Alzheimer Disease Imaging 

The initial criteria for CT scan diagnosis of Alzheimer disease includes diffuse cerebral atrophy with enlargement of the cortical sulci and increased size of the ventricles. A multitude of studies indicated that cerebral atrophy is significantly greater in patients with Alzheimer disease than in patients who are aging without Alzheimer disease.

This concept was soon challenged, however, because cerebral atrophy can be present in elderly and healthy persons, and some patients with dementia may have no cerebral atrophy, at least in the early stages. The extent of cerebral atrophy was determined by using linear measurements; in particular, bifrontal and bicaudate diameters and the diameters of the third and lateral ventricles. Various measurements were adjusted according to the diameter of the skull to account for normal variation.

To complement this modification, volumetric studies of the ventricles were done. Despite these efforts, it is still difficult to distinguish between findings in a healthy elderly patient and those in a patient with dementia.

In addition, a review of serial CT scans obtained over several months was not clinically useful in the primary diagnosis of the disease.

Rate of change of brain atrophy

Changes in the rate of atrophy progression can be useful in diagnosing Alzheimer disease.^[12] Longitudinal changes in brain size are associated with longitudinal progression of cognitive loss,^[13] and enlargement of the third and lateral ventricles is greater in patients with Alzheimer disease than in control subjects.^[14]

Changes in brain structure

Diffuse cerebral atrophy with widened sulci and dilatation of the lateral ventricles can be observed. Disproportionate atrophy of the medial temporal lobe, particularly of the volume of the hippocampal formations (< 50%), can be seen.

Dilatation of the perihippocampal fissure is a useful radiologic marker for the initial diagnosis of Alzheimer disease, with a predictive accuracy of 91%.^[15] The hippocampal fissure is surrounded laterally by the hippocampus, superiorly by the dentate gyrus, and inferiorly by the subiculum. These structures are all involved in the early development of Alzheimer disease and explain the enlargement in the early stages. At the medial aspect, the fissure communicates with the ambient cistern, and its enlargement on CT scans is often seen as hippocampal lucency or hypoattenuation in the temporal area medial to the temporal horn.

The temporal horns of the lateral ventricles may be enlarged. Prominence of the choroid and hippocampal fissures and enlargement of the sylvian fissure may be noted. White matter attenuation is not a feature of Alzheimer disease.

Degree of confidence

CT scan indices of hippocampal atrophy are highly associated with Alzheimer disease, but the specificity is not well established. Use of a nonquantitative rating scale showed a sensitivity of 81% and a specificity of 67% in differentiating 21 patients with Alzheimer disease with moderate dementia from 21 age-matched control subjects.^[16] Hippocampal volumes in a sample of similar size permitted correct classification of 85% of control subjects.^[17]

Many studies have shown that cerebral atrophy is significantly greater in patients with Alzheimer disease (Alzheimer's disease) than in persons without it. However, the variability of atrophy in the normal aging process makes it difficult to use MRI as a definitive diagnostic technique. (See the images below.)

Fox et al used an automated technique that is potentially applicable in the clinical setting to subtract MRI scans obtained an average of 1 year apart. They observed that there was a significant difference between the rate of change in patients with Alzheimer disease and the rate in control subjects. With MRI, sensitivity and specificity were approximately 90% for predicting the decline in dementia.^[18]

Early MRI studies to evaluate the size of the hippocampus in patients with Alzheimer disease relative to control subjects showed large reductions in hippocampal volumes (approximately 50%) and high sensitivity and specificity for classification.^[19] Over time, enlargement of the temporal horns, as well as of the third and lateral ventricles, was noted in patients with Alzheimer disease compared with control subjects.

On structural MRI, atrophy of the entorhinal cortex is already present in MCI. In the autosomal-dominant forms of Alzheimer disease, the rate of atrophy of the medial temporal structures differentiates affected individuals from control subjects as early as 3 years before the clinical onset of cognitive impairment. The accelerated annual rate of brain atrophy is a surrogate tool for evaluating new therapies in small samples that saves time and resources.

