Assessment of a diagnostic technology typically focuses on the following 3 domains; 1) technical performance; 2) diagnostic performance (sensitivity, specificity, and positive and negative predictive value) in relevant populations of patients, such as those with mild cognitive impairment or suspected Alzheimer’s disease (AD); and 3) demonstration that the diagnostic information can be used to improve patient outcomes. The gold standard for the diagnosis of AD is post-mortem neuropathologic examination. In the absence of comparisons with the gold standard, long-term clinical follow-up (e.g., conversion from mild cognitive impairment [MCI] to probable AD) may be used as a surrogate standard to evaluate the diagnostic performance of beta-amyloid imaging with positron emission tomography (PET).
Beta-amyloid imaging may be particularly helpful for the future study of novel therapeutic agents that target amyloid plaques. However, current clinical purposes of testing for beta-amyloid plaque density would be to improve diagnostic accuracy (e.g., rule out AD) or predict conversion from MCI to AD. In general, evidence of a health benefit or clinical utility from testing requires demonstration of:
- incremental improvement in diagnostic or prognostic accuracy over current practice, and
- that incremental improvements lead to improved health outcomes (e.g., by informing clinical management decisions), and
- that these outcomes may be obtained (i.e., are generalizable) outside of the investigational setting.
Technical Performance: Evidence on technical performance of this test should demonstrate that the test measures what it is intended to, i.e., beta-amyloid plaque. The best evidence on this would be direct comparison with a gold standard test for measuring amyloid plaque, which is histopathologic examination of tissue. Other important measures of technical performance are the reliability of testing, including both test-retest reliability and interobserver reliability in reading test results.
Data on technical performance of the test was included in an FDA-regulated study, which was published in 2011. (6, 7) This study was a Phase III multicenter trial with 2 separate cohorts. These cohorts were an autopsy cohort and a young, cognitively intact cohort. The autopsy cohort was drawn from 152 subjects who had a projected life expectancy of 6 months or less. Thirty-five individuals passed away and were autopsied within 12 months of PET imaging; 29 were included in the primary efficacy analysis. This cohort was composed of 9 subjects (31%) who were not cognitively impaired, 2 (7%) who were mildly impaired, 13 (45%) with a clinical diagnosis of AD, and 5 (17%) with a clinical diagnosis of a non-AD dementia.
All patients had direct measurement of amyloid burden by histopathologic examination, and 52% met the pathologic criteria for AD. A significant correlation of 0.78 was found between amyloid burden in the brain measured by Amyvid and the gold standard of histopathology, however, there was not an exact match between the 2 measures. The correlation between quantitative whole-brain florbetapir image scores and post-mortem silver stain was 0.71. In the young controls (specificity cohort to evaluate false positives), the primary efficacy endpoint was the exclusion of amyloid in 47 young subjects who were negative for the apolipoprotein E ε4 (APOE4) allele, randomly interspersed with PET scans of 40 subjects in the autopsy cohort. The study achieved specificity of 100% in this cohort, although it is noted that the young controls are outside of the intended use population.
Reproducibility of the readings was assessed using 3 trained readers who were blinded to the clinical information. Using a binary scale (positive or negative for amyloid), sensitivity ranged from 55% to 90% for the 3 readers, and in 24-45% of the images (depending on the sample), at least one reader would have had a different interpretation of amyloid status from the other readers. (6) Subsequent reanalysis for publication used the majority rating of 3 nuclear medicine physicians as the primary outcome variable, resulting in 96% agreement between florbetapir-PET images and histopathologic results in the 29 subjects in the primary analysis cohort. (7)
Conclusions: Evidence on technical performance is mainly from the FDA-sponsored study. A strength of this study is the comparison of florbetapir imaging with the gold standard of post-mortem histopathology. Limitations include the small sample size, a majority rating for assessing diagnostic accuracy, and having only 2 patients in the mildly impaired category, which is the population for whom the test is most likely to be used. Evidence from this study indicates that the agreement between histopathology and beta-amyloid testing by PET is good but not perfect. There is evidence for inter-observer variability in reading the test; using a majority of 2/3 readers leads to a high agreement with histopathology.
