Summary of Existing Health Technology Assessments
describes the health technology assessments and systematic reviews identified that reviewed functional brain imaging for AD, brain tumours, epilepsy, and Parkinson’s disease. There were no health technology assessments or systematic reviews identified that reviewed functional brain imaging for patients with suspected or confirmed MS. Most health technology assessments and systematic reviews investigated the use of PET; 4 reviews specifically reviewed MRS (3 for patients with suspected brain tumours, and 1 for the differential diagnosis of parkinsonian syndromes), and 1 health technology assessment reviewed the use of MEG in patients with epilepsy.
Description of Systematic Reviews and Health Technology Assessments of Functional Brain Imaging*
There were 4 health technology assessments identified that investigated the role of PET in patients with epilepsy: one by the National Health Service (NHS) in the United Kingdom (9
), one by the Agence d'Évaluation des Technologies et des Modes d'Intervention en Santé, (AETMIS) in Canada (29
) and 2 by the Medical Services Advisory Committee (MSAC) in Australia. (30
The most recent review by the NHS (9
) in 2006 incorporated various neuroimaging technologies including PET, MRS, and SPECT. It did not include MEG or fMRI. Based on their thorough review of the literature, they concluded that the limitations of the neuroimaging studies identified “...did little to inform clinical practice.”
The first health technology assessment by MSAC (31
) was published in 2000, and reviewed PET for various indications, including epilepsy. Based on this review they concluded that some patients with refractory epilepsy may benefit from presurgical PET scans; however, it was unclear if PET scans could benefit all patients with refractory epilepsy. In 2004, MSAC (30
) published a health technology assessment on the use of PET specifically in patients with epilepsy. Despite a lack of high-quality evidence and substantial limitations with the available evidence (including inconsistency in defining a reference standard and assuming that PET results alter patient management without evidence), MSAC concluded that PET scans are clinically useful in patients with refractory epilepsy, where there are inconsistent results on structural imaging and EEG.
The 2001 health technology assessment by AETMIS (29
) made a similar conclusion to MSAC.
In addition to the health technology assessments presented in , in 1999 INAHTA (32
) published a review of systematic reviews investigating the role of PET for various conditions and indications. They grouped the systematic reviews into 3 main categories: neuropsychiatry, cardiology and oncology (non-CNS tumours). The category of neuropsychiatry included AD, PD, epilepsy, brain tumours, cerebrovascular disorders, other neurodegenerative disorders, and other neuropsychiatry conditions. Between 1990 and 1999, they identified 13 systematic reviews by 10 health technology assessment organizations investigating one or a combination of the aforementioned neuropsychiatry conditions. The INAHTA review reported that most of the studies included in the systematic reviews used 18F-labelled-fluorodeoxyglucose (FDG)-PET to study glucose metabolism.
Three of the systematic reviews identified included studies of patients with brain tumours. According to the INAHTA review, the 3 systematic reviews were not able to demonstrate an added benefit of PET in the management of patients with brain tumours. Based on the limited evidence available, they reported that SPECT was superior to PET, while CT and MRI were inferior to PET for the differentiation between brain tumour recurrence and radiation necrosis. There was a paucity of high-quality controlled studies available to review effectively the evidence on the use of PET in patients with suspected brain tumours.
The INAHTA review identified 8 health technology assessments and systematic reviews on the clinical utility of PET in the management of epilepsy. Diagnostic imaging is most frequently used in patients with intractable epilepsy undergoing presurgical evaluation to identify the epileptogenic foci and to determine resectability. Among the 8 reviews, there were conflicting reports on study quality; however, all of the reviews’ authors agreed that there was insufficient evidence to support replacing ICEEG or MRI with PET. The review reported that more high-quality evidence was required on the clinical utility of PET for epilepsy.
The review by Tolosa et al. (44
) summarized the usefulness of various strategies for diagnosing PD. They concluded that drug challenge and MRI were useful in routine practice for making differential diagnoses of parkinsonism, and that SPECT was useful in routine practice in the early diagnosis of PD. Positron emission tomography was deemed to be of limited clinical use at this time, and further research was required to attempt to establish a role for PET in the diagnosis of PD ().
Role of Diagnostic Techniques in Parkinson’s Disease*
Results: Brain Tumours
There were 3 questions this review asked with respect to functional brain imaging for brain tumours:
- What is the role of functional brain imaging in the detection of primary tumours?
- What is the role of functional brain imaging in the differentiation of recurrence from radiation necrosis?
- What is the role of functional brain imaging in the selection of surgical candidates for tumour resection?
What is the Role of Functional Brain Imaging in the Detection of Primary Tumours?
In all of the studies identified that investigated the role of functional brain imaging in the detection of primary tumours, a suspicious lesion had already been identified; thus, the role of functional brain imaging was to determine if the lesion was benign or malignant. In most cases the reference standard was biopsy; however, in some cases, postoperative follow-up was used as the reference standard.
