Extensive evidence indicates that the vascular factors contribute to cerebral neurodegeneration in Alzheimer disease (AD) (1). Multiple vascular risk factors were found to be related to AD in large-scale epidemiological studies (2,3), and control of these vascular risk factors may reduce the likelihood of developing AD (4–6). Global and focal cerebral hypoperfusion measured by transcranial Doppler, single-photon emission computed tomography, and arterial spin MRI is not only evident in AD, but also exists in mild cognitive impairment (MCI) (7,8). Longitudinal studies have demonstrated that decreased cerebral perfusion in MCI can predict progression to AD (9). In comparison with cognitively normal (CN) controls of similar ages, microscopic pathologic studies showed significant structural changes of the cerebral vasculature in AD, such as loose capillary density, capillary kinking, looping, and twisting (10,11). Cerebral amyloid angiopathy is one of the major pathologic changes in AD, besides amyloid plaques and neurofibrillary tangles (12). Despite the evidence suggesting that vascular impairment plays a role in the pathogenesis of AD, it is unknown whether vascular alterations precede and contribute to neural death or whether they are the bystander effect from decreased metabolic demand. The lack of data is due to the difficulty in visualizing and assessing the cerebral microvasculature in vivo.
Similar to the brain, the retina has a highly isolated and protected vascular system and shares similar physiological and anatomical features with the brain (13,14). The transparent ocular media enables a noninvasive evaluation of the retinal vasculature. Epidemiologic studies using fundus photography have demonstrated less complex retinal vasculature, and smaller, more tortuous retinal vessels in patients with AD compared with CN controls (15–17). In addition, the loss of ganglion cells seems to correlate with cerebral atrophy (18).
With advancements in imaging techniques, microvascular networks that are invisible on fundus photographs can be noninvasively imaged (19). Optical coherence tomography (OCT) angiography (OCTA) can image both the superficial and deep retinal microvascular plexuses, which reflect retinal perfusion (19). Compared with previous investigations on the retinal vasculature using fundus images (15,20), OCTA enables visualization of the retinal microvascular network, including the capillary network, at the micrometer level. Studying the retinal microvasculature (precapillary arterioles, capillaries, and postcapillary venules) and retinal neurodegeneration may provide an insight into the role of vascular dysfunction in the pathogenesis of AD. The goal of our study was to characterize retinal microvascular network and its relation to retinal neuronal structure in patients with MCI and AD.
The study was approved by the Institutional Review Board for human research at the University of Miami, and signed informed consent was obtained from each subject. Patients with AD and MCI were recruited from the McKnight Brain Registry and referred from the Division of Cognitive Disorders at the University of Miami to the neuro-ophthalmology clinic at the Bascom Palmer Eye Institute. A group consensus conference that included neurologists, psychiatrists, and neuropsychologists discussed and confirmed the diagnoses of AD (21) and MCI (22) based on National Institute on Aging–Alzheimer's Association (NIA-AA) criteria. The disease duration was calculated from the date of symptoms onset. All subjects were treated in accordance with the Tenets of the Declaration of Helsinki.
We excluded patients with high refractive errors of more than +6.0 or −6.0 diopters (because of the limit of the imaging device) or with any ocular disease, such as age-related macular degeneration, diabetic retinopathy, cystic macular edema, dense cataracts, or corneal disease. To ensure image quality, the cutoff of the signal strength of OCT was set to be 5, which is likely the minimal signal strength for OCT measurements of macular, optic nerve head, and retinal nerve fiber layer parameters (23). Patients with a history of stroke, coagulopathy, uncontrolled hypertension, and uncontrolled diabetes also were excluded. The CN controls were CN individuals who fit the same inclusion and exclusion criteria. Their Mini-Mental State Examination (MMSE) was performed by a trained and qualified neurologist (H.J.) or research associates (Y.S. and Y.W.).
