2.28-fold increased odds of ICI (95% CI 1.08–4.78) after adjusting for risk factors and incident stroke. There was no association of higher pro-NT/NMN quartiles with ICI in the overall group or men. There were no race or age differences in associations.

Conclusion:

In this biracial population-based study, elevated systemic pro-NT/NMN was associated with more than doubled risk of ICI in women but not men. Others reported sex-specific associations in women for cardiovascular mortality and diabetes with higher pro-NT/NMN, supporting a role for future research on sex differences in the neurotensinergic system.

Keywords

Biomarkers case-control studies cognition disorders cognitive dysfunction neuropeptides neurotensin prospective studies

INTRODUCTION

Neurotensin (NT) is a tridecapeptide initially isolated from bovine hypothalami in the 1970s [1]. Its precursor, pro-neurotensin/neuromedin N (pro-NT/NMN), often called “proneurotensin” in clinical literature, is differentially processed into NT and three other byproducts [2]. In addition to its direct effects on cardiac physiology [3], NT is anorexigenic in rats [4], increases post-prandially in humans [2], and likely plays a role in the pathogenesis of obesity [5]. While plasma NT is unstable [6], unfrozen pro-NT/NMN is stable for up to 48 hours [2] and is thus a laboratory proxy to NT concentrations at sampling.

Elevated pro-NT/NMN has been linked to risk of type 2 diabetes [7 –10] and aggregate cardiovascular disease (CVD) including stroke in the middle-aged [9, 11] and elderly [10]. A sex-specific association of pro-NT/NMN with cardiovascular mortality and diabetes in women has been reported [9, 12], leading to its consideration as a biomarker for women’s cardiometabolic health [13]. Nevertheless, the pathophysiologic mechanisms through which neurotensin influences CVD remain elusive [12].

Recent studies have uncovered the contributions of cardiovascular risk factors to the development of cognitive impairment [14 –18]. It is probable that most individuals with cognitive impairment have pathologic contributions from both vascular and neurodegenerative processes [19], and exist on a spectrum between the two archetypes of dementia [15, 20]. The mechanisms linking CVD to cognitive impairment are uncertain, but chronic cerebral hypoperfusion and stroke are likely mediators [21]. CVD biomarkers are emerging as risk factors for cognitive disorders [22 –24] and there is growing interest in biomarkers predictive of cognitive impairment [25]. Given associations of pro-NT/NMN with CVD [9–11 , 22] and risk factors like diabetes [7–10 , 22], we conducted a prospective nested case-control study to examine the association of pro-NT/NMN with incident cognitive impairment (ICI). We hypothesized that increased fasting plasma pro-NT/NMN would be associated with ICI.

MATERIALS AND METHODS

Sample

Participants were selected from within the REasons for Geographic And Racial Differences in Stroke (REGARDS) study, detailed previously [26]. In brief, the study was designed to investigate the factors contributing to excess stroke mortality in the southeastern United States and in Black Americans. From 2003 to 2007, REGARDS enrolled 30,239 Black or White participants aged 45 years or older who were randomly selected from a publicly available list and subsequently contacted by mail and telephone to participate. The study oversampled Black individuals (42%) and those residing in the “stroke belt” of the southeastern U.S. (56%). Individuals were excluded from participation for apparent cognitive impairment on initial telephone interview, active cancer treatment, medical conditions precluding long-term follow up, anticipation of or actual residence in a nursing home, or poor English language ability.

A telephone interview was conducted to gather medical history, at which time all participants provided verbal consent. Participants subsequently underwent an in-home assessment to conduct a medication inventory and obtain biometrics, an electrocardiogram (EKG), fasting phlebotomy, and urine samples; written consent was obtained at this time. The methods of the REGARDS study and this investigation were approved by the institutional review boards of all involved institutions.

Cognitive assessment

Cognitive function was longitudinally assessed in two ways over 3.4 years of follow-up. The Six-Item Screener (SIS) [27], a short screening tool assessing immediate recall and temporal orientation skills, was added to the initial telephone assessment in 2003 and was conducted yearly thereafter. A three-test cognitive battery was added to the follow-up telephone assessments in 2006 and was conducted biennially thereafter. This included three tests: Word List Learning (WLL), Word List Recall (WLR) [28], and Animal Fluency (AF) [29]. The WLR and WLL contain 10 items assessing word list learning and its delayed recall, respectively. Animal Fluency counts the number of animals the participant can name in one minute [29].

REGARDS formed a nested case-control study with controls drawn from a cohort random sample (stratified on age, race, and sex) of 1104 participants [30] and 500 cases of ICI identified as previously described [22]. Cases were identified within the entire REGARDS study using the most recently administered three-test cognitive battery [22] to include participants scoring >1.5 SD below their age, sex, education, and race-predicted score on ≥2 of tests in the cognitive battery, as shown in Fig. 1.

