Pulmonary Function is Associated with Cognitive Decline and Structural Brain Differences

Abstract

The association of poor pulmonary function (PF) with cognitive trajectories and structural brain differences remains unclear. Within the Rush Memory and Aging Project, 1377 dementia-free subjects were followed up to 21 years. PF was assessed with a composite score measured at baseline. Global and domain-specific cognitive function was assessed annually constructed from 19 cognitive tests. A subsample of 351 participants underwent brain magnetic resonance imaging to investigate the cross-sectional association between PF and structural brain volumes. We found that low PF was related to faster decline in global cognition, and domain-specific function including episodic memory, semantic memory, working memory, visuospatial ability, and perceptual speed. In addition, low PF was associated with smaller volumes of total brain, white matter and gray matter, and larger white matter hyperintensities volume. Our results suggest that low PF is associated with faster cognitive decline, and both neurodegeneration and vascular brain lesions may underlie the association.

1. NARRATIVE

1.1. Contextual background

Tests of pulmonary function (PF) are used to assess lung capacity and the potency of the respiratory tract in people with respiratory diseases. PF decreases with age due to modifications in elastic recoil and thorax compliance [1], which slowly lead to a decline in the body’s oxygen supply [2]. In general, the prevalence of pulmonary hypofunction increases sharply with age [3]. Decreased PF has been used in the clinical diagnosis of airway dysfunction and chronic respiratory diseases [4, 5]. Further, poor PF has been associated with lower quality of life and all-cause mortality in older adults, making it a public health priority [6, 7].

PF is an important parameter for complete evaluation of the respiratory system and can be measured by a spirometer. PF is assessed by peak expiratory flow (PEF; i.e. the maximum speed of expiration), forced expiratory volume in 1s (FEV1; i.e. the volume of air exhaled with force for 1 second), and forced vital capacity (FVC; i.e. the maximum amount of air that can be exhaled after a maximum inhalation) [8]. This composite measure reflects not only lung capacity but also the resistance encountered during breathing [8].

Growing evidence suggests that compromised lung health may be linked to faster cognitive decline and greater risk of dementia. Several epidemiological studies have demonstrated the association between indicators of pulmonary function (including FEV1, FVC, and PEF) and global cognitive decline in older individuals [2, 9–11]. However, the relationship between PF indicators and domain-specific cognitive function has been less consistent [12–18]. One study reported an association between poor PEF and faster decline in working memory, episodic memory, and semantic memory [12]. However, other studies have suggested that low FEV1 is associated with declines in spatial ability, processing speed, semantic memory, and working memory [13–16]. In addition, several studies have reported no association between FEV1 or FVC and episodic memory or semantic memory [17, 18]. In light of these conflicting results, the impact of lung function on long-term trajectories of cognitive function remains unclear, particularly since prior studies often used a single PF indicator and had relatively short follow-up (3–19 years).

There are several mechanisms by which pulmonary function could impact cognition. Low PF may result in reduced brain oxygenation, leading to vascular endothelial dysfunction, enhanced blood clotting activity, or reduced cerebral blood flow, ultimately exacerbating ischemia in some brain regions or contributing to the development of white matter lesions and lacunar infarctions [19–21]. Impaired lung function could also affect cognition indirectly by increasing the risk of cardiovascular events [18], which could contribute to cerebral hypoperfusion, hypoxia, emboli, or infarcts, causing cognitive decline by accelerating vascular damage and degenerative lesions [22]. Furthermore, PF is also related to general physical health, which could influence cognitive function by affecting factors such as physical activity level [23].

Of course, these candidate mechanisms are not mutually exclusive. Indeed, vascular and neurodegenerative brain damage can overlap and develop in parallel, leading to more severe impacts on cognition than would have resulted from either pathological process alone [24]. Mixed pathologies are very common in the aging brain [24], and the presence of vascular lesions can accelerate the progression of neurodegeneration [25].

