In addition to fMRI, structural imaging has recently gained more attention. Voxel-based morphometry (VBM) allows voxel-by-voxel comparison of the relative contribution of white or gray matter to the signal between groups. A well-known application to learning was the increase of parietal regions used for visuospatial integration in people who learned to juggle (Draganski et al., 2004). Integrity of tracts may be determined through diffusion tensor imaging (DTI). DTI measures the diffusion vector in each voxel. It is feasible to detect the direction of diffusion across longer distances by analyzing and relating the values in several neighboring voxels. As diffusion is facilitated along axes, in contrast, probabilistic fiber tracking identifies the most likely course of fiber tracts (Kreher et al., 2008). Comparison of diffusivity (i.e., fractional anisotropy) voxel by voxel across the entire brain allows the differentiation of tracts between groups, analogous to VBM. For instance, stutterers have a less smooth connection below a region responsible for preparation of articulation (Sommer et al., 2002).

Positron Emission Tomography

Although the first attempts to use positron emitters for medical imaging were made as early as the 1950s, the concept of modern PET was developed during the 1970s (Phelps et al., 1975). The underlying principle of PET, and also of SPECT, is to image and quantify a physiological function or molecular target of interest (e.g., blood flow, metabolism, receptor binding) in vivo by noninvasively assessing the spatial and temporal distribution of the radiation emitted by an intravenously injected target-specific probe (radiotracer). Importantly, PET and SPECT tracers are administered in a non-pharmacological dose (micrograms or less), so they neither disturb the underlying system nor cause pharmacological or behavioral effects. Because of their ability to visualize molecular targets and functions on a macroscopic level with unsurpassed sensitivity, down to picomolar concentration, PET and SPECT are also called molecular imaging techniques. (See Cherry et al., 2003, for an excellent textbook on PET and SPECT physics.)

In the case of PET, a positron-emitting radiotracer is injected. The emitted positron travels a short distance in tissue (effective range < 1 mm for common PET nuclides) before it encounters an electron, yielding a pair of two 511-keV annihilation photons emitted in opposite directions. This photon pair leaving the body is detected quasi-simultaneously (within a few nanoseconds) by scintillation detectors of the PET detector rings that surround the patient’s head. Assuming that the annihilation site is located on the line connecting both detectors (known as the line of response [LOR]), three-dimensional (3D) PET image data sets of the distribution of the PET tracer and its target in tissue are generated by standard image reconstruction algorithms using the LOR data. To actually gain quantitative PET images (i.e., radioactivity/tracer concentration per unit tissue), the acquired data are corrected for scatter and random coincidences and photon attenuation by tissue absorption (e.g., using a measured CT or ⁶⁸Ge/⁶⁸Ga transmission scan). The spatial resolution of modern PET or combined PET/CT systems is about 3 to 5 mm. Thus, PET is susceptible to partial volume effects if the object or lesion size is below two times the scanner resolution (as a rule of thumb).

Commonly used radionuclides in neurological PET studies are oxygen-15 (¹⁵O, physical half-life = 2.03 minutes), carbon-11 (¹¹C, half-life = 20.4 minutes), and fluorine-18 (¹⁸F, half-life = 109.7 minutes), which are all cyclotron products. Whereas the relatively long half-life of ¹⁸F allows shipping ¹⁸F-labelled tracers from a cyclotron site to a distant PET site, this is not possible in the case of ¹⁵O and ¹¹C. This clearly limits the clinical use of ¹⁵O-labelled water, molecular oxygen, and carbon dioxide for quantification of CBF, cerebral metabolic rate of oxygen, and oxygen extraction fraction. This also applies to clinically very interesting ¹¹C-labelled tracers like [¹¹C]raclopride (dopamine D₂/D₃ receptor), [¹¹C]flumazenil (GABA_A receptor), [¹¹C]methionine (amino acid transport), and [¹¹C]PIB (amyloid beta [Aβ]). Thus, ¹⁸F-labelled substitutes have been proposed and are currently under investigation.

In this chapter on perfusion and metabolism, we will focus on PET studies using the glucose analog, 2-deoxy-2-(¹⁸F)fluoro-d-glucose ([¹⁸F]FDG), to assess cerebral glucose metabolism. With the rate of glucose metabolism being closely related to maintenance of ion gradients and transmitter turnover (in particular, glutamate), [¹⁸F]FDG represents an ideal tracer for assessment of neuronal function and its changes (Sokoloff, 1977). After uptake in cerebral tissue by specific glucose transporters, [¹⁸F]FDG is phosphorylated by hexokinase. Since [¹⁸F]FDG-6-P is neither a substrate for transport back out of the cell nor can it be metabolized further, it is virtually irreversible trapped in cells. Therefore, the distribution of [¹⁸F]FDG in tissue imaged by PET (started 30-60 minutes after injection to allow for sufficient uptake; 5-20 minute scan duration) closely reflects the regional distribution of cerebral glucose metabolism. By use of appropriate pharmacokinetic models and a plasma input function (i.e., [¹⁸F]FDG concentration in arterial or arterialized venous plasma), the absolute cerebral metabolic rate of glucose (CMRglc in µmol/min/100 g tissue) can be estimated. In the case of [¹⁸F]FDG, absolute quantification is usually not necessary for routine clinical studies, since the diagnostic information can often be obtained from the cerebral pattern of [¹⁸F]FDG uptake or relative estimates of regional glucose metabolism gained by normalizing regional [¹⁸F]FDG uptake to the uptake of a suitable reference region unaffected by disease.

Single-Photon Emission Computed Tomography

The first SPECT measurements were performed in the 1960s (Kuhl and Edwards, 1964). As explained earlier, SPECT relies on the same radiotracer principle as PET. Unlike PET, SPECT employs gamma-emitting radionuclides that decay by emitting a single gamma ray. Typical radionuclides employed for neurological SPECT are technetium-99m (^99mTc; half-life = 6.02 hours) and iodine-123 (¹²³I; half-life = 13.2 hours). Gamma cameras are used for SPECT acquisition, whereby usually two or three detector heads rotate around the patient’s head to acquire two-dimensional planar images (projections) of the head from multiple angles (e.g., in 3-degree steps). Whereas radiation collimation is achieved by coincidence detection in PET, hardware collimators with lead septa are placed in front of the detector heads in the case of SPECT scanners. Finally, 3D image data reconstruction is done by conventional reconstruction algorithms. With combined SPECT/CT systems, a CT transmission scan can replace the less accurate calculated attenuation correction.

