Around fifty treatments for AD are in phase II, so the cupboard is certainly not bare. The question is how best to evaluate new agents in relation to their molecular targets. Imaging probably holds the key, Nick Fox, of the Institute of Neurology, University College London, UK, told a well-attended plenary session.

Heterogeneity kills trials

In the past, we have had 10-15% of patients entering pivotal trials without the disease we are supposed to be treating and whose brains – at post mortem – did not express the molecular target our drug is designed to engage. Even when great efforts were made to exclude non-AD patients, we have found that many had vascular or frontotemporal dementia. This gives any targeted treatment a big hurdle to overcome even before the race has started.

The solution will be to use markers, which include those in the CSF but will probably focus on imaging, to enroll patients with homogeneous pathology and defined disease stage. We have imaging of macrostructure, microstructure, function and molecular pathology. We are well along the road with amyloid, and we have started out with tau.

New treatments will be pathology specific and prolonged

Monitoring efficacy and safety

Clinical outcome remains paramount. But scores on clinical scales are “noisy”: they are non-linear, performance in the same patient varies from day to day, and there are floor, ceiling and practice effects.

From what we know at present, no imaging variable is an adequate surrogate. That said, the precision and objectivity of an imaging scan is a welcome additional source of information about likely efficacy. And seeing a relevant change in the molecular or pathological target can give us the courage to continue from a positive proof of concept study to a large, randomized trial. We can also use imaging to evaluate progressive tissue loss, either in the whole brain or specific areas such as the hippocampus.

The new Alzheimer’s agents will be pathology-specific, and -- if we aim to suppress a disease process that has been active over many years -- treatment will be prolonged.

We can use scans to monitor the development of new comorbidities, such as stroke, that may mask treatment effects. And we can see toxicities in the brain before they become clinically evident. The occurrence of amyloid-related imaging abnormalities (ARIAs) is a case in point.

We had not expected that use of antibodies would result in greater permeability in blood vessels affected by amyloid angiopathy. Imaging picked this up in patients who had not yet developed symptoms, leading to interruption of treatment and resolution. Subsequent imaging showed that treatment could safely be resumed in many patients, and even hinted that the occurrence of vasogenic oedema might correlate with treatment efficacy.

PET makes a difference to diagnosis

Amyloid PET alters diagnosis and treatment

In an earlier presentation in the symposium, Philip Scheltens, VU University Medical Centre, Amsterdam, the Netherlands, presented preliminary data from Dutch study showing that knowing whether an amyloid PET scan was negative or positive changed diagnosis and subsequent management.

Traditionally, imaging has been used to exclude treatable conditions such as a brain tumour or hydrocephalus. But these are found in only 1-5% of cases. Now the aim is to have imaging aid a positive diagnosis of AD. And PET imaging using an 18F-labelled tracer is as good at detecting amyloid as post mortem examination.

In the study, 211 patients were diagnosed on clinical grounds as having AD or non-AD dementia. At this stage, clinicians rated their degree of confidence in the diagnosis. The patients were young (mean age 62 years) and had relatively early disease.

Patients then had amyloid PET, and the results were fed back to the clinicians. Of the 145 people initially diagnosed on clinical grounds as having AD, 77% had a positive PET scan, but 23% did not. Knowing the amyloid status of their patients led clinicians to revise their diagnoses in the majority of cases where there was a discrepancy between clinical findings and PET. This effect was particularly strong when clinicians had not been confident about their initial clinical diagnosis.

Opportunity to intervene in early disease

Importantly, knowing the PET scan results led to a change of management in 37% of cases. Some patients who had been put on treatment had it stopped; and some who had been offered no treatment were prescribed drugs.

Making a correct diagnosis is important, not least because patients and their families want to know the prognosis. In an everyday clinical context, knowing the amyloid PET result makes a difference to the diagnosis and management of people who have early dementia.

But correct diagnosis is also crucial to effective research, and imaging is proving crucial. The next question is whether we can use PET to diagnose preclinical AD in people who are at high risk but still asymptomatic.

In population studies charting incidence, there is a 20-25 year lag between tau deposition and AD. So the agent closely linked to neurodegeneration is in place decades before its effects become evident. In theory at least, this long latency gives us ample opportunity to intervene.