Hippocampal and entorhinal cortex atrophy in frontotemporal dementia and Alzheimer’s disease

Methods.

Volumetrics.

MRI was performed at the Radiology Department, University of Verona, with a 1.5-T unit (Magneton; Siemens GmBh, Erlangen, Germany) and a standard head coil. A gradient-echo three-dimensional technique was used for image acquisition (repetition time, 10 msec; echo time, 4 msec; inversion time, 300 msec; flip angle, 10 deg; field of view, 250 mm; acquisitions, 2; matrix, 160 × 256). Total acquisition time was 7:40 minutes.

Volumetric analysis was made by a single rater (M.P.L.), who was blind to the subject’s diagnosis and clinical data. The regions of interest (ROIs) were traced manually from contiguous coronal 2.0-mm–thick T1-weighted images oriented perpendicular to a midsagittal line drawn through the anterior and posterior commissures. Tracing proceeded from anterior to posterior for both the hippocampus and the ERC. The volume of the hippocampus (considered as dentate gyrus, hippocampus proper, and the subicular complex) was measured starting from its appearance below the amygdala (figure). The uncal portion of the rostral hippocampus that is located ventral to the caudal amygdala was included in the hippocampus. The tracing ended posteriorly in the section where the crura of the fornices depart from the lateral wall of the lateral ventricles (see the figure).3 The entorhinal volumes were traced according to the criteria of Insausti et al.,33 with minor modifications. The first slice measured was the one after the appearance of limen insula when the temporal lobe can be first appreciated to be attached to the rest of the brain when proceeding from the anterior position (see the figure). The last slice was the one in which the uncus and gyrus intralimbicus could no longer be appreciated (see the figure).

• Download figure
• Open in new tab
• Download powerpoint

Figure. Tracing of the entorhinal cortex and hippocampus. (A–C) The entorhinal cortex (ERC). (A) The temporal lobe gets attached to the rest of the brain by the limen insula (star). This slice is a reference point for ERC tracings starting from the following posterior slice (not shown). (B) An example of an ERC tracing. The lateral ERC boundary—the border between the entorhinal and perirhinal cortices—is at the midpoint of the medial bank of the collateral sulcus (solid arrow). The medial boundary is at the level of the sulcus semianularis (open arrow). (C) Presented is the slice in which the uncus and gyrus intralimbicus can no longer be separated, which is the last slice in which the ERC is traced. (D–F) The hippocampus. (D) Presented is the most anterior slice in which the hippocampus can be appreciated to the right, just below the amygdala (Am). (E) An example of a hippocampal tracing (same slice as in C). (F) Presented is the last slice in which the hippocampus is traced. The crus of the fornix departing from the lateral wall of the lateral ventricle on the right side can still be appreciated. Top to bottom: anterior to posterior.

When the ROIs had been traced, the volumes were calculated in cubic millimeters by computing the number of voxels within the traced images using specifically developed software for a standard work console. The reported intrarater variability for the measurements is 6.7% for the hippocampus3 and 7.4% for the ERC.33 The intracranial area (in square millimeters) on a coronal section at the level of the anterior commissure was measured and used for normalization of the volumetric data. For data presentation purposes, the volumes reported herein are normalized to the intracranial area according to the formula: (volume/intracranial area) × 100.3 The right, left, and smallest volumes were considered, the smallest being the most atrophic between the right and the left sides for each subject. Volume loss was defined as the percentage difference of patients’ volumes with respect to the mean of control subjects (set at 100).

Results.

Table 1 shows that FTD patients, control subjects, and AD patients had increasingly older age. FTD patients were less frequently women than both AD patients and control subjects. A family history of dementia was more frequent, although not significant, in the FTD group. On historic examination, memory disturbances and topographic disorientation at onset were more frequent in AD patients, whereas behavioral, oral, and dietary changes and language disturbances were more frequent in FTD patients. This symptom pattern is in agreement with recent findings by Miller et al.13 Daily function, as reported by the informant, was (both at onset and at the time of examination) consistent with the reported memory disturbances in the vast majority of AD patients. On the contrary, most FTD patients usually had a level of daily function much better than could be expected on the basis of the memory disturbances reported by the informant. FTD patients scored lower than AD patients on MMSE and language production. However, global indicators of dementia severity were not consistent with different severity in the two patient groups. Although not reaching statistical significance, mean disease duration, the distribution of CDR scores (42% of FTD patients were stage 0.5 versus 27% of AD patients), and instrumental disability seem to indicate milder disease severity in FTD patients (instrumental disability showed a statistical trend toward significance: p = 0.09). The apolipoprotein E ε4 allele was increased only in the AD group. Lastly, EEG results were abnormal in 3 FTD patients (25%) and 9 AD patients (37%; p = not significant), and primitive reflexes were present in 9 FTD patients (75%) and 14 AD patients (47%; p = not significant).