The different acquisition principles outlined above imply that SPECT possesses a considerably lower sensitivity than PET. Thus, rapid temporal sampling (image frames of seconds to minutes) as a prerequisite for pharmacokinetic analyses is the strength of PET, whereas a single SPECT acquisition usually takes 20 to 30 minutes. Furthermore, the spatial resolution of modern SPECT is only about 7 to 10 mm, deteriorating with increasing distance between object and collimator (i.e., higher resolution for cortical than subcortical structures; distance between patient and collimator should be minimized for optimal resolution). SPECT is considerably more susceptible to partial volume effects than PET, which can be a particular drawback when it comes to imaging small structures or lesions (e.g., brain tumors). Nevertheless, brain-dedicated SPECT instruments have been proposed that allow for optimized spatial and temporal sampling and pharmacokinetic data quantification (Meyer et al., 2008), and further technical developments are underway (Jansen and Vanderheyden, 2007). The important advantages of SPECT over PET are the lower costs, the broad availability of SPECT systems, and the availability of SPECT radionuclides in smaller community hospitals and private practices. While ¹²³I-labelled tracers (e.g., [¹²³I]FP-CIT and [¹²³I]IBZM for dopamine transporter and receptor imaging, respectively) can easily be shipped over long distances, technetium-99m can be eluted onsite from molybdenum-99 (⁹⁹Mo)/^99mTc generators and used for labeling commercially available radiopharmaceutical kits. Although ^99mTc is an almost ideal radionuclide from the perspective of imaging physics, chemical incorporation of ^99mTc into tracers is much more demanding than in case of the aforementioned PET radionuclides, limiting the diversity of ^99mTc-labelled tracers.

We will focus on the two most widely used CBF tracers, hexamethylpropyleneamine oxime ([^99mTc]HMPAO) and ethylcysteinate dimer ([^99mTc]ECD). Owing to their lipophilic nature and thus high first-pass extraction, both radiotracers are rapidly taken up by the brain. They are quasi-irreversibly retained after conversion into hydrophilic compounds (enzymatic de-esterification of [^99mTc]ECD; instability, and possibly interaction with glutathione in the case of [^99mTc]HMPAO). Differences in uptake mechanisms may explain slight differences in biological behavior (e.g., in stroke), with [^99mTc]HMPAO being more closely correlated to perfusion, while [^99mTc]ECD uptake is also influenced by metabolic activity. Despite the fact that cerebral radiotracer uptake is virtually complete within just 1 to 2 minutes after injection, SPECT acquisition is usually started after 30 to 60 minutes to allow for sufficient background clearance.

Given the fact that the CBF is closely coupled to cerebral glucose metabolism, and thus to neuronal function (with a few rare exceptions), [^99mTc]HMPAO and [^99mTc]ECD are used to assess neuronal activity. However, since cerebral autoregulation is also affected by many other factors (e.g., carbon dioxide level) and possibly diseases, cerebral glucose metabolism represents a more direct and probably less variable marker of neuronal activity. Given the technical limitations mentioned earlier, [¹⁸F]FDG PET is generally preferred to CBF SPECT. One important exception, however, is the use of ictal CBF SPECT in the assessment of patients with epilepsy.

Clinical Applications

Dementia

Early and accurate diagnosis of dementia is of crucial importance for appropriate treatment (including possible enrollment into investigational trials on novel therapies and avoidance of possible side effects of treatments), for prognosis, and for adequate counseling of patients and caregivers. The diagnostic power of [¹⁸F]FDG PET in this situation is well established (Herholz, 2003). In clinical practice, [¹⁸F]FDG PET studies are interpreted by qualitative visual readings. These readings are commonly assisted by voxel-based statistical analyses in comparison to aged-matched normal controls (Herholz et al., 2002; Minoshima et al., 1995), which have become state of the art in recent years. PET studies should always be interpreted with parallel inspection of a recent CT or MRI scan to detect structural defects (e.g., ischemia, atrophy, subdural hematoma) that cause regional hypometabolism.

The typical finding in Alzheimer disease (AD), the most frequent neurodegenerative dementia, is bilateral hypometabolism of the temporal and parietal association cortices, with the temporoparietal junction being the center of impairment. As the disease progress, frontal association cortices also get involved (Figs. 33C.1 and 33C.2). The magnitude and extent of the hypometabolism increases with progressing disease, with relative sparing of the primary motor and visual cortices, the basal ganglia, and the cerebellum (often used as reference regions). The degree of hypometabolism is usually well correlated with the dementia severity (Herholz et al., 2002; Minoshima et al., 1997; Salmon et al., 2005). Furthermore, cortical hypometabolism is often asymmetrical, corresponding to predominant clinical symptoms (language impairment if dominant or visuospatial impairment if non-dominant hemisphere is affected, respectively). Voxel-based statistical analyses consistently show that the posterior cingulate gyrus and precuneus are also affected even in the earliest AD stages (Minoshima et al., 1997). It is unclear whether the hypometabolism in this rather circumscribed area is in fact particularly pronounced in early AD or more narrowly and consistently localized (compared to other regions) and thus preferentially detected by statistical methods (Herholz et al., 2002). In any case, it is an important diagnostic clue in early AD that should not be missed. The hippocampus is particularly affected by AD pathology and, consequently, neurodegeneration. However, studies on hippocampal metabolism in AD yielded conflicting results, mostly showing no significant hypometabolism. This may be due to the relatively low-normal [¹⁸F]FDG uptake, small size, and AD-related atrophy of this structure, which render visual and voxel-based statistical analyses insensitive. Region-based analyses like a recently proposed automated hippocampal masking technique can help overcome these limitations and provide valuable incremental diagnostic information (Mosconi et al., 2005).