Diagnostic Performance: The FDA-regulated study also included some information on diagnostic performance of the test. Using the majority consensus of three independent reviewers as the final test reading, sensitivity and specificity was calculated compared to the gold standard of histopathology. (7) Of 15 subjects who met pathologic criteria for AD, 14 had positive florbetapir scans (sensitivity of 93%). Of the 14 subjects who did not meet pathologic criteria for AD, all 14 had negative scans (specificity of 100%). Scans from all of the young subjects (27 APOE4+ and 47 APOE-) were negative. Exploratory analysis indicated that in 3 subjects (20%), the clinical diagnosis did not match with the final autopsy diagnosis. These measures of diagnostic accuracy are limited by the patient population, which is not representative of the population that the test is intended to be used for, and the use of a majority reading based on 3 independent experts, which is not likely to be used in clinical care.
An industry-funded multicenter study by Fleisher et al. pooled data from 4 phase I and II trials of florbetapir-PET imaging for a total of 210 participants, including 68 subjects with probable AD, 60 subjects with MCI, and 82 older unimpaired controls. (8) Quantitative standard uptake value ratio (SUVRs) thresholds were determined from the Phase III trial described above. Although there were significant differences in mean SUVRs across groups, there was considerable overlap in the range of values. The percentage of subjects meeting threshold levels of amyloid with clinical AD, MCI and cognitively healthy controls was 80.9%, 40.0%, and 20.7%, respectively. The percentage of subjects with any identifiable florbetapir signal was 85.3%, 46.6%, and 28.1%, respectively. Among healthy controls, the percentage of subjects with any florbetapir positivity increased linearly by age, ranging from 11.8% for subjects 55 to 60 years of age to 41.7% for subjects 81 years of age or older. APOE4 carriers in the control group had about twice the percentage of florbetapir positivity as noncarriers, although this comparison did not reach statistical significance.
In 2012, Camus et al. reported the diagnostic performance of florbetapir-PET in a clinical setting. (9) Included were 13 subjects with AD, 12 with MCI, and 21 older unimpaired controls. PET images were assessed visually by 2 readers who were blinded to any clinical information and quantitatively by the SUVR of cortical regions compared to the cerebellum. Sensitivity and specificity were calculated based on clinical diagnosis as the comparison standard. Agreement in visual analysis between the 2 readers gave a kappa value of 0.71. Comparing visual assessment with the initial clinical diagnosis, 11 of 13 AD patients (85%), 6 subjects with MCI (50%) and 13 of 21 control subjects (60%) had positive scans, resulting in sensitivity of 84.6% and a specificity of 38.1% for discriminating AD patients from control subjects. A quantitative assessment of the global cortex SUVR showed a sensitivity of 92.3% and specificity of 90.5% at a cut-off value of 1.12 (ROC [receiver operating characteristics] area under the curve 0.894). Although the study is limited by the small number of subjects and the use of clinical diagnosis as a reference standard, these results suggest a high number of false positives with visual assessment of the images. In addition, quantitative analysis was not able to differentiate subjects with MCI from unimpaired controls.
Conclusions: Evidence on the diagnostic performance of beta-amyloid testing is limited, and the available studies all have methodologic limitations that limit the validity of reported results. As a result, it is not possible to determine the sensitivity and specificity of testing. Some evidence suggests that there are a high number of false positive results in patients without AD. However, the FDA study reports high specificity, so the true rate of false positives is uncertain. Further high-quality studies using populations of patients that represent those presenting in clinical care are needed to better define the diagnostic performance of this test.
Clinical Outcomes: No trials have been identified that reported health outcomes following florbetapir-PET imaging, thus there is no direct evidence for clinical utility.
Possible clinical uses of beta-amyloid testing could include confirming the diagnosis of AD in order to begin medications at an earlier stage, or ruling out AD, which may lead to further diagnostic testing to determine the etiology of dementia and/or avoidance of anti-Alzheimer’s medications that would be unnecessary.