Fig.1

Identification of cases of cognitive impairment in REGARDS. WLL, Word List Learning; WLR, Word List Recall; AF, Animal Fluency.

Exclusion criteria for cases and controls included those missing cognitive testing data needed to classify participants as a case or noncase and those with prevalent cognitive impairment based on the SIS, prevalent stroke, or stroke prior to the first cognitive assessment.

Variables

Demographic variables included age (years; continuous or in categories [45–54, 55–64, 65–74, 75–84, and ≥85 years]), race, sex, annual income (< $20,000, $20,000–$34,999, $35,000–$74,999, ≥$75,000, or refused), education level (<high school, high school only, some college, or ≥college), region (Stroke Belt, Stroke Buckle [31], or other), exercise (none, 1–3 times weekly, or ≥4 times weekly), and alcohol intake (none, moderate [1–7 weekly drinks in women, 1–14 weekly drinks in men], or heavy [>7 weekly drinks in women, >14 weekly drinks in men]).

Clinical variables included history of coronary artery disease (CAD; self-report of past myocardial infarction (MI), evidence of prior infarct on EKG, or history of coronary revascularization procedure such as coronary artery bypass grafting or percutaneous coronary intervention with angioplasty or stent placement), atrial fibrillation (AF; self-report or evidence on EKG), diabetes mellitus (self-reported use of oral hypoglycemic agents or insulin, fasting glucose ≥126 mg/dL, or random glucose ≥200 mg/dL), and hypertension (self-report of antihypertensive drug use, systolic blood pressure [SBP] > 130 mmHg, or diastolic blood pressure [DBP] > 80 mmHg) [32], perimenopausal hormone replacement therapy (self-reported use at any time in women), and left ventricular hypertrophy (defined using sex-specific Cornell voltage criteria [33] on 12-lead EKGs or modified Cornell voltage criteria [34] on 7-lead EKGs). Body mass index (BMI) was calculated using measured height and weight obtained at baseline. The occurrence of suspected incident stroke was assessed every six months over the telephone [35]; all hospitalizations were reviewed. Medical records for suspected stroke events were retrieved with adjudication by a group of stroke experts. Similarly, suspected coronary heart disease (CHD) events—defined as nonfatal MI or CHD death—were assessed with medical record adjudication by trained clinicians [36].

Fasting blood was collected from each participant at the initial in-home visit [37]. Blood samples were locally centrifuged for 10 min, then shipped overnight on ice packs to the core laboratory at the University of Vermont, where they were recentrifuged at 30,000 g and aliquoted then stored at –80°C. Colorimetric reflectance spectrophotometry was used to measure blood glucose (Ortho Vitros 950 IRC Clinical Analyzer, Johnson & Johnson Clinical Diagnostics).

Pro-NT/NMN was measured using EDTA plasma in the nested case-control sample. Plasma samples were removed from storage, thawed, and run on a single-step sandwich immunoassay in a microtiter plate format with chemiluminescence label (SphingoTec, Hennigsdorf, Germany). The assay was conducted at an independent laboratory (ICI Immunochemical Intelligence GmbH, Berlin, Germany) by technicians who were blinded to clinical data. This method had a mean interassay precision of 3.7% (standard deviation 0.8%) in the measuring range 3–270 pmol/L, and a lower detection limit of 1.9 pmol/L.

TaqMan assays of two single-nucleotide polymorphisms (rs429358, rs7412) [38] of apolipoprotein E ɛ4 (APOE4) informed APOE4 genotype status. The open-source PHASE program [39] was used in haplotype reconstruction, with uncertain haplotypes recorded as the most probable outcome, indicated by a likelihood >85%.

Statistical analysis

All statistical testing was probability-weighted to account for the sampling of the original cohort sample, using strata based on age, sex, and race. Participants with missing values were excluded from the analyses or models in which they were absent. Analyses were two-sided with α= 0.05 and performed with Stata, version 16 (StataCorp, College Station, TX).

Plasma concentration of pro-NT/NMN was log-transformed for some analyses given its positively-skewed distribution [12]; values were binned into quartiles for other analyses. Distributions of baseline demographic and clinical factors among and across pro-NT/NMN quartiles were compared with design-adjusted Pearson chi-squared tests (categorical variables) and linear regressions or adjusted Wald tests (continuous variables).