Fig. 33C.1 [¹⁸F]FDG PET in early Alzheimer disease. Early disease stage is characterized by mild to moderate hypometabolism of temporal and parietal cortices and posterior cingulate gyrus and precuneus. As in this case, distinct asymmetry is often noticed. As disease progresses, frontal cortices also become involved. Upper panel, Transaxial PET images of [¹⁸F]FDG uptake (color coded, see color scale on right; orientation in radiological convention as indicated). Lower panel, Results of voxel-based statistical analysis using Neurostat/3D-SSP. Three-dimensional stereotactic surface projections of [¹⁸F]FDG uptake (upper row) and statistical deviation of the individual’s examination (as z score) from age-matched healthy controls. Data are color coded in rainbow scale (see lower right for z scale). Given are right and left lateral and mesial views.

(Neurostat/3D-SSP analysis based on Minoshima, S., Frey, K.A., Koeppe, R.A., et al., 1995. A diagnostic approach in Alzheimer’s disease using three-dimensional stereotactic surface projections of fluorine-18-FDG PET. J Nucl Med 36, 1238-1248.)

Fig. 33C.2 [¹⁸F]FDG PET in advanced Alzheimer disease. Advanced disease stage is characterized by severe hypometabolism of temporal and parietal cortices and posterior cingulate gyrus and precuneus. Frontal cortex is also involved, while sensorimotor and occipital cortex, basal ganglia, thalamus, and cerebellum are spared. Mesiotemporal hypometabolism is also apparent. Upper panel, Transaxial PET images of [¹⁸F]FDG uptake. Lower panel, Results of voxel-based statistical analysis using Neurostat/3D-SSP. Given are right and left lateral, superior, and posterior views (see Fig. 33C.1 for additional details).

In [¹⁸F]FDG PET studies on autopsy-confirmed AD patients with memory complaints, the pattern of temporoparietal hypometabolism as assessed by visual readings alone showed a high sensitivity of 84% to 94% for detecting pathologically confirmed AD, with a specificity of 63% to 74% (Hoffman et al., 2000; Jagust et al., 2007; Silverman et al., 2001). As a diagnostic tool, visual inspection of [¹⁸F]FDG PET was found to be of similar accuracy as a clinical follow-up examination performed 4 years after PET (Jagust et al., 2007). In two recent large multicenter trials, voxel-based statistical analyses of cortical [¹⁸F]FDG uptake provided a sensitivity of 93% to 99% and a specificity of 93% to 98% for the distinction between mild to moderate AD (clinical diagnosis) and normal controls (Herholz et al., 2002; Mosconi et al., 2008). The specificity of the differentiation between AD and dementia with Lewy bodies (DLB) and frontotemporal dementia (FTD) was lower (71% and 65%, respectively) if done by voxel-based cortical analyses alone. However, the use of an additional hippocampal analysis greatly improved specificity (100% and 94%, respectively), yielding an overall classification accuracy of 96% for the aforementioned patient groups and controls (Mosconi et al., 2008). Patterns of hypoperfusion observed with CBF SPECT in AD are very similar, but according to a SPECT meta-analysis (Dougall et al., 2004) and direct comparisons (Herholz et al., 2002), [¹⁸F]FDG PET provides higher diagnostic accuracy (also see recommendation in Dubois et al., 2007).

The syndrome of mild cognitive impairment (MCI) (Petersen et al., 1999) represents a risk state for dementia. More than half of subjects progress to manifest dementia within 5 years, with AD being the most frequent underlying cause, particularly in the group with amnestic MCI (Gauthier et al., 2006). Several studies demonstrated that an AD-like [¹⁸F]FDG PET pattern can be observed in high frequency among MCI patients (e.g., 79% and 31% of multidomain and amnestic MCI patients, respectively; Mosconi et al., 2008), and that this pattern is highly predictive of subsequent progression to manifest dementia, with a sensitivity of 92% to 93% and a specificity of 82% to 89% in two recent studies in which overall accuracy of [¹⁸F]FDG PET for prediction of progression was higher than of ApoE genotype or in-depth memory assessment (Anchisi et al., 2005; Drzezga et al., 2005). In addition, a recent meta-analysis showed that [¹⁸F]FDG PET performs better on prediction of rapid progression to AD than CBF SPECT and MRI (Yuan et al., 2009). Finally, it has been demonstrated that cognitively normal healthy controls at risk for AD owing to being an ApoE ε4 carrier and/or having a positive family history (maternal in particular) exhibited significantly reduced glucose metabolism in those cortical areas typically affected by AD (Mosconi et al., 2007; Reiman et al., 1996; Small et al., 1995). This was also found in young ApoE ε4 carriers (mean age of 30 years) and thus decades before the possible onset of AD (Reiman et al., 2004). Follow-up studies in subjects at risk for AD also demonstrated that the subsequent decline in cerebral glucose metabolism in AD-typical regions was significantly greater compared to non-at-risk subjects (Mosconi et al., 2009; Reiman et al., 2001; Small et al., 2000). Taken together, these results emphasize that [¹⁸F]FDG PET is not only a very powerful method for accurate diagnosis of manifest AD and prediction of progression in possibly prodromal AD (MCI) but may also be useful to investigate preclinical stages of AD (e.g., in prevention and treatment trials).

Dementia with Lewy bodies is considered the second most frequent cause of dementia. The typical [¹⁸F]FDG PET pattern observed in DLB resembles the pattern observed in AD, with the exception of additional hypometabolism of the primary visual cortex and the occipital association cortex (Albin et al., 1996) (Fig. 33C.3). The latter has been linked to the occurrence of typical visual hallucinations in DLB patients (particularly in those with relatively preserved posterior temporal and parietal metabolism) (Imamura et al., 1999). Occipital hypometabolism was found to be a valuable diagnostic feature to separate clinically diagnosed patients with AD and DLB (sensitivity 86%-92%, specificity 91%-92%) (Higuchi et al., 2000; Ishii et al., 1998a). This was confirmed in a study with autopsy confirmation (sensitivity 90%, specificity 80%) (Minoshima et al., 2001). Recent studies have demonstrated that hippocampal metabolism is preserved in DLB but reduced in AD (Ishii et al., 2007; Mosconi et al., 2008). The aforementioned multicenter study (Mosconi et al., 2008) found considerably enhanced accuracy of [¹⁸F]FDG PET for differentiating DLB from AD, FTD, and controls if an additional hippocampal analyses were included (92%). It has to be mentioned that differences may be hard to appreciate in routine clinical examination of individual patients. In this situation, PET or SPECT examinations of nigrostriatal integrity (e.g., using [¹⁸F]FDOPA or a dopamine transporter ligand like [¹²³I]FP-CIT) can be very helpful in differentiating between AD and DLB (McKeith et al., 2007). One has to bear in mind that nigrostriatal projections may also be damaged in FTD (Rinne et al., 2002) and atypical parkinsonian syndromes with dementia (e.g., PSP and CBD; see below). DLB is clinically distinguished from Parkinson disease with dementia (PDD) by the so-called 1-year rule, referring to the duration of parkinsonism prior to dementia. Although the separation between DLB and PDD is controversial, it is still recommended for several reasons (Lippa et al., 2007; McKeith, 2007). In line with the notion that both diseases most likely represent manifestations of the same disease process (Lewy body disease), [¹⁸F]FDG PET studies in PDD (Peppard et al., 1992; Vander Borght et al., 1997) found very similar results to those in DLB. In a recent direct comparison, there were only minor differences between both groups (barely significant lower metabolism in DLB compared to PDD in the anterior cingulate cortex) (Yong et al., 2007). Of note, however, in a recent [¹¹C]PIB PET study, the majority of DLB patients (11/13) showed increased Aβ binding, while the majority of PDD patients (10/12) did not (Edison et al., 2008).