MRI measurements of the hippocampus, amygdala, cingulate gyrus, head of the caudate nucleus, temporal horn, lateral ventricles, third ventricle, and basal forebrain yield a prediction rate of 77% for conversion to Alzheimer disease from questionable Alzheimer disease.^{[20, 21]}

Functional MRI (fMRI) techniques can be used to measure cerebral perfusion. Dynamic susceptibility contrast (DSC) MRI consists of the passage of a concentrated bolus of a paramagnetic contrast agent that sufficiently distorts the local magnetic field to cause a transient loss of signal with pulse sequences, especially T2-weighted sequences. The passage of contrast material is imaged over time by sequential rapid imaging of the same section. In animal studies, the rate of change of signal intensity over time gives a measure directly proportional to cerebral blood volume. Studies in humans have shown a correlation between PET and DSC MRI scan results, as well as between cerebral blood volumes measured with DSC MRI and perfusion on single-photon emission computed tomography (SPECT) scanning.

Studies have been performed using MRI with echo-planar imaging and signal targeting with attenuation radiofrequency (EPISTAR) in patients with Alzheimer disease. Focal areas of hypoperfusion were in the posterior temporoparietooccipital regions. Ratios of signal intensity in the parieto-occipital and temporo-occipital areas to signal intensity on whole section signal intensity were significantly lower in the patients with Alzheimer disease than in those without it. The parieto-occipital ratios were not correlated with the severity of dementia, as measured by using the Blessed Dementia Scale Information Memory Concentration subset.

With fMRI, structural imaging can be done by using the same imaging plane, field of view, and section thickness. Activational fMRI studies have included blood oxygenation level–dependent (BOLD) imaging, which uses changes in the level of oxygenated hemoglobin in capillary beds to depict areas of regional brain activation. In Alzheimer disease, fMRI activation in the hippocampal and prefrontal regions is decreased.

On fMRI, paradigms activate a larger area of parietotemporal association cortex in persons at high risk for Alzheimer disease than in others, whereas the entorhinal cortex activation is relatively low in MCI.^[22]

The techniques are reasonably sensitive and specific in differentiating Alzheimer disease from changes resulting from normal aging, and studies with pathologic confirmation show good sensitivity and specificity in differentiating Alzheimer disease from other dementias. These techniques can also be used to detect abnormalities in asymptomatic or presymptomatic individuals, and they may help in predicting the decline to dementia.

Degree of confidence

MRI findings of hippocampal atrophy are highly associated with Alzheimer disease (Alzheimer's disease), but the specificity is not well established.^[23] Studies have shown that in patients with Alzheimer disease and moderate dementia, hippocampal volumes permitted correct classification in 85% of patients.^[24] In patients with Alzheimer disease and mild dementia, sensitivity was 77%, and specificity, 80%.^[25] Hippocampal volume was the best discriminator, although a number of medical temporal-lobe structures were studied, including the amygdala and the parahippocampal gyrus.

False positives/negatives

Hippocampal atrophy appears to be a feature of vascular disease (multi-infarct dementia) and Parkinson disease, even in patients with Parkinson disease without dementia. Hippocampal and entorhinal cortical atrophy are features of frontotemporal dementia, but they do not appear to be as profound as atrophy is in Alzheimer disease (Alzheimer's disease).^[26]

Single-photon emission computed tomography (SPECT) scanning uses direct photon-emitting isotopes rather than radioisotopes. SPECT isotopes have an average half-life of 6-12 hours.

SPECT instrumentation is highly variable; therefore, use of a SPECT scanner with poor resolution can result in poor clinical performance. Positron-emission tomography (PET) scanning uses tracers that measure regional glucose metabolism (rCMRGlc). SPECT imaging is most commonly used for blood-flow measurement.

Early SPECT studies of blood flow replicated findings of functional reductions in the posterior temporal and parietal cortex. The severity of temporoparietal hypofunction has been correlated with the severity of dementia in a number of studies.^{[12, 27, 28]}

Reductions of blood flow and oxygen use can be found in the temporal and parietal neocortex in patients with Alzheimer disease and moderate to severe symptoms.^[16] Early reductions of glucose metabolism are seen in the posterior cingulate cortex.

SPECT scanning is not commonly used to assess Alzheimer disease. SPECT scanning is useful in the diagnostic assessment of Alzheimer disease if standardized and semiquantitative techniques are used.