Brain magnetic resonance imaging (MRI) can be used to investigate structural alterations in the brain, offering an opportunity to evaluate the possible mechanisms underlying cognitive changes associated with PF [26]. MRI measures the volume of different brain regions, which reflect different brain pathological changes. A reduction in total brain volume or white matter volume may be reflective of vascular brain changes [25, 27], whereas a reduction in gray matter volume is indicative of neurodegeneration [28]. White matter hyperintensities (WMH) have a multifactorial pathogenesis, but are strongly associated with cerebrovascular disease and vascular risk factors and thus could reflect microvascular lesions [25]. So far, only a few studies have addressed the association between PF and structural brain differences, and with mixed results [29–31]. Two studies have shown that lower FEV1 is related to smaller volumes of gray and white matter and a greater volume of WMH [30, 31]. However, another study has indicated that poor FVC, but not FEV1, is associated with increased WMH [29]. To date, no prior studies have addressed the association between PF and cognitive trajectories in different cognitive domains with supporting results from brain MRI data.

Based on this emerging evidence, we hypothesize that low PF accelerates cognitive decline. We further hypothesize that this is related to both neurodegeneration and the accumulation of vascular lesions in the brain.

We tested our novel hypothesis using 21-year follow-up data from the Rush Memory and Aging Project (MAP). We first analyzed the association between baseline PF and cognitive function using a composite measure of PF that included FEV1, FVC, and PEF in order to assess PF comprehensively from different dimensions. In addition to pure pulmonary function, this measure also reflects the contribution of some respiratory muscles [32, 33], capturing even more aspects of lung health in order to evaluate the PF-cognition association more globally. We assessed cognitive function in terms of five specific cognitive domains: episodic memory, semantic memory, working memory, visuospatial ability, and perceptual speed. These were operationalized as composite scores calculated from the raw scores of 19 individual cognitive tests, an approach that reduces random variability and the impact of floor and ceiling artifacts [34]. We found that low PF significantly accelerated the decline in global cognitive function as well as function in all five specific cognitive domains. To our knowledge, this is the first study to examine the impact of comprehensive PF on long-term cognitive trajectories among older adults.

Since APOE ε4, sex, smoking, and physical activity may affect cognitive function, we also explored whether these factors interacted with PF to impact cognitive decline. We did not uncover any statistically significant interactions, but this may be a reflection of the small sample size in joint effect analysis. Further research is required to clarify the interaction between PF and these factors on cognitive decline.

Additionally, to evaluate possible mechanisms underlying the association between PF and cognitive decline, we analyzed the relationship between baseline PF and structural brain differences in a subsample of Rush MAP participants who underwent MRI. We assessed the volumes of total brain, total gray matter (including cerebellar, cortical, and subcortical gray matter), total white matter (including cerebellar and cortical white matter), WMH, and hippocampus. We found that poor PF was related to smaller total brain, gray matter, and white matter volume and a greater volume of WMHs. Together, our results indicate that poor PF might contribute to cognitive decline through both neurodegenerative and vascular pathologies.

Furthermore, specific structural brain changes may affect cognitive performance in different domains [35]. Gray matter atrophy may underlie deterioration of episodic memory, processing speed, and semantic memory [30, 36–38]. White matter lesions, i.e., a greater volume of WMH, has been linked to impairments in working memory, semantic memory, visuospatial ability, and processing speed [36, 39–41]. Finally, white matter atrophy is related to the decline of working memory and processing speed [42, 43]. Thus, the connection between domain-specific cognitive decline and regional brain volumes suggests that neurodegeneration and vascular changes in the brain may underlie the association between low PF and cognitive decline.

Our findings raise the possibility that pulmonary function could play a role in the development and progression of dementia. To assess this, future studies should clarify whether poor PF is associated with the development of mild cognitive impairment (MCI) and its progression to dementia. Furthermore, PF’s relationship to other relevant brain pathology – including biomarkers from cerebrospinal fluid and brain autopsy (e.g. total tau, β-amyloid, neurofibrillary tangles, and plaques) – should be analyzed in large cohort studies with long-term follow-up in order to better understand the mechanisms linking PF and cognitive function. Furthermore, future studies investigating the underlying pathological mechanisms of the PF-cognitive function association should include a larger sample size that enables the assessment of the mediating role of structural brain differences. Finally, clinical or community trials should be conducted to determine whether improving lung health in older adults through lifestyle interventions (e.g. smoking cessation, healthy diet, exercise, management of chronic conditions, adequate sleep) can improve cognitive function and prevent or delay the onset of dementia [44].

