**6.1 Voxel-wise or ROI approach?**

The MRI analysis can take a voxel-wise approach if the researchers have no particular hypotheses regarding expected areas of impact, or the MRI analysis can take a regions of interest (ROI) approach if there are research-driven hypothesis regarding expected areas of impact from physiological knowledge. There are pros and cons to both methods. It is argued that the voxel-wise approach is less biased, less time consuming, and therefore less costly. The ROI approach is argued to be driven by scientific knowledge of the physiology, more targeted and less "look-see" exploratory type research. While the ROI approach can be more costly due to the labor-intensive outlining of the specific ROI, and costly if ROI mapping is being done by more than one investigator for reliability purposes, if automated ROI templates are available, the cost and time decrease substantially. Unfortunately, not all ROI have templates. The "con" to template mapping is that a small degree of accuracy is sacrificed due to assuming a "standard shape" for a brain region when in fact there is no true standard shape (Scheibel, 2009). Thanks to the NIH Neuroscience Roadmap Initiative, there are a variety of free resources, including software mapping templates, available on the internet to download. One such site is called *MIDAS* and offers brain images as well as toolkits: http://www.insight-journal.org/midas/gallery/?flash=true

#### **6.2 How long does it take for exercise to alter brain structure?**

While rate of decline in brain mass with aging is highly individualized, it is often stated that the "normal" brain gradually decreases in size 10-15% with aging, and this shrinkage becomes particularly evident in octogenarians (Scheibel, 2004). The average rate of brain atrophy is between 0.9 to 1.5% per year after age 52, with the steepest rate of decline occurring in the frontal region with concomitant cognitive decline (Dennis & Cabeza, 2008). While it is not currently known precisely how much time is needed to facilitate structural changes (i.e., improvements) in the brain due to exercise, it is reasonable to hypothesize that relatively short-term structural change may be possible. In two separate studies demonstrating brain plasticity, as little as 1.5 to 3 months of cognitive training resulted in cortical changes in younger populations (Driemeyer et al., 2008; Haier et al., 2009). As for exercise, six months of aerobic exercise using moderate intensity walking 3 days per week

guide. More complete exercise prescription information is available in the ACSM's Guidelines for Exercise Testing and Prescription book (2009), updated every four years. The best advice for researchers using exercise as a research treatment arm, or clinicians using exercise as a therapeutic agent is to make sure one or more ACSM-certified exercise

What type of neuroimaging is best when trying to determine how exercise impacts the [aging] brain? That depends on the question(s) one needs to answer. If the goal is to determine the long-term impact of exercise on brain structure, then structural MRIs are appropriate. Structural imaging can track changes over time due to a stimulus such as exercise. Structural imaging can involve gray and white matter volume determination ("quantity"), cerebral white matter integrity ("quality") using diffusion tensor imaging (DTI), or both. It can also involve investigating the cerebral blood vessels utilizing magnetic

The MRI analysis can take a voxel-wise approach if the researchers have no particular hypotheses regarding expected areas of impact, or the MRI analysis can take a regions of interest (ROI) approach if there are research-driven hypothesis regarding expected areas of impact from physiological knowledge. There are pros and cons to both methods. It is argued that the voxel-wise approach is less biased, less time consuming, and therefore less costly. The ROI approach is argued to be driven by scientific knowledge of the physiology, more targeted and less "look-see" exploratory type research. While the ROI approach can be more costly due to the labor-intensive outlining of the specific ROI, and costly if ROI mapping is being done by more than one investigator for reliability purposes, if automated ROI templates are available, the cost and time decrease substantially. Unfortunately, not all ROI have templates. The "con" to template mapping is that a small degree of accuracy is sacrificed due to assuming a "standard shape" for a brain region when in fact there is no true standard shape (Scheibel, 2009). Thanks to the NIH Neuroscience Roadmap Initiative, there are a variety of free resources, including software mapping templates, available on the internet to download. One such site is called *MIDAS* and offers brain images as well as

While rate of decline in brain mass with aging is highly individualized, it is often stated that the "normal" brain gradually decreases in size 10-15% with aging, and this shrinkage becomes particularly evident in octogenarians (Scheibel, 2004). The average rate of brain atrophy is between 0.9 to 1.5% per year after age 52, with the steepest rate of decline occurring in the frontal region with concomitant cognitive decline (Dennis & Cabeza, 2008). While it is not currently known precisely how much time is needed to facilitate structural changes (i.e., improvements) in the brain due to exercise, it is reasonable to hypothesize that relatively short-term structural change may be possible. In two separate studies demonstrating brain plasticity, as little as 1.5 to 3 months of cognitive training resulted in cortical changes in younger populations (Driemeyer et al., 2008; Haier et al., 2009). As for exercise, six months of aerobic exercise using moderate intensity walking 3 days per week

toolkits: http://www.insight-journal.org/midas/gallery/?flash=true

**6.2 How long does it take for exercise to alter brain structure?** 

physiologists are a part of your team.

resonance angiography (MRA).