Fig. 33C.3 [¹⁸F]FDG PET in dementia with Lewy bodies (DLB). This disorder affects similar areas as those affected by Alzheimer disease (AD). Occipital cortex is also involved, which may distinguish DLB from AD; mesiotemporal lobe is relatively spared in DLB. A very similar, if not identical, pattern is observed in Parkinson disease with dementia (PDD). Upper panel, Transaxial PET images of [¹⁸F]FDG uptake. Lower panel, Results of voxel-based statistical analysis using Neurostat/3D-SSP. Given are right and left lateral, superior, and posterior views (see Fig. 33C.1 for additional details).

Frontotemporal lobar degeneration (FTLD) probably represents the third most common cause of dementia. It comprises the three prototypical syndromes: frontotemporal dementia (FTD), semantic dementia (SD), and progressive nonfluent aphasia (PA) (Neary et al., 1998). FTLD is caused by a spectrum of underlying, possibly related, pathologies (also including corticobasal degeneration [CBD] and progressive supranuclear palsy [PSP]) which lead to a variety of overlapping clinical presentations that hinder predicting the underlying pathology by the clinical phenotype (so-called Pick complex) (Kertesz et al., 2005). FTD is usually associated with a bilateral, sometimes asymmetrical, frontal hypometabolism which is most pronounced in the mesial (polar) frontal cortex (Fig. 33C.4) (Garraux et al., 1999; Salmon et al., 2003). The putamen, thalamus, and temporal and parietal cortices are also affected, although to a lesser extent (Garraux et al., 1999; Ishii et al., 1998). Despite the fact that FTD and AD affect overlapping cortical areas, the predominance of frontal and temporoparietal deficits, respectively, is usually very apparent and allows a clear distinction between FTD and AD. In line with this, a voxel-based statistical analysis provided a diagnostic accuracy of 90% (sensitivity 98%, specificity 86%) for separating FTD and AD in an autopsy-confirmed study, which was clearly superior to clinical diagnosis alone (Foster et al., 2007). As mentioned, additional hippocampal analyses may enable an even higher diagnostic accuracy (94%) for distinguishing FTD from AD, DLB, and control subjects (Mosconi et al., 2008). SD and PA are rather rare disorders, and PET studies are less abundant. Unlike FTD patients, SD patients usually show a predominant hypometabolism of the temporal lobes, most pronounced for the temporal poles, which is usually leftward asymmetrical (Fig. 33C.5) (Diehl et al., 2004; Rabinovici et al., 2008). In contrast, a strikingly greater left than right perisylvian hypometabolism of frontal, temporal, and parietal cortices is found in PA (Fig. 33C.6), with left insular/frontal opercular involvement being particularly associated with nonfluent aphasic features (Josephs et al., 2010; Nestor et al., 2003; Panegyres et al., 2008; Rabinovici et al., 2008). Given the relatively rare incidence of these syndromes and the controversy regarding their classification (Knibb et al., 2006), studies are missing that address the diagnostic value of [¹⁸F]FDG PET to distinguish between FTLD subgroups (including additional entities not discussed here) and to separate these from other dementias.

Fig. 33C.4 [¹⁸F]FDG PET in frontotemporal dementia (FTD). Bifrontal hypometabolism is usually found in FTD, sometimes in a somewhat asymmetrical distribution, as in this case. At early stages, frontomesial and frontopolar involvement is most pronounced, while parietal cortices are affected later in disease course. Upper panel, Transaxial PET images of [¹⁸F]FDG uptake. Lower panel, Results of voxel-based statistical analysis using Neurostat/3D-SSP. Given are right and left lateral, superior, and posterior views (see Fig. 33C.1 for additional details).

Fig. 33C.5 [¹⁸F]FDG PET in semantic dementia (SD). [¹⁸F]FDG PET scans in SD may initially appear normal because SD leads to hypometabolism of temporal poles, sometimes hard to recognize by visual inspection alone. Voxel-based statistical analyses can be particularly helpful. Usually, both temporal poles are affected (left greater than right), but sometimes degeneration is very asymmetrical, as in this case. Upper panel, Transaxial PET images of [¹⁸F]FDG uptake. Lower panel, Results of voxel-based statistical analysis using Neurostat/3D-SSP. Given are right and left lateral, superior, and inferior views (see Fig. 33C.1 for additional details).

Fig. 33C.6 [¹⁸F]FDG PET in progressive aphasia (PA). [¹⁸F]FDG PET finding in PA is a very asymmetrical hypometabolism of left (speech dominant) hemisphere, which is usually already fairly pronounced when PA is first diagnosed. Both more frontal and more temporoparietal involvements have been described. In this case, there is also a pronounced striatal involvement. Upper panel, Transaxial PET images of [¹⁸F]FDG uptake. Lower panel, Results of voxel-based statistical analysis using Neurostat/3D-SSP. Given are right and left lateral, superior, and posterior views (see Fig. 33C.1 for additional details).