Trollor et al examined 18 patients with early Alzheimer disease and 10 healthy, elderly control subjects with high-resolution SPECT scanning during their performance of a simple word-discrimination task and observed a gradation of regional cerebral blood flow (rCBF) values in both groups. The lowest values were in the hippocampus and the highest in the striatum, thalamus, and cerebellum. In the study, SPECT images were coregistered with individual MRI scans, allowing for the delineation of predetermined neuroanatomic regions of interest (ROI).^[17]

Compared with healthy control subjects, patients with Alzheimer disease had low relative rCBF in the parietal and prefrontal cortices. Analysis of individual the ROI demonstrated bilateral reduction of rCBF in the prefrontal poles and posterior temporal and anterior parietal cortex, with unilateral reduction of rCBF in the left dorsolateral prefrontal cortex, right posterior parietal cortex, and left cingulate body. No significant differences in hippocampal, occipital, or basal ganglia rCBF were found. Discriminant function analysis indicated that rCBF in the prefrontal polar regions permitted the best classification.^[17]

In class II studies, the sensitivity of SPECT scanning was lower than that of the clinical diagnosis.^[29] Sensitivity increased as the severity of dementia worsened, but the pretest probability of Alzheimer disease increased as well.^[30]

The added value of SPECT scanning was greatest for a positive test among patients with mild dementia in whom the diagnosis of Alzheimer disease was substantially doubted.^[31] In this situation, a positive SPECT scan result would have increased the posttest probability of Alzheimer disease by 30%, whereas a negative test result would have increased the likelihood of the absence of Alzheimer disease by only 10%.^[32]

Degree of confidence

Without surprise, clinically validated SPECT scan studies showing differences between patients with Alzheimer disease (Alzheimer's disease) and control subjects reveal high sensitivities and specificities of 80-90%.^[32]

In one study, investigators compared patients from a dementia clinic with a community sample of control subjects using quantitative SPECT scanning and reported a 63% sensitivity and an 87% specificity. Alzheimer disease was defined in the study as temporal-lobe perfusion more than 2 standard deviations below control values.

Holman et al found that bilateral temporoparietal hypoperfusion had a positive predictive value of 82% for Alzheimer disease.^[33] Using inhaled xenon-133 (¹³³ Xe) and injected technetium-99m [^99m Tc]hexamethylpropyleneamine oxime, researchers reported a sensitivity of 76% and a specificity of 73%, with a positive predictive value of 92% and a negative predictive value of 57%.^[34] These studies may assist in the early and late diagnosis of Alzheimer disease and with the differential diagnosis of dementias.

PET scanning is a powerful imaging technique that enables in vivo examination of brain functions. It allows for noninvasive quantification of cerebral blood flow, metabolism, and receptor binding. PET scanning helps in understanding the disease's pathogenesis, making the correct diagnosis, and monitoring the disease's progression and response to treatment.^[35]

PET scanning involves the introduction of a radioactive tracer into the human body, usually with an intravenous injection. A tracer is essentially a biologic compound of interest that is labeled with a positron-emitting isotope, such as carbon-11 (¹⁵ O). These isotopes are used because they have relatively short half-lives (from minutes to < 2h), allowing the tracers to reach equilibrium in the body without exposing the subjects to prolonged radiation.

The 2 most common physiologic process measurements performed using PET scanning are glucose with fluorine-18 [¹⁸ F]FDG and cerebral blood flow using water.^[23]

FDG-PET has been used extensively to study Alzheimer disease, and it is evolving into an effective tool for early diagnosis and for differentiation of Alzheimer disease from other types of dementia. FDG-PET has been used to detect persons at risk for Alzheimer disease even before the onset of symptoms.^[36]

Patients with Alzheimer disease have characteristic temporoparietal glucose hypometabolism, the degree of which is correlated with the severity of dementia.^[37] (Temporal and parietal glucose hypometabolism is widely seen on PET images in patients with Alzheimer disease.) With disease progression, frontal involvement may be evident. Glucose hypometabolism in Alzheimer disease is likely to be caused by a combination of neuronal cell loss and decreased synaptic activity.^[38]

In control subjects, entorhinal cortex hypometabolism on FDG-PET has predictive value in the progression of dementia to MCI or, even, to Alzheimer disease.^{[39, 40]} The identification of asymptomatic individuals at risk will have an enormous role in the treatment strategy for Alzheimer disease.^[41]

