Article Text

Download PDFPDF
Aerobic fitness and obesity: relationship to cerebral white matter integrity in the brain of active and sedentary older adults
  1. BL Marks1,2,3,
  2. LM Katz1,3,
  3. M Styner4,
  4. JK Smith5
  1. 1Department of Exercise and Sport Science, University of North Carolina at Chapel Hill, Chapel Hill, North Carolina, USA
  2. 2Department of Allied Health Sciences, University of North Carolina at Chapel Hill, Chapel Hill, North Carolina, USA
  3. 3Department of Emergency Medicine, University of North Carolina at Chapel Hill, Chapel Hill, North Carolina, USA
  4. 4Department of Computer Science and Psychiatry, University of North Carolina at Chapel Hill, Chapel Hill, North Carolina, USA
  5. 5Departments of Radiology and Neuroimaging, University of North Carolina at Chapel Hill, Chapel Hill, North Carolina, USA
  1. Correspondence to Dr BL Marks, Department of Exercise and Sport Science, Fetzer Gym, Campus Box #8700, University of North Carolina at Chapel Hill, Chapel Hill, NC 27599-8700, USA; marks{at}


Objective Aerobic fitness (VO2 peak) and obesity risk (OR) may impact brain health. This study examined hemispheric and segment specific relationships between VO2 peak, OR and cerebral white-matter (CWM) integrity in the cingulum brain region in healthy older adults.

Methods Fifteen subjects (66±6 years) completed VO2 peak testing and MRI of the brain. OR was determined via body mass index (BMI) and abdominal girth. MRI analysis was performed with a structural 3D T1 MP-Rage and diffusion tensor imaging technique (DTI, 21 directions, repeated four times) on a 3.0 T MR imaging unit. CWM integrity indices, fractional anisotropy (FA) and mean diffusivity (MD), were computed from the tensors. The anterior, middle and posterior cingulum segments were analysed on both sides of the brain. Partial correlations (age and gender controlled) and standard multiple regressions were used to determine significant associations and unique contributions to CWM integrity.

Results VO2 peak was moderately related to FA in the left middle cingulum segment (r partial=0.573, p=0.041) and explained 28.5% of FA's total variance (p=0.10). Abdominal girth (r partial=−0.764, p=0.002) and BMI (r partial=−0.690, p=0.009) were inversely related to FA in the right posterior cingulum (RPC) segment. Abdominal girth and BMI uniquely explained 53.9% of FA's total variance (p=0.012) and 43.9% (p=0.040), respectively, in the RPC.

Conclusion Higher aerobic fitness and lower obesity risk are related to greater CWM integrity but not in the same cingulum segments.

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.


Participation in regular physical activity is related to better executive function (planning, scheduling, working memory, task coordination) and successful brain ageing.1,,7 However, the specific mechanisms by which this occurs are unclear. A healthy lifestyle incorporating moderate to vigorous exercise is necessary for cardiovascular disease risk reduction and may be vital for optimal brain health.8,,10 Declining cardiovascular function is strongly related to cognitive decline.11,,13 Decreased brain volumes in regions related to executive function and memory in older adults with lower aerobic fitness have also been reported.14 15 In addition, depression has been associated with hippocampal atrophy16 and now obesity, a known metabolic and cardiovascular disease risk factor,8 9 is suspected as a risk factor for brain atrophy and cognitive decline.16,,22

Cerebral white-matter (CWM) fibres are responsible for neuronal transmission and integration of cognition between brain regions and across cerebral hemispheres, but there is limited information on whether exercise affects these fibres. Diffusion tensor imaging (DTI) can detect degradation of CWM fibre bundles, decreased axonal integrity and potential demylination, and therefore may provide a tool for early detection of impending cognitive decline not possible with standard MRIs.23,,25 DTI provides detailed delineation of the CWM pathways based on rates of microscopic water diffusion. A higher degree of CWM integrity is represented by a greater degree of fractional anisotropy (FA) of diffusion and a lower mean diffusivity (MD), thereby indirectly implying the maintenance of adequate neural connectivity.26 Hence, a disruption in neural transmissions may contribute to a decline in cognitive function.27

CWM integrity begins to decline after age 60 in the anterior prefrontal regions associated with executive function and memory and exhibits an anterior–posterior gradient in the cingulum.28 29 The cingulum is a bundle of white-matter fibres underlying the cingulate cortex anatomically positioned above the corpus callosum. Neural connections traverse through the cingulum bundle and continue on to other cortical structures.30 Thus, the cingulum's interconnectivity with the amygdala, frontal motor areas, hippocampus and hypothalamus renders its involvement with emotional regulation, motor movement, cognitive function and cardiovascular autonomic control.30