Finally, pure vascular dementia (VD) seems to be rather rare in North America and Europe and more prevalent in Japan; [¹⁸F]FDG PET adds little to the diagnosis of VD. In agreement with CT and MRI, PET may show defects of [¹⁸F]FDG uptake corresponding to ischemic infarcts in all cerebral regions, including primary cortices, striatum/thalamus, and cerebellum. Since the latter are usually well preserved in AD, defects in these regions can be an important diagnostic clue. Deficits due to vascular lesions can be considerably larger or cause remote deficits of [¹⁸F]FDG uptake due to diaschisis. Furthermore, cerebral glucose metabolism was reported to be globally reduced (Mielke et al., 1992), but without absolute quantification this finding cannot be reliably assessed.

Parkinsonism

The current application of [¹⁸F]FDG PET and CBF SPECT in movement disorders focuses mostly on the diagnostic workup of parkinsonism, whereas their use for investigations of other movement disorders like Huntington or Wilson diseases is of little practical clinical importance in the era of gene analyses. Similar conclusions can also be drawn from CBF SPECT, but as noted earlier, the diagnostic accuracy is expected to be lower.

Most studies using quantitative [¹⁸F]FDG PET in idiopathic Parkinson disease (PD) reported a global decrease of gray-matter CMRglc in nondemented PD patients compared to controls (Hu et al., 2000; Kuhl et al., 1984; Peppard et al., 1992). This global hypometabolism was more pronounced in demented PD patients, with accentuation in temporo-parieto-occipital cortices (Kuhl et al., 1984; Peppard et al., 1992; Vander Borght et al., 1997). Cerebral hypometabolism correlated with dementia severity (Piert et al., 1996). Interestingly, temporo-parieto-occipital hypometabolism may also been seen in nondemented PD patients (Hu et al., 2000), raising the important and so far unresolved question of whether this finding indicates an increased risk of subsequent dementia.

In addition, several studies investigated abnormalities in relative regional glucose metabolism (regional [¹⁸F]FDG uptake normalized to global [¹⁸F]FDG uptake), either by direct comparison between PD patients and controls or by more sophisticated methods like spatial covariance analyses. Both approaches yielded very similar findings in terms of affected regions and local effects (Eckert et al., 2005; Eidelberg et al., 1994; Tang et al., 2010a): relatively increased activity is usually observed in putamen, globus pallidus, thalamus, pons, cerebellum, and primary motor cortex, whereas decreased activity is detected in bilateral parietal, occipital, and frontal cortices (dorsolateral prefrontal, premotor, and supplementary motor areas) (Fig. 33C.7). Of note, however, since global normalization is performed, the aforementioned relative changes do not necessarily imply corresponding changes in absolute glucose metabolism (Borghammer et al., 2009). Spatial covariance analyses can be powerful methods to identify abnormal cerebral networks as biomarkers for assessment of disease severity, progression, treatment efficacy, and differential diagnosis (Hirano et al., 2009). The expression of two distinctive spatial covariance patterns characterize PD: one related to motor manifestations and one related to cognitive manifestations. Alteration of metabolic brain network activity has been found to correlate with the cardinal motor symptoms of PD (PDRP), and a PD-related cognitive pattern (PDCP) has been described that correlates with cognitive impairment and affective disorder. A recent study showed that abnormal PDRP activity may precede motor symptom onset by 2 years (Tang et al., 2010b). Such analyses can also be useful for early diagnosis of suspected PD (separating PD from controls) (Eckert et al., 2005; Eidelberg et al., 1994, 1995), but it is unlikely that they will be superior to PET and SPECT examinations using molecular probes to investigate integrity of nigrostriatal dopaminergic projections like [¹⁸F]FDOPA or [¹²³I]FP-CIT (Brooks, 2010; Piccini and Whone, 2004). The strength of [¹⁸F]FDG PET in parkinsonism lies in the differential diagnosis of PD and atypical parkinsonian syndromes (APS) like multiple system atrophy (MSA), PSP, and CBD, that cannot be reliably differentiated on an individual patient basis by the aforementioned studies of nigrostriatal integrity, but present with distinct metabolic patterns on [¹⁸F]FDG PET.

Fig. 33C.7 [¹⁸F]FDG PET in Parkinson disease (PD). PD is typically characterized by (relative) striatal hypermetabolism. Temporoparietal hypometabolism can be observed in a significant fraction of PD patients without apparent cognitive impairment. As shown here, hypometabolism can be fairly pronounced, strongly resembling the typical finding in dementia with Lewy bodies/PD with dementia. However, it still unclear whether this represents preclinical dementia. Upper panel, Transaxial PET images of [¹⁸F]FDG uptake. Lower panel, Results of voxel-based statistical analysis using Neurostat/3D-SSP. Given are right and left lateral, superior, and posterior views (see Fig. 33C.1 for additional details).

In a very recent large clinicopathological study (Ling et al., 2010) with extensive review of all medical records available, the sensitivity the clinical diagnosis was report to be 92.8% for PD, while the sensitivity was only 70.1% for MSA, 73.1% for PSP, and 26.3% for CBD. Specificity was 85.8% for PD and better than 95% for MSA, PSP, and CBD (Ling et al., 2010). In line with the fact that decisive diagnostic features tend to develop over time (often years) and that a considerable fraction of APS patients show some initial response to l-dopa, the sensitivity of the initial clinical diagnosis was reported to be lower: 73.5% for PD (Litvan et al., 1998), approximately 60% for MSA (Litvan et al., 1997, Osaki et al., 2002), and less than 50% for PSP (Osaki et al., 2004). Consequently, for appropriate therapy selection (including therapy trials), prognosis, and counseling of patients and caregivers, improving early differential diagnosis is essential, and [¹⁸F]FDG PET can significantly contribute to this endeavor by providing disease-specific patterns of cerebral glucose metabolism. In MSA, decreased glucose metabolism is commonly found in striatum, brainstem, and cerebellum (Fig. 33C.8), which can be used as a reliable discriminator from PD (Antonini et al., 1997; Eidelberg et al., 1993; Ghaemi et al., 2002; Otsuka et al., 1997; Taniwaki et al., 2002). While patients with olivopontocerebellar atrophy have predominant cerebellar hypometabolism, predominant striatal hypometabolism is found in striatonigral degeneration (Perani et al., 1995). In addition, there is a reduction of cerebral glucose metabolism in the frontal cortices, which appears to spread to temporal and parietal cortices during the disease course, with subsequent cognitive decline (Lyoo et al., 2008; Otsuka et al., 1996). In PSP, glucose metabolism was consistently reported to be reduced in caudate nucleus and putamen, thalamus, pons/midbrain, and the mesial and dorsal frontal cortex (most notably in anterior cingulate cortex, precentral, dorso- and ventrolateral premotor and prefrontal areas) (Foster et al., 1988; Garraux et al., 1999; Hosaka et al., 2002; Juh et al., 2005; Karbe et al., 1992) (Fig. 33C.9). Comparing FTD and PSP, which both show frontal lobe involvement, striatofrontal metabolic impairment is greater in FTD, whereas mesencephalothalamic impairment was only observed in PSP (Garraux et al., 1999). Finally, the hallmark [¹⁸F]FDG PET of clinically diagnosed CBD is a highly asymmetrical cerebral hypometabolism in the thalamus, striatum, and predominantly parietal cortex but also frontal cortex (including cingulate cortex, precentral, premotor, and prefrontal areas) of the hemisphere contralateral to the side most clinically affected (Eidelberg et al., 1991; Garraux et al., 2000; Hosaka et al., 2002; Juh et al., 2005; Nagahama et al., 1997) (Fig. 33C.10). In a direct comparison of PSP and CBD, midbrain, thalamus, and cingulate cortex hypometabolism is greater in PSP, whereas asymmetry of the parietal lobes, sensorimotor area, and striatal involvement was more pronounced in clinical CBD (Garraux et al., 2000; Hosaka et al., 2002; Juh et al., 2005; Nagahama et al., 1997).