Individuals at high risk for Alzheimer disease (asymptomatic carriers of the APOE*E4 allele) exhibit a pattern of glucose hypometabolism similar to that of patients with Alzheimer disease. After a mean follow-up of 2 years, the cortical metabolic abnormality continues to decline despite preservation of cognitive performance.^{[42, 40]}

In patients with Alzheimer disease, PET performed with ligand PK11195 labeled with carbon-11, or (R)-[¹¹ C] PK11195, showed increased binding in the entorhinal, temporoparietal, and cingulate cortices. This finding corresponded to the postmortem distribution of Alzheimer disease pathology.^[43]

Degree of confidence

Despite the technical differences, results from PET and SPECT scanning are comparable, although data suggest that PET scanning is more sensitive than SPECT scanning.^[44] On PET or SPECT scanning, mild Alzheimer disease may be more difficult to detect than moderate or severe disease.

In Alzheimer disease, FDG-PET has a sensitivity of 94% and a specificity of 73%. It can also be used to correctly predict a progressive course of dementia with a 91% sensitivity and a nonprogressive course with a 75% specificity.^[45]

Efforts to develop a specific ligand for Aβ plaques may further enhance the sensitivity of PET scanning for early diagnosis of Alzheimer disease and may provide a biologic marker of disease progression.^[43]

In their study, Boxer et al reported that different amyloid-binding PET scan agents—Pittsburgh Compound-B and FDDNP—may have differential sensitivity to prion-related brain pathology and that a combination of amyloid imaging agents may be useful in the diagnosis of early onset dementia.^[46]

Single-photon emission computed tomography (SPECT) scanning uses direct photon-emitting isotopes rather than radioisotopes. SPECT isotopes have an average half-life of 6-12 hours.

SPECT instrumentation is highly variable; therefore, use of a SPECT scanner with poor resolution can result in poor clinical performance. Positron-emission tomography (PET) scanning uses tracers that measure regional glucose metabolism (rCMRGlc). SPECT imaging is most commonly used for blood-flow measurement.

Early SPECT studies of blood flow replicated findings of functional reductions in the posterior temporal and parietal cortex. The severity of temporoparietal hypofunction has been correlated with the severity of dementia in a number of studies.^{[12, 27, 28]}

Reductions of blood flow and oxygen use can be found in the temporal and parietal neocortex in patients with Alzheimer disease and moderate to severe symptoms.^[16] Early reductions of glucose metabolism are seen in the posterior cingulate cortex.

SPECT scanning is not commonly used to assess Alzheimer disease. SPECT scanning is useful in the diagnostic assessment of Alzheimer disease if standardized and semiquantitative techniques are used.

Trollor et al examined 18 patients with early Alzheimer disease and 10 healthy, elderly control subjects with high-resolution SPECT scanning during their performance of a simple word-discrimination task and observed a gradation of regional cerebral blood flow (rCBF) values in both groups. The lowest values were in the hippocampus and the highest in the striatum, thalamus, and cerebellum. In the study, SPECT images were coregistered with individual MRI scans, allowing for the delineation of predetermined neuroanatomic regions of interest (ROI).^[17]

Compared with healthy control subjects, patients with Alzheimer disease had low relative rCBF in the parietal and prefrontal cortices. Analysis of individual the ROI demonstrated bilateral reduction of rCBF in the prefrontal poles and posterior temporal and anterior parietal cortex, with unilateral reduction of rCBF in the left dorsolateral prefrontal cortex, right posterior parietal cortex, and left cingulate body. No significant differences in hippocampal, occipital, or basal ganglia rCBF were found. Discriminant function analysis indicated that rCBF in the prefrontal polar regions permitted the best classification.^[17]

In class II studies, the sensitivity of SPECT scanning was lower than that of the clinical diagnosis.^[29] Sensitivity increased as the severity of dementia worsened, but the pretest probability of Alzheimer disease increased as well.^[30]

The added value of SPECT scanning was greatest for a positive test among patients with mild dementia in whom the diagnosis of Alzheimer disease was substantially doubted.^[31] In this situation, a positive SPECT scan result would have increased the posttest probability of Alzheimer disease by 30%, whereas a negative test result would have increased the likelihood of the absence of Alzheimer disease by only 10%.^[32]

Degree of confidence

Without surprise, clinically validated SPECT scan studies showing differences between patients with Alzheimer disease (Alzheimer's disease) and control subjects reveal high sensitivities and specificities of 80-90%.^[32]

In one study, investigators compared patients from a dementia clinic with a community sample of control subjects using quantitative SPECT scanning and reported a 63% sensitivity and an 87% specificity. Alzheimer disease was defined in the study as temporal-lobe perfusion more than 2 standard deviations below control values.