Findings by Marks and others31 reported that greater aerobic fitness in both younger and older adults, independent of age or gender, was significantly related to greater CWM integrity (FA) in the cingulum. However, that study was a retrospective analysis, and aerobic fitness was estimated32 33 rather than actually measured. The brain mapping analyses used a very basic six-directional DTI process, MD was not reported,34 and no hemispheric or gradient-specific relationships were evaluated. In fact, human brain mapping has suggested that the human cingulum needs to be mapped more precisely, evaluating not only the anterior and posterior segments but also the middle segment (rather than absorbing it into both the anterior and posterior portions).35 36

Therefore, the first aim of this study was to verify the relationship between aerobic fitness and CWM integrity in the cingulum of healthy older adults via actual measurement of oxygen consumption (VO2) and using a more complex DTI sequence. A second aim was to determine if a relationship exists between obesity risk and CWM integrity in the cingulum. A third aim was to determine if specific segments were related to aerobic fitness and/or obesity risk. The final aim was to determine if hemispheric relationships exist. Since executive function has been associated with the prefrontal (anterior) brain region, it was hypothesised that a greater CWM integrity would be found in the anterior region of the cingulum but that there would be no hemispheric associations.



This study was approved by the University of North Carolina at Chapel Hill's Committee on the Protection of the Rights of Human Subjects. All subjects provided informed written consent for participation. One hundred and twenty community-dwelling older adults were prescreened over the telephone. Subjects were purposefully recruited per their self-reported health and physical activity histories in order to maximise chances to determine the differences in brain structure (10 active=engaged in aerobic activities over 180 min/week for past 10 years; 10 sedentary=engaged in any physical activity under 90 min/week for the past 10 years).8 Those accepted into the study were healthy without orthopaedic, cardiopulmonary, metabolic, cognitive or MRI testing limitations.8 37 None were on antidepressants or on medications that would alter the heart-rate response to exercise. Women were postmenopausal and not on hormone replacement therapy. MRI exclusionary criteria included non-removable metallic and electronic implants, claustrophobia and/or a past history of inability to complete an MRI scan. Although 20 subjects were accepted into the study (60 to 76 years old; 11 males, nine females), only 15 completed all phases of the study (table 1). Testing was completed within 4 weeks.

Table 1

Subject exclusion information


Cognitive status, depression risk, physical exam, aerobic exercise test, obesity risk.

Cognitive status and depression risk

The Telephone Interview for Cognitive Status (TICS) was used to screen for general cognitive function.38 Scoring guidelines suggested scores below 26 (on a point scale of 0–41) should be considered mildly impaired and scores between 26 and 32 should be considered ‘ambiguous’; however, the validation research suggested that a score cut-off of 30 was accurate for detecting dementia.38 39 Thus, a score below 32 was pre-established as the exclusionary score for this study. Depression and medications used to treat depression can confound the physiological, cognitive and MRI test results, so subjects were screened for depression using the Beck Depression Inventory (BDI).39 40 A score of 15 (on a point scale of 0–63) is considered clinical depression, so a score of 15 or greater was used as the exclusionary score.40

Physical exam

A physician reviewed the subjects' TICS, BDI and medical history questionnaires, and then conducted the general physical exams. Those with no active medical diseases or any other condition that would exclude them from participating in an aerobic exercise test or MRI were approved for the study.

Aerobic exercise test

Subjects underwent a physician-supervised peak exercise (walking) test on a treadmill utilising Duke's ramped Pepper Treadmill Protocol.41 VO2 peak was measured using the PARVo TrueMax VO2 Metabolic Cart System. Heart rate was continuously monitored via a 12-lead electrocardiogram. Blood pressure was measured manually every 3 min. Ratings of perceived exertion (6–20 scale) were obtained at the end of each stage.8 41 Exercise test termination criteria included signs and symptoms of exertional intolerance or any symptom the supervising physician deemed warranted stopping the test.8

Obesity risk

Subjects had their height, weight and abdominal girth measured for obesity risk determination. Height and weight were measured without shoes on a calibrated balance-beam weight scale with a height rod. Body mass index (BMI) was calculated from height and weight. The abdominal girth was measured three times with a fibreglass spring-loaded tape measure at the level of the iliac crest, in line with the umbilicus.42 The mean of the two lowest girth measurements within 1 cm was used for analyses.


