Asymmetric connectivity reduction and its relationship to “HAROLD” in aging brain

Abstract

Hemispheric asymmetry reduction in older adults (HAROLD) has been frequently reported in studies of functional brain aging. It is commonly considered to be a plastic brain reorganization that provides compensation for declining unilateral neural efficiency. However, plastic functional alterations may also be associated with neural connectivity changes. Using activation and resting state functional magnetic resonance imaging (fMRI) as well as diffusion tensor imaging (DTI), we examined whether functional and structural connectivity related to prefrontal working memory function is asymmetrically reduced in the two hemispheres of the aging brain; and if yes, whether these asymmetric connectivity declines are correlated with asymmetry reduction in functional activation. With regions of interests defined by verbal working memory activations, it is revealed that although neural connectivity is generally reduced in the aging brain, prefrontal–parietal resting functional connectivity is better preserved in the left hemisphere while prefrontal DTI fiber pathways are better preserved in the right hemisphere. In addition, the laterality change of the functional activation is negatively correlated with that of the resting connectivity and positively correlated with that of the structural connectivity. These results reveal additional aspects of the neuronal alterations of aging and suggest a link between asymmetric connectivity reduction and HAROLD.

Introduction

Cognitive neuroimaging studies have shown consistently that prefrontal activation tends to be less lateralized in old adults compared with young (Cabeza et al., 1997, Cabeza et al., 2002, Cabeza et al., 2004, Dennis et al., 2007, Morcom et al., 2003, Reuter-Lorenz et al., 2000). This phenomenon is known as the HAROLD (hemispheric asymmetry reduction in older adults) effect (Cabeza, 2002), which has also been documented with electroencephalography (Bellis et al., 2000), near-infrared spectroscopy (Herrmann et al., 2006) and behavioral (Reuter-Lorenz et al., 1999) studies.

Cabeza et al. (2002) interpreted the HAROLD effect as a compensatory neural mechanism that can counteract age-related unilateral working efficiency decline. Their data showed that in a memory task, low-performing old subjects exhibited similar unilateral prefrontal activation as young adults, while high-performing old subjects engaged prefrontal regions bilaterally. Together with additional support from studies that suggest bilateral engagement facilitates language recovery from unilateral brain damage (Cao et al., 1999, Thulborn et al., 1999), contralateral compensation is considered a likely interpretation of neuroimaging evidence of HAROLD.

However, there are also reports of increased bilateral prefrontal engagement in old subjects where the contralateral involvement is not universally beneficial. For example, Colcombe et al. (2005) found significantly greater bilateral frontal recruitment in old (than young) adults during an inhibition task but the additional contralateral activation predicted poor, rather than good performance. Similarly, in a language study, Wierenga et al. (2008) found in old adults a negative correlation between picture naming accuracy and fMRI response in the non-dominant hemisphere. These data suggest that contralateral compensation is consistent with some neuroimaging findings, but there are likely other factors contributing to the HAROLD phenomenon.

The aforementioned aging studies all focused on associations between behavior and activation. However, as brain functions are supported by coordinated activations across distributed neural networks, it is highly possible that plastic functional reorganization in the aging brain is also associated with neural connectivity changes. Behavioral compensation aside, the degree to which a region activates may also be affected by how this region is connected to other collaborating areas. Specifically in the context of HAROLD, one important yet unexplored question is whether the observed activation asymmetry reduction is associated with an asymmetric connectivity decline.

This question is valid because previous studies have demonstrated that functional (Greicius et al., 2004, Rombouts et al., 2005, Sorg et al., 2007) and structural connectivity (Grieve et al., 2007, Madden et al., 2007, Makris et al., 2007, Pefefferbaum et al., 2005, Salat et al., 2005) were both non-uniformly (more pronounced in anterior regions than posterior) reduced in aging brain. With fMRI and DTI data collected in multiple sclerosis patients, Bonzano et al. (2009) also found that reduced structural connectivity could influence brain activation asymmetry indicating that changes in activation pattern may have a connectivity basis. With regard to HAROLD, the activation asymmetry reduction may also be a consequence of functional and/or structural connectivity changes. For example, for typically left dominant activations, the increased right hemisphere activity in old subjects could be due to more reduced neural connectivity in the left side such that some of the processing load needs to be shifted to the relatively intact right homologue. To examine this possible activation–connectivity association behind HAROLD, the present study attempted to answer two questions: (1) whether functional and structural connectivity associated with prefrontal verbal working memory function are asymmetrically reduced in older adults; and if yes, (2) whether these asymmetric connectivity reductions are correlated with prefrontal activation asymmetry reduction.

We collected verbal working memory and resting-state fMRI as well as DTI data from groups of old and young subjects. With the task fMRI data, we defined regions of interest (ROI) in bilateral prefrontal areas based on memory activation, and the prefrontal HAROLD effect was assessed through the activation map and laterality index. As the prefrontal and parietal regions usually co-activate (functionally linked) in working memory tasks (D'Esposito et al., 1998) and they both are known to be part of the “executive-control network” (Seeley et al., 2007), resting fMRI was used to assess hemispheric decline of this frontal–parietal functional connectivity (low frequency signal correlation). With DTI, because the precise fiber connections between the prefrontal verbal working memory area and other brain regions are difficult to ascertain, the DTI measurement of structural connectivity between prefrontal ROIs and other specific brain regions may not be valid. Instead, a fiber tracking algorithm that can calculate all possible fiber pathways passing through the bilateral prefrontal ROIs was used. It provided pathway number/amount as the connectivity strength index for the assessment of hemispheric reduction of structural connectivity. With the ROIs and connectivity indices determined, we examined the hemisphere differences of these connectivity reductions and their correlations with the prefrontal hemisphere activation changes in old subjects.