Fig. 33C.8 [¹⁸F]FDG PET in multiple system atrophy (MSA). In contrast to Parkinson disease, striatal hypometabolism is commonly found in MSA, particularly in those patients with striatonigral degeneration (SND, or MSA-P). In patients with olivopontocerebellar degeneration (OPCA, or MSA-C), cerebellar hypometabolism particularly evident, as shown here. Upper panel, Transaxial PET images of [¹⁸F]FDG uptake. Lower panel, Results of voxel-based statistical analysis using Neurostat/3D-SSP. Given are right and left lateral, superior, and inferior views (see Fig. 33C.1 for additional details).

Fig. 33C.9 [¹⁸F]FDG PET in progressive supranuclear palsy (PSP). Typical finding in PSP is bilateral hypometabolism of mesial frontal and premotor/prefrontal areas. Thalamic and midbrain hypometabolism is usually also present. In line with overlapping pathologies in FTD and PSP, patients with clinical FTD can show a very PSP-like pattern, and vice versa. Upper panel, Transaxial PET images of [¹⁸F]FDG uptake. Lower panel, Results of voxel-based statistical analysis using Neurostat/3D-SSP. Given are right and left lateral and mesial views (see Fig. 33C.1 for additional details).

Fig. 33C.10 [¹⁸F]FDG PET in corticobasal degeneration (CBD). In line with the clinical presentation, CBD is characterized by a strongly asymmetrical hypometabolism of frontal and parietal areas, striatum, and thalamus. Most pronounced hypometabolism is often found in parietal lobe, but patients with the clinical syndrome of CBD may also show a one-sided PSP-like pattern on [¹⁸F]FDG PET. Upper panel, Transaxial PET images of [¹⁸F]FDG uptake. Lower panel, Results of voxel-based statistical analysis using Neurostat/3D-SSP. Given are right and left lateral and superior views (see Fig. 33C.1 for additional details).

Few large-scale studies have explored the diagnostic value of [¹⁸F]FDG PET for differential diagnosis of uncertain parkinsonism. In one study, qualitative readings assisted by voxel-based statistical analyses were used to differentiate between PD, MSA, PSP, CBD, and healthy controls, using clinical diagnosis after a 2-year follow-up as reference. Disease-characteristic templates were used for data interpretation, which were generated by direct comparison of subsets of patients and controls (diagnostically most relevant features: PD, hypermetabolism in putamen; MSA, hypometabolism in putamen and cerebellum; PSP, hypometabolism in brainstem and midline frontal cortex; CBD, cortex and basal ganglia hypometabolism contralateral to the most affected body side) (Eckert et al., 2005). The following diagnostic accuracies were achieved in the individual groups: 97.7% in early PD, 91.6% in late PD, 96.0% in MSA, 85.0% in PSP, 90.1% in CBD, and 86.5% in controls. Except for the sensitivity in classifying PSP (85%) and controls (86%), sensitivity and specificity exceeded 90% in all other groups (Eckert et al., 2005). In a subsequent very recent study by the same group (Tang et al., 2010a), an automated image-based classification procedure relying on spatial covariance analyses was used to classify a large cohort of clinically diagnosed patients with PD, MSA, and PSP (follow-up > 6 months; on average, 2.6 years). The authors used a two-level analysis (first level: APS versus PD; second level: MSA versus PSP) to automatically classify patients according to disease-related metabolic patterns (see earlier discussion). Again, a high diagnostic accuracy was achieved in classifying PD (sensitivity/specificity 84%/97%), MSA (85%/96%), and PSP (88%/94%) (Tang et al., 2010a). However, it has to be acknowledged that this does not necessarily reflect the clinical situation. A typical clinical population also includes patients with CBD and possibly also patients with DLB, AD, and FTD, and finally, a relevant fraction of subjects without neurodegenerative parkinsonism (if not excluded beforehand by a preceding [¹²³I]FP-CIT SPECT or [¹⁸F]FDOPA PET). Thus, an automated classification approach will probably result in a less optimal outcome in a typical clinical population. The overall classification accuracy in a four-class categorization (76%-84%; PD, MSA, PSP, and controls) was inferior to a three-class categorization (92%-94%; PD, MSA, and PSP only) using an automated classification based on principal component analysis (Spetsieris et al., 2009). On the other hand, an experienced observer will probably correctly recognize the majority of “unexpected” cases as such (e.g., AD and non-neurodegenerative patients).

