Re: Aducanumab-Related ARIA

To the Editor:

We enjoyed reading the recent commentary by Drs Wassef and Colletti¹ in Clinical Nuclear Medicine to an opinion article published shortly before in the same journal.² Not only because the authors support some of the stated opinions, but in particular because their communication is an impressive concentrate of existing literature in the field, which very few have presented in so little space, making it a recommended read for new researchers interested in dementia and Alzheimer disease (AD). Moreover, the commentary points to new developments focusing on presymptomatic detection and development of effective prevention and therapy.

On a single point, however, we do not agree with Wassef and Colletti. This applies to the sentence: “Negative amyloid PET can reliably rule out AD.” It is far away from the impression we have. They refer to an article in which authors Scheltens et al³ state that amyloid PET has a high negative predictive value with regard to exclude the diagnosis AD. However, they do this without providing documentation for their claim, meaning that the statement can only be interpreted as the authors' own opinion.

A number of prerequisites must be met for a study of diagnostic accuracy to provide valid predictive diagnostic values.⁴ One of them is a fairly reliable reference to compare with. However, when it comes to dementia and AD, such a reference does not exist. On the contrary, no one knows the correct context, including the cause of the disease or to what degree amyloid, tau, or other circumstances play an active role in formation and development of AD, or whether they are by-products or waste products. Therefore, all talk of specific predictive values should be silenced, as they are inherently unknown and will remain so until a greater understanding of AD and how it can be prevented or treated has been provided. And, frequently encountered statements that procedures, including amyloid PET, are “suitable,” “good,” “extremely useful,” and so on for diagnosing or ruling out the presence of AD are examples of circular reasoning⁵ and should not be taken for anything other than what they are: postulates.

Reported AD diagnostic values are on uncertain grounds. If for no other reason, then it is for the fact that the diagnosis of mild cognitive impairment and AD, which the authors typically want to distinguish between and separate from normal cognition, has been based on variable national or international consensus criteria. These have constantly and concurrently been changed, from emphasis initially on clinical and pathological findings, then more on biomarkers and recently with a reinforcement of the clinical manifestations,^6–9 all testifying to the uncertainty and ambiguity of the diagnosis of AD and the fact that the reference the authors of diagnostic trials have used is rarely the same.

Actual values for the negative predictive value are rarely reported, but several studies indicate that a significant proportion of patients with clinical AD are amyloid negative by PET imaging; numbers range between 2% and 61%.^10,11 This has often made authors look for other reasons of dementia than AD,^12–16 suggesting that in their minds not the clinical impression but the presence or absence of amyloid PET positivity is what defines AD—yet another example of circular reasoning. On the whole, reports differ widely with regard to methodology, patient population, and clinical assessment, hampering comparison and leaving doubts as to whether AD-like cognitive decline is one or different diseases.¹⁷

In a recent large study across the AD clinical spectrum, amyloid abnormality was present in 84% of patients with clinical AD at age 50 years and decreased to 72% at age 90 years, meaning that the negative predictive value of amyloid PET is far from 100%.¹⁸ Our own, still relatively modest, experience supports the impression of a low negative predictive value of amyloid PET with regard to making the diagnosis of AD. In a cohort of the first 195 patients imaged with ¹⁸F-flutemetamol PET at the Division of Radiology and Nuclear Medicine, Oslo University Hospital, Norway, 64 patients were examined with both amyloid PET and lumbar puncture. ¹⁸F-flutemetamol PET is validated only for binary visual classification. A positive PET scan is one showing increased cortical uptake in 1 of 5 regions (frontal lobe, posterior cingulate and precuneus combined, lateral parietal lobe, lateral temporal lobe, or striatum). As many as 9 (30%) of 30 patients with a clinical diagnosis of AD had a negative amyloid PET.¹⁹ The AD diagnosis was made according to a comprehensive research protocol including detailed information from patients and their caregivers about symptoms, previous disorders, use of medication, demographic information, and, in addition, a battery of cognitive tests performed by 2 experienced memory clinic physicians: Mini-Mental State Examination, the Consortium to Establish a Registry of Alzheimer's Disease 10-item word list and constructional praxis exercise, the Clock Drawing Test, the Trail Making Tests A and B and the animal-naming test, the Controlled Oral Word Association Test, and the 15-word short form of the Boston Naming Test. The severity of the cognitive impairment was scored by an experienced rater using the Clinical Dementia Rating scale. These many tests are not part of the clinical routine, but together they give us a rather confident impression that these patients suffer from dementia of the AD type, although we know that this diagnosis cannot be made with certainty, either clinically or using biomarkers. Thus, our conclusion, which is supported by the mentioned studies,^10,11 is that the negative predictive value of amyloid PET is rather approximately 70% to 75% than close to 100%, and hence, it is debatable whether amyloid PET has a meaningful place in the workup of patients with suspected or known dementia.

As we see it, there is every reason to clarify exactly what are the predictive values of amyloid imaging in new and well-designed prospective studies because this is the key question. Not an easy task—because how does one make a certain diagnosis of AD? Nevertheless, one can solve the problem by doing something different, namely, perform FDG and amyloid PET in a treatment randomized controlled trial of well-characterized demented patients. Only if cerebral FDG uptake increases, while cerebral amyloid deposits decrease, is there evidence for a favorable effect of immunotherapy. If, on the other hand, FDG uptake decreases more in the treated group than in the placebo group, all further attempts with immunotherapy should be stopped.

We support Wassef and Colletti¹ in pointing to substrates other than amyloid and tau in a renewed effort to achieve a better management of AD. For a start, we have suggested looking more closely at the respective importance of infection, low-dose radiation, and not least atherosclerosis, the latter assessed by ¹⁸F-sodium fluoride PET imaging.²⁰

Poul F. Høilund-Carlsen, MD, DMSc∗†
Abass Alavi, MD‡
Mona-Elisabeth Revheim, MD, PhD, MHA§
∗Department of Nuclear Medicine
Odense University Hospital
University of Southern Denmark
Odense, Denmark
†Department of Clinical Research
University of Southern Denmark
Odense, Denmark
‡Department of Radiology
Hospital of the University of Pennsylvania
Philadelphia, PA
§Division of Radiology and Nuclear Medicine
Oslo University Hospital; Institute of Clinical
Medicine, University of Oslo
Oslo, Norway
mona.elisabeth.revheim@ous-hf.no
m.e.rootwelt-revheim@medisin.uio.no