Five weighted logistic regression models were employed to study the association of pro-NT/NMN with ICI, yielding odds ratios (ORs) and their 95% confidence intervals (CIs), calculated using a Taylor series as a finite population correction. In these analyses, pro-NT/NMN concentrations were considered continuously (log-transformed) or in quartiles. Model 1 included demographics: age, race, sex, region, income, and education. Model 2 included the Model 1 covariates and added stroke risk factors [40] (systolic BP, antihypertensive therapy, prevalent diabetes, smoking, prevalent CAD, AF, and LVH) and ICI risk factors (alcohol use, physical activity, and BMI). Model 3 added APOE4 genotype to Model 2. Model 4 added incident stroke to Model 3. Model 5 added incident CAD to model 3. Multiplicative interaction terms of log pro-NT/NMN with race or sex were tested in each model, considered significant at p < 0.10. Stratified analyses were conducted in race, age, and sex groups in each model; sex, age, and race were removed as covariates when used in stratification. Stratified analysis in the ≥85 age group was not performed due to small subgroup size.

RESULTS

Sample characteristics

After excluding potential participants who were not fasting at the time of phlebotomy (67 controls, 68 cases), we identified 490 controls and 393 cases of ICI. In controls, fasting plasma pro-NT/NMN ranged from 33 pmol/L to 1151 pmol/L, with a median [IQR] of 169 [106, 261] pmol/L, mean 211 pmol/L, and standard deviation (SD) 165 pmol/L. No significant differences in demographics or risk factors were observed among fasting and non-fasting (excluded) participants (see Supplementary Table 1).

Baseline characteristics by case-control status

Table 1 shows baseline characteristics of controls and ICI cases. Cases were more often White, female, younger, current smokers, resided in the Stroke Belt or Stroke Buckle, and were more likely to have diabetes. Log pro-NT/NMN did not differ among cases and controls (p > 0.05).

Table 1

Baseline Characteristics of ICI Cases and Controls

	Controls* (n = 490)	ICI Cases (n = 393)
Age (years; mean, 95% CI)^†	70.0 (68.8, 71.1)	65.1 (63.5, 66.6)
Black race (%)^‡	54.8	35.7
Male sex (%)^‡	53.7	41.9
Region (%)^‡
Stroke belt	31.4	44.5
Stroke buckle	19.3	20.4
Other	49.4	35.2
Annual Income (%)
< $20,000	14.8	24.7
$20,000 – $34,999	25.2	21.0
$35,000 – $74,999	32.5	26.8
≥$75,000	15.2	13.6
Refused	12.4	13.9
Education (%)
<High School	12.7	9.3
High School Grad	20.5	24.5
Some College	26.3	25.6
≥College	40.5	40.6
Weekly Exercise (%)
None	39.2	36.8
1–3 Times	34.9	36.0
≥4 Times	25.9	27.2
Tobacco (%)
Never Smoker	52.1	48.5
Past smoker	37.5	33.6
Current Smoker^‡	10.4	18.0
Current Alcohol Use (%)
None	62.5	69.6
Moderate	33.3	27.8
Heavy	4.2	2.6
Diabetes Mellitus (%)^‡	15.4	27.3
Coronary Artery Disease (%)	16.5	21.4
Hormone Replacement Therapy (Women; %)	52.1	54.6
Hypertension (%)	77.7	75.4
BMI Category (%)
<18.5 Underweight	0.4	1.3
18.5–24.9 Normal Weight	30.4	19.6
25.0–29.9 Overweight	37.4	34.9
≥30.0 Obesity	31.7	44.2
Left Ventricular Hypertrophy (%)	9.6	10.7
Atrial Fibrillation (%)	7.8	8.0
APOE4 homo/heterozygotes (%)	30.0	31.8
Pro-NT/NMN (pmol/L) (mean, 95% CI)^‡	207 (188–225)	232 (209–254)
Log (pro-NT/NMN) ([pmol/L]) (mean, 95% CI)^‡	5.1 (5.0–5.2)	5.2 (5.1–5.2)

*Weighted to analytical cohort; ^†Adjusted Wald Test p < 0.05; ^‡Pearson Chi-Squared p < 0.05.

Baseline characteristics of controls across pro-NT/NMN quartiles

Table 2 compares baseline characteristics of controls across quartiles of pro-NT/NMN. Higher quartiles contained greater proportions of Black participants, diabetes, and hypertension; greater proportions of participants with no active alcohol use were found in the highest quartile (p < 0.05).

