End of the road for amyloid

In an opinion piece in 2012 I wrote about the spectacular failure of yet another anti-amyloid vaccine trial for Alzheimer’s disease (AD), calling out the idea as detrimental to progress in the field.

Fast forward 8-years and what has changed? The amyloid wrecks have piled up and last week we may have witnessed the final, and long overdue, end of the road.

For therapeutic purposes, at its core the hypothesis has been that excess amyloid proteins (of competing different flavours) are bad for brain cells, causing their loss and dysfunction, in turn leading to worsening cognitive and functional impairment (dementia).

So let’s recap what’s happened. Solanezumab (Eli Lilly’s amyloid vaccine) went into a large Phase III trial in patients diagnosed with mild AD dementia and found to have no evidence of clinical impact, causing the program to be scrapped. That was 2016.

A second anti-amyloid strategy, enzymatically interfering with biological production of amyloid using secretase inhibitors also flamed out. A Phase III trial of verubecestat in patients with mild-moderate AD was halted because of clinical futility in 2017. In fact, troubling observations were reported of dose-dependent increases in adverse events and deaths. 

A reflex refrain from amyloid proponents is “too late”, and so a different Phase III trial was run, this time much earlier in the AD process before definitive dementia diagnosis. This trial was again halted due to futility in 2018. Alarmingly, those in the treatment arm showed faster rates of cognitive and functional decline than those in the placebo group.

In February of 2020 the 8-year DIAN study read out, the most clear-cut test of the amyloid hypothesis to date. Patients with familial Alzheimer’s were treated – a form of AD where we know abnormal amyloid processing is implicated because of an inherited genetic defect. Patients with this rare but devasting form of AD develop symptoms very young (in their 30-50s), and once symptoms begin, progress is often very aggressive.

DIAN results were therefore highly anticipated but again unambiguously negative. Anti-amyloid vaccine treatment, even before symptoms had begun had no impact on the primary cognitive endpoint. In other words, individuals with familial AD experienced no clinical benefit to monthly injections of an amyloid vaccine with known risks and side effects.

Finally, we come to Biogen’s saga over their amyloid vaccine, aducanumab. Biogen launched two large Phase III trials called EMERGE and ENGAGE. In 2019 midway through the trials they were abandoned because a pre-programmed interim analysis found the net results to be so negative as to make continuation pointless.

Yet at the end of 2019 Biogen stunned the field, and financial markets, by resurrecting aducanumab following a data mining exercise. Basically, the company argued that whilst one interrupted trial was clearly negative (ENGAGE), the other interrupted trial (EMERGE) was clinically positive if analysis was restricted to those patients who received the high vaccine dose.

However, a closer look reveals several problems with this argument.

First, what if we restrict analysis to just those who received high dose in the other (ENGAGE) trial? Well, they actually got worse compared to placebo controls. And what of the so-called positive clinical outcome in high dosers in EMERGE? Actually, the benefit was marginal, and if taken at face value, of unlikely clinical meaningfulness. Third, active treatment in both trials, and in the high dose groups specifically, had high dropout rates compared to placebo. This leaves open the possibility that the “positive response” was actually an arithmetic artefact of non-responders excluding themselves from the trial and by so doing pulling up the group average.

The issues are not academic. Aducanumab at higher doses causes brain oedema in about 35% of patients and microhaemorraghes in about 18%, not trivial side effects in the context of a medicine estimated to cost about $US50,000 a year.

Moreover, the treatment is not a pill or one-off shot; Aducanumab means intravenous infusion in a supervised medical environment every 4 weeks, for years, potentially for life.

Any new treatment of sporadic AD with that kind of price tag and capacity for health system disruption better work clinically.

Nevertheless, BIOGEN put this data forward to the FDA as the basis for a new treatment for AD. And last month a very confusing answer came back.

First, on Monday of a roller-coaster week (Nov 2), a highly respected AD researcher from the Mayo clinic, who served as site investigator on the BIOGEN trial, recused himself from the FDA Advisory Committee’s (independent external expert panel) formal review because of potential for perceived conflict of interest.

Rather, he and colleagues published a highly critical viewpoint in a leading scientific journal, concluding with a seemingly self-evident point: “Aducanumab's efficacy as a treatment for the cognitive dysfunction in Alzheimer's disease cannot be proven by clinical trials with divergent outcomes.”

Yet the point appeared lost within the FDA because on Wednesday a pre-emptive internal review was released with commentary that Biogen had shown “robust and exceptionally persuasive” evidence.

Strangely, this seemed to ignore analysis by their own in-house statistician who observed, “…there is no compelling, substantial evidence of treatment effect or disease slowing and that another study is needed to confirm or deny the positive study and the negative study.”

Despite the inconsistency, markets were buoyed by the prospect of a likely positive outcome at the “Adcomm” meeting scheduled for a few days later. Biogen’s shares rose ~40% in anticipation, an $US17B boost to their market cap.

But Friday’s Adcomm was a blood bath for BIOGEN and aducanumab. Inexplicably, the FDA presented their independent experts with a scenario in which interpretation of the partially-positive results of the halted and incomplete EMERGE trial could occur in isolation, pretending as if the ENGAGE trial didn’t exist.

Virtually all Adcomm members took exception to this and voted down the question of evidence of clinical efficacy of aducanumab 10-0 (one member voting “uncertain”).  At the same time, many on the panel openly challenged the FDA’s biased presentation of the data in favour of Biogen.

The FDA may yet choose to authorise adubcanumab but it would do so against the near-unanimous dissent of their own expert panel, a massive risk for their reputation and for the integrity of medical research more generally.

Also on the floor was the amyloid hypothesis itself.

Almost all amyloid trials to date have shown they are actually quite good at mopping up amyloid from the brain. The disconnect at the heart of the amyloid hypothesis is there for all to see: eliminating amyloid from the brain, with varying levels of success, at varying timepoints in the disease process, is of no clinical value.

The trials mentioned in this narrative alone (that is not exhaustive) represent more than 16,000 patient-years and, conservatively, more than $US15B of investment.

The essence of science is to be able and willing to refute your own hypothesis on the basis of observable fact. Those continuing to trumpet amyloid must now articulate what experiment is necessary and sufficient to negate the idea.

There is no question that Alzheimer’s is one of the most pressing and urgent challenges we face, but approval of an unproven treatment is no solution. As noted by members of the Adcomm, that would only set back development of genuine new treatments for years. 

The way forward must now be to retire the amyloid hypothesis and focus the considerable talent, energy and resources of the biotech sector on development of all-new Alzheimer treatments.