It is becoming easier, cheaper and quicker to diagnose dementia

OF THE ESTIMATED 55m people living with dementia around the world, only one-quarter have been formally diagnosed with the condition. There are many reasons for this. Two are enduring: many patients and clinicians alike wrongly believe that dementia is an inevitable part of the ageing-human condition and, being incurable, is hardly worth diagnosing; and some people experiencing cognitive impairment fear hearing what sounds like a sentence of brain-death, and so do not seek help.

Some of the reasons for non-diagnosis, however, may be about to change. During the covid-19 pandemic many people have delayed consulting their doctors about non-urgent conditions, and as lockdowns ease, they may begin to ask for professional guidance (moreover, evidence suggests that covid itself heightens the risk of dementia). In addition, diagnostic techniques, hitherto unreliable, time-consuming and costly, are becoming available, and for some forms of dementia hopes are emerging of more effective treatments.

Dementia is normally diagnosed by testing cognitive functions such as memory. If mild cognitive impairment (MCI), often a precursor to dementia, is detected, a patient may then be referred for tests to identify which of the dozens of causes of dementia are to blame. By far the most common is Alzheimer’s disease, accounting for 60-80% of cases.

Fear of the needle

Identifying Alzheimer’s normally requires a brain scan, and perhaps a lumbar puncture (the insertion of a needle into the lower spine), to extract cerebrospinal fluid, so as to measure its levels of two proteins that build up in the brains of people with Alzheimer’s, known as beta-amyloid and tau. Some patients are reluctant to undergo the intrusive procedure. The scans are usually by magnetic-resonance imaging (MRI), to look at the size of the brain, along with a positron-emission tomography (PET) scan to measure the build-up of beta-amyloid. PET and MRI scanners are expensive pieces of kit, running into the hundreds of thousands of dollars. For most people in the world, they are unobtainable.

That explains the excitement at the development of simple blood tests to distinguish Alzheimer’s from other neurodegenerative conditions. One, announced last year and likely to be validated for routine use within 12 months, according to Serge Gauthier, professor of neurology and neurosurgery at McGill University in Montreal, measures a form of tau called p-tau217. It has been found to predict Alzheimer’s with 96% accuracy.

It is already possible using mass spectrometry, which detects how atoms and molecules are deflected by magnetic fields, to measure the level of beta-amyloid in the blood. But it is not certain how this relates to levels in the brain. The accuracy of the procedure rises to 94% if two other risk factors are considered: age and the presence of a form of the APOE gene, known as APOE4, which heightens the risk of developing Alzheimer’s (and also appears to increase vulnerability to heart disease and covid). This can also be detected by a blood test, so Dr Gauthier envisages symptomatic patients giving blood samples for simultaneous tests for both tau and APOE4.

Other approaches aim to detect asymptomatic people years or even decades before they begin to show obvious symptoms. George Stothart at the University of Bath in Britain leads a team that has developed very quick passive tests that hold great promise for detecting cognitive impairment early. These tests use electroencephalographic caps which are worn on the head to compare brainwave responses to a series of images. The caps are relatively cheap and the test can be conducted using a tablet computer.

Taking the exam

Meanwhile, machine-learning and AI are enabling big improvements in cognitive testing, which, like other sorts of pencil-and-paper exams, has been prone to cultural and educational bias, and to a “learning bias” (lessened accuracy as practice improves the participants’ results). Cognetivity Neurosciences, for example, a firm launched by two academics at the University of Cambridge, produces an “integrated cognitive assessment” that has already been deployed by some regions of Britain’s National Health Service. It has also been approved by America’s Food and Drug Administration (FDA) and by regulators in the European Union.

The Cognetivity test relies on a series of flashing images, among which some animals have been embedded for the person being tested to identify—rather like the CAPTCHA tests used by some websites to weed out robots trying to log on, only these are conducted at rapid speeds. So rapid, indeed, that the tests cannot be conducted remotely, but are done on tablets in memory clinics or at doctors’ surgeries. Already, however, a variety of do-it-yourself cognitive tests of various standards are available online, and some of these could become important diagnostic tools.

Even before the explosion in the use of AI, scientists could detect evidence of dementia from how people use words. A study in 2011, for example, found clear retrospective evidence in the writings in her 40s and 50s of Iris Murdoch, a novelist, of the Alzheimer’s she was to die with in 1999, aged 79. The research arm of IBM, a computing giant, has used data from the Framingham heart study, which has tracked three generations of people in a town in Massachusetts since 1948 to improve knowledge of cardiovascular health. Ajay Royyuru, who heads IBM’s health-care and life-sciences research, says that studying the use of language by participants in the study suggests that changes over time can be used to predict which of them will acquire dementia, seven and a half years before they are diagnosed even with MCI.

Such data—and the massive amounts people compile every day on their smartphones, using various services such as messaging or navigation—could help enable much earlier detection of dementia. This might be possible with apps, although the ethics of any non-consensual diagnosis and the willingness of people to use such services are another matter. One way or another, though, Paola Barbarino, chief executive of Alzheimer’s Disease International, an advocacy group, expects a “tsunami” of demand for treatment.

For health-care systems around the world all this may pose more problems than it solves. The main one is the lack of proven treatments. In June the FDA approved Aduhelm, the first drug to treat Alzheimer’s. A monoclonal antibody shown to reduce accumulations of beta-amyloid, it has so far been little used, because it is expensive and insurers are reluctant to approve reimbursement when there are doubts as to whether it actually slows cognitive degeneration. But it is the first in a queue of drugs, for which Alzheimer’s specialists hold out great hopes.

The second difficulty is in assessing when MCI requires medical intervention. As Dr Gauthier points out, some mental decline is indeed part of ageing. People find ways of coping. Telling the difference is hard to do through blood tests or the use of AI. It still requires time and human intervention. This is part of the third and biggest difficulty: that, as the world ages, the number of people with dementia is going to rise rapidly, to more than 80m by 2030 and more than 140m by 2050. Even today health services are buckling. Who knows how they will cope in the decades ahead. ■

To enjoy more of our mind-expanding science coverage, sign up to Simply Science, our weekly newsletter.