Structural images were acquired with a head only 3.0 T MRI unit (Allegra, Siemens Medical Systems, Erlangen, Germany) with a maximum gradient strength of 40 mT/m and a maximum slew rate of 40 mT/m/ms. A spin echo diffusion weighted imaging (DWI) sequence was used to acquire the MR images. A baseline image (b=0, no diffusion gradient) and 21 directional images (b=1000 s/mm2) were acquired (four repetitions) at an isotropic resolution of 2×2×2 mm. DTI tensor data were estimated from the raw DWI data via a weighted least-squares fitting. FA and MD maps were then computed from the tensors. Regions of interest were outlined using the Insight SNAP tool ( An investigator blinded to participant characteristics outlined and analysed the anterior, middle, posterior and all segments combined (anterior+middle+posterior) on a slice-by-slice basis for each subject on each side of the brain (figure 1).44 High anisotropy measurement stability (coefficient of variance=3.9%) was determined via three repeated measurements on the overall datasets.45

Figure 1

Segmented cingulum fibre bundle, sagittal view, left hemisphere. A, anterior; I, inferior; P, posterior; S, superior. Red area, anterior cingulum segment; teal area, middle cingulum segment; fuschia area, posterior cingulum segment; lavender area, genu of the corpus callosum.

Statistical analyses

Means and SDs were used to describe the subjects' descriptive characteristics. Differences in descriptive characteristics and CWM integrity between the three segmented regions of the cingulum (anterior, middle, posterior) were investigated with a one-way repeated-measures ANOVA with a Bonferroni adjustment for multiple comparisons. Bivariate correlations were used to determine relationships between descriptive characteristics (age, gender, abdominal girth, BMI, VO2 peak) and CWM integrity (FA and MD in the various cingulum segments). Due to the influence of age and gender on many of these variables, partial correlations (correcting for age and gender) were also employed. Significant partial correlations were followed up with standard multiple linear regressions to determine the variables' unique contributions to CWM integrity. Significance was set at p≤0.01 to reduce Type I errors, but since this was a small preliminary study, trends (p<0.10) were also reported, and those exhibiting p<0.05 were explored.46 47 Analyses were completed with the SPSS Software System (V 17.0, SPSS, Chicago, Illinois).


Descriptive characteristics are listed in table 2. There were no significant gender differences in age, education, or abdominal girths; nor were there any significant age or gender differences for the BMI, TICS or BDI scores. Neither the TICS nor BDI scores suggested cognitive impairment or clinical depression. The active group exhibited significantly greater VO2 peaks than the sedentary group, and as expected, males and ‘younger’ older adults had greater VO2 peaks than their counterparts.

Table 2

Descriptive characteristics of the subjects (n=15)

The independent variables exhibited varying degrees of association (table 3). As shown in table 4, CWM integrity peaked in the cingulum's middle segment, with FA being consistently higher and MD lower in comparison with the FA and MD values in either the anterior or posterior cingulum segments.

Table 3

Correlation matrix displaying bivariate relationships between the independent variables

Table 4

Fractional anisotropy (FA) and mean diffusivity (MD) values for the cingulum segments per cerebral hemisphere (right or left side) and both sides combined uncorrected for descriptive characteristics (n=15)

Aerobic fitness (VO2 peak) and obesity risk (abdominal girth, BMI).

There were no adverse events during exercise testing. All subjects attained or slightly exceeded their age-predicted maximum heart rates (220–age), RPE scores met or exceeded a rating of 17, but a VO2 plateau and/or a respiratory exchange ratio >1.15 was not achieved in all cases. Since their physiological results did not meet the above criteria to be considered a ‘true maximal’ test, the exercise tests were classified as ‘peak’ rather than ‘maximal’ efforts.48

Bivariate relationships listed in table 5 suggested that a greater VO2 peak was moderately associated with greater FA on the left side of the brain, specifically in the anterior and middle cingulum segments. However, once corrected for age and gender influences, the anterior relationship diminished, and a middle–posterior relationship emerged (table 6). Similarly, controlling for age and gender lessened the relationship between MD and VO2 peak in the left posterior segment but strengthened an inverse relationship in the left middle segment. All of the noted VO2 peak associations with CWM integrity only attained a statistical significance suggestive of ‘trends’ (p=0.094–0.048).