Since the sensitivity and specificity of beta-amyloid testing has not yet been established, it is not possible to determine an indirect chain of evidence that would indicate that health outcomes are improved. Because of the presence of beta-amyloid in elderly patients who do not have AD, it is not likely that the test will have a high positive predictive value, and therefore it may have limited utility in confirming AD. It is possible that the negative predictive value of testing may be high and that the test may be useful in ruling out AD. If this is true, it is not certain how many patients would benefit from additional testing to determine etiology, or whether a substantial number of patients would avoid unnecessary medications that would otherwise be given.
Conclusions. Evidence on clinical utility, i.e. that health outcomes are improved by testing, is lacking. There are no studies that report on clinical outcomes following testing. The diagnostic accuracy of testing is too uncertain to determine whether testing is likely to impact management and/or lead to improved outcomes.
Ongoing Clinical Trials
A search of the online site www.ClinicalTrials.gov in May 2012 identified a number of trials on amyloid imaging with PET. Of particular interest are the following:
- An industry-sponsored Phase III open-label study to evaluate the efficacy and safety of florbetaben (BAY94-9172) PET imaging for detection/exclusion of cerebral beta-amyloid compared to postmortem histopathology (NCT01020838). This study has an estimated enrollment of 216 subjects with completion of the primary outcome measure in 2011 and final study completion in 2014.
- An industry-sponsored Phase III open-label study to compare the brain uptake of flutemetamol with brain amyloid levels determined post-mortem (NCT01165554). The study has an estimated enrollment of 100 subjects with completion in 2012.
- An industry-sponsored Phase III open-label study to assess the prognostic usefulness of flutemetamol for identifying subjects with amnestic MCI who will convert to clinically probable AD (NCT01028053). The study has an estimated enrollment of 225 subjects with completion estimated for January 2013.
Literature on the use of florbetapir-PET imaging to aid in the diagnosis of patients with suspected Alzheimer’s disease is limited. The pivotal Phase III trial, although to be commended for its use of the gold standard of histopathology, has a number of limitations including small sample size, use of a majority rating of 3 physicians, and having few patients in the mildly impaired category. This study reported a moderately high correlation of amyloid plaque with histopathologic examination. The sensitivity and specificity of this test have not yet been adequately determined in an appropriate population, including a larger number of patients with mild cognitive impairment.
The clinical utility of this technology is uncertain. The test is not likely to be useful for confirming AD in patients who present with cognitive impairment. It may have a role in ruling out AD, but this has yet to be established with certainty. Questions also remain about the use of this test outside of the investigational setting, particularly regarding the accuracy of visual interpretation of images and how best to apply this test in routine clinical practice.
Practice Guidelines and Position Statements
2011 Guidelines from the National Institute on Aging and Alzheimer’s Association on the diagnosis of mild cognitive impairment and dementia due to Alzheimer’s disease recommend the use of biomarkers, including beta-amyloid imaging with PET, only in research settings. (1, 3) Reasons for this recommendation are that more research needs to be done to ensure that the criteria that include the use of biomarkers have been appropriately designed, there is limited standardization of biomarkers from one locale to another; and access to biomarkers may be limited in community settings.
The Alzheimer’s Association has indicated qualified support for the availability of florbetapir. (10) The statement includes the following: “On one hand, FDA approval of this product will expand the clinical and research opportunities for amyloid imaging by making this brain imaging tool more widely available to the field. On the other hand, the fact that all of the potential uses of this product are not crystal clear tempers our enthusiasm. Again, additional research is needed to clarify the role of florbetapir-PET imaging in Alzheimer’s.” The Alzheimer’s Association has convened a task force with the Society of Nuclear Medicine to develop recommendations for the use of amyloid imaging.
Disclaimer for coding information on Medical Policies
Procedure and diagnosis codes on Medical Policy documents are included only as a general reference tool for each policy. They may not be all-inclusive.
The presence or absence of procedure, service, supply, device or diagnosis codes in a Medical Policy document has no relevance for determination of benefit coverage for members or reimbursement for providers. Only the written coverage position in a medical policy should be used for such determinations.
Benefit coverage determinations based on written Medical Policy coverage positions must include review of the member’s benefit contract or Summary Plan Description (SPD) for defined coverage vs. non-coverage, benefit exclusions, and benefit limitations such as dollar or duration caps.