Large Overlap Between Healthy Controls and Alzheimer Patients in Amyloid PET Study
Avid*/Lilly continues to flood the medical literature with data to ostensibly promote its PET amyloid tracer florbetapir (Amyvid). And while the company wants to give the world the impression that its tracer has diagnostic use, a repeated heaping helping of caveat emptor is appropriate.
The latest case of industry-funded promotion as peer-reviewed research: Another brain-amyloid imaging study—while showing significant differences among subjects with probable Alzheimer disease (AD), mild cognitive impairment (MCI), and other healthy controls—nevertheless also demonstrated considerable overlap in the burden of brain amyloid among these clinical subgroups. The PET data, pooled from four of the company’s phase 1 or 2 clinical studies, showed that the mean value for brain amyloid (described as the cortical-to-whole-cerebellar SUVR) was significantly different among the subjects, but that also about 20% of older healthy controls demonstrate an amyloid load consistent with pathologically defined AD. And about 30% of these healthy elderly (>55 years) exhibit any level of brain amyloid, as detected with florbetapir. Conversely a substantial number of patients with clinical AD did not demonstrate pathologic levels of amyloid. (Lilly would argue that this is where its PET tracer is most useful—in ruling out AD on the basis of a low amyloid burden in suspect patients.) The amyloid burden was also significantly more likely among carriers of the APOE4 gene, for which the AD subgroup was enriched, and with each decade of life in older healthy controls.
The amyloid-overlap data are consistent with those from autopsy and other PET studies, which show, on a cumulative basis, that about 30% of cognitively normal elderly will demonstrate some level of AD-consistent brain pathology, and that many of these subjects meet the neuropathologic criteria for AD despite the absence of cognitive impairment (during life in the case of the autopsy studies).**
The $64 question for Avid/Lilly is how PET imaging for brain amyloid can be practically integrated into clinical practice, if at all. Currently the tracer appears to have its greatest use in assessing the clearing of protein in trials of investigational anti-amyloid compounds (which have their own safety problems). In the context of practice and beyond the issues of cost, PET access, and inter-reader reliability (see here, for example), the most important question is what to do when PET images demonstrate brain amyloid—given the substantial percentage of older healthy adults who demonstrate the AD-linked substance. By my read of this study, unless the florbetapir-enriched PET scan shows an amyloid burden that is undeniably above that of healthy controls or really, really low, the results are functionally useless.
* Avid is a wholly owned subsidiary of Lilly, which bought the company for $300 million and will fork over another $500 million if florbetapir ever becomes FDA approved.
** The issue of what to do with these data becomes a bit zen-like. Are these positive amyloid scans an indicator of impending AD, if the patient lives long enough? And if so, so what (especially if clinical AD is unlikely to emerge until 10 years later and/or there are no useful interventions)?
PET = positron emission tomography; SUVR = standard uptake value ratios.
Photograph: Atrophied brain from person with AD from National Institute on Alcohol Abuse and Alcoholism.