The relation between hippocampal and ERC volumes with age in the control subjects was assessed with both correlation and locally weighted regression analyses (data not shown). Neither method was able to show any association of hippocampal and ERC volumes with age. The relation of volumes with gender and age was assessed with ANOVA, with hippocampal and ERC volumes as dependent variables and gender and age (<69 years and ≥70 years) as factors. Main effects and interactions of gender and age never approached significance. This observation allowed us to perform the following analyses with no need to control for the effect of age and gender.

Table 2 shows that both hippocampal and ERC volumes were smaller in FTD and AD patients than control subjects. The greatest loss was present in the hippocampi of the AD group (28 to 32% volume loss relative to control subjects), whereas the least loss was in the hippocampi of the FTD group (16 to 21%). Thus, there was a significant trend toward decreasing hippocampal volumes from control subjects through FTD to AD patients, with the FTD patients having hippocampal volumes intermediate between control subjects and AD patients. This was not true for the ERC, the volumes of which were similarly atrophied in both FTD and AD patients. Table 2 also shows that although the left hippocampus was smaller than the right hippocampus in the control subjects, this was not true in either patient group. The ERC volumes were symmetric in all groups.

The distribution of hippocampal and ERC volumes by levels of dementia severity is shown in table 3. Hippocampal atrophy tended to be greater in the more severe cases both in FTD and AD patients, as indicated by the statistical significance of the severity main effect on ANOVA. The significance of the diagnosis main effect confirmed what was suggested by the data reported in table 2—hippocampal atrophy was greater in AD patients than in FTD patients. The lack of significance of the interaction terms indicates that the effect of diagnosis on atrophy was present and of similar magnitude in both severity levels.

ERC atrophy was distributed differently from hippocampal atrophy. Greater ERC atrophy was again present in the more severe cases (the severity main effect was significant), but ERC atrophy was similar in FTD and AD patients (the diagnosis main effect was not significant). Once again, the interaction terms were not significant, indicating that these effects were not modified by each other.

The last step of the analysis addressed the issue of precocity of detectable atrophic changes in FTD and AD. The underlying assumption is that, on a cross-sectional basis, the brain structure in which atrophy first develops is also the structure with the volumetric measures that are furthest from normal, and as a consequence is also the one most efficient in the discrimination of those patients from normal control subjects. Table 4 shows the results of discriminant analysis and Gaussian modeling of volume values. Of the right, left, and smallest hippocampi, the right was the best in discriminating both FTD and AD patients from control subjects, whereas for the ERC the most atrophic between the right and the left side was the best discriminator. The accuracy of the classification based on the discriminant analysis algorithm was variable, but consistently lower for the ERC for both patient groups. When specificity was set at 90%, the sensitivity for the detection of FTD patients from control subjects was approximately 50% for both hippocampal and ERC volumes. Under the same circumstances, the hippocampus performed better than the ERC only in the detection of AD patients, with sensitivity reaching 80%.

Discussion.

We have shown that 1) the hippocampus and the ERC are atrophic in FTD, 2) atrophy of the hippocampus in FTD is less severe than in AD but the atrophy of the ERC is equally severe, 3) hippocampal and ERC atrophy follow the clinical progression of dementia severity in both patient groups, and 4) neither hippocampal nor ERC atrophy was particularly sensitive in discriminating FTD patients from control subjects, whereas the hippocampal volumes could differentiate AD patients from control subjects with remarkable sensitivity. The data lead us to hypothesize that detectable in vivo atrophy of the hippocampus is not an early event in FTD, or as severe in nature, and confirms that this is the case in AD.

These data are consistent with the view that it is possible to identify patterns of atrophy that characterize the primary degenerative dementias. We have shown previously4 that FTD patients can be differentiated simultaneously from AD patients and control subjects with 79% sensitivity and 83% specificity on the basis of measures of atrophy in different brain areas. Enlargement of the temporal horn (mainly the left), frontal atrophy, and sparing of the parahippocampal region were characteristic of FTD, whereas milder frontal atrophy and more severe parahippocampal atrophy was characteristic of AD. The current study indicates that this pattern might be enriched by the findings of milder hippocampal atrophy in FTD and equally severe ERC atrophy.