Six studies were identified that met the inclusion criteria for this review and that had been published since the most recent health technology assessments on functional brain imaging for brain tumours. Five investigated the role of PET using various radiotracers (18F-FDG, 18F-FLT, 11C-MET), and the other study investigated PET and MRS. describes the characteristics of these studies. It is important to note that the half-lives for the radiotracers are variable, making some more appropriate for clinical practice than others. ( lists the half-lives of commonly used radiopharmaceuticals in PET scans)
Characteristics of Studies Investigating the Detection and Grading of Primary Tumours With Functional Brain Imaging*
Only the study by Pauleit et al. (52
) blinded the clinicians from the patients’ clinical histories. All of the studies established tumour diagnosis with biopsy or resection. Not all of the patients in all of the studies had suspected primary tumours, because there were patients in some of the studies who had been treated previously.
In the prospective study by Wang et al., (50
) patients were included if they had positive but inconclusive MRI or CT results. The majority of patients in their study had brain metastases (n = 78) compared with primary brain tumours (n = 39); however, since they stratified the results for primary diagnoses and metastases, this study was eligible for inclusion.
The results of the studies investigating functional brain imaging for diagnosing malignant versus benign brain tumours are listed in . The thresholds defined by the studies are variable. The thresholds in the PET studies are ratios between normal tissue uptake of a radiotracer versus abnormal uptake. The thresholds in MRS studies are ratios between the metabolites.
Results of Studies Investigating Functional Brain Imaging for Diagnosing Malignant Versus Benign Lesions*
Two of the studies relied on visual inspection to determine if a tumour was present, while all of the other studies defined quantitative thresholds. The threshold in the study by Wang et al. (50
) was by visual inspection; however, they stated that if the FDG uptake in or near the lesion was lower than the surrounding tissue, the PET scan was considered to be negative. Alternatively, if the FDG uptake was higher in or near the lesion, the PET scan was considered positive. The study by Choi et al. (51
) also made similar specifications for their visual inspection of images with FDG and FLT.
In all studies, most patients underwent MRI in addition to the MRS or PET scan. Only 2 of the studies compared the accuracy of MRS or PET and MRI, with MRI alone. (52
) Gadolinium-enhanced MRI is the gold standard for the diagnosis of primary tumours (personal communication, clinical expert, November 15, 2006), thus a comparison to Gd-enhanced MRI is very worthwhile to establish if the results of the MRS or PET scans improve the accuracy of tumour diagnosis.
In the prospective study by Floeth et al., (61
) patients with suspected primary tumours underwent MRI (Gd-enhanced), MRS, and FET-PET. Ninety-one patients received FET-PET; 50 of these also received a MRS analysis. Of the 50 patients undergoing PET, MRS, and MRI, 34 (68%) had malignant tumours confirmed after biopsies. Compared with MRS and FET-PET, Gd-enhanced MRI alone had lower sensitivity and specificity. Floeth et al. did not report if this difference in sensitivity and specificity was significant; however, they did report that by adding MRS and FET-PET to MRI the accuracy increased from 68% for MRI alone to 97% for MRS, FET-PET, and MRI. They did not report the accuracy of MRS and MRI without FET-PET, nor did they report the accuracy of FET-PET and MRI without MRS. There are some limitations of this study, including that the 2 neurosurgeons who reviewed the test results were not blinded, and that not all patients received the same intervention. It was unclear why some patients had MRS analyses and others did not.
The prospective study by Pauleit et al. (52
) studied the use of FET-PET in 31 newly diagnosed patients with primary gliomas. They reported that compared with MRI (Gd-enhanced) alone, combining FET-PET with MRI (Gd-enhanced) resulted in a similar sensitivity (96% for MRI alone, 93% for FET-PET + MRI) and higher specificity (53% for MRI alone, 94% for FET-PET + MRI) for tumour diagnosis. The 3 observers who reviewed the results were blinded to the clinical information of the patients included in the study. Twenty (71%) of 28 patients analyzed (3 patients were excluded from the analysis) had malignant tumours confirmed by biopsies.
groups the results for sensitivity and specificity by imaging modality and radiotracer and threshold. In both 11C-MET PET and 18F-FET PET, sensitivities and specificities are consistently high.
Sensitivity and Specificity for Functional Brain Imaging for the Diagnosis of Brain Tumours*
Unfortunately, none of these authors commented on whether the accurate diagnosis of brain tumours had an impact on treatment or clinical outcomes in the patients in these studies. Patients with primary tumours are likely to have biopsies despite imaging results (personal communication, clinical expert, December 5, 2006).
What is the Role of Functional Brain Imaging in the Differentiation of Recurrence From Radiation Necrosis?
There were 6 studies identified since the most recent health technology assessments were published that examined the use of functional brain imaging in the differentiation of recurrence and radiation necrosis (3 RS’ and 3 prospective studies). Four used MRS and the other 2 used PET (18F-FDG, 18F-FDOPA and 18F-FET). describes the characteristics of the studies.
