Predicting Alzheimer's Disease

Welcome to the first Stanford Neuroblog live blogging event of 2011, covering a panel discusion entitled:

Predicting Alzheimer’s Disease: Intersection of Science, Medicine, and Society

From the official description:

We may soon be able, with good accuracy, to tell a healthy 60 year old whether or not he or she will have Alzheimer’s Disease - and, perhaps, how soon the symptoms will appear.  Join our panel of clinicians, lawyers, and health care workers in a discussion about the sciences that may allow us to predict the onset of alzheimer’s disease, and its implications for the health care system, patients, families, and society.

Tonight’s panelists are:

  • Hank Greely, J.D. Deane F. and Kate Edelman Johnson Professor of Law; Professor, by courtesy, of Genetics; Director, Center for Law and the Biosciences; Stanford University
  • Michael D. Greicius, M.D., M.P.H., Assistant Professor, Department of Neurology and Neurological Diseases, Stanford Medical School; Medical Director, Stanford Memory Clinic
  • Michael Gilfix, J.D., expert in Elder Law, Gilfix & La Poll, Palo Alto
  • Frank M. Longo, M.D., Ph.D., George E. and Lucy Becker Professor in Medicine; Chair, Department of Neurology and Neurological Diseases; Stanford University

The talk is sponsored by the Stanford Interdisciplinary Group in Neuroscience and Society, the same group that put on the previously blogged panel on Neuroscience in the Courtroom.

The talk begins with Hank Greely opening with a soliloqy regarding the prediction of disease. He notes that disease prediction is nothing new for the human race, and that Alzheimer's Disease (AD) holds a certain resonance, attacking our minds (a scary prospect for most people) as well as being remarkably common. He previews tonights discussion by stating that our ability to predict AD is increasing due to scientific advances. Tonights panel will cover those advances, as well as the effect on patients of this new ability, and both the legal and social implications of an ability to predict AD.

First to speak is:

Dr. Michael Greicius;  Novel Diagnostic and Predictive Tests for Alzheimer's Disease.

Currently, the clinical criteria for probable AD includes: dementia (change in cognition), deficits in two or more cognitive area (visual spatial, memory, etc), progressive decline in memory and other realms (such as executive function), no disturbance of consciousness, onset age 40-90 (mostly 65 and up), and no other explanation. Using these clinical criteria, sensitivity is fairly good , however specificity is rather poor - the distinction being that there are many false positives, but few false negatives. In speciality clinics, diagnosis is accurate 85-90% of the time, which leaves plenty of room for uncertainty. Indeed, the differential diagnosis is rather broad, including mild cognitive impairment, dementia with Lewy bodies, mass lesion, frontotemporal dementia, vascular dementia, depression, and others. Teasing these different possibilites apart can be tricky. For example, take mild cognitive impairment, which can be thought of as a middle ground between normal aging and AD. So a patient with MCI would have memory troubles, but would be able to overcome them. Conversion from MCI to AD is thought to be approximately 12%, and it would be beneficial to have the capability to predict which patients would convert.

Revising AD criteria is ongoing, mostly involving adding biological markers. These markers are preferably diagnostic, predictive, dynamic (tracking the course of the disease, useful as a measure of treatment efficacy), and a gold standard, that it could act as a surrogate for the condition (aka becomes something to treat itself).

Some examples of biological markers:

  • Apolipoprotein E (ApoE): polymorphisms, some of which confer risk for AD. It is a susceptibility gene, and is therefore not useful as a general screening tool. However, it may be predictive for how patients will respond to treatment.
  • Plaques and Tangles: Beta-Amyloid and Tau. These two proteins are positive markers of AD, that are used in post-mordem confirmation of an AD diagnosis. A group from Pittsburg, Klunk et al, developed a radioactive tracer that can be used to selectively image amyloid plaques in patients. Some limitations are that healthy controls can be positive for the tracer (as high as 51% positivity in health older controls). In addition, amyloid imaging is not dynamic, with signal peaking early in the condition, and not necessarily getting worse during disease progression. In terms of prediction, amyloid imaging is strongly predictive of which MCI patients will convert to AD.
  • Spinal fluid markers: Tau is increased and Beta-amyloid is decreased within the spinal fluid of patients - this pattern is predictive in MCI to AD conversion, as well as being sensitive and specific for AD. In addition, a recent study (Fagan et al, Arch Neurol 2007) showed that spinal levels of tau/beta-amyloid can predict the conversion from healthy aging to mild dementia.

Predictive use of biological markers: amyloid imaging is used in selecting people for clinical trials, will soon be used in clinical settings. More ethically dubious is the use of predictive screens for healthy control (this will be covered later in the discussion). Greicius concludes by stating that all the clinical advances are paving the way toward development and early initiation of a definitive AD treatment.