1.2. Strengths and limitations

Strengths of this study include the community-based cohort study design with a relatively large sample and long follow-up, the comprehensive assessment of PF and domain-specific cognitive function, and the use of brain MRI data to assess the mechanisms underlying the association between PF and cognitive function. However, some limitations must be acknowledged. First, the Rush MAP participants were volunteers who were not randomly selected from the community. Compared to the general population, they were well-educated and scored highly on cognitive tests. This may have contributed to an underestimation of the magnitude of the association between PF and cognitive function. Furthermore, our findings may only be generalizable to demographically similar cohorts, and this limitation precludes the generalization of these findings to the general population. Second, the association between PF and structural brain markers was assessed based on cross-sectional data, thus the temporality of the observed association is unclear. Third, given the small sample size of MRI data available, we could not assess the mediating role of brain MRI changes in the PF-cognitive function association. Fourth, selection bias may have occurred due to missing data. However, we addressed this by repeating the analysis after multiple imputations, and the results were largely unchanged. Fifth, since PF was measured at baseline, there may be a reverse causality between cognitive function and PF. However, we addressed this by repeating the analysis after excluding participants with baseline MCI, and the PF-cognitive decline association remained significant. Thus, the temporality of the observed association is clear. Finally, potential confounding caused by unmeasured factors (such as an unhealthy diet) could not be completely ruled out.

1.3. Study conclusions

Our study provides further evidence that poor PF is associated with faster decline in global cognition and several domain-specific cognitive functions including episodic memory, working memory, semantic memory, visuospatial ability, and perceptual speed. Both neurodegeneration and vascular lesions in the brain may underlie this association. Our results underscore the importance of maintaining pulmonary health to prevent cognitive decline and subsequent dementia among older adults.

2. DETAILED METHODS AND RESULTS

2.1. Study design, setting, and participants

The Rush Memory and Aging Project (MAP) is an ongoing community-based prospective study with an emphasis on investigating the risk factors for common chronic neurodegenerative conditions in older adults [45]. Details regarding the MAP study design and the evaluation protocol have been described previously [45]. Briefly, 2,155 participants were followed up annually from 1997 until 2019, for a maximum of 21 years. Beginning in 2009, 401 participants were invited to undergo brain MRI scans. As MAP is an ongoing longitudinal cohort study, the baseline MRI scans used in these analyses could be conducted at study entry or any follow-up examination. We excluded 115 participants with prevalent dementia as well as those with missing data on baseline PF (n=591) or cognitive function during the follow-up (n=282). Therefore, the present study included 1,377 participants, 351 of whom had MRI data available (Figure 1).

Figure 1.

Flow chart of the study population.

Abbreviations: APOE ε4 = Apolipoprotein E epsilon 4; PF = Pulmonary Function; MRI = Magnetic resonance imaging; FU = Follow-up; D = Died; M = Missing.

All participants provided written informed consent and well as repository consent to allow their data to be shared. MAP was approved by the Institutional Review Board of Rush University Medical Center and was performed in accordance with the ethical standards outlined in the 1964 Declaration of Helsinki and its later amendments.

2.2. Assessment of PF

At baseline, measures of FVC, FEV1, and PEF were collected using a hand-held spirometer in order to assess PF (MicroPlus Spirometer MS03, MicroMedical LTC. Kent, UK) [46]. Two measurements were collected and averaged for each parameter. Raw scores from the three averaged measures were then converted to z-scores based on the means and standard deviations for the entire cohort. As previously reported [32], a composite PF score was created by averaging the z-scores for FVC, FEV1, and PEF, respectively. Composite PF scores were then tertiled into three categories: lowest PF (tertile 1), middle PF (tertile 2), and highest PF (tertile 3). Participants with an FEV1/FVC ratio less than or equal to 0.7 were considered to have possible chronic obstructive pulmonary disease (COPD) [32].

2.3. Assessment of cognitive function, mild cognitive impairment, and dementia

Cognitive function was assessed at baseline and each follow-up using a comprehensive battery of cognitive tests. Cognitive assessments have been described in detail in a previous study [47]. Briefly, a composite score for global cognitive function was derived by averaging the z-scores from 19 cognitive tests in the battery [48]. Scores were also derived for domain-specific cognitive functions including episodic memory, semantic memory, working memory, visuospatial ability, and perceptual speed (Supplementary Methods A).

Dementia and mild cognitive impairment (MCI) were diagnosed based on a combination of cognitive test scores, clinical judgment by a neuropsychologist, and diagnostic classification by a clinician (Supplementary Methods A) [48, 49].