Retinal microvasculature was acquired using Zeiss Angioplex OCTA (Carl Zeiss Meditec, Dublin, CA) (19), covering a retinal area of 3 × 3 mm2 centered on the fovea. There are 2 layers of the interconnecting retinal capillary network: the superficial vascular plexus (SVP), located in the retinal nerve fiber and ganglion cell layers, containing arterioles, venules, and capillaries, and the deep vascular plexus (DVP) located in the inner nuclear and outer plexiform layers that contain mainly capillary-sized vessels (Fig. 1) (24). The total retinal vascular network (RVN) is defined as the vasculature in the retina, including both SVP and DVP (19). The enface images of RVN, SVP, and DVP were exported. These images were analyzed using custom software, including separation of large and small vessels, partition of quadrantal sectors and annular zones, and fractal analysis of the processed images (25,26). Briefly, images with a size of 245 × 245 pixels were resized to 1,024 × 1,024 pixels for vessel segmentation (27,28). The segmentation removed the large vessels from the microvascular network and extracted the microvascular network using image processing procedures, such as inverting, equalizing, and removing background noise and nonvessel structures to create a binary image. Vessels with a diameter of ≥25 μm were defined as large vessels and separated from the remaining vessels, which were considered as microvessels (25,26). This procedure also was used to eliminate the shadow graphic projection artifact of the large vessels in the SVP, which was projected onto the DVP (29).
The binary images of the extracted microvessels were skeletonized. The center of the foveal avascular zone (FAZ) was located, and the center of the FAZ was used for all the subsequent partitions. As in our previous studies (25,26), a disc with a diameter of 0.6 mm (which roughly represents the avascular zone) was not used in the analysis. The area between circles with diameters of 0.6–2.5 mm was defined as the annular zone (Fig. 2). The annular zone was then divided into 4 quadrantal sectors, named the superior temporal (ST), inferior temporal (IT), superior nasal (SN), and inferior nasal (IN). The annular zone was also divided into 6 thin annuli with a width of ∼0.16 mm because of the uneven distribution of retinal ganglion cells and corresponding vascular supply (30). Because of the symmetric structure of the retina between the left and right eyes, the results of the nasal or temporal sectors of both the left and right eyes were averaged.
Standard OCTs were used to quantify the thickness of macular ganglion cell–inner plexiform layer (GC-IPL) imaged using Zeiss 200 × 200 macular cube scan protocol (Fig. 3) (31). Measurements of GC-IPL thickness were obtained from the whole annulus with removal of the center elliptic zone and 6 sectors of the annulus, including the ST, superior (S), SN, IN, inferior (I), and IT sectors of the annulus.
One eye of each patient or CN control was imaged. The right eye was the first choice for imaging. The left eye was selected if the right eye did not meet the eligibility criteria. Patients and CN controls were asked to avoid large meals and to not drink alcohol or coffee before ophthalmic imaging. They were also advised to avoid physical exercise for 24 hours before the study.
Fractal analysis was performed in each sector or annular zone using the box counting method with the fractal analysis toolbox (TruSoft Benoit Pro 2.0, TruSoft International, Inc, St. Petersburg, FL) (25,26). The fractal dimension (Dbox) was obtained, representing the vessel density in each zone.
Analyzed using SAS (ver. 9.4; SAS Institute, Cary, NC), the analysis of covariance (ANCOVA) was used to test for a trend from AD to MCI to control. The Spearman rank-order correlation was used to evaluate the relationship among the parameters, and the Spearman correlation coefficient (ρ) is reported. Chi-square test was used to test the confounding factors. A result of P < 0.05 was considered significantly different.
The baseline characteristics of the patients are listed in Supplemental Digital Content 1 (see Table E1, http://links.lww.com/WNO/A283). There were no significant differences in demographics and vascular risk factors among the groups (P > 0.05). The AD patient cohort, as expected, had a significantly worse MMSE compared with the MCI group (P < 0.01). Large vessels and microvessels were clearly visualized in the OCTA images in both SVP and RVN (Fig. 4). The majority of vessels in the DVP were small vessels. The large vessels in the DVP were projection artifacts of the large vessels in SVP (29). Compared with the CN controls, the large vessels in patients with AD and MCI appeared to have similar densities. Compared with controls, patients with AD had lower densities of RVN, SVP, and DVP in the annulus from 0.6 to 2.5 mm in diameter (See Supplemental Digital Content 1, Figure E1, http://links.lww.com/WNO/A266; P < 0.05). The quadrantal analysis showed that AD had lower vessel densities in all quadrants of RVN, ST quadrant of SVP, and ST and IN quadrants of DVP (See Supplemental Digital Content 1, Figure E1, http://links.lww.com/WNO/A266; P < 0.05). Analysis of thin annuli showed that AD had a lower density of RVN in annuli C2–C6 (from 0.92 to 2.5 mm) (See Supplemental Digital Content 2, Figure E2, http://links.lww.com/WNO/A267; P < 0.05). In addition, the annuli C3–4 (from 1.55 to 1.87 mm) showed a lower density of SVP and DVP in AD (See Supplemental Digital Content 2, Figure E2, http://links.lww.com/WNO/A267; P < 0.05), compared with controls.
