The aim of our study was to compare DSC-MR perfusion and FDG-PET studies based on the assessment of: (1) hypoperfusion and hypometabolism patterns in the selected brain areas in AD and MCI, (2) correlation between the results of these two techniques and their accuracy in diagnosis of AD and MCI, and (3) correlation between the results of DSC-MRI and FDG-PET with the severity of cognitive impairment in AD and MCI.
In our study AD patients, compared to the control group, showed significant hypoperfusion in all examined cortical locations. Our results are consistent with the typical pattern of Alzheimerʼs degeneration and hypoperfusion reported in numerous publications within the PCG, temporo-parietal cortices, and in later stages also frontal cortices with relative sparing of the sensorimotor cortex [12, 13, 15, 31, 32]. In the MCI group, compared to controls, we found significantly decreased rCBV values within the cortex of both parietal lobes, temporal lobes and left PCG, while by using the rCBV z-scores, significant hypoperfusion was detected in the right parietal cortex, which is also consistent with the pattern of very early alterations in the course of AD pathology [13, 17, 18, 32, 33]. In the MCI group perfusion alterations were less severe than in the AD group, which supports the theory that hypoperfusion is a marker of neuronal damage and becomes more prominent in the later stages of AD.
In our study the FDG-PET results in the AD group showed significant glucose hypometabolism in all investigated locations of the cerebral cortex reported before [24] and the most pronounced in the parietal, temporal and left PCG regions, followed by hypometabolism in the frontal cortices. These results are in accordance with the commonly accepted metabolic pattern in the course of AD, thanks to which, as demonstrated by Mosconi et al. it is possible to differentiate AD from DLB (Dementia with Lewy bodies) and FTLD (frontotemporal lobar degeneration) even in the advanced forms [20, 22]. MCI subjects showed less severe hypometabolism mainly in the parieto-temporal regions and left PCG, which is in line with the existing literature [23, 24]. Mosconi et al. suggests that FDG-PET is a good diagnostic method in detecting the early stages of dementia already at the MCI level. In her work, the typical AD pattern of glucose hypometabolism was observed in 79% of MCI subjects with deficits in multiple cognitive domains and in 31% of patients with amnestic MCI [20].
In our study, in AD patients we found significant correlations between the results of DSC-MRI and FDG-PET in almost all the evaluated locations, apart from the right parietal cortex, while in the MCI group there was only a single correlation within the left PCG. The single correlation in the case of MCI was probably due to the small sample of subjects. After combing AD and MCI subjects in one group, significant correlations between DSC-MR perfusion and FDG-PET studies were revealed in all evaluated locations. The strongest correlations were revealed within temporal (r = 0.55–0.6) and PCG (r = 0.53–0.63) regions followed by parietal (r = 0.45–0.46) cortices, which are the regions of the most pronounced and typical changes in the course of AD degeneration. To our knowledge in the literature there are only two reports comparing DSC-MRI with FDG-PET in AD and MCI. In the first paper by Gonzales et al. the authors performed their study only on 10 patients with dementia (6 with AD) using visual evaluation of rCBV and brain glucose metabolism maps [15]. They compared the results within 8 brain layers and demonstrated a significant correlation (r = 0.62) at the levels of the upper and supraventricular layers. The mean correlation from all layers was r = 0.53, with the temporal area and the posterior fossa showing the weakest correlations (r = 0.24–0.33), which was explained by artifacts related to vessel pulsation. Our results do not fully agree with these findings, but it has to be stressed that our analyzes were conducted on a larger number of subjects and were based on parametrical values of rCBV and glucose metabolism, and thus seem to be more accurate than a visual assessment. In the second report, Zimny et al. showed a statistically significant correlation (r = 0.44) of rCBV measurements and FDG-PET results in PCG [18]. The results of this report are partially consistent with our findings in the left PCG (r = 0.4). However, the authors did not compare other regions of the brain and did not separate PCG into right and left regions.
To compare DSC-MR perfusion and FDG-PET results, we evaluated sensitivity and specificity and the accuracy of these two studies in distinguishing AD and MCI from healthy controls. We found a very similar high accuracy of DSC-MR perfusion and FDG-PET in distinguishing AD from the control group (0.98 and 0.97, respectively), and markedly higher accuracy of FDG-PET than DSC-MR perfusion in the differentiation of MCI from the control group (0.96 and 0.68–0.77, respectively). When distinguishing AD from MCI, both methods showed intermediate accuracy around 0.84 for MR and 0.81 for PET studies. It has to be stressed that though there are many reports in the literature showing the results of sensitivity, specificity and accuracy of DSC-MRI or FDG-PET in the diagnosis of AD or MCI, none of them were performed on the same groups of patients [12, 13, 16, 17, 20, 34,35,36,37].