Characteristics of Studies Investigating the Differentiation of Recurrence From Radiation Ncrosis Using Functional Brain Imaging*
outlines the results reported by each of the 6 studies identified that used MRS or PET to differentiate recurrence from radiation necrosis. Similar to the studies identifying primary brain tumours, various thresholds were used to determine presence of recurrence.
Results of Studies Investigating Functional Brain Imaging for Brain Tumours*
The study by Rachinger et al. (56
) compared FET-PET with MRI (with and without Gd enhancement) to MRI alone in 45 patients with suspected tumour recurrence. They found that MRI with FET-PET had higher specificity than MRI alone (93% versus 50%, respectively). Sensitivity was also higher for PET with MRI versus MRI alone (100% versus 94%). Rachinger et al. reported that the difference in accuracy between PET and MRI versus MRI alone was significant (P
≤.05). It is unclear whether the reviewers of the PET and MRI results were blinded to the patients’ clinical information.
The study by Lichy et al. (57
) was the only study identified that compared MRS with Gd-enhanced MRI to Gd-enhanced MRI alone. They reported the highest sensitivity and specificity when MRS was combined with T1 (Gd-enhanced) and T2 MRI (100% and 86%, respectively); however, T1 and T2 MRI also had high sensitivity and specificity without MRS (94% and 86%, respectively). It is unclear if the added benefit of MRS in terms of sensitivity is significant.
The sensitivity and specificity in the study by Sundgren et al. (58
) were calculated by the Medical Advisory Secretariat using the following assumptions:
- True positive: Patients with recurrence on MRS confirmed through clinical, neuroradiologic or neuropathologic follow-up.
- True negative: Patients with radiation injury on MRS confirmed through clinical, neuroradiologic or neuropathologic follow-up.
- False positive: Patients with recurrence or inconclusive results on MRS but radiation injury confirmed through clinical, neuroradiologic or neuropathologic follow-up.
- False negative: Patients with radiation injury on MRS but recurrence confirmed through clinical, neuroradiologic or neuropathologic follow-up.
Based on these assumptions, the sensitivity and specificity for MRS in this study were 88% and 70%, respectively.
groups the results for sensitivity and specificity by imaging modality and radiotracer and threshold. Both of the studies by Palumbo et al. (60
) and Lichy et al. (57
) reported accuracy for MRS using the Cho/Cr ratio threshold of 2.0. The study by Palumbo et al. reported a higher specificity than sensitivity (90% sensitivity, 100% specificity) while the study by Lichy et al. reported a higher sensitivity than specificity (81% sensitivity, 71% specificity). The study by Weybright et al. (59
) also reported accuracy of MRS using Cho/Cr ratio; however, they used a different threshold (1.8), but they reported similar results to Palumbo et al. (94% sensitivity and 100% specificity).
Sensitivity and Specificity for Functional Brain Imaging for the Diagnosis of Brain Tumours*
Rachinger et al. reported the highest sensitivity (100%) using 18F-FET-PET. The highest specificity (100%) was reported by Palumbo et al. and by Weybright et al. for MRS using the Cho/Cr ratio.
As in those studies that investigated the use of PET and MRS in patients with primary tumours, none of these studies reported how the accuracy of diagnosis had an impact on treatment or clinical outcomes.
What is the Role of Functional Brain Imaging in the Selection of Surgical Candidates for Tumour Resection?
There is considerable evidence indicating that fMRI and MEG can accurately identify the sensorimotor complex; (88
) however, the impact of this on surgical outcomes is not reported in the studies identified. The role of fMRI is to identify the language, sensory, and motor areas of the brain, with the aim of establishing whether tumour resection will affect language, sensory, or motor function. The primary purpose of fMRI is not to identify tumours, as it is with PET and MRS.
One study, by Ganslandt et al., (98
) reported on surgical outcome after fMRI or MEG, but did not provide a comparison of what the surgical outcome would have been without fMRI or MEG.
Two studies (85
) were identified that described the role of functional brain imaging in the selection of surgical candidates for tumour resection. Both investigated the role of fMRI in the selection of patients for surgery.
The case-control study by Winkler et al. (86
) compared surgical outcomes in patients with brain tumours who had preoperative imaging with fMRI (n = 49) with those who underwent surgery without preoperative fMRI (historical controls, n = 55). In both groups there were patients with meningiomas (18 in experimental group, 19 in control group), metastases (9 in experimental group, 19 in control group), and stage II to IV gliomas (22 in experimental group, 17 in control group). The mean age of the patients was 53.9 years (range, 14–78 years) in the experimental group, and 52.3 years (range 21–76 years) in the control group.
At 6 months postoperatively, patients were categorized as “improved,” “unchanged,” or “deteriorated” in terms of neurological function; tumour outcome results were not reported. outlines the results. Winkler et al. reported the results separately for the various tumour types.