As mentioned, an important caveat exists regarding the use of the clinical diagnosis of MSA, PSP, and even more so, CBD (Ling et al., 2010) as gold standards in diagnostic trials. In particular, the clinical distinction of CBD from PSP is questionable, since PSP is a common clinical misdiagnosis of pathologically verified CBD and a common pathological substrate of clinical CBD (Ling et al., 2010; Wadia and Lang, 2007). This issue gets even more complex if one considers that FTLD is often caused by PSP and CBD pathology (Kertesz et al., 2005). Thus, autopsy-confirmed imaging studies are ultimately needed to unravel this uncertainty and validate the use of [¹⁸F]FDG PET in parkinsonism.

Brain Tumors

Imaging with [¹⁸F]FDG PET or PET/CT is a well-established and often indispensable modality for diagnosis, staging, treatment monitoring, and follow-up of oncological patients with various malignancies of virtually all organs. Imaging of brain tumors was actually the first oncological application of [¹⁸F]FDG PET (Chen, 2007; Di Chiro et al., 1982; Herholz et al., 2007; Patronas et al., 1982). The degree of [¹⁸F]FDG uptake in brain tumors correlates with the grade of malignancy. The basis of this observation is not fully understood. As with other malignancies, both increased hexokinase activity and an increased glucose transport into the cell have been proposed. The use of [¹⁸F]FDG PET in brain tumors was recently reviewed by a National Comprehensive Cancer Network (NCCN) panel. Based on current lower-level evidence (lack of randomized studies) and panel consensus (corresponding to category 2A), a role for [¹⁸F]FDG PET in managing brain tumors was proposed for diagnosis/staging, restaging/recurrence, prognosis, and possibly also for treatment planning/response monitoring (Podoloff et al., 2009).

The use of [¹⁸F]FDG PET in brain tumor imaging is complicated by the high physiological uptake of [¹⁸F]FDG in gray matter. Thus, unlike tumors of other organs that are usually visualized as areas of intense [¹⁸F]FDG uptake compared to the surrounding tissue, brain tumors with little [¹⁸F]FDG uptake present as uptake deficits in cortex or are hardly discernable when localized in white matter. In turn, tumors with high [¹⁸F]FDG uptake may be masked by physiological [¹⁸F]FDG uptake of the cortex. Thus, co-registration of the [¹⁸F]FDG PET scan to a recent MRI scan is mandatory for sufficient [¹⁸F]FDG PET interpretation to accurately delineate the area of interest (Borgwardt et al., 2005; Chao et al., 2001; Wong et al., 2004). This is of particular importance in tumors with low and heterogenous uptake as is often the case after therapy. Other PET radiotracers like 3’-deoxy-3’-¹⁸F-fluorothymidine ([¹⁸F]FLT; a marker of cell proliferation/DNA synthesis) or [¹¹C]methionine and O-(2-[¹⁸F]fluoroethyl)-l-tyrosine ([¹⁸F]FET; markers of amino acid transport into the cell) offer the advantage of very low physiological brain uptake. The radiotracers [¹¹C]methionine and O-(2-[¹⁸F]fluoroethyl)-l-tyrosine ([¹⁸F]FET) have been particularly well evaluated and represent very promising alternatives to [¹⁸F]FDG. They enable a very high lesion–to–normal brain contrast, particularly in high-grade gliomas but also in the majority of low-grade gliomas without contrast enhancement on MRI. Thus, they are particularly well suited for tumor delineation for treatment planning (surgery and radiation therapy) and biopsy guidance, comparing favorably with [¹⁸F]FDG (Goldman et al., 1997; Kaschten et al., 1998; Pauleit et al., 2009; Pirotte et al., 2004). Although some studies suggests that [¹⁸F]FET may also be a suitable tracer of brain tumor grading and therefore of prognostic relevance, this is still a matter of debate, particularly in comparison to [¹⁸F]FDG. Finally, the aforementioned tracers appear to be very suitable for the detection of persistent or recurrent brain tumors after therapy. However, given the limited availability of these tracers in routine clinical practice, we direct the interested reader to recent reviews (Chen, 2007; Herholz et al., 2007; Langen et al., 2006).

From the very early days of [¹⁸F]FDG PET scanning, it was known that high-grade gliomas (World Health Organization [WHO] grade III-IV) show a significantly higher [¹⁸F]FDG uptake than low-grade gliomas (WHO grade I-II) (Di Chiro et al., 1982). Low-grade gliomas usually have an [¹⁸F]FDG uptake that is below or roughly comparable to white matter (Fig. 33C.11), while high-grade gliomas commonly exhibit an [¹⁸F]FDG uptake that is distinctly higher than white matter. In fact, glioblastomas (WHO grade IV) commonly show areas of increased [¹⁸F]FDG uptake similar to or above that of gray matter, while uptake of anaplastic gliomas (WHO grade III) often falls below gray matter but noticeably above white matter (Fig. 33C.12). Several studies reported a high accuracy of [¹⁸F]FDG PET in differentiating low- and high-grade brain tumors, with a diagnostic sensitivity and specificity ranging from 84% to 94% and 77% to 95%, respectively (Delbeke et al., 1995; Meyer et al., 2001; Padma et al., 2003). Qualitative readings were found to be at least as accurate as semiquantitative region-of-interest analyses (Meyer et al., 2001), and such differentiation is also possible in gliomatous and nongliomatous tumors (e.g., meningioma, angioma, etc.) (Delbeke et al., 1995; Meyer et al., 2001). Common causes of false-positive [¹⁸F]FDG PET scans mimicking malignancy include brain abscesses, pituitary adenomas, juvenile pilocytic astrocytomas, choroid plexus papillomas, and gangliogliomas. High [¹⁸F]FDG uptake in benign brain tumors is particularly often found in children, but [¹⁸F]FDG PET has nevertheless been demonstrated to be a powerful method for tumor grading of childhood CNS tumors (Borgwardt et al., 2005). Finally, cerebral lymphomas usually possess extraordinary high [¹⁸F]FDG uptake, making [¹⁸F]FDG PET a powerful method to detect cerebral lymphoma (Fig. 33C.13). This can be particularly useful in differentiating nonmalignant CNS lesions in patients with acquired immunodeficiency syndrome (AIDS) (Hoffman et al., 1993).

Fig. 33C.11 [¹⁸F]FDG and [¹⁸F]FET PET in a left frontal low-grade oligodendroglioma (WHO grade II). [¹⁸F]FDG uptake (middle) of low-grade gliomas is usually comparable to white-matter uptake, prohibiting a clear demarcation of tumor borders. In contrast, the majority of low-grade gliomas (particularly oligodendroglioma) show intense and well-defined uptake of radioactive amino acids like [¹⁸F]FET (right) even without contrast enhancement on MRI (left).