In patients with MCI, the density of retinal microvessels of DVP in the SN quadrant was significantly lower (See Supplemental Digital Content 1, Figure E1, http://links.lww.com/WNO/A266; P < 0.05), compared with controls. The thin annulus analysis showed that the density of DVP in annuli C1–6 (from 0.60 to 2.5 mm) was significantly lower (See Supplemental Digital Content 2, Figure E2, http://links.lww.com/WNO/A267; P < 0.05), compared with controls.
GC-IPL thickness was measured as the averaged values of the whole annulus and 6 sectors. There were no significant differences among groups in the average values and sectors (See Supplemental Digital Content 3, Figure E3, http://links.lww.com/WNO/A268; P > 0.05). The ANCOVA results indicated that there was a statistically significant trend with higher density in the control patients compared with the patients with MCI, and a higher density in the patients with MCI compared with the patients in AD for all RVN variables except the C1 (innermost) annular zone. The same statistically significant trend was observed in the SVP whole annulus, ST quadrant, and the C3 and C4 annular zones, and in the DVP whole annulus, ST and IN quadrants, and the C3 and C4 annular zones. No significant trends were observed in the GC-IPL thickness of the whole annulus or the 6 sectors.
In patients with AD, the microvascular density in the DVP was related to the GC-IPL annular thickness (See Supplemental Digital Content 4, Figure E4, http://links.lww.com/WNO/A269; ρ = 058, P < 0.05). None of microvascular measurements was related to disease duration and MMSE (P > 0.05). In patients with MCI, none of the microvascular measurements was related to the GCIPL annular thickness (See Supplemental Digital Content 5, Figure E5, http://links.lww.com/WNO/A270; P > 0.05) and disease duration (P > 0.05). However, the microvascular network density in the retinal microvascular network was positively related to MMSE in patients with MCI (ρ = 0.49, P < 0.05). In the CN controls, none of the microvascular measurements was related to the GCIPL thickness (P > 0.05).
The loss of the retinal microvascular density in patients with AD found in our study indicates impairment of retinal microvasculature. The alterations that we detected support the hypothesis that vascular impairment resulting in tissue hypoperfusion may contribute to disease onset and progression and may echo the findings in the brain (7–9). The trend of the retinal microvascular loss from MCI to AD may indicate retinal vascular impairment during disease progression, which may contribute to the potential conversion from MCI to AD. Interestingly, the relation between the impaired macular microvasculature and macular ganglion cell layer thickness was not established, which may be due to the different regions of interest that we measured. The vessel density was measured in a 3 × 3 mm2, whereas the GC-IPL thickness was measured in a 6 × 6 mm2 (analyzed in an ellipsoidal area). In addition, the macular RNFL (axonal fibers) was not analyzed. Further studies investigating the relationship between the macular microvasculature and macular RNFL are needed for better understanding the effect of the impaired macular microvasculature on retinal neurodegeneration. As the window to the brain, monitoring retinal microvasculature may add insightful information in the pathophysiology of AD, particularly the role of vascular contribution to neurodegeneration in disease progression and treatment efficacy. Monitoring retinal microvasculature in addition to retinal structural measurements may prove valuable in better understanding the mechanism of cerebral neural loss in AD.
Although there were no correlations between retinal microvasculature and GC-IPL thickness in MCI group and CN control group, the correlation between the loss of retinal microvasculature in DVP and the GC-IPL thinning was established in the AD group. Although the relation (ρ = 0.55) between SVP and GCIPL did not reach a significant level, a similar relation (ρ = 0.58) between DVP and GC-IPL was observed. The SVP is located in the retinal nerve fiber and ganglion cell layers, whereas the DVP is located in the inner nuclear and outer plexiform layers. One possible explanation could be that the DVP primarily is composed of capillaries that may be affected earlier and to a greater extent than the SVP that is composed of relatively larger vessels (precapillary arterioles, capillaries, and postcapillary venules). Future studies with a large sample size may establish the relation between SVP and GC-IPL.