**6.1 Voxel-wise or ROI approach?** 

**6. Structural neuroimaging and exercise impact** 

for an hour each session not only prevented brain volume atrophy but resulted in brain volume improvement in older adults (Colcombe et al., 2006). Using voxel-based morphology, improvements in brain volume were noted in both the gray and white matter regions associated with executive function, long term memory, and general intelligence (i.e., the prefrontal and temporal cortices). These improvements were cautiously reported in terms of brain atrophy risk reduction in comparison to a stretching/toning control group such that a 16% improvement in aerobic fitness resulted in a 27 to 42% risk reduction of brain atrophy. The greatest risk reduction was in the anterior cingulate cortex. The stretching/toning group experienced a non-significant 5% increase in aerobic fitness but no volumetric information was reported for them. Although it is not known if the 5% improvement in aerobic fitness also resulted in some volumetric improvement, it might be surmised that embarking upon a moderate-intensity aerobic exercise program which produces at least a 1% increase in aerobic fitness may attenuate aging-related brain atrophy. This was one of the first longitudinal outcome studies reporting the impact of aerobic versus musculoskeletal-type exercise on the aging brain.

Currently, little neuroimaging information is available on other modes or durations of exercise training; nor is there information regarding how quickly the human brain structure detrains. But if the brain/cerebrovasculature mirrors the heart/cardiovasculature in exercise adaptations, then like the cardiovascular system, the cerebrovascular system may lose that 11% gain in as little as 3 weeks of no training (Coyle et al., 1984). Thus, the protective effect against brain atrophy may be lost in one short month if one is unable to exercise sufficiently. An intriguing question remains, can cognitive brain training (e.g. suduko, puzzles, playing chess, Wii-games) supplant physical activity during periods of physical inactivity in order to maintain brain structure and function?

#### **6.3 Exercise neuroimaging study shortcomings**

Being a pioneer in exercise and aging neuroscience research also means there will likely be design flaws in the research. For instance, in Colcombe et al's study (2006), the age range was wide, 60 to 79 years with a mean age of 66 years. The study age range spanned two decades with three standard aging cohorts: the older end of middle-aged (45-64 years old), the young-old (65-74 years old), and the younger end of old (75-84 years old). No mention was made regarding how many subjects fell within each of these age cohorts, therefore it is not known if these age cohorts responded differently to the exercise programs. With no variance measure or age range provided *per group* on any variable, it is difficult to assume the study did not have a few inadvertent biases. A potential younger-age bias may have preexisted in the aerobic treatment group (the treatment group was on average 1.4 years younger). The stretching/toning control group had a slightly higher percentage of females (4% more), creating a potential gender bias. Further, the actual pre-aerobic fitness distribution per grouping was unclear. Although the mean aerobic fitness (VO2) values were not significantly different between groups (~ 23 ml/kg/min), the pre-intervention VO2 values ranged from 12.9 to 49.9 ml/kg/min. Thus there were some older individuals with pre-intervention VO2 values who would be considered highly fit and therefore have less room for improvement from any type of intervention. It is not known if an attempt was made to balance the placement of these higher-fit individuals into the two groups since the methods claim group assignment was totally randomized. Furthermore, it was reported that the aerobic group was previously sedentary, however older adults with VO2's exceeding 40 ml/kg/min are not likely habitually sedentary. Individually, the between group differences highlighted here are small and were

MRI Techniques to Evaluate Exercise Impact on the Aging Human Brain 241

smaller vessels. In a separate conference paper, although cerebral blood flow velocity did not change, Rahman et al. (2008) reported less variance in the cerebral blood flow velocity in those with higher physical activity levels. To examine both the cerebral vasculature as well as cerebral blood flow, arterial-spin labeling (ASL) would be required. For either the blood oxygen level-dependent effect (BOLD) or ASL methods, intravenously (IV) injected contrast agents will produce more distinct images. However decent (but not great) images can be obtained without the IV injections. Not using invasive procedures is certainly more

If one is interested in determining which regions of the brain are being activated /oxygenated during an exercise or cognitve task, a functional MRI (fMRI) using the BOLD response would be needed. For exercise studies, the obvious hurdle to overcome is movement as most movement causes disruption in the scanning process and poor images are created. Whereas cognitive psychology has forged numerous research pathways using fMRI with BOLD contrasts to determine regions of activation in the brain during various cognitive tasks, this has not been the case with exercise training interventions. Clearly there is a need for this type of research if one desires to investigate changes in cerebral blood flow or neural hormonal factors due to an exercise stimulus from either an acute exercise bout or in response to a chronic adaptation. The stumbling block to overcome is the exercise test itself. Most exercise studies use upright testing protocols on equipment that are large and bulky with both metal and electronic parts. All of this precludes testing within the scanning room due to the magnetic field. Furthermore, by the time the subject could be transferred from the exercise apparatus to the scanning bed, critical time would be lost such that the exercise impact on the cerebrovascular system would likely be missed in all but the most