Outcomes of Patients With Preoperative Functional Magnetic Resonance Imaging Versus Without Preoperative Functional Magnetic Resonance Imaging 6 Months Postoperatively*
There appears to be more inconsistency across outcomes in the patients with gliomas compared with the patients with metastases or meningiomas. However, Winkler et al. reported that there were no statistically significant differences in neurological function outcomes between patients who had undergone preoperative fMRI and those who had not. However, due to small sample subsets the study may not have been powered to detect a significant difference in outcomes (Type II error).
This study suffered from several limitations. There was no blinding, nor was there a sample size calculation reported explaining how the authors chose their sample size. There was also limited information regarding eligibility criteria and whether the patients were enrolled consecutively. Nonetheless, this was the only study identified that compared outcomes after surgery based on fMRI.
In the prospective cohort study by Petrella et al., (85
) 39 patients with potentially resectable tumours were imaged preoperatively with fMRI. Three neurosurgeons completed a questionnaire regarding the treatment plan for the patients before and after seeing the fMRI results. Petrella et al. did not report what clinical information was provided to the neurosurgeons before the fMRI to develop treatment plans. It is important to note that the role of fMRI is to identify the sensorimotor cortex, not to identify tumours. describes the neurosurgeons’ responses before and after fMRI.
Responses of Neurosurgeons Before and After Functional Magnetic Resonance Imaging Results Were Analyzed*
Before fMRI the neurosurgeons thought that 9 patients were not suitable for any surgical intervention. After fMRI, the treatment plan for 7 of the 9 patients changed (2 biopsy, 5 craniotomy with mapping). All of the 8 patients who had a biopsy as their treatment plan before fMRI had their treatment plan changed after fMRI. All 8 patients were changed to craniotomy (7 with intraoperative mapping, 1 under general anesthesia). Overall, the treatment plan was altered in 19 of 39 patients (49%; 95% CI, 33%–64%), with 18 of the 19 treatment plans being more aggressive after fMRI.
In addition to the treatment plan, 4 patients also had Wada tests recommended prior to fMRI. After fMRI none of the patients required this invasive test.
It was the clinical impression of the neurosurgeons in the study by Petrella et al. that fMRI shortened surgical time by an estimated 15 to 60 minutes in 60% of patients. However, they did not report how this was calculated.
There is no long-term follow-up reported for the patients in the study by Petrella et al. (85
) The authors reported that the “[a]ctual intervention agreed with the treatment plan after functional MR imaging in all 39 patients.”
In summary, fMRI and MEG can accurately identify the sensorimotor cortex, as has been demonstrated in several studies. According to the study by Winkler et al., (86
) there is no difference in surgical outcome in patients who have had preoperative fMRI compared with those who have not. However, this was a low-quality study that may not have been powered to detect a significant difference between groups. The study from Petrella et al. (85
) demonstrated that the treatment plan for patients changed with the information provided with preoperative fMRI results and that preoperative fMRI may also decrease surgery time.
What is the Role of Functional Brain Imaging in the Localization of Seizure Foci?
As described earlier in this review, the current gold standard for localizing seizure foci includes video scalp EEG (which is noninvasive, but time consuming, usually involving taking patients off their medications and several days in hospital), MRI, and neuropsychological evaluation. If the results are inconclusive, then invasive ICEEG is considered. The literature has identified 2 possible roles for functional brain imaging in localizing seizure foci:
- an imaging test to be used instead of or in addition to noninvasive EEG, and
- an imaging test to replace ICEEG.
Functional Brain Imaging Versus Intracranial Electroencephalogram
Intracranial electroencephalogram is the gold standard for localizing seizure foci in patients with inconclusive noninvasive results. Based on 3 studies (described in the Alternative Technologies section of this review), (26
) the range of surgical success (seizure-free or significant improvement in seizures) based on ICEEG is from 57% to 89%.
There were 3 studies identified that investigated the localization of seizure foci with MEG compared with ICEEG. An additional 3 studies from the Blue Cross Blue Shield health technology assessment (43
) are also described below to enhance the results of the newly identified studies. In addition, since the Blue Cross and Blue Shield health technology assessment reported that due to insufficient evidence they could not recommend using MEG, the Medical Advisory Secretariat wanted to further investigate previous MEG studies in order to draw some conclusions from the results of the studies on MEG. The Medical Advisory Secretariat extracted the 3 studies from the Blue Cross Blue Shield health technology assessment that included 20 or more patients and compared MEG to ICEEG relative to surgical outcomes. There were no studies identified comparing other functional brain imaging modalities besides MEG to ICEEG. describes the characteristics of these studies.
