Lecanemab doesn’t appear to work in women

Data from the CLARITY trial this week was supposed to be the crowning glory of the amyloid hypothesis, vindication for proponents of this long-held but much-maligned theory of Alzheimer’s disease.

Yet the results left many feeling underwhelmed, and even the study authors noncommittal.

The amyloid hypothesis has been under a cloud for well over a decade and of late only suffered more setbacks.

First came the aducanumab saga, an anti-amyloid treatment that received regulatory fast track approval in the US despite halting of two large clinical trials prematurely for futility, and the FDA’s own independent expert advisory panel unanimously voting against approval.

The resulting conflagration saw the FDA, Biogen and Alzheimer lobby groups’ reputations damaged and a stern reaction from the largest health payor in the US, the Center for Medicare Services (CMS).

CMS refused a traditional funding route for aducanumab because of lack of evidence of “clinical benefit”, leaving it commercially dead in the water. In fact, CMS made clear that any new anti-amyloid now needs to meet this criterion (“meaningfully improve health outcomes”) in order to access public funding. We’ll come back to this critical point.

Next, came what the field assumed to be the definitive test of the amyloid hypothesis. The Alzheimer Prevention Initiative (API) enrolled exclusively from a Colombian kindred with genetically-determined Alzheimer’s, where there is no question that the disease is caused by excess amyloid protein in the brain leading to people to develop symptoms whilst relatively young.

Because in the API study anti-amyloid treatment was commenced years before clinical onset, the idea was that dementia should be delayed or even prevented altogether. However, contrary to these expectations, the group treated by Roche’s crenezumab experienced no clinical advantage whatsoever.

For this reason, all amyloid bets were on Eisai’s lecanemab, originally developed by the Swedish company BioArtic with the premise of better targeting of a particular kind of amyloid, the soluble amyloid protofibrils – in contrast to insoluble amyloid that is visible under the microscope as classic fibrillar plaques. Whether soluble vs insoluble amyloid is the main culprit (if at all) remains of course a matter of intense debate in the Alzheimer field, but let’s just go with the narrative flow here.

The CLARITY trial has many admirable features. It recruited ~1800 people from around the world and managed to avoid the universal white person syndrome of previous trials. 17% of the cohort was Asian and 12% Latino, making it more representative of modern populations. Yet despite more US sites than any other, less than <5% of the cohort were black. Alzheimer trials in the US will need a radical shift in strategy in order to overcome this glaring deficit.

Choice of primary and secondary outcomes were impeccable. The primary outcome was the Clinical Dementia Rating sum of boxes (CDR-SOB), a scorecard of sorts rated by a clinician across the domains of memory, orientation, problem solving, community affairs, home duties and personal care. Secondary outcomes included standard measures of global cognition, daily function, as well as biomarkers in the brain, from the CSF and blood.

So what happened? In short, the group who received fortnightly infusion of amyloid antibodies deteriorated by 1.21 points on the CDR-SOB whilst the placebo group deteriorated by 1.66 points, the average difference being 0.45 points. That’s right, less than half a point on the CDR-SOB! Less than the smallest change practically possible at an individual level and only calculable by comparing group averages.

At the same time, participants’ PET scans showed massive reductions of amyloid. Almost 75% of baseline cerebral fibrillar amyloid was removed; the majority of individuals moving from amyloid “positive” to amyloid “negative”. Whilst amyloid removal has been seen in prior trials using different anti-amyloids, this is the most convincing biological result yet and indeed difficult to imagine a better brain imaging result.

So we may well have witnessed ‘peak’ amyloid clearance but a clinical result of less than half a point. Does it matter? This is where the debate has quickly moved to. There are three reasons to think CLARITY endpoints will not meet the meaningful health impact criterion set by CMS.

First, the authors’ own analysis showed that lecanemab treatment did not statistically alter chances of transition between dementia and pre-dementia states. They cite a hazard ratio of 0.69 as ‘numerically’ better but the lack of any statistic speaks for itself.

Second, independent to this trial a thorough analysis has considered what threshold may be considered meaningful for different clinical measures in Alzheimer’s disease. These researchers proposed a relative improvement of >1.0 CDR-SOB, or >3-points on the ADAS-COG global cognitive measure. In CLARITY, neither the primary outcome (0.45 CDR-SOB points) or secondary outcome (1.44 ADAS-COG points) met such thresholds.

Third, the most troubling aspect was no statistically significant result –  let alone clinically meaningful outcome –  in women on the primary outcome or any clinical secondary outcome (Supplementary Figures 1-4). In other words, the headline readout was driven by positive outcomes in men and it is beyond remarkable this was not discussed in the paper. This can not be explained by a lack of statistical power and suggests a more fundamental biological interaction at the level of therapeutic potency or mechanism of action. This will need a lot of careful analysis and further investigation to sort out.

Given half the target population may be non-responsive, lecanemab is going to struggle to get to the clinic.

Finally, recall these ‘meh’ results are in the face of not inconsiderable or inconsequential risk of side effects. 26% of those receiving their fortnightly IV dose experienced an infusion reaction, and a 21.5% incidence of ARIA, a form of brain imaging abnormality related to bleeding and oedema unique to those treated by anti-amyloids. Whilst most turn out to be harmless, about a quarter are clinically significant, including stroke, and in any case ARIA abnormalities need expert diagnosis and management.

Altogether, lecanemab is the most potent anti-amyloid to date but like all its therapeutic cousins has not been able to show a clinically important difference compared to placebo.