2.4. Assessment of structural brain volumes

The MAP MRI protocol is described in Supplementary Methods B. Volumetric measurements (in mm³) of total brain, total gray matter (including cerebellar, cortical, and subcortical gray matter), total white matter (including cerebellar and cortical white matter), white matter hyperintensities (WMH), and hippocampus were derived. Values for WMH volume had a skewed distribution and were therefore log-transformed. All MRI data used in the present study came from stroke-free participants in order to exclude brain damage caused by stroke.

2.5. Assessment of other variables

Information on demographic characteristics, socioeconomic status, and lifestyle factors were collected at baseline [50]. Data on race was based on self-report and was classified as white or non-white. Education was represented as the number of years of formal schooling. Body mass index (BMI) was calculated as weight (kg) divided by squared height (m²). Alcohol consumption was measured as the average amount of alcohol (grams) consumed per day over the past year. Smoking status was categorized as never, former, or current smoker. Physical activity was operationalized as participants’ total number of hours of physical activity per week (according to the National Health Interview Survey) and further classified as high or low according to whether this value fell above or below the median weekly physical activity in the study population [47].

Blood pressure was measured using a mercury sphygmomanometer at baseline and each follow-up examination. Two consecutive readings were taken from the right arm while participants were in a seated position with the arm resting at heart level. A third reading was taken from a standing position after 1 min. The mean of the three values was recorded. Hypertension was defined as systolic blood pressure ≥140 mm Hg, diastolic blood pressure ≥90 mm Hg or the use of antihypertensive drugs. Heart disease and congestive heart failure were assessed based on self-reported medical history. Stroke was diagnosed by a clinician through a review of self-reported information on medical history, neurological exams, cognitive testing, and participant interviews. Diabetes was defined as hemoglobin A1c ≥6.5%, fasting plasma glucose ≥126 mg/dl, random blood glucose ≥200 mg/dl, have a history of diabetes, or the use of glucose-lowering medications [51]. Depression was diagnosed based on clinician examination or the use of anti-depressive medication.

Blood samples were taken from all participants at study entry. Apolipoprotein E (APOE) was genotyped by Polymorphic DNA Technologies (Alameda, CA) and genotypes were dichotomized as ε4 carriers vs. non-carriers in the present analysis. Additional details about the MAP data collection protocol can be found at the Rush Alzheimer’s Disease Center Resource Sharing Hub (https://www.radc.rush.edu/).

2.6. Statistical analysis

Baseline characteristics of the study population by PF category were compared using chi-square tests for categorical variables and one-way ANOVA or Wilcoxon rank sum tests for continuous variables.

The associations between PF (as both a continuous and a categorical variable) and annual changes in global cognition and domain-specific cognitive function were analyzed using linear mixed-effects models to estimate the β-coefficients and 95% confidence intervals (CIs). The fixed effect included PF, follow-up time (year), and their interaction. The random effect included random intercept and slope, allowing us to assess individual differences at baseline and across follow-up. Linear regression models were used to estimate the cross-sectional relationship between PF and structural brain volumes. The basic models were adjusted for age, sex, and education. Multivariable-adjusted models were further adjusted for race, BMI, stroke, hypertension, diabetes, depression, heart disease, congestive heart failure, COPD, alcohol consumption, smoking, physical activity, and APOE ε4. Statistical interactions were examined by creating indicator variables containing the cross-product of PF and other variables (APOE ε4, sex, smoking, and physical activity).

In sensitivity analysis, we repeated the original analyses after 1) excluding participants with MCI (n = 352) and COPD (n = 65) and 2) using multiple imputation to estimate missing values for PF and other covariates (n = 413). P-values less than 0.05 were considered statistically significant. All statistical analyses were performed using Stata SE 16.0 for Windows (StataCorp, College Station, Texas).

3. RESULTS

3.1. Baseline characteristics

Among the 1,377 dementia-free participants in the study (mean age=79.41±7.71 years, 75.74% female), baseline PF score ranged from −2.36 to 3.47. Compared to participants with the highest PF (PF > 0.33), those with middle (−0.46 < PF ≤ 0.33) or lowest (PF ≤ −0.46) PF were more likely to be older, female, have fewer years of education, poorer cognitive performance, less alcohol consumption, lower levels of physical activity, higher BMI, and have hypertension, COPD, stroke, congestive heart failure, or depression (Table 1). Subjects who did not undergo tests of pulmonary function at baseline had better global cognitive function (Supplementary Table 1).

