Predicting Alzheimer’s Disease: Potential Ethical, Legal, and Social Consequences

(This post oriignally appeared, in very slightly different form,  in the Neuroethics Blog on June 17, 2014, here.)

Would you want to know the date and time of your death? Life-Line, the first published fiction by Robert A. Heinlein, one of the giants of 20th century science fiction, explored that question. The story’s protagonist, Hugo Pinero, had invented a machine that could tell precisely when individuals would die, but, as Pinero found to his distress, he could not intervene to change their fates.

Would you want to know whether you would be diagnosed with Alzheimer disease (AD)? This question is rapidly leaving the realm of science fiction; indeed, it already has for some unlucky people. Our ability to predict who will suffer from this evil (and I chose that word carefully) condition is proceeding on several fronts and may already be coming into clinical use.

This post will briefly note the ways in which AD prediction is advancing and what some of the ethical, legal, and social implications of such an ability would be, before asking “should we care?”

Science

Several different techniques are providing information about an individual’s risk of being diagnosed with AD, including genetics, biomarkers, and neuroimaging.

Genetics can predict AD with great confidence for about one person in a thousand. People who carry a mutated version of the PS1 gene (or, much more rarely, mutated versions of the PS2 or APP genes) are nearly certain to be diagnosed with AD, unless they die earlier from something else, and with an early onset version that typically strikes in one’s 40s or 50s. People with two copies of the APOe4 allele, about one to two percent of the population, have a very high risk (at least 50 percent, perhaps as high as 80 percent) of being diagnosed with AD in their sixties or seventies. People with one APOe4 allele and one APOe2 or APOe3 allele – that’s about 20 percent of the population – have two or three times the AD risk of people without an APOe4 allele. Other alleles of other genes have also been found to confer higher risk of AD, and some single nucleotide polymorphisms have been associated with higher risk of the disease.

Other researchers have identified biomarkers that are associated with risk of AD, both in the cerebral spinal fluid (CSF) and in the blood serum. Thus far, the CSF methods have looked for levels of the protein beta amyloid (42), which forms plaques on dead and dying neurons of those with AD, and the protein named “tau,” which forms tangles in the bodies of those neurons. Some of the blood work has looked at those biomarkers; others have looked at a range of different proteins in the blood. None of these methods is ready for clinical use; some of the published research has claimed nearly 90 percent accuracy in predicting relatively near-term AD diagnosis.

Recently, the FDA approved a radio ligand that attaches to the amyloid plaque in the brain and allows the existence of amyloid plaque to be seen by positron emission tomography (PET). The approved labeling is for use in diagnosing AD, not in predicting it, but the off-label use doctrine allows doctors to prescribe it for any purpose. Other researchers have been trying to find ways to image tau tangles, though currently the only method for detecting them is through a brain biopsy (not an easy technique!).

Note: When the original version of this post went on-line, I was at a neuroscience meeting in Washington, D.C. where I  learned that there are now radioligands that allow PET scanning of tau tangles, though only being used for research so far – not approved for clinical use.)  Some early evidence that it may eventually add still more power to diagnosis and prediction of AD.

It is also known that magnetic resonance imaging (MRI) scans of brains can see changes in grey matter density in certain parts of the brains of people with AD; efforts are under way to use that method to predict AD diagnoses.

These various methods need not be used in isolation. They could be used together, in an effort to provide greater accuracy than any one test would do on its own. We are only at the beginning of efforts to assess those possibilities.

The FDA has not approved any of these methods (yet) for clinical use in predicting AD and professional groups have recommended against such use. It remains unclear how good any of these methods are alone or in combination, or at what age or ages they are useful. (A genetic cause may be strongly predictive even before birth; amyloid plaque levels may – or may not – be relevant only for people over 60.) Their accuracy might also vary between completely cognitively normal and those showing some minor signs of cognitive problems (which, for many people, would not progress to AD).

Importantly, these methods were not discovered in order to use them for clinical prediction. They are the results of basic research, of efforts to understand the natural history of the disease, in hopes of ultimately finding preventions or treatments. Their first use in humans has been in AD research, stratifying research subjects into high and low risk groups in the hope of making clinical trials faster and cheaper. But nothing prevents a physician from ordering the tests for a worried patient (with money to pay for tests that insurance will not reimburse).

Effects

Let’s assume that people did begin to get fairly accurate tests for their AD risks. What would follow?

If we had good interventions to prevent or treat the disease, much good might come from such testing, but we don’t (beyond “chicken soup” kinds of recommendations like “exercise”.) So how and why will people use these predictions and what non-medical consequences can we expect?

Some people will use the information for financial planning. A friend of mine is an “elder lawyer,” who spends a good amount of his time in financial planning for the elderly. He says that if we had a test that was 90% accurate, he would urge all of his clients to get such a test so they can plan how to use (and preserve) their assets for their struggle with AD.