Next to speak is:

Frank Longo, who is discussing the impact of predictive ability on patients.

Longo comments that the ability to predict onset of neurodegenerative diseases is currently available, in the case of Huntington's disease (HD). In the case of HD, genetic testing can confirm with 100% accuracy whether a patient will develop HD during the course of their lifetime. So HD can serve as a model for the ability to apply AD prediction within a clinical setting. Longo describes the protocol for HD genetic testing, saying that it includes genetic counseling, speaking with a trained neurologist, and finally DNA testing, with predetermined follow-up testing. Given a definite protocol, Longo describes his experience with patients, and with the process of giving news. He describes that process that people given a positive prediction for a neurodegenerative disease go through several stages, including resignation, relief because they at least know what the future is bringing, but also highly individualized stress. He describes some of the effects on family and significant others, who are also affected by a positive predictive diagnosis. Longo describes studies on stress in patients given good and bad news regarding their Huntington's diagnosis, saying that generally, data show a general reduction in stress. Longo says that this reduction in stress holds true for positive diagnoses of abnormal BRCA1/2 genes. Longo concludes by stating that the ability to predict AD must be considered in the context of the effect it will have on patients, and patient-family dynamics.

And next is Michael Gilfix.

He will speak as a practicing attorney, who will discuss his experience working with patients and planning of their legal affairs. He notes that the first reaction many patients have is planning for long term medical costs. Proactivity is important, as is applying for government benefits. There appears to be many legal ways to protect personal assets, as well as protocols for transferring legal responsibilities. Gilfix counsels the importance of taking proactive steps while they are still proactive - the sooner a legal plan is in place the better, appears to be his message. He concludes that the advent of predictive testing would allow much more sophisticated planning.

And lastly, Hank Greely takes the podium.

He will discuss the wider societal implications of AD. He starts by noting the differences between HD and AD, primarily the rates of incidence (very low in HD and common in AD) and the distinction between having a risk and knowing for sure (AD or HD, respectively). Now he speaks regarding the topic of regulation of predictive testing. Who will administer the test, who will decide whether any particular test works, who will regulate how the healthcare system will use this predictive information. As he has commented in other discussions, discrimination against those with positive predictive tests will assuredly be an issue. For example, spinal fluid tests are not protected against discriminate as predictive genetic tests are. And even so, employment discrimination against those with positive genetic tests is illegal, which does not mean that it will not happen (see racial and gender discrimination). Furthermore, genetic tests cannot be discriminated against in the realm of health insurance, but not long-term care insurance. In terms of predictive tests that denote "high risk", there are no specific protections against discrimination. He concludes that our society is really unready to handle the consequences of predictive tests for a variety of neurodegenerative conditions.

Question and Answer Session:

Question one, directed to Greicius, regards the cost of the biomarkers for AD, and why he did not recommend more cost-effective tests (such as hippocampal volume)?

Greicius respondes that many of the more cost-effective tests are, in his opinion (and in the opinion of the American Academy of Neurologists), not as accurate as the more expensive diagnostic tests mentioned during his presentation, and cannot be used as a predictive assay for AD.

Question two, mentions the use of calcium channel blockers in an AD mouse model, and asks whether any members of the panel can discuss the study.

Longo responds, saying the calcium channel blockers is just one of many, many drugs that have been shown to be effective in mouse models of AD. Should we use these medications (which are often used in humans to treat other conditions) on the basis of mouse models alone? He notes that no medications identified in mouse models have been shown to be ineffectual in humans, in addition carrying significant side effects.

Question three, what is the youngest age at which tangles/plaques have been identified?

Answer: Perhaps late 40s, early 50s. Patients with Down's Syndrome will have AD by the time they are in their 30's, and will therefore have tangles/plaques at that age. A neuropathologist in the audience points out that although neuropathology is considered the gold standard, there is still a good deal of variation between people in terms of the amount of pathological markers and cognitive deficits. This is probably due to individual differences in susceptibility to the cellular dysfunction and its translation into cognitive phenotype.

Question four: The audience member is concerned about the possibility that not having the test will end up being frowned upon. He asks whether people have thought about mandating privacy regarding results? Greely notes that in terms of genetic testing, employers/insurance companies are not allowed to take into account genetic testing no matter what their status is - and that something similar would probably be required in terms of predictive tests. However he notes that in terms of AD, employers might not care so much, given that people will probably develop AD long after they are no longer working at the company in question.

Question five: How much do these test help us improve accuracy over the current clinical diagnostic tools?

Answer: It ups your likelihood for a correct diagnosis in atypical presentations.

The questions continue as more a discussion, with speakers going too quickly for this typist to capture in any detail. Therefore, I'm ending my coverage, encouraging readers to discuss the panel discussion in the comments.