Table 2

Baseline Characteristics of Controls by proNT Quartile

Pro -NT/NMN Quartiles	Q1 (33 – 106 pmol/L)	Q2 (106 – 169 pmol/L)	Q3 (169 – 261 pmol/L)	Q4 (261 – 1151 pmol/L)	p
Age (years; mean, 95% CI)	70.1 (66.8, 73.5)	70.9 (66.3, 75.5)	69.3 (65.9, 72.7)	69.7 (65.7, 73.7)	0.97*
Black race (%)	42.0	45.9	61.8	70.5	0.01^†
Male sex (%)	54.3	60.2	46.2	55.8	0.47^†
Region (%)
Stroke belt	35.2	27.4	32.4	29.6	0.9^†
Stroke buckle	15.7	24.6	17.4	20.5
Other	49.1	48.0	50.2	49.9
Annual Income (%)
< $20,000	7.5	20.8	11.7	21.3	0.19^†
$20,000 – $34,999	16.5	28.1	35.7	19.8
$35,000 – $74,999	42.0	27.5	27.6	31.9
≥$75,000	21.4	11.6	15.0	11.5
Refused	12.6	12.0	10.1	15.5
Education (%)
<High School	10.2	18.4	5.4	18.9	0.17^†
High School Grad	15.2	13.7	28.6	24.0
Some College	28.9	27.1	20.7	29.4
≥College	45.7	40.8	45.3	27.8
Weekly Exercise (%)
None	27.6	38.4	40.7	52.1	0.13^†
1–3 Times	38.2	40.1	37.3	23.0
≥4 Times	34.2	21.5	22.0	24.9
Tobacco (%)
Never Smoker	56.7	48.5	54.4	47.4	0.45^†
Past smoker	38.8	35.3	35.4	40.8
Current Smoker	4.5	16.3	10.2	11.8
Current Alcohol Use (%)
None	53.8	52.1	70.1	74.1	0.01^†
Moderate	36.6	44.8	28.0	24.2
Heavy	9.6	3.1	1.8	1.7
Diabetes Mellitus (%)	8.1	11.1	18.8	24.4	0.03^†
Coronary artery disease (%)	17.1	16.1	14.9	18.0	0.98^†
Hormone Replacement Therapy (Women; %)	56.2	56.3	54.2	40.0	0.51^†
Hypertension (%)	69.3	80.0	75.8	87.9	0.04^†
BMI Category (%)
<18.5 Underweight	0.4	0.0	0.7	0.5	0.4^†
18.5–24.9 Normal Weight	29.5	43.6	24.8	24.6
25.0–29.9 Overweight	39.5	32.0	42.1	34.9
≥30.0 Obesity	30.7	24.4	32.3	40.0
Left Ventricular Hypertrophy (%)	7.7	15.2	2.0	15.3	0.10^†
Atrial Fibrillation (%)	8.9	6.7	10.9	3.5	0.37^†
APOE4 homo/heterozygotes (%)	27.8	21.6	31.0	40.3	0.18^†

*Linear Regression, ^†Pearson Chi-Squared.

Association of pro-NT/NMN with ICI

Log pro-NT/NMN was not associated with ICI in bivariate logistic regression (OR per SD 1.08, 95% CI 0.89–1.32). Table 3 shows multivariable logistic regression models 1–5 considering pro-NT/NMN concentrations in quartiles, relative to Q1. Quartiles of pro-NT/NMN versus 1st quartile were not associated with ICI in any model. Log pro-NT/NMN was not associated with ICI in any multivariable model (Model 2 OR per SD 0.94, 95% CI 0.76–1.17; other models not shown).

Table 3

Association of pro-NT/NMN With Incident Cognitive Impairment, by Quartiles

	pro-NT/NMN
	Q1 (33–106 pmol/L)		Q2 (106–169 pmol/L)		Q3 (169–261 pmol/L)		Q4 (261–1151 pmol/L)
n	98/124		95/124		86/135		114/107
cases/controls	OR		OR	95% CI	OR	95% CI	OR	95% CI
Model 1	1	(ref)	1.09	(0.67, 1.80)	0.80	(0.49, 1.30)	1.56	(0.96, 2.51)
Model 2	1	(ref)	1.10	(0.64, 1.87)	0.73	(0.43, 1.23)	1.24	(0.72, 2.15)
Model 3	1	(ref)	1.10	(0.64, 1.89)	0.69	(0.41, 1.17)	1.23	(0.71, 2.12)
Model 4	1	(ref)	1.10	(0.64, 1.89)	0.69	(0.40, 1.17)	1.22	(0.71, 2.12)
Model 5	1	(ref)	1.11	(0.65, 1.90)	0.69	(0.41, 1.17)	1.23	(0.71, 2.13)

Covariates in multivariable models: Model 1: age, race, sex, region, income, education. Model 2 added systolic blood pressure, antihypertensive therapy, prevalent diabetes, smoking, prevalent coronary artery disease, prevalent atrial fibrillation, and baseline left ventricular hypertrophy to Model 1. Model 3 added APOE4 genotype to Model 2. Model 4 added incident stroke to Model 3. Model 5 added incident coronary artery disease to Model 3.

Table 4 shows the sex, age, or race-stratified odds of ICI by quartiles of pro-NT/NMN in Models 1–5. Race and age did not interact with pro-NT/NMN in any of the models (all p > 0.10). Pro-NT/NMN was not associated with ICI in any race or age-stratified model.