On the other hand, some will worry about the effects of getting tested. Being at high risk for AD might lead to all the usual discrimination suspects – employment, health, life, and disability, plus one special one, long term care insurance. The relatively old ages at which AD strikes (except for the roughly 1% of cases that are early onset) mitigate, but do not eliminate, the number of people who would risk employment and health insurance discrimination. Most people will not be employed when they are diagnosed with AD. And, at time of diagnosis (and hence of increased health care costs), most of those affected will be over 65, and thus will have Medicare for health coverage (whatever may happen to Obamacare). Ironically, though, whether GINA, the Genetic Information Non-discrimination Act, protects them will depend on whether their risks were predicted using genetic methods or other methods. (The consequences of the use of mixed methods are not clear.)

A few special cases of possible “employment” discrimination might be noted. Every four years Americans “employ” someone as President. Would the public want to know the AD risks of the candidates? Not too long ago, President Ronald Reagan was diagnosed with AD only a few years after the end of his second term. The public, acting largely through the press, might want AD risk information from future candidates. (Teneille Brown has explored these issues in more depth1.)

Similarly, sitting presidents may well want that kind of information about candidates for appointment to jobs with life tenure – federal judges, and particularly Supreme Court justices. In 2009, Judge Karen Williams, Chief Judge of the United States Court of Appeals for the Fourth Circuit, retired from the bench at the age of 57 because of early onset AD. All things being equal, presidents want the judges they appoint to sit, and influence the law, for decades after the president’s term is over.

It is not clear that life insurers would care much about AD risk; the disease process is so long that the age at death, though somewhat reduced, may not be change significantly. But private disability insurers should care, as AD patients who are employed at the time of diagnosis may end up claiming on such policies.

And long-term care insurers, should care, a lot. AD patients will often need years of long-term care. The private long-term care market is relatively new and small. It is a policy initiative to try to deal with the upcoming huge cost of long term care for Baby Boomers, care that is not significantly covered by Medicare or private health insurance. If people were able to test for their AD risk and then, if they test positive, buy long-term care insurance on the same terms, the resulting “adverse selection” will cause insurers either to lose money or to raise their rates. Either outcome, in this young and relatively fragile market, could end long-term care insurance. On the other hand, if insurers can take AD risks into account (at least when the customer knows those risks), people at high risk will often find long term care insurance unaffordable, even though – and especially because – they will need it.

But other, less tangible, consequences may follow. Consider the effects on family dynamics. Will the children take away Dad’s car keys sooner if he has been predicted to be at high risk for AD? Will they take away his checkbook, and control over his finances? How will the relationships within the family change when spouses, partners, and children expect an AD diagnosis?

And, of course, what will be the effects on people predicted to be at high risk? They may face depression or other psychological consequences. They might even make plans for suicide.

These issues, of course, are not new – they occur already with an AD diagnosis. But an AD prediction may move the opening point of these concerns forward several years, years that otherwise might not have been clouded by the knowledge, or fear, of AD.

Should We Care?

In a different sense, none of these issues is new. They already exist with fatal diseases that can be confidently predicted, like Huntington disease, as well as fatal diseases once they are diagnosed. But AD is, in some ways, distinctive. Instead of striking one person in 20,000, like Huntington disease, it will strike an estimated 10% to 15% of the population. And its memory, and ultimately personality, destroying characteristics lead to special challenges, as well as, for some people, to special horror. What, if anything, should we do about it? For now, I will make only two suggestions: assurance of the accuracy of the predictions and a requirement for counseling.

The accuracy of the tests, alone and in combination, needs to be assessed carefully, and for people of different sexes, ethnicities, and other possibly relevant possibilities. I believe some kind of public assessment of accuracy, akin to (and possibly including) FDA approval, should be required before the testing is allowed.

Then, both before the test is taken as well as after any positive results are returned, we should require skilled counseling. The first session will help make sure that the individual understands the advantages and risks of taking the test. The second will help high risk people deal with the shock of the prediction – and with its longer-term consequences.

Conclusion

Pinero’s “life predictor” never existed and never will. That would have been good for the fictional Pinero: in the short story thugs paid by life insurance companies murdered him, on the very date his machine had predicted

Widespread, accurate (or even inaccurate) AD prediction is not yet here. It will be soon. As a common, expensive, and severe disease, its predictability will bring some foreseeable challenges, as well, no doubt, as some unforeseeable ones. We need to work to understand, and cope with, those challenges. And we need to start yesterday.

Hank Greely

Reference

1) Teneille Brown, Double Helix Double Standards: Private Matters and Public People, J. Hlth Care L. & Pol. 11:295-376 (2008).