MMSE has been reported to not be sensitive enough to differentiate MCI from AD because many factors influence cognitive tests (32,33). This also may explain why we did not find correlations between the MMSE score and retinal microvascular alterations in patients with AD. Interestingly, a positive relation between the RVN and MMSE was found in MCI, and further large sample studies are needed to validate this relationship.
The OCTA angiogram shows the details of the retinal vasculature, including large vessels that are visible on fundus photographs and microvessels that are not visible on fundus photographs (19). Removal of the large vessels helps to analyze the microvascular network (25–27). Our automated quantitative software for evaluating the capillary network imaged by OCTA was evolved from our previous analysis of the microvascular network obtained from the Retinal Function Imager and has the features with detailed partitions and large vessel removal in addition to fractal analyses (27,28). The repeatability tests of the data acquired using OCTA were confirmed to be valid (25,26). Careful analysis of the retinal microvasculature may be the key for revealing early signs of microvascular impairment in MCI and AD. The fractal dimension of the annulus (0.6–2.5 mm) of the microvascular network is ∼1.75–1.78 in both the SVP and DVP of healthy young adults (25,26). In the present study, the fractal dimension in the elderly CN controls was ∼1.74 in SVP and ∼1.73 in DVP, which is slightly lower than previously reported, likely due to normal aging. The disease process may further induce the microvascular network loss on top of the normal aging by approximately 0.05 (fractal dimension) in patients with AD and 0.03 in patients with MCI. In addition to the overall analysis of the retinal microvasculature, detailed partitioning may add another dimension in understanding early disease changes. When only the total annulus was analyzed, sectorial changes may be missed. This may be due to the nonuniform distribution of retinal ganglion cells and corresponding vascular supply (30). The density of microvasculature was highest from 0.5 to 1.25 mm from the fovea as per our previous study (25).
There are limitations to the present study. First, our small sample size, especially of the AD group, may have been a factor in not being able to show significant changes in all quadrants and annuli. However, significant changes in certain annuli and sectors were determined in AD, indicating that analyzing the retinal microvascular network is sensitive for detecting the changes. Second, age matching is another limitation of our study. The mean age was 73.3 years in the AD group, 69.6 years in the MCI group, and 67.6 years in the CN control group with a trend of younger age in the MCI and CN groups. Although statistically there were no significant differences of ages among groups, the slight difference in the mean age (∼6 years) may contribute to the difference of macular vessel density of the AD group in comparison with the CN group. Third, we studied the correlations of retinal microvascular alterations with GC-IPL, disease duration, and MMSE but not with other clinical or neuroimaging features of AD and MCI. Fourth, the cross-sectional nature of our study cannot provide information about temporal causality of vascular alterations and neurodegeneration. Future longitudinal studies are needed. Last, the analyses of adjacent tissue sections cannot be considered statistically independent because the microvasculature of adjacent tissue sections is more likely to be similar than the microvasculature of nonadjacent tissue sections. Yet, this does not compromise the statistical validity of our analyses because, within all of our analyses, the individual data (eyes) were independent, and this is what is required by the assumptions of statistical tests.
In conclusion, patients with AD had less density of the retinal microvascular networks than controls. The trend of retinal microvascular loss may indicate progressive retinal vascular impairment during disease progression.
STATEMENT OF AUTHORSHIP
Category 1: a. Conception and design: H. Jiang, C. B. Wright, T. Rundek, and J. Wang; b. Acquisition of data: Y. Wei, Y. Shi, C. B. Wright, X. Sun, G. Gregori, F. Zheng, H. Jiang, J. Wang, and B. L. Lam; c. Analysis and interpretation of data: H. Jiang, Y. Wei, Y. Shi, C. B. Wright, X. Sun, G. Gregori, F. Zheng, and E. A. Vanner. Category 2: a. Drafting the manuscript: H. Jiang and J. Wang; b. Revising it for intellectual content: C. B. Wright, G. Gregori, E. A. Vanner, B. L. Lam, T. Rundek, H. Jiang, and J. Wang. Category 3: a. Final approval of the completed manuscript: Y. Wei, Y. Shi, C. B. Wright, X. Sun, G. Gregori, F. Zheng, H. Jiang, J. Wang, and B. L. Lam.