Table 4

Association of 4th versus 1st Quartile pro-NT/NMN with ICI by Race, Sex, and Age*

Group	Model 1		Model 2		Model 3		Model 4		Model 5
	OR	95% CI	OR	95% CI	OR	95% CI	OR	95% CI	OR	95% CI
Sex
Female	2.23	(1.20, 4.14)	2.22	(1.08, 4.57)	2.21	(1.06, 4.63)	2.28	(1.08, 4.78)	2.17	(1.03, 4.56)
Male	0.88	(0.42, 1.86)	0.60	(0.25, 1.47)	0.56	(0.23, 1.38)	0.56	(0.23, 1.39)	0.58	(0.24, 1.41)
Interaction p^†	0.10		0.07		0.06		0.10		0.06
Race
White	1.89	(1.03, 3.45)	1.69	(0.81, 3.53)	1.63	(0.77, 3.43)	1.63	(0.77, 3.43)	1.60	(0.76, 3.38)
Black	0.99	(0.45, 2.16)	0.86	(0.35, 2.09)	0.89	(0.36, 2.18)	0.88	(0.36, 2.17)	0.87	(0.36, 2.13)
Interaction p^‡	0.26		0.36		0.47		0.47		0.46
Age (years)
<55	1.73	(0.58, 5.16)	1.95	(0.45, 8.40)	1.76	(0.41, 7.57)	1.72	(0.39, 7.52)	1.84	(0.43, 7.79)
55–64	1.61	(0.74, 3.51)	2.00	(0.79, 5.09)	2.00	(0.78, 5.13)	1.95	(0.75, 5.03)	1.97	(0.77, 5.12)
65–74	1.37	(0.61, 3.08)	0.92	(0.37, 2.13)	0.79	(0.30, 2.07)	0.78	(0.29, 2.07)	0.79	(0.30, 2.07)
75–84	0.47	(0.18, 1.26)	0.58	(0.16, 2.02)	0.66	(0.19, 2.28)	0.60	(0.17, 2.13)	0.67	(0.19, 2.34)
Interaction p^§	0.89		0.77		0.98		0.97		0.99

*Model covariates shown in Table 3; ^†Multiplicative interaction of log pro-NT/NMN and sex; ^‡Multiplicative interaction of log pro-NT/NMN and race; ^§Multiplicative interaction of log pro-NT/NMN and age.

Pro-NT/NMN interacted with sex in all models (Model 4 interaction p = 0.10). Women with a 4th versus 1st-quartile pro-NT/NMN had significantly increased odds of ICI in Models 1–5, across which ORs ranged from 2.17 to 2.28 without attenuation by adjustment across the 5 models. There was no significant association among men, with ORs ranging from 1.2 to 1.4. Figure 2 presents sex-stratified restricted cubic splines for the odds of incident cognitive impairment across log pro-NT/NMN, which shows no association among men and a gradually increasing association among women above the median log pro-NT/NMN value.

Fig.2

Sex-stratified restricted cubic splines showing the odds of incident cognitive impairment (left y axis) across log pro-NT/NMN concentration (x axis), referencing the sex-specific median and adjusting for other Model 1 covariates (age, race, region, income, and education). Kernel density plots (right y axis) show distributions of pro-NT/NMN, stratified by sex and case-control status.

DISCUSSION

In this prospective study of Black and White Americans aged 45 years and older, we observed a sex-specific association of baseline pro-NT/NMN with risk of cognitive impairment over 3.5 years follow-up. Specifically, women with a 4th versus 1st-quartile pro-NT/NMN had a more than doubled odds of ICI, with little evidence of confounding by other ICI risk factors or interval coronary and stroke events after baseline. These findings support the notion of a “heart-brain continuum” for the development of dementia and add support to the previously reported concept of pro-NT/NMN as a biomarker of women’s cardiometabolic health.

This sex-specific finding is unsurprising, given prior studies reporting sex-specific or higher-magnitude associations of higher pro-NT/NMN with cardiovascular, metabolic, and mortality outcomes in women [9 –12]. Women did not have higher concentrations of pro-NT/NMN, consistent with prior analysis in REGARDS [12], although greater pro-NT/NMN concentrations in women were previously reported [7]. Variation in the association of pro-NT/NMN with ICI by sex may be attributable to differences in lifetime exposure to estrogens, considering that transcription of the gene encoding NT is augmented by estradiol [41 –43]. It is unlikely that exogenous estrogen use could have influenced our finding of an association specific to women, given that history of perimenopausal hormone replacement therapy did not differ among cases and controls or across quartiles of pro-NT/NMN. We are unable to account for parity or age at menarche or menopause in the REGARDS study.