One relevant question arising from these results is whether the patterns of atrophy are specific for a given pathologic entity or whether they are specific for a syndrome. FTD and AD are pathologically distinct. FTD consists of nonspecific neuronal loss and gliosis affecting more severely the anterior regions (frontal and temporal) of the brain, whereas AD is characteristically a plaque-and-tangle disease of the medial temporal lobe and limbic areas. In general, in FTD, behavioral and language disturbances are particularly striking features early in the course of the disease, whereas learning abilities are relatively spared; the opposite is true of AD. The relatively more severe frontal atrophy and enlargement of the left temporal horn, and the milder hippocampal atrophy in FTD patients might at least partly account for these clinical features. However, AD may have an atypical presentation with prominent behavioral and language disturbances, and sparing of memory also in its early phases.34 It is currently unknown whether the atrophic pattern of atypical cases of AD resembles more closely FTD or typical AD. Therefore, how much of the observed atrophic pattern is due to underlying different pathology or to different affected brain regions remains to be investigated further. It should also be recognized that the presence of some patients with AD pathology in our FTD group might have a significant effect on the current results.

One of the unique findings of this study regards atrophy of the ERC. Although early attempts to quantify ERC atrophy with MRI date back to at least 1991,2 detailed guidelines of MR anatomy of the ERC have been only recently introduced.33 Based on these guidelines, Juottonen et al.35 have shown that the ERC is atrophic in AD patients of mild to moderate severity when compared with control subjects. These findings are confirmed in our own AD patients. However, at least two questions have arisen from our data on ERC atrophy.

First, the ERC is believed to be affected by AD pathology in the very early stages of the disease, even before changes in the hippocampus can be appreciated.19 Our AD data suggest that, in vivo, the atrophic changes that can be appreciated most easily are those of the hippocampus rather than those of the ERC. Compared with the hippocampus, this may be due to limitations of MRI to measure accurately the complicated anatomy of smaller structures, such as the ERC, particularly due to occasionally poor gray–white matter contrast within the parahippocampal gyrus. It may, however, also reflect the true pattern of volumetric pathology. It is noteworthy that, although pathologic hallmarks of AD such as neurofibrillary tangles may first appear in the ERC,19 the presence of these changes does not equal volumetric atrophy per se. This view is supported by evidence from studies of demented patients and cognitively normal elders. In autopsied brains of AD patients, Mizutani and Kasahara36 have shown that at least part of the macroscopic hippocampal atrophy of AD can be secondary to microscopic changes of the ERC. Furthermore, although plaques and tangles were present in biopsy material taken from the hippocampi of cognitively intact older individuals,37 some volumetric MRI studies have not found any significant correlation with hippocampal38,39 or ERC33 volumes with age. These observations suggest that microscopic changes that are believed to be pathologic are not mirrored invariably by macroscopic changes.

Second, the reason why the ERC should be atrophic in FTD needs explanation, because the primary site of the pathology is believed to lie elsewhere (i.e., in the anterior frontal and temporal cortices). The ERC was involved heavily in the pathologic process in some sporadic and familial cases of FTD.21-23 Sima et al.23 have speculated that in the chromosome 17-associated form of FTD the involvement of the ERC might be interpreted in the light of the more widespread involvement of corticobasal ganglionic circuits encompassing the entorhinal, anterior temporal, and cingulate projections to the ventral striatum, as well as of the projections from the pyriform cortex and amygdala to the ERC. Alternatively, it can be hypothesized that ERC atrophy simply might be a deafferentation phenomenon following frontal and temporal damage. In fact, it has been shown in the monkey that the major afferents to the ERC come from the ventral frontal and temporal cortex.40

The hippocampal atrophy of our FTD patients also deserves some discussion. Pathologic descriptions of the hippocampal region are discordant regarding the presence and severity of hippocampal atrophy in FTD, but some observations indicate that hippocampal involvement can occur. Knopman et al.8 have reported hippocampal atrophy in dementia lacking distinctive histology, a condition linked closely to FTD. Although this finding has not been confirmed,9 Mann et al.14 and Mann and South15 also reported hippocampal atrophy in FTD. An independent confirmation of our findings of greater hippocampal atrophy in FTD patients than control subjects has been reported recently.41 Furthermore, Mann et al.14 pointed out that the hippocampus was not involved in their cases of FTD associated with motor neuron disease. None of our FTD patients had motor neuron disease at the time of evaluation, although one developed motor neuron involvement approximately 2 years after onset of the cognitive and behavioral changes, but his atrophic pattern was similar to that of the other FTD patients. Lastly, it should be stressed that the pattern of milder involvement of the hippocampus with more severe involvement of the ERC is similar to that reported on pathologic examination of the brains of patients affected by chromosome 17-associated FTD.23