Characteristics of Studies Investigating the Localization of Seizure Foci With Functional Brain Imaging Instead of Intracranial Electroencephalogram*
The PCS by Knowlton et al. (71
) was designed to address whether MEG can replace ICEEG in the presurgical evaluation of patients with refractory epilepsy. This study has clinical relevance because it compared MEG to ICEEG, the current gold standard. When MEG was compared with ICEEG and surgical outcome, the values for sensitivity and specificity were 75% and 70%, respectively. They also reported subgroup analyses, the results of which suggested that MEG may be less beneficial compared with ICEEG in patients with extratemporal lobe epilepsy (ETLE) compared with those with temporal lobe epilepsy (TLE). This result was based on a subset analysis, and the study was not designed to compare differences between ETLE and TLE.
Knowlton et al. (71
) concluded that MEG could potentially replace the invasive ICEEG. However, there were 2 letters to editor (103
) regarding this study which stated that the results by Knowlton et al. are promising, but must be considered with caution for the following reasons:
- Knowlton et al. (71) did not clearly differentiate neocortical from temporolimbic epilepsy (although the inclusion criteria would have limited the number of patients with bilateral temporolimbic epilepsy).
- For patients with temporolimbic epilepsy, the need to identify the intracranial epileptic (ictal) onsets is more important than the identification of interictal spikes captured by MEG, thus MEG does not aid in the surgical planning of patients with temporolimbic epilepsy.
- Potential for bias in favour of MEG, in cases where resection success was primarily due to ICEEG but included MEG dipoles in the planned resection area.
- In 7 of 49 (14%), cases were localized by ICEEG, but not MEG. Three (6%) cases were localized by MEG, but not by ICEEG.
The small retrospective study by Oishi et al. (78
) compared MEG to ICEEG in 20 patients with medically refractory neocortical epilepsy. Based on postoperative outcomes using Engel’s classification, (105
) they reported that MEG results were significantly correlated with surgical outcomes in patients in which MEG detected single clusters compared with patients in which MEG detected multiple clusters (P =
.049). Patients with single clusters were more likely to be seizure-free after surgery than patients with multiple clusters detected on MEG. There was also a significant correlation between the results for MEG and ICEEG (P =
.014). However, it is important to note that patients with multiple foci are less likely to benefit from surgical resection compared with patients with localized foci.
In the prospective study by Gallen et al., (102
) 33 patients underwent presurgical evaluation for the resection of seizure foci. After the presurgical evaluation, 4 patients were considered inappropriate surgical candidates. Of the remaining 29, 21 (72%) were seizure-free at a mean follow-up of 42.4 months (standard deviation [SD], 8.9 months; range 25–57 months). Twenty-one patients required ICEEG monitoring prior to surgery. They reported on sensitivity and specificity for MEG, ICEEG, noninvasive EEG (video monitoring), neuropsychological testing, and MRI. They reported higher sensitivity and specificity for ICEEG compared with all the other modalities, including MEG.
The retrospective study by Papanicolaou et al. (75
) compared ICEEG to MEG in 41 patients with refractory epilepsy. Surgical focal resection was based on the results of the ICEEG, MRI, neuropsychological evaluation, SPECT, PET, and Wada testing. Magnetoencephalography and ICEEG results were compared with postoperative outcomes (using the Wieser classification (106
)). Compared with postoperative outcomes, MEG was correct in 23 (56%) of 41 patients, and ICEEG was correct in 22 (54%) of 41 patients. Predictions were incorrect for 15 patients with MEG and 16 patients with ICEEG, and there were 3 indeterminate cases for each MEG and EEG.
All 5 of the studies reported concordance or agreement with the resected area for both MEG and ICEEG. In all of the studies, the decision as to where to resect was based on a multitude of tests, including MRI, noninvasive EEG, neuropsychological testing, and ICEEG. Some patients also underwent PET, SPECT, and/or Wada testing. Only in the study by Knowlton et al. could the MEG results impact the decision as to where to resect; however, MEG could only add to increase the selected area, it could not change the area to resect. summarizes the agreement of MEG and ICEEG to the actual resected area.
Agreement Between Resected Area and Magnetoencephalography and Intracranial Electroencephalogram*
Based on the results of these studies, the agreement between MEG and the resected area ranged from 49% to 71%, and the agreement between ICEEG and the actual resected area ranged from 44% to 74%. It is important to note that 4 of the 5 studies reported that the rate of agreement between ICEEG and the resected area was between 68% and 74%. The study by Wheless et al. (101
) reported an agreement of only 44% for ICEEG. It is unclear why agreement was lower in this study compared with the others. If this study were removed, the range for MEG would remain the same (49%–71%) and the range for ICEEG would be improved (68%–74%).