This investigation is strong in the large size, prospective design, large proportion of Black participants, and standardized data collection methods of its parent study. Several limitations must be considered. Firstly, a number of selection biases cannot be excluded, given the race-dichotomous and targeted nature of the sampling; this was minimized with the use of sampling weights. Secondly, that a participant’s score on the SIS, and not the three-test battery (not administered at baseline), was the criterion for baseline cognitive impairment, raises the possibility that misclassification bias of baseline function exists, although this type of bias would likely lead to underestimation of true associations. Furthermore, to standardize measurement of the peptide for analysis, we excluded participants who were not fasting at time of phlebotomy, but because following a meal, NT has been shown to increase up to three times its pre-prandial concentration [2], the possibility that inter-participant variability in post-prandial pro-NT/NMN elevation exists and has an effect on the outcome cannot be excluded. Lastly, pro-NT/NMN was measured at various times following phlebotomy (participants were recruited over a 4-year period). While the stability of the peptide beyond 48 hours post-phlebotomy [2] has not been evaluated, we have shown in general that proteins do not deteriorate in storage at –80°C or colder [44]. That no differences in baseline characteristics were observed when comparing fasting and non-fasting (excluded) participants makes omission bias as a result of this exclusion unlikely.

Our finding of an association between higher pro-NT/NMN and cognitive impairment specific to women reinforces the concept that cardiovascular risk factors play an important role in cognitive health and supports further study of this biomarker in disease outcomes. Specifically, predictive modeling of the association of pro-NT/NMN with outcomes is needed to establish its clinical utility as a biomarker. Furthermore, that significantly increased odds of ICI in women were not attenuated by cardio-and-cerebrovascular risk factors or events suggests that the neurotensinergic system may exert influence on the pathogenesis of ICI beyond that mediated by these risk factors or mediators. Lastly, because the mechanisms linking pro-NT/NMN to cardiovascular and metabolic diseases have not been established, further basic and translational research is needed to establish the molecular and physiological mechanisms of neurotensinergic interaction with the pathogenesis of disease.

In conclusion, we observed a strong, independent association of pro-neurotensin/neuromedin N with incident cognitive impairment in women, such that when controlling for demographic variables and cardio-and-cerebrovascular risk factors, a highest-quartile concentration was associated with a more than doubled risk of cognitive impairment relative to the lowest quartile. We found no significant influence of APOE4 genotype, incident stroke, or incident CAD on this association. This association did not differ by race or age. Our findings underline the contributions of cardiovascular risk factors to the pathogenesis of cognitive impairment and add further support to pro-NT/NMN as a biomarker of women’s cardiovascular health.

Footnotes

ACKNOWLEDGMENTS

This research project was funded by cooperative agreement U01 NS041588 co-funded by the National Institute of Neurological Disorders and Stroke (NINDS) and the National Institute on Aging (NIA), National Institutes of Health (NIH), and United States Department of Health and Human Services. Additional laboratory support was provided by SphingoTec GmbH (Hennigsdorf, Germany). The content is solely the responsibility of the authors and does not necessarily represent the official views of the NINDS or the NIA. The authors thank the other investigators, the staff, and the participants of the REGARDS study for their valuable contributions. A full list of participating REGARDS investigators and institutions can be found at:

Authors’ disclosures available online ().

The supplementary material is available in the electronic version of this article: .

References

Carraway

, Leeman

(1973) The isolation of a new hypotensive peptide, neurotensin, from bovine hypothalami. J Biol Chem 248, 6854–6861.

Ernst

, Hellmich

, Bergmann

(2006) Proneurotensin 1-117, a stable neurotensin precursor fragment identified in human circulation. Peptides 27, 1787–1793.

Osadchii

(2015) Emerging role of neurotensin in regulation of the cardiovascular system. Eur J Pharmacol 762, 184–192.

Cooke

, Patterson

, Patel

, Smith

, Ghatei

, Bloom

, Murphy

(2009) Peripheral and central administration of xenin and neurotensin suppress food intake in rodents. Obesity (Silver Spring) 17, 1135–1143.

, Song

, Zaytseva

, Liu

, Rychahou

, Jiang

, Starr

, Kim

, Harris

, Yiannikouris

, Katz

, Nilsson

, Orho-Melander

, Chen

, Zhu

, Fahrenholz

, Higashi

, Gao

, Morris

, Cassis

, Fan

, Weiss

, Dobner

, Melander

, Jia

, Evers

(2016) An obligatory role for neurotensin in high-fat-diet-induced obesity. Nature 533, 411–415.

Friry

, Feliciangeli

, Richard

, Kitabgi

, Rovere

(2002) Production of recombinant large proneurotensin/neuromedin N-derived peptides and characterization of their binding and biological activity. Biochem Biophys Res Commun 290, 1161–1168.

Barchetta

, Cimini

, Leonetti

, Capoccia

, Di Cristofano

, Silecchia

, Orho-Melander

, Melander

, Cavallo

(2018) Increased plasma proneurotensin levels identify NAFLD in adults with and without type 2 diabetes. J Clin Endocrinol Metab 103, 2253–2260.

