This question was examined in a large community based autopsy sample of Japanese American men who participated in the Honolulu Asia Aging Study [HAAS 1991 – ongoing]. The HAAS, and this sub-study, have been extensively described previously [6
]. This current report is based on 439 men autopsied between April 1992 and October 2001. Permission to be included in the autopsy study was obtained from the next of kin. The Institutional Review Board of the Kuakini Hospital approved the study.
Evaluation of the neuropathologic material has been described previously [6
]. Briefly, a neuropathologist identified gross lesions from coronal sections of cerebral hemispheres, and transverse sections of brainstem and cerebellum. Lacunar infarcts were defined as circumscribed, cystic, cavitary lesions that were <1 centimeter in maximum dimension; infarcts >1 centimeter were classified as large infarcts. In a microscopic exam, micro infarcts were evaluated bilaterally in the four cortical lobes, basal ganglia, hippocampus, thalamus, brainstem and cerebellum. They were defined as focal ischemic areas with reduced number of neurons, pale appearance and noticeable gliosis. All acute (<10 days) CV lesions were excluded. A modified Bielschowsky method was used to count neocortical NP and NFT in five fields in each of the four neocortical regions. There was no upper limit to the number of NFT counted; NP counts were truncated at 17/mm2
. Cerebral amyloid angiopathy was quantified by Aβ immunohistochemistry of parenchymal blood vessels; for the analysis CAA was classified as absent (all vessels non-reactive) or present.
For the analysis, we created mutually exclusive groups of vascular pathology: No lesions (reference group), only microinfarcts, only lacunaes, both lacunaes and microinfarcts, only infarcts or infarcts plus one other lesion (combined to increase sample size), and all three lesions. Multiple regression analysis was used to estimate the risk for neuropathology, controlling for age at death and whether the subject carried the Apolipoprotein E ε4 allele. Regression models for plaque and tangle pathology were based on a poisson distribution, and those for CAA were based on a binomial distribution. We examined the association of NP and NPT to vascular lesions in any location, and to lesions in the neocortex only. For these latter analyses we collapsed the vascular pathology groups to increase sample size.