Regional Cerebral Blood Flow in the Posterior Cingulate and Precuneus and the Entorhinal Cortical Atrophy Score Differentiate Mild Cognitive Impairment and Dementia Due to Alzheimer Disease

DISCUSSION

ASL is a noninvasive technique for cerebral perfusion measurement using electromagnetically labeled arterial water as a diffusible tracer. Several ASL perfusion studies have highlighted regional hypo- and hyperperfusion in multiple areas of the brain in MCI-AD.^{7⇓⇓⇓⇓⇓–13} However, due to the variability of clinical subsets studied and the usage of nonreplicable sequences, these studies were not directly comparable.¹³ In the past, differences in technical implementation schemes and lack of standardized protocols hindered the routine use of ASL as a robust perfusion method in the clinical setting. International consensus guidelines have been developed recently for standardized ASL acquisitions.²² A 3D fast spin-echo pseudocontinuous ASL with a spiral readout is now generally considered the standard method of ASL in clinical research. In this study, we used a 3D fast spin-echo pseudocontinuous ASL technique on 3T MR imaging to investigate regional perfusion differences in patients with MCI and AD compared with HC. ROIs were extracted using atlas-based semiautomated techniques to avoid errors due to manual contouring.

Although several brain areas have been shown to be involved, one of the most consistent regions of hypoperfusion reported in MCI-AD is the PCG and anterior PC.^10,11,13 Identification of hypoperfusion in the PCG + PC has been reported to have a sensitivity of 91% and a specificity of 80% for diagnosing AD.²³ This method is in contrast to inclusion of a combination of different regions in subjects with amnestic and dysexecutive MCI, in which an accuracy of only 60%–70% was observed.²⁴ From the existing literature, it is not clear whether the hypoperfusion is the cause or consequence of the disease process.²⁵ In our study, the regional mean CBF in PCG + PC was 46.64, 39.08, and 36.51 mL/100/min in HC, MCI, and AD, respectively, and it performed well in distinguishing AD-HC and MCI-HC, but not between MCI and AD. This finding may be because the perfusion deficit in the PCG + PC occurs early in the course of this dementing illness spectrum, and this particular feature may be used as a reliable marker to delineate patients with MCI and AD from HC. However, from MCI to AD transformation, the perfusion reduction in these regions may be less remarkable and hence may not have significant diagnostic utility as a stand-alone measure.

There is no consistent correlation established between CBF and gray matter changes in the corresponding regions in patients with AD and MCI.¹³ In the initial stages of AD, changes such as intraneuronal alterations in the form of neurofibrillary tangles and neuropil threads are observed more frequently in the entorhinal cortex compared with hippocampus.²⁶ This accumulation may be noted in other conditions, including normal aging, and when this is associated with neuronal loss, it leads to global cognitive decline. MR volumetric studies have also shown entorhinal cortex volume loss in dementia due to Alzheimer spectrum diseases.^27,28 It has been shown in studies that in early stages of AD, certain areas such as the prefrontal cortex or medial temporal lobe structures would eventually have the ability to compensate for the cognitive decline.²⁹ Simultaneously, perfusion deficits might also be present from the very early preclinical phases of AD, including MCI at risk of conversion, persisting into the advanced stages of the disease, demonstrating a progressive hypoperfusion with disease development, leading to brain atrophy and underlying neuronal loss that correlates with cognitive and functional decline.¹³ With regard to the pathophysiologic interpretation of the regional hypoperfusion consistently found in the PCG + PC, in comparison with ERICA, controversy exists about whether this is a cause or consequence of the disease process.²⁵ It is well known that risk factors such as ischemic stroke, amyloid angiopathy, atherosclerosis, hypertension, diabetes, and cardiac disease are repeatedly implicated in the risk of MCI converting to AD.³⁰ While AD and vascular dementia have traditionally been considered as independent entities, there is growing evidence implicating vascular pathology in AD to the extent of suggesting an overlap between AD and vascular dementia.³¹ Moreover, evidence from aging and stroke studies suggests that chronic brain hypoperfusion is the key pathophysiology that leads to neurodegeneration and consequent cognitive decline.²⁵ Neuropsychologic correlations have also been established between perfusion measures in the parieto-occipital region and the parietal cortex along with the PCG + PC.³² Hypoperfusion may lead to changes in cortical thickness, as measured on structural MR imaging scans, in areas most vulnerable to aging (medial prefrontal and pericentral cortices) as well as in areas associated with amyloid aggregation (eg, occipitotemporal and basal temporal cortices), more so in apolipoprotein E4 carriers.³³ Hence, there is the necessity to address disparate or nonconcordant topographic areas rather than looking at concordant brain volumes underlying areas of hypoperfusion while attempting to combine hypoperfusion and volumetric measures as with ERICA.