Table 1.

Characteristics of the Rush MAP Population by Tertiles of Pulmonary Function (PF) at Baseline (N =1,377).

Characteristics	Pulmonary function (PF)*			P value
	Lowest (n = 457)	Middle (n = 464)	Highest (n = 456)
Age (y), mean (SD)	82.61 (6.78)	79.24 (7.22)	76.35 (7.99)	<0.001
Female	422 (92.34)	404 (87.07)	217 (47.59)	<0.001
Race-White	419 (91.68)	426 (91.81)	430 (92.73)	0.199
Education (y), mean (SD)	14.12 (3.26)	14.49 (2.99)	15.40 (3.36)	<0.001
BMI (kg/m2), mean (SD)	27.51 (5.29)	27.40 (5.28)	27.38 (5.45)	<0.001
Alcohol consumption (g), median (IQR)	0.00 (0.00–4.32)	0.00 (0.00–4.32)	2.16 (1.84–10.80)	<0.001
Smoking status				0.267
Never	256 (56.26)	288 (62.07)	252 (55.26)
Former smoker	184 (40.44)	164 (35.34)	190 (41.67)
Current smoker	15 (3.30)	12 (2.59)	14 (3.07)
Physical activity (h/week), median (IQR)	1.83 (0.50–4.00)	2.33 (0.75–4.60)	3.02 (1.50–5.50)	<0.001
Hypertension	367 (80.31)	342 (73.71)	309 (67.76)	<0.001
Diabetes	64 (14.00)	58 (12.50)	73 (16.01)	0.306
COPD	46 (10.07)	13 (2.80)	6 (1.32)	<0.001
Stroke	54 (12.77)	28 (6.76)	26 (6.44)	0.001
Congestive heart failure	30 (6.56)	17 (3.66)	11 (2.41)	0.006
Heart disease	44 (9.63)	43 (9.29)	39 (8.55)	0.847
Depression	102 (22.32)	81 (17.46)	60 (13.16)	0.001
APOE ε4 carrier	95 (21.16)	89 (19.52)	119 (26.50)	0.031
MMSE, median (IQR)	28.00 (27.00–29.00)	29.00 (27.00–29.00)	29.00 (27.50–30.00)	<0.001
Global cognition, mean (SD)	−0.11 (0.53)	0.11 (0.51)	0.22 (0.56)	<0.001
Episodic memory, mean (SD)	−0.01 (0.64)	0.16 (0.66)	0.17 (0.67)	<0.001
Visuospatial ability, mean (SD)	−0.23 (0.76)	−0.00 (0.77)	0.34 (0.78)	<0.001
Perceptual speed, mean (SD)	−0.22 (0.76)	0.11 (0.70)	0.29 (0.75)	<0.001
Semantic memory, mean (SD)	−0.12 (0.67)	0.12 (0.60)	0.20 (0.67)	<0.001
Working memory, mean (SD)	−0.12 (0.73)	0.07 (0.74)	0.20 (0.79)	<0.001

^*PF category: lowest group (PF ≤ −0.46); middle group (−0.46 < PF ≤ 0.33); highest group (PF > 0.33).

Abbreviations: APOE ε4 = apolipoprotein E epsilon; BMI = body mass index; COPD = chronic obstructive pulmonary disease; MMSE = Mini-Mental State Examination.

Missing data: Race = 2; Alcohol consumption = 2; APOE ε4 genotype = 23; BMI = 26; Heart disease = 1; Stroke = 136; Smoking status = 2.

3.2. Association between PF and cognitive decline

The median follow-up time was 7 years (range: 1 to 21 years). Better PF (as a continuous variable) was related to slower decline in global cognition, controlling for the estimated intercept at baseline (Table 2). In subsequent analyses, better PF was also associated with declines in all five domain-specific cognitive functions – episodic memory, semantic memory, working memory, visuospatial ability, and perceptual speed (Table 2).

Table 2.

Association between Baseline Pulmonary Function (PF) and Changes in Global Cognition and Domain-Specific Cognitive Functions.