Barchetta

, Cimini

, Capoccia

, Bertoccini

, Ceccarelli

, Chiappetta

, Leonetti

, Di Cristofano

, Silecchia

, Orho-Melander

, Melander

, Cavallo

(2018) Neurotensin is a lipid-induced gastrointestinal peptide associated with visceral adipose tissue inflammation in obesity. Nutrients 10, 526.

Melander

, Maisel

, Almgren

, Manjer

, Belting

, Hedblad

, Engstrom

, Kilger

, Nilsson

, Bergmann

, Orho-Melander

(2012) Plasma proneurotensin and incidence of diabetes, cardiovascular disease, breast cancer, and mortality. JAMA 308, 1469–1475.

10.

Fawad

, Bergmann

, Struck

, Nilsson

, Orho-Melander

, Melander

(2018) Proneurotensin predicts cardiovascular disease in an elderly population. J Clin Endocrinol Metab 103, 1940–1947.

11.

Januzzi

Jr , Lyass

, Liu

, Gaggin

, Trebnick

, Maisel

, D’Agostino

Sr , Wang

, Massaro

, Vasan

(2016) Circulating proneurotensin concentrations and cardiovascular disease events in the community: The framingham heart study. Arterioscler Thromb Vasc Biol 36, 1692–1697.

12.

Wettersten

, Cushman

, Howard

, Hartmann

, Filippatos

, Beri

, Clopton

, Howard

, Safford

, Judd

, Bergmann

, Struck

, Maisel

(2018) Usefulness of proneurotensin to predict cardiovascular and all-cause mortality in a united states population (from the reasons for geographic and racial differences in stroke study). Am J Cardiol 122, 26–32.

13.

Daniels

, Maisel

(2015) Cardiovascular biomarkers and sex: The case for women. Nat Rev Cardiol 12, 588–596.

14.

Saver

, Cushman

(2018) Striving for ideal cardiovascular and brain health: It is never too early or too late. JAMA 320, 645–647.

15.

de Toledo Ferraz Alves

, Ferreira

, Wajngarten

, Busatto

(2010) Cardiac disorders as risk factors for Alzheimer’s disease. J Alzheimers Dis 20, 749–763.

16.

Abete

, Della-Morte

, Gargiulo

, Basile

, Langellotto

, Galizia

, Testa

, Canonico

, Bonaduce

, Cacciatore

(2014) Cognitive impairment and cardiovascular diseases in the elderly. A heart-brain continuum hypothesis. Ageing Res Rev 18, 41–52.

17.

Muqtadar

, Testai

, Gorelick

(2012) The dementia of cardiac disease. Curr Cardiol Rep 14, 732–740.

18.

Ferreira

, Tamashiro-Duran

, Squarzoni

, Duran

, Alves

, Buchpiguel

, Busatto

(2014) The link between cardiovascular risk, Alzheimer’s disease, and mild cognitive impairment: Support from recent functional neuroimaging studies. Braz J Psychiatry 36, 344–357.

19.

, Seo

, Kim

, Cho

, Noh

, Kim

, Yoon

, Woo

, Kim

(2015) Effects of amyloid and vascular markers on cognitive decline in subcortical vascular dementia. Neurology 85, 1687–1693.

20.

Barbay

, Taillia

, Nedelec-Ciceri

, Arnoux

, Puy

, Wiener

, Canaple

, Lamy

, Godefroy

, Roussel

(2017) Vascular cognitive impairment: Advances and trends. Rev Neurol (Paris) 173, 473–480.

21.

Kalaria

(2012) Cerebrovascular disease and mechanisms of cognitive impairment: Evidence from clinicopathological studies in humans. Stroke 43, 2526–2534.

22.

Gillett

, Thacker

, Letter

, McClure

, Wadley

, Unverzagt

, Kissela

, Kennedy

, Glasser

, Levine

, Cushman

(2015) Correlates of incident cognitive impairment in the reasons for geographic and racial differences in stroke (REGARDS) Study. Clin Neuropsychol 29, 466–486.

23.

Cushman

, Callas

, McClure

, Unverzagt

, Howard

, Gillett

, Thacker

, Wadley

(2016) N-terminal pro-B-type natriuretic peptide and risk of future cognitive impairment in the REGARDS cohort. J Alzheimers Dis 54, 497–503.

24.

Gillett

, McClure

, Callas

, Thacker

, Unverzagt

, Wadley

, Letter

, Cushman

(2018) Hemostasis biomarkers and incident cognitive impairment: The REGARDS study. J Thromb Haemost 16, 1259–1267.

25.

Cipollini

, Troili

, Giubilei

(2019) Emerging biomarkers in vascular cognitive impairment and dementia: From pathophysiological pathways to clinical application. Int J Mol Sci 20, 2821.

26.