In addition to the studies identified that described the accuracy of MEG in identifying the epileptic foci compared with ICEEG or noninvasive EEG, the retrospective study by Stefan et al. (83
) investigated the clinical use of MEG results. They reported on 455 patients who underwent MEG for refractory epilepsy. As part of this review, 2 expert clinicians reviewed the 104 patients’ charts and rated the contribution of the MEG results compared with other modalities used to localize the epileptic foci (video EEG, MRI, SPECT, MRI, etc.). The results were as follows:
- Disagreement between MEG and other modalities: 2 (2%) patients
- MEG results made no contribution: 10(10%) patients
- MEG results in agreement with other modalities: 56 (54%) patients
- MEG provided additional information: 25 (24%) patients
- MEG had influence on surgical procedures: 11(11%) patients
Based on the results of MEG compared with ICEEG and noninvasive EEG, it is difficult to draw conclusions regarding the effectiveness of MEG due to limitations in the evidence. The studies identified are mostly small case-series with heterogeneous samples. The patients in the studies have variable types of epilepsy, ages, duration of seizures, location of seizures, and variable presurgical evaluations. Moreover, the study designs and outcomes reported vary across studies.
Functional Brain Imaging Versus Noninvasive Electroencephalogram
There were 8 studies identified that investigated the localization of seizure foci with MEG or PET instead of or in addition to noninvasive (scalp) EEG. The characteristics of the studies are described in . Four of the studies compared MEG to EEG, and 4 compared PET to EEG. The 4 studies investigating the role of PET in the localization of seizure foci were conducted by the same core group of authors; thus, there may be overlap in the patients included in these studies. Patients were heterogeneous across the studies on several variables including age and seizure origin.
Characteristics of Studies Investigating the Localization of Seizure Foci With Functional Brain Imaging Instead of Noninvasive (Scalp) Electroencephalogram*
Due to the heterogeneity of the studies in the sample population, study design, and methods for reporting outcomes, the results of the studies could not be pooled. The results of the studies are described below.
The retrospective study by Iwasaki et al. (73
) compared the results of simultaneous noninvasive EEG and MEG recordings in 43 patients with intractable focal epilepsy. Patients underwent focal resection based on an independent diagnosis not related to the study. Three expert reviewers were blinded to the clinical information of patients.
In 31 patients, spikes were detected in both MEG and EEG; in 8 patients, spikes were only detected with MEG, and in 1 patient spikes were detected by EEG alone. There were 3 patients who did not have any spikes detected with either modality. The authors did not report if one modality was better at identifying spikes in certain locations. Iwasaki et al. reported that there was not a significant difference in the total number of spikes detected by MEG compared with EEG (P= .81). Thirty-five of 43 patients were reported to be seizure-free postoperatively. Of these patients, MEG detected interictal spikes in 32 (91%) compared with 27 (77%) patients with EEG alone. In the 32 patients with MEG spikes detected, 25 (78.1%) were localized to the resection site. In the 27 patients with EEG spikes detected, 23 (85.2%) were localized to the resection site.
The retrospective study by Ochi et al. (72
) compared the concordance between lateralization of MEG and noninvasive EEG with video monitoring in 41 children. They found there was concordance between MEG and EEG in 34 patients (83%). None of the patients with nonconcordant results underwent surgery.
In the prospective study by Assaf et al., (74
) video scalp EEG recordings were compared with MEG results in 26 patients with intractable TLE. Twenty-two patients underwent surgery. The surgeons were blinded to the MEG results. Surgical decisions were based on scalp EEG recordings, in addition to other presurgical analyses (MRI, neuropsychological testing), in 15 patients. The other 7 patients who had surgery also had ICEEG in addition to the video scalp EEG and other presurgical tests. The authors reported statistically significant correlation between MEG and surgical resections (P
< .001). Twenty-one of the 22 patients who underwent surgery were seizure-free postoperatively (duration of follow-up was not reported).
The prospective study by Pataraia et al. (77
) compared MEG to noninvasive video EEG. They included 113 patients with refractory epilepsy in their study, but they only reported results for 82 patients. Sixteen patients were excluded due to large magnetic artifacts in their MEG results, and another 15 patients were excluded because their results were damaged in a flood. The decision to undergo surgery was based on video EEG, MRI, neuropsychological evaluation, SPECT, PET and Wada testing. The MEG results were not considered in the surgical decision. Both the MEG and the video EEG results were compared with the actual resected region. There was agreement between MEG and EEG in 32.3% of cases. Seventeen patients (20.7%) had no spikes on MEG, and were excluded from the analysis. There was perfect overlap for MEG in 47 of 65 patients compared to the actual resected area and 26 of 65 patients with EEG. There were 5 patients where they had partial overlap or no overlap in MEG, but perfect overlap in EEG. Alternatively, there were 26 patients with partial overlap or were nonlocalizable in EEG, but had perfect overlap in MEG.
Pataraia et al. reported 1-year postoperative outcomes for the patients in their study based on the Wieser classification. (106
) A Wieser classification of 1 or 2 (no seizures or only aura) was considered to be success. A Wieser classification between 3 and 4 indicates that the patient has had a worthwhile improvement from seizures, and a classification of 5 indicates that there has been no or minimal improvement in the frequency of seizures. Based on data provided in the Pataraia et al. study, the Medical Advisory Secretariat created and to illustrate the postsurgical outcome based on MEG and video EEG results. Of the 43 patients that were seizure-free 1 year after surgery, MEG localized the resected area in 25 patients (58%), compared with 42% for EEG. However, for the 19 patients who experienced no or minimal improvement in seizures after 1 year, MEG agreed with the resected area in 63% of cases, compared with 32% for EEG. Based on these results, it would appear that MEG has a higher sensitivity than video EEG, but a lower specificity. Pataraia et al. concluded that MEG may have a role in cases where video EEG is partially or nonlocalizing.
