DISCUSSION
To reiterate, in the present study using FDG PET scans, we found significantly impaired P300 evoked potentials in those subjects with HM compared to those subjects with NM. The P300 latency was significantly longer and the amplitude was significantly lower in the HM compared to the NM group. While other reports have suggested the importance of event related potentials of the brain as a biomarker for decline in cognitive processes including MCI and AD,22-25 this is the first report that has used FDG PET scans to demonstrate a significant association between brain metabolism and event related potentials. We are encouraged that with additional confirmation of these results, our initial hypothesis will be borne out.
There have been many patterns used to diagnose MCI. Ferris2 classified multiple stages, but the most appropriate classification system is the subdivision into amnestic single or multi-domain and non-amnestic single or multi-domain. This 4 category subdivision is accomplished through the following series of tests: attention; reaction time; judgment; learning ability; delayed recall; linguistic function; verbal IQ; performance IQ; abstract IQ; processing speed; immediate memory; and general cognitive functioning.26,27 Subjects who are positive in more than one of these areas are placed into regions, either single or multi-domain. The annual conversion rate of MCI to dementia is approximately 4.2% in the general population.28
In the present study, 76.7% of the HM patients were multi-domain. The most common brain region affected in these patients was the parietal lobe (n=18), affecting 41.8% of the HM group. According to Jacobs29 parietal lobe activity characterizes early MCI. With a new concept of the regional spread of Alzheimer’s or tauopathy, it is likely that some MCI patients progress with abnormalities in parietal regions, toward abnormalities in the temporal and frontal lobes, similar to a disease that progresses forward.28 Disease begins with the loss of electrophysiological processing speed followed by voltage, thought to action gaps (e.g., TOVA, P300), and several other forms of cognitive domains. Once progression occurs to hypometabolic loss in the parietal lobes, it is simultaneously marked by demyelination, atrophy, and micro stroke, as seen on MRI scans.30 This progression is unpredictable because once the process begins, it is extremely difficult to reverse even with the incorporation of new neurogenesis approaches (including serotonin-norepinephrine reuptake inhibitors, insulin-like growth factor 1, selective serotonin re-uptake inhibitors, fish oil, dehydroepiandrosterone, and a variety of other neuroendocrine and nutritional advances). The correlation and supplementation of this primary care data with MRI hippocampal volume is still ongoing.31
The value of using MMSE as a screening tool to identify cognitive decline has been praised32 and questioned.33 Our lab suggested that TOVA also showed promise as a validator of early cognitive decline and/or dementia. Our previous research correlated TOVA abnormalities with impaired WMS scores of early dementia. Coupling of TOVA assessment findings with results of P300, MMSE, and WMS-III may allow for enhanced accuracy in the diagnosis and evaluation of the complex pathways of failing attention, memory, and cognition that lead to dementia.16 In addition, Gualtieri and Johnson reported on the reliability and validity of the computerized neurocognitive test battery, CNS Vital Signs (CNSVS).34 Among the cognitive scores, the MMSE, TOVA, and CNSM scores were significantly lower in the HM compared to the NM group, thus supporting earlier reports.16,32 Comparing NM/HM to non-impaired/impaired CNSM was significant, with more HM patients presenting poor memory than those with NM.
Logistic regression modeling significantly validated membership in the HM group with memory complaints vs. the NM group without memory complaints (correctly classifying 81.3% of all patients) using the absolute value of a difference score calculated between each subject’s TOVA RT and P300 latency, and scores on the CNSM. Based on our earlier work,17 it was not surprising to find that logistic regression modeling did not retain MMSE as a valuable validator in this study. This supports other studies in the literature that suggest it has limited predictive validity.17,35-40
We are cognizant of certain limitations in the present study (e.g., MCI/dementia not specifically evaluated for the current population).41 It is tempting to speculate that coupling both the impaired electrophysiological and neuropsychological parameters may provide a less expensive test than FDG PET to assist the primary care physician in diagnosing memory deficits. Thus, large independent studies including additional work coupling impaired P300, validated neuropsychological memory tests, and activities for daily living with FDG PET is encouraged. Another limitation of the current study is that the interpretation of FDG PET scans as hypometabolic or normal were made by visual inspection of the reconstructed scans instead of quantitative analysis of the actual voxel values, either globally or refined by a region of interest analysis. The current study may be extended to include such a quantitative analysis. Since brain volume decreases after age 30 at 0.2% per year,42,43 another relative weakness of the study was excluding other imaging modalities such as PET beta amyloid tracers, VILIP-1, F-ML-10 apoptosis measures, MRI hippocampal volume, cortical thickness, demyelination, and micro ischemia, which can be combined to earlier detect pre-clinical states of MCI.44-46 We are cognizant that the depressed subjects showing NM (n=54) being treated with anti-depressants may have skewed the data because of medication-induced normalization of brain metabolism.
Importantly, the finding of a high specificity of the variables in validating HM is not surprising. However, while the sensitivity appears low by comparison to the specificity, we propose that the sensitivity remains beneficial: If indeed HM is validated in half of the patients, it is highly likely to be present and warrants further investigation for the presence of MCI or dementia. Forty-eight percent (48%) of the HM subjects were false negatives. It is important to utilize the PET scan more selectively, indicating that these 48% receive a PET scan. Fifty-two percent (52%) are ruled out because they do not require the PET scan for early diagnosis.
Results of the present study demonstrated for the first time that electrophysiological parameters (e.g., P300) coupled with neuropsychological measures (e.g., TOVA and CNSM) validate brain HM with clinically useful sensitivity and specificity. The P300 latency was significantly longer and the amplitude was significantly lower in the HM compared to the NM group, and these effects were retained after age correction. Neurocognitive measures (e.g., the MMSE [p=.01], TOVA [p=.00006], and CNSM [p=.0001] scores) were also significantly lower in the HM compared to the NM group. NM/HM compared to non-impaired/impaired CNSM were significant (p=.003), with more HM patients presenting poor memory than patients with NM.
It also appears that patients with single-domain cognitive decline have less progression to dementia, followed by those who are multi-domain amnestic or non-amnestic, causing the use of electrophysiology to deteriorate.29,47,48 As MCI atrophy sets in and advances to dementia, positive hypometabolic FDG PET scans are helpful indicators. Brain metabolism then creates a cognitive tipping point that is indicative of dementia or AD.31 Our subjects, although incompletely characterized, had MCI by a series of neuropsychological measures. Current characterizations suggest that MCI needs greater stratification considering the many patients who already have features of dementia. Most patients with hypometabolic FDG PET scans do have early dementia, even if their current symptoms resemble only MCI.15
MCI is heterogeneous, with electrophysiological decline, memory and attention failure, multiple domains of cognitive deterioration with and without losses of hippocampal volume, and cortical atrophy steering patients from MCI into dementia. There are numerous clinical variants of MCI that are antecedents to dementia, where progression is altered dependent on different cognitive phenotypes (e.g., individuals who emphasize working memory vs. auditory memory, abstract IQ vs. emotional IQ).49 A similar progression occurs in heart disease based on risk factors (e.g., cholesterol [HDL/LDL], electrophysiological dysfunction, valve and coronary artery disease, hormonal and vascular factors) which may occur in any