There are several reports evaluating MR perfusion in the differentiation of AD from CG based on temporo-parietal areas and the results are slightly worse than in our study. For example, Harris et al. defined sensitivity as 0.95 in moderately affected patients with AD and 0.88 in mild cases of AD, whereas specificity as 0.96 [13]. In turn, Bozzao et al. in distinguishing AD from CG, achieved sensitivity of 0.91 and specificity of 0.87, while Maas et al. achieved 0.8 and 0.88 for sensitivity and specificity, respectively [12, 16]. On the other hand, Zimny et al. in the regions of PCG alone showed the accuracy of AD diagnosis as 0.87 [17], so lower than in our study (accuracy 1.0). Our results of FDG-PET in differentiating AD from CG (sensitivity 0.97, specificity 1.0, accuracy 0.97) are similar to other publications by Gambir et al. (sensitivity of 0.9–0.96, specificity of 0.67–0.97 and accuracy of 0.89), Mosconi et al. (sensitivity of 0.99, specificity of 0.98, accuracy of 0.98) or Gupta et al. (sensitivity of 0.9, specificity of 0.9 and accuracy of 0.92) and much higher compared to other studies reporting their sensitivity, specificity and accuracy results below 0.9 [20, 34,35,36,37]. It should be emphasized that in MCI subjects cognitive functions are impaired to an intermediate degree between proper aging and dementia, and there are so-called overlap periods, so distinguishing AD from MCI is a more difficult task than AD from CG [2]. To our knowledge, there are no reports in the literature in which authors could provide the accuracy values of MR perfusion in differentiating AD from MCI. In the differentiation of AD from MCI using the FDG-PET method, our results are similar to the literature. Gupta et al. when distinguishing AD from converting MCI assessed sensitivity, specificity and accuracy as 0.67, 0.88, and 0.81, respectively, (in our study 0.82, 0.77, and 0.81, respectively) [35]. According to De Santi et al. it is best to differentiate AD from MCI based on results of glucose metabolism in the temporal lobes, which is consistent with the results of our study, where accuracy from this cortical location was greater (0.9–0.92) than in other regions [38]. Regarding the differentiation of MCI from CG using DSC-MRI, our study showed better results compared to several previous reports, for example by Zimny et al. who, based on evaluation of PCG, determined sensitivity, specificity and accuracy as 0.72, 0.8 and 0.7, respectively, and in the next study accuracy as 0.67 [17, 18]. In the differentiation of MCI from CG using the FDG-PET method, our results are similar to the literature [20, 22, 35]. For example, Gupta et al. in analyzing MCI converting to AD from CG, showed the sensitivity, specificity and accuracy as 0.98, 1.0, and 0.8, respectively, (in our study 0.95, 1.0, and 0.95, respectively).
In the last part of our study we evaluated correlations between the results of DSC-MRI or FDG-PET studies and the results of the MMSE test. In AD, statistically significant correlations were found with the results of DSC-MR perfusion from the left parietal and left temporal lobes, while in MCI in PCG and both parietal cortices. In the FDG-PET study, statistically significant correlations with MMSE were found only in AD patients with the results from the left frontal cortex. However, it should be emphasized that after putting together all AD and MCI subjects in one bigger group, a statistically significant correlation between MR perfusion or FDG-PET results and MMSE test was found in all the examined locations. Summarizing, it should be stated that in a larger group the results of these correlations are very similar for DSC-MR perfusion and FDG-PET. In the literature the results of correlation of MMSE test with DSC-MRI are ambiguous. Some authors showed no correlation of rCBV parameter with the MMSE test in AD or MCI patients [13, 31, 32] and several other authors found such correlations [16, 17]. The lack of correlation of psychological tests in separate groups of AD and MCI with FDG-PET results is in contradiction with several literature reports [37, 39,40,41].
Recently, more reports have focused on the comparison of FDG-PET with a non-contrast MR perfusion technique such as ASL. Fällmar et al. demonstrated a positive predictive value of ASL MR in AD and FTLD patients using visually analyzed perfusion maps and high specificity (0.84) of diagnoses, despite lower sensitivity (0.53) compared to FDG-PET (0.96) [42]. Similarly, Musiek et al. demonstrated, using visual inspection of perfusion and glucose metabolism maps, that both methods showed alterations in parieto-temporal areas, while the FDG-PET examination also depicted hypometabolism within the frontal lobes [43]. Johnson et al. comparing ASL MR and FDG-PET techniques in the AD group, showed that in both techniques the lower parts of the parietal lobes, PCG, superior and middle frontal gyrus were involved [44]. On the other hand, in the MCI group Johnson et al. showed a reduction in perfusion in the lower part of the right parietal lobe, which was slightly consistent with the pattern of glucose hypometabolism [44]. Riederer et al. also using the ASL MR method in MCI, showed no statistically significant differences in ASL perfusion rCBF parameter between aMCI and CG, contrary to FDG-PET studies, which showed hypometabolism on both sides of inferior parietal, superior temporal, right prefrontal dorsolateral cortex, precuneus, PCG and MTL [45]. All the above studies were performed using only visual inspection of ASL MRI and FDG-PET maps. Despite a growing interest in ASL perfusion due to the lack of contrast material needed during the examination, this MR method has several drawbacks. One of them is a prolonged acquisition time, which makes ASL impossible to be used in non-cooperative patients (e.g., those with advanced dementia). Other disadvantages are the necessity of three Tesla MR scanners to obtain reliable data, which are not widely available and a low signal-to-noise ratio (SNR).
The important advantage of the FDG-PET study is that its results may be partially evaluated based only on visual inspection, which is possible thanks to existing software such as CORTEX ID, which calculates the glucose metabolism normalized to the cerebellum and the z-score in relation to the database of healthy people, and automatically generates color-coded 3D maps of the cerebral cortex. There is no such software to post-process MR perfusion studies, which makes it impossible to visually assess the degree of hypoperfusion based on raw CBV maps. Assessment of DSC-MR perfusion results requires manual ROI placement and calculations of CBV values. Absolute CBV values cannot be evaluated, since they are dependent on several factors, such as blood hemodynamics or capillary permeability. This is why the relative value of CBV (compared to the cerebellar CBV) was used.
There are a few limitations of our study. Firstly, manual determination of ROIs is somewhat subjective and makes the method operator-dependent. Secondly, rather small groups of subjects may have had an impact on some results. We assessed more significant correlations after combining patients in a larger group of AD and MCI subjects. Another drawback is the cross-sectional character of the study. We have not evaluated longitudinal results regarding follow-up studies of aMCI subjects and the rate of their progression to dementia. It would be very interesting to check if DSC-MR perfusion has a similar strength as FDG-PET in predicting such a conversion.