Magnetoencephalography Results Compared With 1-Year Postoperative Outcome (Based on Wieser Classification)
Noninvasive Video-Electroencephalogram Results Compared With 1-Year Postoperative Outcome (Based on Wieser Classification)
Since the publication of the NHS health technology assessment on PET in patients with epilepsy, 4 studies were identified that investigated the accuracy of PET in patients with epilepsy. (62
) These 4 studies were conducted by the same core group of authors, thus there is likely overlap in the patients in the studies.
outlines the percentage of accurate localizations of seizure foci by imaging modality. Accuracy was based the imaging modality localizing the resected area and on patients being seizure-free 1 year postoperatively. Across the 4 studies, there was variability in the sample size and type of epilepsy that was included. Single photon emission computed tomography appeared to be the least accurate, however, when observing the ranges covered by each of the modalities in the 4 studies, there was overlap across all modalities in terms of accuracy rates. Thus, it is not possible to conclude that 1 modality is better than another at localizing seizure foci based on these 4 studies.
Percentage of Accurate Localizations by Modality (Based on 1-year Postoperative Outcomes)*
The study by Ollenberger et al. (82
) assessed the role of FDG-PET in the diagnosis and management of children with refractory epilepsy through surveying epileptologists about the management of their patients’ epilepsy. The purpose of the survey was to assess if the FDG-PET results changed patient management. Three epileptologists responded to the survey in reference to 108 patients. The mean age of patients was 7.3 years (SD, 3.4 years; range, 0.5–12.5 years). For surgical candidates, PET resulted in surgery being excluded (major change) in 39% of patients, and resulted in a modification of surgery (minor change) in 19% of patients. Positron emission tomography did not influence surgery in 39% of patients. Most patients had undergone both MRI and EEG (97% and 92%, respectively) prior to the PET scan, which was initially used to determine surgical management of the patients. About 39% of patients had undergone ictal SPECT, and 19% underwent interictal SPECT prior to the PET scan.
What is the Role of Functional Brain Imaging in Presurgical Functional Mapping?
As described in the alternative technologies section of this report, the current gold standard for presurgical functional mapping is the Wada test. Briefly, the Wada test is a procedure that involves inserting a catheter through the femoral artery to the carotid artery to numb each side of the brain individually to assess function. A potential role of functional brain imaging is to replace the somewhat invasive Wada test with a noninvasive test.
There were 6 studies identified that investigated functional brain imaging in presurgical functional mapping: 5 studies using fMRI and 1 using MEG. (66
) The characteristics of these studies are described in .
Characteristics of Studies Investigating the Localization of Seizure Foci With Functional Brain Imaging*
The prospective study by Medina et al. (67
) evaluated the effect of fMRI on the diagnosis and treatment planning of 60 patients with seizure disorders who were candidates for surgery. A combination of language, motor, and visual mapping were performed in each of the patients: 53 patients had language mapping, 33 had motor mapping, and 7 had visual mapping.
Four clinical experts reviewed the patients’ charts without the fMRI results and completed a questionnaire regarding the diagnosis and treatment plan for the patients; then, the 4 experts completed a similar questionnaire for each of the patients with their fMRI results included with the patient charts. The fMRI results altered the counseling for the patients and their families in 35 (58%) patients, helped to avoid Wada procedure (invasive procedure) in 38 (68%) patients, altered intraoperative mapping in 31 (52%) patients, and altered surgical planning in 25 (42%) patients, including altering the extent of resection in 4 (7%) patients.
The prospective study by Benke et al. (66
) compared fMRI to the Wada test in 68 patients with epilepsy who were candidates for surgery. Using the Wada test as the reference standard, they reported the following sensitivity and specificity:
- 89% sensitivity and 50% specificity for frontal fMRI; and
- 78% sensitivity and 71 % specificity for temporoparietal fMRI.
The study by Sabsevitz et al. (68
) compared preoperative fMRI to the Wada test results in 24 patients with left anterior temporal lobectomy. The sensitivity and specificity reported for fMRI were 100% and 73%, respectively.
The health technology assessment by Blue Cross Blue Shield from 2003 (43
) investigated the role of MEG in presurgical functional mapping. Since then, 1 additional study was identified that met the inclusion criteria for this review that compared MEG to the invasive Wada procedure.