Recent studies have shown hippocampal or medial temporal lobe atrophy also in other primary degenerative brain diseases such as PD with no dementia,3,5 PD with dementia,3 and Lewy body dementia,5 in addition to the well-known atrophic changes of the hippocampus in AD.1,2,18,38 Our data of decreased hippocampal volume in FTD indicate that this might be one more degenerative brain disease showing hippocampal involvement. This anatomic finding should be interpreted in the light of the clinical features of the disease. It has been argued that hippocampal and medial temporal lobe atrophy might be the biological substrate for the memory disturbance that is often present in some non-AD primary degenerative brain diseases.3,5 We have already pointed out that the characteristic sparing of learning and memory in FTD might be partly explained by the relatively less severe, compared with AD, hippocampal atrophy that we found in our FTD patients. However, the preservation of some brain areas involved in the memory processes, such as the parahippocampal region,4 might also be contributing factors. The contributing role of parahippocampal atrophy to memory deficits is supported by human and animal data. In fact, relevant atrophic changes of the parahippocampal region have been demonstrated in vivo in AD patients with memory disturbances42 of even a mild severity.20 In the monkey, it has been shown that bilateral lesions of the ERC, which spare the parahippocampal cortices, produce only minor and reversible impairment on a task of visual recognition memory.43 In contrast, animals with bilateral lesions of the parahippocampal cortices, which spare the ERC and the hippocampal formation, produce a robust and enduring memory deficit on this task.44

Some limitations of this study should be noted. Age, gender, and disease severity were different in our FTD and AD patients. However, although some investigators have found greater hippocampal atrophy with advancing age in normal elders,18 neither hippocampal nor ERC volumes were associated with age in the group of normal control subjects of our own and other series.33,38,39,45 Moreover, hippocampal and ERC volumes have been shown in our own and other series38 of normal elders to be similar in men and women, once adjusted for head size. For these reasons, different age and gender in patient groups should not have contributed to the results of the study.

The matching of dementia severity between FTD and AD patients is a more uncertain issue. In fact, although the MMSE score was clearly indicative of greater severity in the FTD group, other disease severity indicators (disease duration, CDR, and basic and instrumental disability) pointed toward either no difference or a slightly greater severity in the AD group. Because the MMSE is based mainly on verbal tasks, it may not be well suited to measure global dementia severity in conditions such as FTD, in which language disturbances are particularly severe. Indeed, our FTD patients had language disturbance more severe than that of our AD patients, as shown by a test specifically aimed to assess language production (verbal fluency test). It should also be noted that even if global dementia severity were greater in FTD patients, this would enhance rather than decrease the magnitude and significance of most findings of this study (lower hippocampal atrophy in FTD, and poor discrimination of FTD patients from control subjects by hippocampal and ERC measures). It could be argued that disease duration is a better indicator of dementia severity in conditions in which cognitive or functional instruments are not applicable. Indeed, although not significant, disease duration was almost 1 year shorter in our FTD patients than in our AD patients, which might partly account for the milder hippocampal atrophy of FTD patients. Future volumetric studies of FTD patients might try and match them carefully with AD patients of equal disease duration.

Pathologic specimens were not available for any of these patients. However, all patients were followed clinically for at least 2 years. Although this does not provide certainty of diagnosis, it should at least help to reduce the diagnostic error. Indeed, one of the 14 original FTD patients4 was excluded from the current study because he did not show any cognitive and functional deterioration over the following 2 years. All analyses previously published4 were rerun with the exclusion of this patient, but the results showed only minimal changes, confirming the stability of the findings. Moreover, we tried to reduce the pathologic heterogeneity of the FTD group by excluding those patients who might have had Pick’s disease, but the accuracy of this procedure can hardly be checked. The only supportive argument is that the proportion of Pick’s to non-Pick’s forms in our series (4 of 14, 29%) is close to that reported in other autopsy-proved series (25%).11 Indeed, this is the first study of brain volumetric analysis of FTD patients, and confirmation of these findings in other series is certainly warranted.

We believe that the results of this study are relevant because they add a piece of information to the outline of the volumetric patterns of brain atrophy of the degenerative diseases that are being defined by current research worldwide.