Compared with conventional MR imaging–medial temporal lobe atrophy scoring, which reported a sensitivity of 81% and a specificity of 67%, ERICA had a higher sensitivity of 83% and specificity of 98% in differentiating AD.^17,34 ERICA is also easier to comprehend and score compared with the more complex medial temporal lobe atrophy score. It is also reliable and reproducible as confirmed in our study. The tentorial cleft sign (Fig 1C), which indicated an ERICA score of ≥2, had a high diagnostic accuracy for AD.¹⁷ In our study, the average ERICA score of patients with AD was 2.15. ERICA performed well in differentiating AD from MCI and AD from HC, but not MCI from HC. This performance may mean that significant parahippocampal atrophy does not happen early in MCI and probably is a much later occurrence compared with CBF reduction in PCG + PC in the course of progression of dementia. This hypothesis is further validated by the observations of neuropathology studies by Braak and Braak published in 1991.²⁶

ASL hypoperfusion abnormality in PCG + PC and parahippocampal atrophy may be considered as complementary imaging biomarkers to distinguish MCI and AD. In our study combining the ERICA score with regional CBF in PCG + PC showed a promising trend to differentiate MCI, AD, and HC groups rather than individual parameters. However, MCI-AD is a more heterogeneous disease process, and not all patients with MCI are converters to AD.¹⁷ Further prospective studies are needed to validate the usefulness of combined ERICA and regional CBF in predicting this conversion. Although the pattern of results in MCI-HC comparison indicates better utility of CBF PCG + PC as opposed to the ERICA score, nosologically, there are important considerations while stratifying MCI at risk of AD using biomarkers. Per available criteria, to identify MCI at an intermediate or higher likelihood of AD requires combining biomarkers of -β amyloid accumulation (PET or CSF) with markers of neuronal injury (τ, FDG-PET, structural MR imaging).³⁵ A combined instrument using 2 MR imaging markers of neuronal injury rather than -β amyloid accumulation precludes prediction of the proportion of the MCI cohort who are likely to convert in our series.

One also needs to consider that MCI is a heterogeneous condition and potentially could be due to other causes like vascular cognitive impairment. There is evidence in the literature available from MCI negative for β amyloid, where common etiologies include vascular diseases and depression.³⁶ It is possible that a subset of our MCI cohort may, with time, evolve as vascular cognitive impairment or other alternative trajectories or may potentially remain stable without converting to AD. This possibility is perhaps part of the reason that the ERICA score did not show a significant difference between MCI and HC as opposed to the other 2 comparisons. Lack of longitudinal data to ascertain the trajectory of the MCI group with time precludes confirmation of the same in the present study and warrants future prospective follow-up studies. A relatively small sample size in our study also might have resulted in a lower ability to detect differences between MCI and AD using CBF alone at baseline. All these facts point to the relevance of including multiple MR imaging markers as implemented in one of our previous studies on multimodality neuroimaging.³⁷

The study was performed on a prospectively acquired dataset with measurement of only 2 parameters, which could be easily reproduced in a clinical setting. We preferred to use computed segmentation of the ROI CBF to avoid any bias due to manual drawing of contours. However, this study had several limitations. Though ASL is a robust technique to measure cerebral perfusion, technical parameters and patient-specific factors could variably influence the CBF calculation and confound the results. Specific to dementia, one of the major confounders is the arterial transit time, which can be different for each subject due to associated vascular or nonvascular comorbidities. The possible vascular risk contributors like diabetes mellitus and hypertension were not considerably different in each subgroup in the present study (Table 1). Again, transit delay can be higher in some of the ROIs in dementia imaging like the posterior parietal lobes, leading to erroneous CBF measurements.³⁸ The appropriately increased postlabeling delay time could be used to reduce this error. Alternatively, a multidelay ASL can potentially solve this problem, but it is not ready for prime time imaging yet. Also, a small sample size and cross-sectional study design did not permit analyzing a causal relationship between the biomarkers and the development of MCI and AD, and further large longitudinal studies are needed to confirm its role in predicting progression of the disease. Additionally, head-to-head comparisons between Pittsburgh compound PET imaging and CSF biomarkers (which are currently not available at our center) and MR imaging markers would have enabled us to ascertain the sensitivity and specificity; however, this comparison remains a concern across many centers in the developing world, and only multicentric collaborative ventures can transcend the issue.