Table 5

Segmental and hemispheric bivariate relationships between cerebral white-matter (CWM) integrity in the cingulum (fractional anisotropy, mean diffusivity), aerobic fitness (VO2 peak), abdominal girth, BMI, age and gender

Table 6

Segmental and hemispheric relationships (partial correlations correcting for age and gender) between cerebral white-matter integrity in the cingulum (fractional anisotropy, mean diffusivity), aerobic fitness (VO2 peak), abdominal girth and BMI

On the other hand, the obesity risk variables demonstrated strong right hemispheric specificity with no overlap into the left hemisphere when corrected for age and gender (table 6). Both higher abdominal girths and higher BMI scores were significantly associated with lower FA, specifically in the right posterior cingulum. To determine if these two partial correlations were significantly different from each other, their partial r scores were converted into standardised z scores using an ‘r to z’ transformation table.46 This resulted in a calculated Zobs score of 0.385 (see formula in table 6), indicating no statistically significant difference between the abdominal girth and BMI correlations as the Zobs score fell well within the −1.96 to +1.96 range.46 There was also a trend (p=0.05) for higher abdominal girths to be associated with higher FAs in the anterior segment of the right cingulum.

Because both abdominal girth and BMI had significant partial correlations for FA in the right posterior cingulum segment (p≤0.01), a standard multiple regression was used to determine each of their unique contributions to CWM integrity. Due to their high correlation with each other (r=0.824, p<0.0005), separate regression analyses were run to avoid multicollinearity issues. As shown in tables 7 and 8, both abdominal girth and BMI contributed significantly to explaining the total variance in their respective models. Abdominal girth uniquely explained 53.9% of the total variance found in FA in the age–gender-girth regression model, whereas BMI uniquely explained 43.9% of the total variance found in FA in the age–gender–BMI regression model. Interesting, while age remained a significant contributor in both models, gender did not; gender was important only for the BMI model, trailing slightly behind age.

Table 7

Standard multiple regression for gender, age, and abdominal girth on cerebral white-matter integrity (fractional anisotropy) in the right posterior cingulum segment

Table 8

Standard multiple regression for gender, age and BMI on cerebral white-matter integrity in the right posterior cingulum segment

Three partial correlation ‘trends’ (p<0.05) were also explored with regression analyses due to the pilot nature of this study. As shown in table 9, age uniquely explained almost half of the total variance of FA in the right anterior cingulum (41.7%, p=0.01) with abdominal girth contributing 26% (p=0.03). The regression models incorporating VO2 peak (tables 10, 11) to explain total variance in the left middle and posterior cingulum (L+R sides combined) were statistically weak, and so the potential unique contribution of VO2 peak (28.5 to 30.7%, p<0.05) must be viewed with caution.

Table 9

Standard multiple regression for gender, age and abdominal girth on cerebral white-matter integrity in the right anterior cingulum segment

Table 10

Standard multiple regression for gender, age and aerobic fitness (VO2 peak) on cerebral white-matter integrity (FA) in the left middle cingulum segment

Table 11

Standard multiple regression for gender, age and aerobic fitness (VO2 peak) on cerebral white-matter integrity (FA) in the posterior cingulum segments (left and right sides combined)


The results of this investigation further elucidate factors that may be associated with CWM integrity in the cingulum and suggest segment and hemispheric specific affinities. The middle segment has traditionally been mapped to be shared by both the anterior and posterior segments, which then masks the middle segment's potential contribution. This study more precisely identified the potential location of impact of aerobic fitness as well as obesity risk on the cingulum structure.

This study extended the earlier work of Marks and others31 by finding that a positive aerobic fitness:FA relationship may be specific to the middle cingulum segment and perhaps overflows into the posterior segment rather than impacting the entire cingulum or only the anterior portion. This finding suggests that aerobic fitness may contribute to the anterior–posterior gradient postulated to be in effect within the cingulum structure.28 29 Not only may the degradation of CWM integrity in the anterior cingulum be due to the ‘last in–first out’ theory of cognitive development, but greater aerobic fitness may allow for more blood circulation in the brain, particularly in the middle and posterior cingulum segments, which may then help preserve CWM integrity in that region.