The study by Papanicolaou et al. (76
) used MEG to identify the sensorimotor cortex in 100 patients with epilepsy being considered for surgery. The Wada procedure provides information on which hemisphere has primary control of language, but it does not identify the specific location within a hemisphere. When Papnaicolaou et al. compared MEG to the Wada procedure, they found that there was agreement between them in 79 (93%) of 85 cases. They did not explain why they did not have results for the remaining 15 patients in the study. Using the Wada procedure as the gold standard, the sensitivity and specificity of MEG was 98% and 83%, respectively.
The results for presurgical functional mapping were weak in the Blue Cross Blue Shield health technology assessment. (43
) There were 11 studies identified investigating MEG in presurgical functional mapping. Nine of these compared MEG to invasive functional mapping. Of these, 2 included more than 20 patients. Magnetoencephalography was compared with other invasive functional mapping measures (various technologies and tasks), making a comparison among the studies impossible. However, in 7 of the 9 studies, there was 100% agreement between MEG and invasive functional mapping. There was more agreement between MEG and invasive functional mapping than there was between MEG and ICEEG.
The results of the studies identified were added to the results of the Blue Cross Blue Shield health technology assessment in . These results are consistent with the results of the previous studies comparing MEG with invasive functional mapping.
Studies of Presurgical Functional Mapping*
Thus, presurgical functional mapping with MEG seems to agree consistently with invasive functional mapping measurements. Functional MRI appears to have high agreement with invasive functional mapping as well. There were no studies identified that compared fMRI to MEG for presurgical functional mapping.
Results: Multiple Sclerosis
What is the Role of Functional Brain Imaging in the Diagnosis of Multiple Sclerosis?
There has been a tremendous quantity of literature published on imaging of MS in order to further understand the pathogenesis of the disease; however, there were no studies identified that were eligible for inclusion on the clinical utility of functional brain imaging in the diagnosis of MS.
Results: Parkinson’s Disease
This review addressed 2 questions regarding the role of functional brain imaging for patients with PD. The first involves using functional brain imaging in the initial diagnosis of PD; the second question investigates the role of functional brain imaging in the diagnosis of parkinsonian syndromes.
What is the Role of Functional Brain Imaging in the Diagnosis of Parkinson’s Disease?
The study by Eckert et al. (79
) investigated the role of FDG-PET in the diagnosis of PD and in the diagnosis of parkinsonian syndromes. They included 43 patients with early PD and compared them using FDG-PET with 22 normal controls. Compared with clinical diagnosis, the sensitivity and specificity to diagnose PD was 100% and 91%, respectively (). This study did not report how the use of PET in the diagnosis of PD changes patient treatment or clinical outcomes.
Characteristics of Functional Brain Imaging Studies for the Diagnosis of Parkinson’s Disease*
What is the Role of Functional Brain Imaging in the Diagnosis of Parkinsonian Syndromes?
Studies investigating the diagnosis of parkinsonism used PET imaging with glucose metabolism or receptor binding. (116
) For glucose metabolism FDG PET is used. Patients with PD show increased metabolism in the lentiform nucleus, thalamus, pons, and cerebellum, and there is decreased metabolism in the lateral frontal, paracentral and parietal areas. (116
) Patients with MSA show decreased metabolism in the lentiform nucleus and cerebellum. (116
) In patients with progressive supranuclear palsy, glucose metabolism is decreased in the frontal cortex. (116
FDOPA PET allows for assessment of the functionality of the presynaptic nigrostriatal dopaminergic projections. FDOPA is a surrogate measure of endogenous dopamine synthesis. FDOPA uptake is reduced in parkinsonian patients and can differentiate PD from healthy subjects (even in early disease states). (116
There were 3 studies identified that used PET in the assessment of parkinsonian syndromes that met the inclusion criteria for this review. (79
) describes the characteristics of these 3 studies.
Characteristics of Functional Brain Imaging Studies in the Diagnosis of Parkinsonian Syndromes*
Values referring to the accuracy of PET in the 3 studies are listed in . The sensitivity and specificity for PET is quite high for all 3 studies, even though all 3 studies used different radiotracers. The study by Eckert et al. (79
) reported the use of FDG-PET in the diagnosis of patients with PD and in the diagnosis of parkinsonian syndromes. It is not clear if the addition of PET in the clinical diagnosis of PD or parkinsonian syndromes changed the treatment outcomes in patients with these conditions.
Accuracy of Functional Brain Imaging in the Evaluation and Diagnosis of Parkinsonian Syndromes*
There were no studies investigating the role of MRS that met the inclusion criteria because they did not compare MRS to a reference standard nor did they compare clinical outcomes in patients diagnosed with different modalities. For instance, in the prospective study by Watanabe et al., (117
) they measured NAA/Cr) ratios with MRS to diagnose MSA. They found that patients with MSA had significantly lower NAA/Cr ratios than the normal controls or patients with PD (P
≤ .001). They reported that using MRS they were able to differentiate MSA from PD sooner than with MRI, however, they did not report treatment outcomes in patients who have MSA distinguished from PD sooner compared with those with a delayed diagnosis.