This study demonstrated a significant relationship between cortical blood supply (assessed via cortical CBF) and subcortical white-matter integrity (assessed via fractional anisotropy and diffusivity). Importantly, these associations were observed in healthy-appearing white matter. Moreover, these effects were not entirely explained by age or by vascular risk. In addition, these associations were spatially selective and only partially coincide with the white matter regions demonstrating the most significant age-associations. It is important to note that the current data do not permit us to ascribe causal or directional mechanisms of these associations. However, these findings provide compelling evidence for a link between cortical neurovascular physiology and white-matter health, both as a part of normal aging as well as a part of the intrinsic inter-individual variability that is not accounted for by age.
Age-associated reductions in white-matter integrity were widespread, but were more prominent anteriorly in structures such as the corpus callosum, as consistently reported in prior work 
. In contrast, the association between cortical CBF and white-matter integrity was more evenly distributed between anterior and posterior regions of the callosum, suggesting potentially distinct mechanisms underlying than found in aging. Our findings also raise the possibility that the measured associations reflect to a large degree physiological variability among healthy individuals, not dominated by age but instead influenced by other biological and environmental factors. In contrast, the associations between cortical thickness and white-matter microstructure were dominated by age-associated variance. Ongoing work is exploring potentially mediating factors behind these distinct observations.
Previous studies have found an association between white-matter lesion formation and risk for neurovascular diseases 
. Also, periventricular white-matter lucency has been linked to impaired cerebral autoregulation 
, potentially associated with age-related arterial stiffening and attenuation of autoregulatory capacity 
. Although such associations have been experimentally demonstrated in patients with type 1 diabetes 
, the relationship between vascular health and white-matter integrity has yet to be demonstrated in normal aging. Interestingly, we noted that while certain vascular risk factors, such as hypertension and diabetes, are associated with reduced CBF 
, the inclusion of subjects with mild vascular risks did not significantly alter the trends in the DTI-CBF relationship we observed. This finding supports our view that vascular function and white-matter integrity co-vary in a healthy adult population irrespective of vascular impairment. Nonetheless, it is important to note that CBF is one of many indices of cerebrovascular health. For instance, while CBF is correlated with blood pressure 
, CBF alone does not determine vascular reactivity or neurovascular interactions. In our future work, we will examine more direct mechanisms underlying the results reported here.
While all white-matter structural parameters exhibited strong associations with age and with CBF, radial diffusivity (RD), a potential indicator of myelo-degeneration in aging 
, was more markedly associated with CBF (than with age) than the remaining DTI parameters (results controlled for age). This suggests that reductions in perfusion, or more generally, compromised cerebrovascular health, may contribute to white-matter damage through a “low-flow” or “low-grade” ischemic mechanism 
. This mechanism may be distinct from the primary mechanism driving age-related axial diffusivity (AD) increases, as axial diffusivity was more strongly related to age than to CBF. Also, the larger extent of RD association with CBF (when compared to AD-associations) may indicate a preponderance of myelin degeneration with advancing age. Additional work is necessary to disentangle the potential mechanistic distinctions, and to determine the link between these empirical observations and specific histopathology.
It should be noted that CBF measurements reflect both neuronal metabolic activity and vascular physiology, giving rise to two alternative interpretations to our findings. Under the metabolic hypothesis, white-matter integrity would impact neuronal function in regions projected to by the affected tracts, leading to a covariation between CBF and white-matter microstructure only in anatomically connected tissue regions. While ongoing work is examining this mechanism, we noted that the spatial patterns of the measured DTI-CBF associations did not follow the fibre structure of the affected white-matter regions, as demonstrated in the corpus callosum, the structural connectivity of which is well documented 
. On the other hand, under a vascular hypothesis, there may be a more general association between cerebral perfusion and white-matter integrity, stemming from vascular damage associated with normal aging 
. However, associations between callosal microstructure and cortical CBF do not reflect the known vascular supply routes to the various callosal regions 
. It is of note that the regions exhibiting significant CBF-DTI links, namely the superior-frontal, medial-frontal, temporal and precuneal regions, are also primary components of the “default-mode network”, associated with high glucose and oxygen metabolism, as well as vulnerability to degeneration in aging and in age-related diseases 
. It is possible that the high baseline perfusion in these regions biased the detection of the associations. It is also possible that these highly metabolizing regions are most sensitive to changes in global blood flow, which impact the white-matter integrity in regions most vulnerable to vascular insult (e.g. regions supplied by deep penetrating vessels) 
. Future work will probe associations among metabolic demands, CBF, and white-matter integrity in detail through the simultaneous measurement of hemodynamic and metabolic variables. Irrespective of the mechanism underlying the observed associations, our findings may provide important mechanistic insight into understanding the age-associated decline in connective tissue integrity 
A recent study examining the relationship between diffusivity and CBF, both measured in the white-matter, found lower perfusion in apparently healthier fibres, attributed to their higher degree of myelination, hence lower energy demand and greater impedance for vascular penetration 
. However, as histological studies of vascular anatomy demonstrate a continuous vascular supply path between grey and white matter 
, highly perfused white-matter should underlie highly perfused cortex in the same vascular territory. Such a scenario would predict greater white-matter integrity with higher CBF, as demonstrated here and in prior PET work 
. These contradicting findings represent alternate models of how CBF relates to neural health, and remain to be reconciled.
The current results should be interpreted in view of the potential caveats. First, given the nature of diffusion-weighted contrast, white-matter diffusivity may potentially contain contributions from microvascular blood flow. However, within the white-matter, blood vessels are oriented along the main fibre direction 
; hence, if the diffusivity measures had significant perfusion contribution, one would expect high CBF to correspond to high axial diffusivity, which was not the case here, precluding significant vascular contribution. Secondly, both ASL and DTI are intrinsically sensitive to subject-motion. However, motion would be expected to result in global rather than these spatially specific biases, suggesting that artifacts have limited influence. Thirdly, the older adults in the current sample contained more women than men, with the men being older than the women and blood pressure trends were slightly lower than normative values in this age range (compared with 
). These factors may affect the generalizability of these results. The levels of education seen in our older population exceeded those of our other age groups, but not at a statistically significant level. Fourthly, while prior literature associated AD and RD to different histological features, such interpretations can be influenced by technical limitations with regard to DTI acquisition and analysis, therefore should be made with caution. Fifth, as a caveat to the interpretation, while we have shown that certain regions are more strongly associated with CBF than age, and vice versa, our study does not demonstrate a quantitative causality. It is possible, as we alluded to earlier, global cortical CBF is in part a measure of cerebrovascular health, and the most vulnerable white-matter regions would be associated with an overall decline in CBF, irrespective of the spatial distribution of the latter. Finally, we note that the reported associations were found in cross-sectional sample. The longitudinal trajectories of CBF and DTI measures is investigated in our ongoing work, and may provide information about the causality of the associations investigated here.
Using DTI in conjunction with pulsed ASL perfusion imaging, we found a link between cortical CBF and subcortical white-matter microstructural health. The associations were regionally specific, not simply accounted for by age or by vascular risk. These findings provide support for a connection between cortical vascular physiology and subcortical white-matter health, and may have important implications for understanding the basic mechanisms of neurodegeneration.