Aerobic fitness facilitates improved oxygen delivery and utilisation in the cardiocerebral vascular systems, and improved oxidative capacity has been shown to upregulate expression of other important neuronal growth factors such as insulin-like growth factor I (IGF-I), brain-derived neurotrophic factor (BDNF) and related protein precursors in animal models.49 50 Although neural-growth factors were not measured in this study, it is speculated that aerobic fitness may mediate improved CWM integrity via the intricate adaptations that take place on the neural–humoral level during exercise. Future studies need to incorporate longitudinal exercise trials that include not only structural imaging but also hormonal measurements and perfusion imaging.

Critchley and others51 reported evidence of a relationship between the anterior cingulate gyrus and cardiovascular autonomic control. Initially, we too noted a moderate positive relationship in the anterior cingulum segment between CWM integrity and aerobic fitness. However, after controlling for both age and gender, that anterior relationship diminished, and the middle and posterior segments surfaced as the probable cingulum sites related to aerobic fitness. Thus, it is crucial to control for both age and gender in order to obtain a truer picture of aerobic fitness's relationship to brain health. The premise that aerobic fitness may induce CWM integrity adaptations in the cingulum, which in turn may track to the cardiovascular control centre in the brainstem, needs to be investigated further via fibre tractology.

It is known that the posterior region of the cingulate gyrus is important for both motor movement and memory function, and its association with obesity is becoming a critical line of investigation. The finding that obesity risk, specifically large abdominal girths and higher BMI scores, is significantly related to less CWM integrity in the right posterior cingulum is an important novel discovery. Translating this finding into a public health perspective suggests that older adults with expanding abdominal girths and creeping BMI scores yet not deemed to be clinically obese may be experiencing preclinical CWM decline, and this decline may target the right side of the brain. There are several cross-sectional studies supporting these findings.

Obesity negatively impacts brain matter volume in early postmenopausal women52 and adverse neurological outcomes coupled with brain volume loss have been associated with larger waist-to-hip ratios as well as greater BMI scores.19 53 54 Raji and others21 reported that those with a BMI of 25–30 exhibited atrophy in the basal ganglia and corona radiate of the white matter, while those with BMIs exceeding 30 exhibited brain atrophy in the frontal lobe, anterior cingulate gyrus, hippocampus and thalamus. Most recently, weight loss was shown to potentially reverse obesity's negative impact on CWM volume,55 and posterior cingulum abnormalities have been linked to the progression of Alzheimer disease.56 Thus, intervention studies are needed to determine the causal nature of these relationships and their impact on cognitive function.

In this study, hemispheric associations between aerobic fitness, obesity risk and CWM integrity were evident. A higher VO2 peak was associated with a higher FA and lower MD on the left side of the brain in the cingulum, whereas higher BMI and abdominal girths were associated with lower FA on the right side of the brain. Theorising the clinical implications is difficult, as the theory of lateralisation is subject to debate, and the extent of lateralisation in older adults is unclear due to potential compensatory responses.57,,60 Therefore, hemispheric dominance patterns for aerobic fitness and obesity in relation to cognitive decline warrant further research.

Limitations and conclusions

The sample size was small due to the strict requirements for healthy, relatively non-medicated, active and sedentary older adults. No hormonal measures, tracer perfusion or exercise interventions were performed during this study. Although the findings have limited generalizability, the preliminary results are encouraging. Longitudinal diet and exercise intervention studies examining CWM integrity, neural–humoral factors and cerebral blood flow indices are needed to determine causal impact and clinical relevance.

What is already known on this topic

Low aerobic fitness and obesity are associated with brain volume atrophy in the prefrontal brain region associated with executive function and memory.

What this study adds

Aerobic fitness and obesity may selectively impact different brain hemispheres as well as different regions within the cingulum structure. Greater aerobic fitness is moderately associated with greater cerebral white-matter integrity in the middle cinculum region on the left side of the brain, whereas higher BMIs and higher abdominal girths are highly associated with lower cerebral white-matter integrity in the posterior cingulum region on the right side of the brain.


The authors are indebted to K Wilber and C Vachet, for their MRI assistance, and J Hawkins and M Babbitt, for their exercise testing assistance.



  • Funding This pilot study was made possible by a seed grant awarded to Dr Marks by the Biomedical Research Imaging Center (BRIC), School of Medicine, University of North Carolina at Chapel Hill.

  • Competing interests None.

  • Ethics approval Ethics approval was provided by the University of North Carolina at Chapel Hill, Biomedical Internal Review Board, Study #05-3151.

  • Provenance and peer review Not commissioned; externally peer reviewed.