To the best of our knowledge, this is the first study to examine in vivo
the relationship between regional brain iron content and age-related differences in regional brain volumes in healthy adults. An excellent correspondence between the T2* values and regional concentrations of non-heme iron (from , Haacke et al., 2005
), supports the validity of T2* as an index of iron content (see ). Although we observed age-related increase in iron content, the regional pattern of T2* shortening differed from that of age-related shrinkage.
Figure 6 Calculated regional T2* correspond well to concentrations of iron (C) found in postmortem material (medians from multiple studies summarized in Haacke et al., 2005, ). Note, that because there are no reports on the entorhinal cortex iron content, (more ...)
In the neostriatum and the prefrontal cortex, age-related differences in volume and iron content were of similar magnitude, and no age differences were found in the primary visual cortex by either method. In contrast, within medial temporal lobe, age-related differences in T2* exceeded the differences in volume. It is unclear, whether this reflects increased sensitivity of T2* to aging, a response to age-related changes, or an early indication of incipient pathology.
T2* is affected by local content of two types of iron: heme, mainly blood-born, and non-heme, mainly sequestered, in ferritin within glia and endothelium and in transferrin within the neurons (Andrews & Schmidt, 2007
; Benarroch, 2009
; Haacke, et al., 2005
). Thus, the differences in T2* may stem from alterations of cerebral microvascular dynamics and tissue oxygenation and/or from the changes in non-heme iron homeostasis.
Non-heme iron is distributed unevenly throughout the brain (Haacke et al., 2005
; Hallgren & Sourander, 1958
; Morris et al., 1992
), and its regional contents increase with age (Hallgren & Sourander, 1958
). Concentration of iron-binding proteins increases with age in the frontal white matter, basal ganglia, and substantia nigra (Bartzokis et al., 2007
; Connor et al., 1995
; Zecca, et al., 2001
). Although, in contrast to the basal ganglia, non-heme iron content in the neocortex is relatively low (Hardy et al., 2005
; Ogg, 1999
), direct experimental manipulations of iron content show that the influence of deoxyhemoglobin and its heme iron on neocortical T2* is negligible in comparison to that of ferritin, i.e. non-heme iron (Fukunaga et al., 2010
; Lee et al., 2010
). It is possible, however, that the role of heme iron in cortical T2* increases with age because of changes in cortical vasculature. The findings by several research groups (Asslani et al., 2009
; Heo et al., 2010
; Small et al, 2004
) suggest that for the medial temporal lobe this may be the case. However, because that work was limited to the hippocampal formation and the entorhinal cortex, it is still unclear whether their findings generalize across neocortical regions.
In comparison to the neocortex and the striatum, the medial temporal regions present additional difficulty in interpretation of T2* differences, because the they may contain iron trapped in the amyloid plaques (Bartzokis, et al., 2007
; Collingwood et al., 2008
; Quintana et al., 2006
). Indeed, hippocampal ferritin content is elevated with age (Bartzokis, et al. 2007
). Postmortem (Price et al., 2009
) and in vivo
(Bourgeat et al., 2010
; Rowe et al., 2007
) studies show significant amyloid load in 22–34% of cognitively normal persons (for review see Rodrigue, et al., 2009
). Amyloid burden in the inferior temporal cortex correlates inversely with hippocampal volume (Bourgeat et al., 2010
); limbic and neocortical plaque counts obtained postmortem are associated with suppressed memory scores measured before death (Price et al., 2009
). It is possible therefore, that differential age effects on T2* of the medial temporal structures reflect elevation in the amyloid plaque load therein. Early accumulation of amyloid plaques involves only the entorhinal and perirhinal cortices, with later proliferation into the hippocampus and subsequent involvement of the neocortical sites (Braak & Braak, 1991
). It is noteworthy in that respect that we observed a significant association between EC volume and T2* only in the older participants, and no such association was found in the neocortical regions that are not expected to contain a significant number of plaques. This pattern suggests the possibility that amyloid plaques in the EC significantly contribute to T2* shortening in that structure. However, only concurrent assessment of brain iron content and amyloid burden in vivo
can test that proposition.