1. de la Torre JC. Vascular risk factors: a ticking time bomb to Alzheimer's disease. Am J Alzheimers Dis Other Demen. 2013;28:551–559.
2. Ott A, Stolk RP, van HF, Pols HA, Hofman A, Breteler MM. Diabetes mellitus and the risk of dementia: the Rotterdam Study. Neurology. 1999;53:1937–1942.
3. Hofman A, Ott A, Breteler MM, Bots ML, Slooter AJ, Van HF, van Duijn CN, Van BC, Grobbee DE. Atherosclerosis, apolipoprotein E, and prevalence of dementia and Alzheimer's disease in the Rotterdam Study. Lancet. 1997;349:151–154.
4. de la Torre JC. Vascular risk factor detection and control may prevent Alzheimer's disease. Ageing Res Rev. 2010;9:218–225.
5. de la Torre JC. Cerebral hemodynamics and vascular risk factors: setting the stage for Alzheimer's disease. J Alzheimers Dis. 2012;32:553–567.
6. de la Torre JC. A turning point for Alzheimer's disease? Biofactors. 2012;38:78–83.
7. Ruitenberg A, den HT, Bakker SL, van Swieten JC, Koudstaal PJ, Hofman A, Breteler MM. Cerebral hypoperfusion and clinical onset of dementia: the Rotterdam Study. Ann Neurol. 2005;57:789–794.
8. Austin BP, Nair VA, Meier TB, Xu G, Rowley HA, Carlsson CM, Johnson SC, Prabhakaran V. Effects of hypoperfusion in Alzheimer's disease. J Alzheimers Dis. 2011;26(suppl 3):123–133.
9. Park KW, Yoon HJ, Kang DY, Kim BC, Kim S, Kim JW. Regional cerebral blood flow differences in patients with mild cognitive impairment between those who did and did not develop Alzheimer's disease. Psychiatry Res. 2012;203:201–206.
10. Buee L, Hof PR, Delacourte A. Brain microvascular changes in Alzheimer's disease and other dementias. Ann N Y Acad Sci. 1997;826:7–24.
11. de la Torre JC. Hemodynamic consequences of deformed microvessels in the brain in Alzheimer's disease. Ann N Y Acad Sci. 1997;826:75–91.
12. Knight MJ, McCann B, Kauppinen RA, Coulthard EJ. Magnetic resonance imaging to detect early molecular and cellular changes in Alzheimer's disease. Front Aging Neurosci. 2016;8:139.
13. Patton N, Aslam T, MacGillivray T, Pattie A, Deary IJ, Dhillon B. Retinal vascular image analysis as a potential screening tool for cerebrovascular disease: a rationale based on homology between cerebral and retinal microvasculatures. J Anat. 2005;206:319–348.
14. London A, Benhar I, Schwartz M. The retina as a window to the brain-from eye research to CNS disorders. Nat Rev Neurol. 2013;9:44–53.
15. Cheung CY, Ong YT, Ikram MK, Ong SY, Li X, Hilal S, Catindig JA, Venketasubramanian N, Yap P, Seow D, Chen CP, Wong TY. Microvascular network alterations in the retina of patients with Alzheimer's disease. Alzheimers Dement. 2014;10:135–142.
16. Williams MA, McGowan AJ, Cardwell CR, Cheung CY, Craig D, Passmore P, Silvestri G, Maxwell AP, McKay GJ. Retinal microvascular network attenuation in Alzheimer's disease. Alzheimers Dement (Amst). 2015;1:229–235.
17. Frost S, Kanagasingam Y, Sohrabi H, Vignarajan J, Bourgeat P, Salvado O, Villemagne V, Rowe CC, Macaulay SL, Szoeke C, Ellis KA, Ames D, Masters CL, Rainey-Smith S, Martins RN; AIBL Research Group. Retinal vascular biomarkers for early detection and monitoring of Alzheimer's disease. Transl Psychiatry. 2013;3:e233.
18. Cheung CY, Ong YT, Hilal S, Ikram MK, Low S, Ong YL, Venketasubramanian N, Yap P, Seow D, Chen CL, Wong TY. Retinal ganglion cell analysis using high-definition optical coherence tomography in patients with mild cognitive impairment and Alzheimer's disease. J Alzheimers Dis. 2015;45:45–56.