Therefore, up to this point, the more feasible methodology for cerebral blood flow investigations with exercise have been with using transcranial dopplers (TCD) and/or electroencephalography (EEG). Although these methods also have difficulty with accurate measures during movement, they are in comparison, lower in cost and easier to administer than an fMRI study. However, for approximately \$75,000 (US\$), the Lode MRI-compatible recumbent leg cycle ergometry system can now be purchased from ELECTRAMED Corporation, located in Flint, Michigan, http://www.electramed.com/MRI%20ERGOMETER %20CARDIOLOGY%20\_Details.htm. This would enable the researcher to conduct exercise tests while the subject remains in the MRI unit. Also available are MRI-compatible electrocardiography and blood pressure measurement units, thereby solving the equipment issue. Unfortunately, this particular equipment model is only compatible with a 1.5 T MRI scanner and only with select manufacturers. Given that most research is now being done on 3.0 T or 4.0 T scanners, this ergometer may not be useable for many research protocols. The final issue left to resolve is an acceptable exercise protocol that would be taxing enough yet involve minimal movement from the torso up during scanning sequences. One potential resolution would be to develop an intermittent exercise test protocol so that exercise bouts would take place during the imaging sequence changes, akin to an event-related design. Since stimuli in an event-related design are presented as isolated events of short duration, a brief

appealing to the volunteer subject and helps to contain the imaging costs as well.

**7. Functional neuroimaging and exercise** 

deconditioned subjects.

**7.1 MRI-friendly leg cycle ergometer** 

reported to be non-significant, but considered collectively, these small biases could contribute to confounding the interpretation of the results.

Sociological studies have shown that women outlive men by 4 to 10 years, thereby partially explaining why there are usually more women in research studies involving older adults. Although women tend to live about a decade longer than men, they also experience an accelerated pace of physiological decline between their seventh and eighth decade of life. Older men tend to weigh more, be more physically active and have a higher degree of aerobic fitness than older women of the same age (Spriduso et al. 2005) and women's brain volumes are smaller than men (Allen et al., 2003). Thus care must be taken when studying variables with inherent age or gender differences. Colcombe et al. (2006) made no mention of controlling for age or gender in the statistical analyses of their data in order to determine if improvements attributed to aerobic fitness change was independent of age or gender influences. It is well known that both age and gender can impact brain volumes and cognition independently (Madden et al, 2009). Failure to control for these variables can lead to potentially erroneous conclusions. For example, Marks et al. (2010) initially noted moderate positive relationships in the anterior cingulum segment between cerebral white matter integrity and aerobic fitness as measured by diffusion tensor imaging (DTI). However, upon controlling for both age and gender, only the middle and posterior cingulum segments remained significantly related to aerobic fitness. Similarly, this pattern of reduced significance was repeated in a voxel-wise brain analysis on the same data (Liu et al. 2009). Thus it is critical to control for factors that are known to impact the brain and/or aerobic fitness parameters. Lastly, neither the exercise test protocol nor the "stretching and toning" prescription was ever fully described by Colcombe et al. (2006). This lack of information makes it difficult to determine the validity of the aerobic fitness and strength training outcomes, and it is even more difficult to impart health recommendations with confidence. Hence the encouraging conclusions regarding aerobic fitness and brain improvement from this study by Colcombe et al. (2006) must be viewed with cautious optimism. There are currently a few new NIH-funded exercise trials involving both healthy and diseased older adults in progress (http://projectreporter.nih.gov/reporter.cfm), with intervention timelines and exercise protocols seemingly mirroring Colcombe et al. 's initial study (2006) . Hopefully, these newer studies will not only control for potential confounding variables but also provide sufficient exercise testing and training details to render their studies replicable.

#### **6.4 Magnetic resonance angiography and exercise impact**

Magnetic resonance angiography (MRA) in conjunction with DTI is helpful in determining the status of one's cerebral blood vessels. Using a process known as arterial spin labeling (ASL), the quality and quantity of the cerebral blood flow can be determined with or without perfusion. It is believed that the progressive reduction in cerebral blood flow attributed to the aging process may be caused by a reduced metabolic demand due to a reduction in neurotransmitter synthesis (Orlandi & Murri, 1996) and/or underlying microvascular disease (Bullitt et al., 2010). The consequential neural atrophy results in smaller cerebral arteries, increased intracranial resistance and slower arteriole vasomotor reactivity (Orlandi & Murri, 1996). Even though studies suggested aging may be associated with smaller cerebral vessels, Bullitt et al. (2010) reported that vessel diameter reductions may be compensated for by an increase in vessel number and that both larger and smaller vessels were impacted. In a sub-study comparing active versus inactive older adults, Bullitt et al. (2009) reported significantly lower vessel tortuosity along with a higher number of smaller vessels. In a separate conference paper, although cerebral blood flow velocity did not change, Rahman et al. (2008) reported less variance in the cerebral blood flow velocity in those with higher physical activity levels. To examine both the cerebral vasculature as well as cerebral blood flow, arterial-spin labeling (ASL) would be required. For either the blood oxygen level-dependent effect (BOLD) or ASL methods, intravenously (IV) injected contrast agents will produce more distinct images. However decent (but not great) images can be obtained without the IV injections. Not using invasive procedures is certainly more appealing to the volunteer subject and helps to contain the imaging costs as well.
