Full publication of pivotal trial data outlines Akili's case for FDA regulation

The large, multi-site investigation described clear improvement in objective symptoms of pediatric attention deficit, but was a bit muddier when describing subjective measures and long-term efficacy.
By Dave Muoio
07:50 pm

A full transcript of STARS-ADHD — the pivotal trial of Akili Interactive’s video game-like pediatric attention deficit treatment — has been published in The Lancet Digital Health journal, providing onlookers a clear view of data fueling the digital therapeutic company’s pending FDA submission.

The writeup echoes topline findings first unveiled by the company in December 2017: compared to a prescribed sham game, AKL-T01 significantly improved objective measures of attention among study participants after four weeks treatment. However, the transcript also offers specifics on the study’s more subjective measures of interest — many of which improved from baseline, but did not significantly outpace similar gains posted by control patients.

Significant improvement in pediatric attention disorder symptoms

From July 2016 to November 2017, the randomized, double-blind, active-controlled study enrolled 348 children with ADHD diagnosis across 20 study sites. The participants, aged eight to 12 years, must have discontinued any other medication for ADHD at least three days before baseline, and have demonstrated a Test of Variables of Attention (TOVA) Attention Performance Index (API) score of − 1.8 or below.

Those in the intervention group (n = 180) received a tablet with AKL-T01, which requires players to focus on navigating an alien avatar through a course while responding to on-screen prompts. The control group (n = 168) received an educational word game that was also designed by the company.

The research team told parents and children that there were two different interventions being tested, so that both study groups would believe that they had received the investigational treatment. All participants received a recommendation to play their game for 25 minutes per day, five times a week, for four weeks.

The primary outcome of interest was the mean change in TOVA API — an objective, validated measure of attention and inhibitory control cleared by the FDA — after the intervention period. In this, the researchers saw a greater mean change in scores among the intervention group (0.93) than in the control group (0.03 ;adjusted = .006). Post-hoc within-group analyses also showed a significant improvement among those receiving the intervention (< .0001), but not the control.

AKL-T01’s positive results also persisted during responder analyses. Here, 47% of the intervention group met the pre-specified threshold for TOVA API score improvement versus 32% of the controls (= .0058). Additionally, Akili’s experimental therapy’s improvement brought 36% of the intervention group into the normative range of one or more objective measurements of attention, as opposed to 21% of the controls (= .0027).

“Because our main mechanism of action is really the attention aspect of cognition, it made sense to us to use a very specific cognitive measure of attention as our primary,” Dr. Anil Jina, chief medical officer at Akili, explained to MobiHealthNews. “You can’t really lie on it or cheat on it. We got very clear separation on that, and that’s an obvious first pass when we look and say ‘Is this having a cognitive effect?’”

AKL-T01 improved secondary measurements — but so did the sham

The performance of Akili’s treatment become a little more conditional when reviewing the STARS-ADHD trial’s secondary endpoints. These included a variety of scales and tools that Jina described as measurements of behavioral or subjective cognitive impact, rather than objective symptom changes.

Patients who received AKL-T01 demonstrated significant improvements across each of these measures — but not enough of a boost to outpace their peers receiving the sham therapy.

“What we saw across all the secondary endpoints was an improvement compared to baseline, but the separation from control did not happen across any of them,” Jina said. “It’s very difficult to control for a video game-based treatment. We had an educational word game which was designed to really try to show the attention differences, [but] we hadn’t really studied how it would affect other scales.”

While Jina said there was no “obvious answer” for why the dummy game generated comparable results to the digital therapeutic, he suspects that its misleading design might have led patients and their parents to believe in — and subsequently report to researchers — its efficacy.

“We used a control. There’s were expectancies at baseline and, surprisingly, because we chose an educational word game, the expectancy from those children and parents was that the word game was going to give them more benefit in their ADHD than our treatment,” he said. “So you could make an argument that the placebo potential for something with higher expectancy was higher for the control, in this case.”

In spite of this, Jina said that the secondary findings they received still show promise. Differing results from the Impairment Rating Scale (IRS) and ADHD Rating Scale-Inattentive (ADHD-RS-Inattentive) measures showed a trend of improvement among intervention patients that he said might have become more pronounced with a longer treatment period. In addition, focused analyses among responding participants weighed more heavily in favor of AKL-T01, with the researchers highlighting a significant difference in IRS among the responders (= .05).

“When you get a clinical scenario like that — where the means both show an improvement versus baseline and there were some numerical differences — you then move onto the next way of looking at it, which is a responder analysis. In this case, [it was looking at] children who responded to treatment across these different measures,” Jina said.

“When we look at that and we do the responder analysis, pretty much every measure, especially those related to attention, … favor our treatment versus the control. Even on some of the measures where the numerical separation wasn’t obvious, when we look at the responder analysis, everything favored our treatment.”

A family of data for the FDA, clinicians to consider

When Akili first announced STARS-ADHD’s topline data, the large study’s results were intended to serve as the backbone of its De Novo device clearance application.

A few years later and there’s still no clearance, but the findings now have additional support from the STARS-ADHD Adjunctive trial, a follow-on investigation that provided the treatment for a longer period of time. Although smaller in scale that the pivotal trial (n = 206), this study provided AKL-T01 to patients already receiving stimulant medication and saw significants improvements one month after baseline in parent-reported IRS scores (p < .001).

“Understanding the benefit of our technology when used alongside ADHD medications has been a research priority for us. Importantly, parents see improvements in their children regardless of whether they are using the treatment alone or alongside stimulants,” Eddie Martucci, CEO of Akili, said last month in an announcement of the adjunct trial’s results. 

Jina said that these studies, taken in conjunction with the company’s other trials, should paint a clear enough picture of AKL-T01 for regulators and clinicians alike. To this point, he said that the company has “no other ongoing work at the moment” investigating the safety and efficacy of AKL-T01 with one notable exception: a project concluded two weeks ago that uses EEG to explore the AKL-T01’s mechanisms of action (which Jina said is “unnecessary” from a clinical and regulatory perspective).

Now, he said, the main question for critics is whether the treatment will maintain its effect over longer periods of time. With adverse events limited only to occasional reports of headaches and frustration, he said the company is waiting until the product has hit the market before it kicks off any work in this area.

“Six months, a year, what happens in the longer run? We believe we can get a lot of that information once we’re on the market because the safety risk is so low,” he said. “There’s not really any other boundary.”

But of course, Akili won’t be able to release its prescription digital therapeutic without the FDA’s blessing. Jina acknowledged some “timeline frustration” ongoing within his company, but also acknowledged that AKL-T01 represents an unusual challenge for regulators. Without any precedents to look toward, all Akili can do is present its case and keep an open line of dialogue with the agency.

“The data is definitely supporting us in terms of showing clinical meaningfulness for improving attention within ADHD, and we just have to keep working with the FDA to keep progressing our status forward,” he said. “I’d love to be able to give you a timeline as to what that looks like — believe me, we’d love to know a timeline ourselves and what that looks like. … We’re still pursuing it, and we have to believe that it’s meaningful.”


The latest news in digital health delivered daily to your inbox.

Thank you for subscribing!
Error! Something went wrong!