PF	Global Cognition	Episodic Memory	Semantic memory	Working memory	Visuospatial Ability	Perceptual speed
	β (95% CI) *	β (95% CI) *	β (95% CI) *	β (95% CI) *	β (95% CI) *	β (95% CI) *
Baseline
Continuous PF	0.12† (0.08~0.16)	0.06† (0.01~0.11)	0.13† (0.08~0.18)	0.12† (0.07~0.18)	0.14† (0.08~0.19)	0.22† (0.16~0.27)
Categories PF
Highest	Reference	Reference	Reference	Reference	Reference	Reference
Middle	−0.06 (−0.13~0.02)	−0.01 (−0.08~0.10)	−0.02 (−0.11~0.06)	−0.07 (−0.18~0.03)	−0.07 (−0.17~0.03)	−0.13† (−0.24~−0.03)
Lowest	−0.21† (−0.29~−0.13)	−0.09† (−0.19~−0.01)	−0.23† (−0.32~−0.13)	−0.19† (−0.30~−0.08)	−0.21† (−0.32~−0.10)	−0.36† (−0.47~−0.25)
Longitudinal
Continuous PF × time	0.02† (0.01~0.03)	0.02† (0.01~0.03)	0.02† (0.01~0.02)	0.01† (0.01~0.02)	0.01† (0.01~0.02)	0.02† (0.01~0.02)
Categories PF × time
Highest	Reference	Reference	Reference	Reference	Reference	Reference
Middle	−0.01 (−0.03~0.00)	−0.01 (−0.03~0.01)	−0.01 (−0.02~0.01)	−0.00 (−0.02~0.01)	−0.01 (−0.03~0.00)	−0.02† (−0.03~−0.00)
Lowest	−0.05† (−0.06~−0.04)	−0.05† (−0.07~−0.03)	−0.03† (−0.05~−0.02)	−0.03† (−0.04~−0.01)	−0.03† (−0.04~−0.01)	−0.04† (−0.05~−0.02)

^*Model adjusted for age, sex, education, race, BMI, stroke, hypertension, diabetes, depression, heart disease, congestive heart failure, COPD, alcohol consumption, smoking, physical activity, and APOE ε4.

^†P < 0.05.

When PF was used as a categorical variable, compared to those with the highest PF, participants with the lowest PF had a significantly faster decline in global cognition as well as all five cognitive domains after adjustment for age, sex, education, race, BMI, stroke, hypertension, diabetes, depression, heart disease, congestive heart failure, COPD, alcohol consumption, smoking, physical activity, and APOE ε4 carrier status. Furthermore, participants with a moderate level of PF had a significantly faster decline in perceptual speed compared to those with the highest PF (Table 2 and Figure 1). The results from the basic-adjusted models were similar to those from multivariable-adjusted models (Supplementary Table 2).

While APOE ε4 is associated with both cognitive decline and PF decline in aging [52, 53], we did not detect any significant interactions between PF and APOE ε4 carrier status with respect to global or domain-specific cognitive decline (all P-values >0.05). This may be due to the small sample size of APOE ε4 carriers in the study. Additionally, APOE ε4 carriers are more likely to develop dementia and therefore may have been excluded from the study population at baseline. Moreover, no significant interactions were identified between PF and sex, smoking, or physical exercise on cognitive decline. While sex, smoking, and physical activity may affect both PF [54–56] and brain function [57–59] simultaneously, we attribute the lack of interaction to the small sample size in this joint effect analysis.

3.3. Association between PF and structural brain differences

Participants with the lowest PF had smaller volumes of total brain, gray matter, white matter, and greater WMH (Supplementary Table 3). In linear regression analysis, compared to participants with the highest PF, those with the lowest PF showed smaller volumes of total brain, gray matter, and white matter as well as greater WMH volume after adjustment for age, sex, education, race, BMI, stroke, hypertension, diabetes, depression, heart disease, congestive heart failure, COPD, alcohol consumption, smoking, physical activity, and APOE ε4 (Table 3, Supplementary Figure 1). There was no significant difference in hippocampus volume between participants with the lowest vs. highest levels of PF. Results from the basic-adjusted models were similar to those from the multivariable-adjusted models (Supplementary Table 4).

Table 3.

Association between Pulmonary Function (PF) and Regional Brain Volumes on Magnetic Resonance Imaging (MRI).