Howard

, Cushman

, Pulley

, Gomez

, Go

, Prineas

, Graham

, Moy

, Howard

(2005) The reasons for geographic and racial differences in stroke study: Objectives and design. Neuroepidemiology 25, 135–143.

27.

Callahan

, Unverzagt

, Hui

, Perkins

, Hendrie

(2002) Six-item screener to identify cognitive impairment among potential subjects for clinical research. Med Care 40, 771–781.

28.

Moms

, Heyman

, Mohs

, Hughes

, van Belle

, Fillenbaum

, Mellits

, Clark

(1989) The consortium to establish a registry for Alzheimer’s disease (CERAD). Part I. Clinical and neuropsychological assesment of Alzheimer’s disease. Neurology 39, 1159.

29.

Rosen

(1980) Verbal fluency in aging and dementia. J Clin Exp Neuropsychol 2, 135–146.

30.

Alexander

, Zakai

, Gillett

, McClure

, Wadley

, Unverzagt

, Cushman

(2014) ABO blood type, factor VIII, and incident cognitive impairment in the REGARDS cohort. Neurology 83, 1271–1276.

31.

Howard

, Anderson

, Johnson

, Sorlie

, Russell

, Howard

(1997) Evaluation of social status as a contributing factor to the stroke belt region of the United States. Stroke 28, 936–940.

32.

Whelton

, Carey

, Aronow

, Casey

Jr , Collins

, Dennison Himmelfarb

, DePalma

, Gidding

, Jamerson

, Jones

, MacLaughlin

, Muntner

, Ovbiagele

, Smith

Jr , Spencer

, Stafford

, Taler

, Thomas

, Williams

Sr. , Williamson

, Wright

Jr (2018) 2017 ACC/AHA/AAPA/ABC/ACPM/AGS/APhA/ASH/ASPC/NMA/PCNA Guideline for the prevention, detection, evaluation, and management of high blood pressure in adults: A report of the American College of Cardiology/American Heart Association task force on clinical practice guidelines. Hypertension 71, e13–e115.

33.

Casale

, Devereux

, Kligfield

, Eisenberg

, Miller

, Chaudhary

, Phillips

(1985) Electrocardiographic detection of left ventricular hypertrophy: Development and prospective validation of improved criteria. J Am Coll Cardiol 6, 572–580.

34.

Soliman

, Howard

, Prineas

, McClure

, Howard

(2010) Calculating Cornell voltage from nonstandard chest electrode recording site in the Reasons for Geographic And Racial Differences in Stroke study. J Electrocardiol 43, 209–214.

35.

Howard

, Kleindorfer

, Judd

, McClure

, Safford

, Rhodes

, Cushman

, Moy

, Soliman

, Kissela

, Howard

(2011) Disparities in stroke incidence contributing to disparities in stroke mortality. Ann Neurol 69, 619–627.

36.

Safford

, Brown

, Muntner

, Durant

, Glasser

, Halanych

, Shikany

, Prineas

, Samdarshi

, Bittner

(2012) Association of race and sex with risk of incident acute coronary heart disease events. JAMA 308, 1768–1774.

37.

Tracy

, Bovill

, Stump

, Lin

, Gomol

, Collen

, Mann

(1988) Reduction of in vitro artifact during blood collection in TIMI II. Blood 78, 376.

38.

Koch

, Ehrenhaft

, Griesser

, Pfeufer

, Müller

, Schömig

, Kastrati

(2002) TaqMan systems for genotyping of disease-related polymorphisms present in the gene encoding apolipoprotein E. Clin Chem Lab Med 40, 1123–1131.

39.

Stephens

, Smith

, Donnelly

(2001) A new statistical method for haplotype reconstruction from population data. Am J Hum Genet 68, 978–989.

40.

D’Agostino

, Wolf

, Belanger

, Kannel

(1994) Stroke risk profile: Adjustment for antihypertensive medication. The Framingham Study. Stroke 25, 40–43.

41.

Dupouy

, Viardot-Foucault

, Alifano

, Souaze

, Plu-Bureau

, Chaouat

, Lavaur

, Hugol

, Gespach

, Gompel

, Forgez

(2009) The neurotensin receptor-1 pathway contributes to human ductal breast cancer progression. PLoS One 4, e4223.

42.

Alexander

(1993) Estrogen-regulated synthesis of neurotensin in neurosecretory cells of the hypothalamic arcuate nucleus in the female rat. Endocrinology 133, 1809–1816.

43.

Watters

, Dorsa

(1998) Transcriptional effects of estrogen on neuronal neurotensin gene expression involve cAMP/protein kinase A-dependent signaling mechanisms. J Neurosci 18, 6672–6680.

44.

Lewis

, Callas

, Jenny

, Tracy

(2001) Longitudinal stability of coagulation, fibrinolysis, and inflammation factors in stored plasma samples. Thromb Haemost 86, 1495–1500.