19. Rosenfeld PJ, Durbin MK, Roisman L, Zheng F, Miller A, Robbins G, Schaal KB, Gregori G. ZEISS angioplex spectral domain optical coherence tomography angiography: technical aspects. Dev Ophthalmol. 2016;56:18–29.
20. Cheung CY, Ong YT, Ikram MK, Chen C, Wong TY. Retinal microvasculature in Alzheimer's disease. J Alzheimers Dis. 2014;42(suppl 4):S339–S352.
21. McKhann GM, Knopman DS, Chertkow H, Hyman BT, Jack CR Jr, Kawas CH, Klunk WE, Koroshetz WJ, Manly JJ, Mayeux R, Mohs RC, Morris JC, Rossor MN, Scheltens P, Carrillo MC, Thies B, Weintraub S, Phelps CH. The diagnosis of dementia due to Alzheimer's disease: recommendations from the National Institute on Aging-Alzheimer's Association workgroups on diagnostic guidelines for Alzheimer's disease. Alzheimers Dement. 2011;7:263–269.
22. Albert MS, DeKosky ST, Dickson D, Dubois B, Feldman HH, Fox NC, Gamst A, Holtzman DM, Jagust WJ, Petersen RC, Snyder PJ, Carrillo MC, Thies B, Phelps CH. The diagnosis of mild cognitive impairment due to Alzheimer's disease: recommendations from the National Institute on Aging-Alzheimer's Association workgroups on diagnostic guidelines for Alzheimer's disease. Alzheimers Dement. 2011;7:270–279.
23. Samarawickrama C, Pai A, Huynh SC, Burlutsky G, Wong TY, Mitchell P. Influence of OCT signal strength on macular, optic nerve head, and retinal nerve fiber layer parameters. Invest Ophthalmol Vis Sci. 2010;51:4471–4475.
24. Kur J, Newman EA, Chan-Ling T. Cellular and physiological mechanisms underlying blood flow regulation in the retina and choroid in health and disease. Prog Retin Eye Res. 2012;31:377–406.
25. Li M, Yang Y, Jiang H, Gregori G, Roisman L, Zheng F, Ke B, Qu D, Wang J. Retinal microvascular network and microcirculation assessments in high myopia. Am J Ophthalmol. 2017;174:56–67.
26. Yang Y, Wang J, Jiang H, Yang X, Feng L, Hu L, Wang L, Lu F, Shen M. Retinal microvasculature alteration in high myopia. Invest Ophthalmol Vis Sci. 2016;57:6020–6030.
27. Jiang H, Delgado S, Liu C, Rammohan KW, DeBuc DC, Lam BL, Wang J. In vivo characterization of retinal microvascular network in multiple sclerosis. Ophthalmology. 2016;123:437–438.
28. Jiang H, DeBuc DC, Rundek T, Lam BL, Wright CB, Shen M, Tao A, Wang J. Automated segmentation and fractal analysis of high-resolution non-invasive capillary perfusion maps of the human retina. Microvasc Res. 2013;89:172–175.
29. Zhang M, Hwang TS, Campbell JP, Bailey ST, Wilson DJ, Huang D, Jia Y. Projection-resolved optical coherence tomographic angiography. Biomed Opt Express. 2016;7:816–828.
30. Curcio CA, Allen KA. Topography of ganglion cells in human retina. J Comp Neurol. 1990;300:5–25.
31. Mwanza JC, Durbin MK, Budenz DL, Girkin CA, Leung CK, Liebmann JM, Peace JH, Werner JS, Wollstein G. Profile and predictors of normal ganglion cell-inner plexiform layer thickness measured with frequency-domain optical coherence tomography. Invest Ophthalmol Vis Sci. 2011;52:7872–7879.
32. Tsoi KK, Chan JY, Hirai HW, Wong SY, Kwok TC. Cognitive tests to detect dementia: a systematic review and meta-analysis. JAMA Intern Med. 2015;175:1450–1458.
33. Arevalo-Rodriguez I, Smailagic N, Roque IF, Ciapponi A, Sanchez-Perez E, Giannakou A, Pedraza OL, Bonfill CX, Cullum S. Mini-Mental State Examination (MMSE) for the detection of Alzheimer's disease and other dementias in people with mild cognitive impairment (MCI). Cochrane Database Syst Rev. 2015;CD010783.