Holman et al found that bilateral temporoparietal hypoperfusion had a positive predictive value of 82% for Alzheimer disease.^[33] Using inhaled xenon-133 (¹³³ Xe) and injected technetium-99m [^99m Tc]hexamethylpropyleneamine oxime, researchers reported a sensitivity of 76% and a specificity of 73%, with a positive predictive value of 92% and a negative predictive value of 57%.^[34] These studies may assist in the early and late diagnosis of Alzheimer disease and with the differential diagnosis of dementias.

PET scanning is a powerful imaging technique that enables in vivo examination of brain functions. It allows for noninvasive quantification of cerebral blood flow, metabolism, and receptor binding. PET scanning helps in understanding the disease's pathogenesis, making the correct diagnosis, and monitoring the disease's progression and response to treatment.^[35]

PET scanning involves the introduction of a radioactive tracer into the human body, usually with an intravenous injection. A tracer is essentially a biologic compound of interest that is labeled with a positron-emitting isotope, such as carbon-11 (¹⁵ O). These isotopes are used because they have relatively short half-lives (from minutes to < 2h), allowing the tracers to reach equilibrium in the body without exposing the subjects to prolonged radiation.

The 2 most common physiologic process measurements performed using PET scanning are glucose with fluorine-18 [¹⁸ F]FDG and cerebral blood flow using water.^[23]

FDG-PET has been used extensively to study Alzheimer disease, and it is evolving into an effective tool for early diagnosis and for differentiation of Alzheimer disease from other types of dementia. FDG-PET has been used to detect persons at risk for Alzheimer disease even before the onset of symptoms.^[36]

Patients with Alzheimer disease have characteristic temporoparietal glucose hypometabolism, the degree of which is correlated with the severity of dementia.^[37] (Temporal and parietal glucose hypometabolism is widely seen on PET images in patients with Alzheimer disease.) With disease progression, frontal involvement may be evident. Glucose hypometabolism in Alzheimer disease is likely to be caused by a combination of neuronal cell loss and decreased synaptic activity.^[38]

In control subjects, entorhinal cortex hypometabolism on FDG-PET has predictive value in the progression of dementia to MCI or, even, to Alzheimer disease.^{[39, 40]} The identification of asymptomatic individuals at risk will have an enormous role in the treatment strategy for Alzheimer disease.^[41]

Individuals at high risk for Alzheimer disease (asymptomatic carriers of the APOE*E4 allele) exhibit a pattern of glucose hypometabolism similar to that of patients with Alzheimer disease. After a mean follow-up of 2 years, the cortical metabolic abnormality continues to decline despite preservation of cognitive performance.^{[42, 40]}

In patients with Alzheimer disease, PET performed with ligand PK11195 labeled with carbon-11, or (R)-[¹¹ C] PK11195, showed increased binding in the entorhinal, temporoparietal, and cingulate cortices. This finding corresponded to the postmortem distribution of Alzheimer disease pathology.^[43]

Degree of confidence

Despite the technical differences, results from PET and SPECT scanning are comparable, although data suggest that PET scanning is more sensitive than SPECT scanning.^[44] On PET or SPECT scanning, mild Alzheimer disease may be more difficult to detect than moderate or severe disease.

In Alzheimer disease, FDG-PET has a sensitivity of 94% and a specificity of 73%. It can also be used to correctly predict a progressive course of dementia with a 91% sensitivity and a nonprogressive course with a 75% specificity.^[45]

Efforts to develop a specific ligand for Aβ plaques may further enhance the sensitivity of PET scanning for early diagnosis of Alzheimer disease and may provide a biologic marker of disease progression.^[43]

In their study, Boxer et al reported that different amyloid-binding PET scan agents—Pittsburgh Compound-B and FDDNP—may have differential sensitivity to prion-related brain pathology and that a combination of amyloid imaging agents may be useful in the diagnosis of early onset dementia.^[46]