The third goal of the current study was to investigate the modifying effect of hypertension on T2*. We found that groups of hypertensive individuals carefully matched to their normotensive peers evidenced comparatively shortened T2* (i.e., greater iron presence) in all examined regions. Thus, hypertension may be associated with increased content of heme and non-heme iron contents beyond the normal effects of age. The addition of a vascular risk factor on top of the normal effects of aging resulted in two distinct findings. First, the hypertensive group displayed shorter T2* than the control group. Second, the differentiated pattern of brain aging (as measured by iron content inferred from T2* measures) was lost in the hypertensive group. Thus, this study replicates and extends previous findings which suggest that vascular risk exacerbates brain aging and is associated with an anterior-to-posterior gradient of regional decline (Raz et al., 2007b
), and extends this pattern from the structural level to the level of brain iron homeostasis. Notably, all hypertensive participants were treated well with anti-hypertensive agents. It is unclear, however, what if any effect different antihypertensive medication might have had on brain iron content. This important question should be addressed in future studies with larger number of participants undergoing various types of anti-hypertensive therapy.
Although the presence of non-heme iron may partly explain the effect of hypertension, T2* variations across the regions may also reflect differences in deoxyhemoglobin content in the local vasculature. Hypertension is associated with increased vascular resistance (Seals et al., 2006
), and a particular vulnerability to hypoperfusion (Bang et al., 2008
; Dai et al., 2008
; Nobili et al, 1993
) and ischemia (Ay et al., 2005
; Baumbach & Heistad, 1988
; Yao et al., 1991
; Schwartz et al., 2007
), which may account for a significant share of the T2* shortening. Although most of the hypertensive participants were taking anti-hypertensive medications that could improve cerebral perfusion (Bertel et al., 1987
; Efimova et al., 2008
), such improvement was likely to be only partial (Nobili et al., 1993
). Notably, the negative effect of hypertension on age-related differences was observed in regional T2* values but not in regional volumes. Thus, it is possible that T2* is a more sensitive index of brain aging than regional volumes and that changes in T2* due to iron accumulation precede local shrinkage.
Finally, age-related shortening of T2* may reflect hemosiderin in CMB. Although CMB are relatively common in the general population of elderly (Greenberg et al., 2009
), it is unlikely that many are present in this selective sample. In a sample with less stringent exclusion criteria, only 11% of healthy participants had CMB and none had more than three (Ayaz et al., 2010
). Nonetheless, as CMB are more likely in persons with hypertension (Ochi et al., 2009
; Sun et al, 2009
; Viswanathan & Chabriat, 2006
) they could have contributed to the observed T2* differences between hypertensive and normotensive participants. Even in that case, however, CMB would be observed in the regions that, with the exception of neostriatum, were not sampled in this study: deep white matter, thalamus, cerebellum, and brainstem (Kato et al., 2002
T2* maps used in this study included no phase information and thus cannot distinguish between iron and other inorganic materials, such as calcium (Chavhan et al., 2009
). In normal adults, calcification is common in the pineal, choroid plexus, and leptomeninges (Vigh et al., 1998
) as well as in the basal ganglia. In this study, only in the latter could calcification confound the effect attributed to iron. In light of the excellent correspondence between regional iron concentration values and the observed T2*, this confound unlikely.
Some previous studies (Siemonsen et al. 2008
) indicate that because T2* is a combination of T2 (spin-spin relaxation) and T2′ (inhomogeneity effect), the latter may be a better index of iron content. Because T2 values were not available in this study, we could not compute T2′. However, age differences in T2* and T2′ in gray matter (e.g., striatum) are equivalent (Siemonsen et al., 2008
) and it is unlikely that use of T2′ would alter the results.
This study has additional limitations. First, we excluded the participants with many common age-related pathologies and risk factors (e.g., heart disease and diabetes). Thus, our sample is not representative of aging in the general population. Second, the relatively low field strength employed in this study could have been insufficient for detection of more subtle differences in T2* that are now possible with higher-field magnets, which in spite of the added susceptibility artifacts, afford a substantially greater contrast-to-noise ratio (Duyn et al., 2007
; Shmueli et al., 2009
In conclusion, the results of this study show that regional iron concentration is sensitive index of brain aging. In longitudinal study, in vivo
assessment of regional content of brain iron may shed light on the origins of age-related regional brain shrinkage. Moreover, altered iron metabolism may mediate exacerbation of brain aging by vascular risk factors. The mechanisms underpinning the relative contribution of various types of iron to age-related T2* shortening remain to be determined. Developing more sensitive and specific MRI methods of estimating brain iron will allow a better understanding of processes that link structural and metabolic aging of the brain. The results reported here underscore the importance of closely monitoring age-related changes in vascular health, as they can be modified by behavioral and pharmacological means (Nilsson, 2008