Regional Brain Volumes (mm3)	Continuous PF	PF Categories
		Middle vs. Highest	Lowest vs. Highest
	β (95% CI) *	β (95% CI) *	β (95% CI) *
Total brain	15.06 (7.23~22.89) †	−20.49 (−33.87~−7.11) †	−30.28 (−45.92~−14.64) †
Total gray matter	9.67 (3.69~15.66) †	−9.01 (−19.28~1.27)	−17.17 (−29.19~−5.16) †
Cerebellar gray matter	1.54 (0.48~2.59) †	−1.48 (−3.29~0.33)	−2.78 (−4.90~−0.67) †
Cortical gray matter	5.80 (1.43~10.17) †	−4.78 (−12.27~2.71)	−10.54 (−19.29~−1.79) †
Subcortical gray matter	2.34 (0.85~3.83) †	−2.75 (−5.31~−0.19) †	−3.86 (−6.85~−0.86) †
Hippocampal	0.03(−0.07~0.14)	−0.11 (−0.29~0.06)	−0.05 (−0.26~0.15)
Total white matter	6.92 (0.51~13.34) †	−12.96 (−23.90~−2.02) †	−15.89 (−28.67~−3.10) †
Cerebellar white matter	0.04 (−0.34~0.42)	−0.27 (−0.91~0.37)	0.10 (−0.66~0.85)
Cortical white matter	6.88 (0.59~13.18) †	−12.69 (−23.42~−1.96) †	−15.98 (−28.52~−3.44) †
White matter hyperintensities‡	−0.12 (−0.18~−0.06) †	0.11 (0.00~0.21) †	0.22 (0.10~0.34) †

^*Model adjusted for age, sex, education, race, BMI, hypertension, diabetes, depression, heart disease, congestive heart failure, COPD, alcohol consumption, smoking, physical activity, and APOE ε4.

^†P < 0.05

^‡The volume of white matter hyperintensities was transformed by taking the logarithm.

3.4. Sensitivity analysis

Similar results were obtained after we repeated the main analyses after 1) excluding participants with MCI (n = 352) or COPD (n = 65) (Supplementary Tables 5, 6) and 2) using multiple imputation to estimate missing values for PF and some covariates (n = 413) (Supplementary Table 7).

Figure 2.

Cognitive Trajectories in Global Cognition and Specific Domains by Pulmonary Function (PF) Tertiles.

Note: Trajectories represent β-coefficients from linear mixed-effect models adjusted for age, sex, education, BMI, stroke, hypertension, diabetes, depression, heart disease, congestive heart failure, COPD, alcohol consumption, smoking, physical activity, and APOE ε4, with the highest PF group as reference group.

Highlights.

• Poor pulmonary function (PF) is associated with faster decline in global cognition and domain-specific cognitive functions.

• Low PF is further related to both neurodegeneration and vascular lesions in the brain.

• Interventions designed to maintain pulmonary health may represent significant opportunities for maintaining cognitive health in late life.

RESEARCH IN CONTEXT.

1. Systematic review:

We have reviewed the literature on pulmonary function (PF), cognitive function and structural brain volumes by accessing PubMed and Web of Science databases. Several studies have examining the association of PF with domain-specific cognitive decline have shown inconsistent findings. Moreover, studies examining the association of PF with structural brain differences were sparse. It is still unclear about the impact of poor PF on cognitive trajectories and structural brain changes.

2. Interpretation:

In this community-based cohort study, we found that poor PF was associated with faster decline in episodic memory, semantic memory, working memory, visuospatial ability, and perceptual speed, as well as global cognition. Both neurodegeneration and vascular brain lesions may underlie the association. Our findings underscore the detrimental impact of poor pulmonary health on the brain, and suggest that interventions designed to maintain pulmonary health may represent significant opportunities for preventing cognitive decline in late life.

3. Future directions:

Further longitudinal studies are warranted to examine the effect of PF on cognitive decline and explore the mechanisms linking PF, structural brain differences, and cognitive trajectories.

Abbreviations:

APOE

Apolipoprotein E

PF

pulmonary function

FEV1

forced expiratory volume in 1s

FVC

forced vital capacity

PEF

peak expiratory flow

COPD

chronic obstructive pulmonary disease

MRI

brain magnetic resonance imaging

WMH

white matter hyperintensities

AD

Alzheimer disease

BMI

body mass index

CI

Confidence interval

MCI

mild cognitive impairment

MMSE

Mini-Mental State Examination