(Courtesy Karl-Josef Langen, MD, Institute of Neuroscience and Medicine, Research Center Juelich, Germany.)

Fig. 33C.12 [¹⁸F]FDG and [¹⁸F]FET PET in a right mesial temporal high-grade astrocytoma (WHO grade III). In contrast to low-grade gliomas, high-grade tumors usually have [¹⁸F]FDG uptake (middle) that is distinctly higher than white matter and sometimes even above gray matter, as in this case. Nevertheless, the [¹⁸F]FET scan (right) clearly depicts a rostral tumor extension that cannot be identified by [¹⁸F]FDG, owing to high physiological [¹⁸F]FDG uptake by adjacent gray matter. Tumor delineation is clearer on [¹⁸F]FET PET than on MRI (left).

(Courtesy Karl-Josef Langen, MD, Institute of Neuroscience and Medicine, Research Center Juelich, Germany.)

Fig. 33C.13 [¹⁸F]FDG and [¹⁸F]FET PET in cerebral lymphoma. Cerebral lymphoma usually shows very intense [¹⁸F]FDG uptake (middle), while metabolism of surrounding brain tissue is suppressed by extensive tumor edema (see MRI, left). [¹⁸F]FET uptake (right) of cerebral lymphoma can also be high.

(Courtesy Karl-Josef Langen, MD, Institute of Neuroscience and Medicine, Research Center Juelich, Germany.)

It has been shown that the degree of [¹⁸F]FDG uptake strongly predicts overall survival of patients with gliomas. Patients with increased tumor [¹⁸F]FDG uptake had a significantly shorter survival (Alavi et al., 1988; Kim et al., 1991; Patronas et al., 1985). Importantly, this was also true within histological groups of high-grade tumors (in particular glioblastoma multiforme), indicating that [¹⁸F]FDG PET provides prognostic information beyond histological grading and may represent a more accurate marker of tumor aggressiveness (De Witte et al., 2000; Hölzer et al., 1993; Kim et al., 1991; Patronas et al., 1985). In a more recent large retrospective analysis, 94% of patients with low uptake (less or equal to white matter) survived for more than 1 year (median survival 28 months), while only 29% of patients with higher uptake (more than white matter) survived for more than 1 year (median survival 11 months) (Padma et al., 2003). Patients with primary brain tumors that exhibited persistent low [¹⁸F]FDG uptake on follow-up studies had a significantly longer survival than those with increased [¹⁸F]FDG uptake (Schifter et al., 1993). This is of particular relevance for follow-up and management of patients with low-grade gliomas that commonly show a malignant transformation during follow-up. Earlier studies demonstrated that malignant transformation of low-grade glioma is accompanied by focal hypermetabolism (Francavilla et al., 1989), which is associated with a considerably worse prognosis (De Witte et al., 1996). Thus, the diagnostic and prognostic information gained from [¹⁸F]FDG PET can be of high therapeutic relevance (e.g., by selecting a more aggressive treatment).

Based on the observation that tumor areas with the highest [¹⁸F]FDG uptake represent the most aggressive portions of the tumor, [¹⁸F]FDG PET can also be used to direct stereotactic biopsies and tumor resections. It was demonstrated that incorporation of [¹⁸F]FDG PET in biopsy planning considerably increased the diagnostic yield of biopsies, particularly in high-grade tumors with heterogenic tissue composition (Goldman et al., 1997; Pirotte et al., 1994). Furthermore, [¹⁸F]FDG and [¹¹C]methionine PET was found to be helpful in targeting the resection to hypermetabolic/anaplastic tumor areas in high-grade gliomas (Pirotte et al., 2006). Complete removal of hypermetabolic tissue was associated with an increased survival in anaplastic gliomas and glioblastomas, whereas a total resection of MRI contrast-enhancing areas was not (Pirotte et al., 2009). With the same rationale, information on active tumor tissue from [¹⁸F]FDG PET was also incorporated into radiation treatment planning (Douglas et al., 2006; Gross et al., 1998; Solberg et al., 2004; Tralins et al., 2002). Based on the few available studies, however, neither tumor volume definition nor targeting dose escalation by [¹⁸F]FDG PET had a noticeable impact on survival of patients with high-grade gliomas, compared to literature data and historical controls (Douglas et al., 2006; Gross et al., 1998). However, volumes of abnormal [¹⁸F]FDG uptake were of higher predictive value regarding time to progression and survival than MRI (Tralins et al., 2002).

Another important, albeit controversial, use of [¹⁸F]FDG PET imaging in brain tumors is the evaluation of persistent or recurrent disease—in particular, differentiation from radiation necrosis. In earlier studies, the sensitivity and specificity of differentiating recurrent brain tumors (including gliomas and metastases of various origins) from radiation necrosis was 80% to 100% and 40% to 100%, respectively (Langleben und Segall, 2000). False-negative findings may result from very recent radiation therapy, pretreatment low FDG uptake (e.g. in low-grade gliomas or metastases with low FDG avidity), and small tumor volumes (due to partial volume effects). Conversely, intense inflammatory reaction after (stereotactic) radiation therapy and seizure activity (possibly unnoticed) may result in false-positive PET findings. Earlier studies were done without PET/MRI co-registration, which can be a relevant source of error. Since [¹⁸F]FDG uptake in recurrent brain tumors and neighboring tissue is often lower than the uptake before radiation therapy, it is crucial to evaluate tumor uptake with reference to the expected background uptake in adjacent brain tissue (Chen, 2007). This is hardly possible without accurate anatomical information from MRI. In more recent reports, the sensitivity and specificity for detecting recurrent brain tumors was reported to be 76% to 78% and 95% to 96%, respectively, for gliomas (low and high grade) (Gómez-Río et al., 2008; Wang et al., 2006) and 82% to 86% and 80% to 97%, respectively, for brain metastases (Chao et al., 2001; Wang et al., 2006). Despite these fairly good results, the use of [¹⁸F]FDG PET for the purpose of detecting recurrent brain tumors remains controversial, and PET studies with amino acid tracers are probably superior (Fig. 33C.14).