Belite Bio, Inc American Depositary Shares 12th Annual Cantor Fitzgerald Global Healthcare Conference

Hao, what do you think?

Sure. I can start. Yes, so far, I think everything has been very smooth. The FDA shows a lot of support throughout the last few years, and I think the whole thing as we expected. We don't want to comment too much about potentially what is coming, but definitely we see a lot of friendly, constructive processes going on with the FDA. We will update the market whenever we have any things that we can actually confirm.

Happy to, and I can only agree with Hao about the constructiveness of the FDA and how educated the regulatory agency actually is when it comes to endpoints in the field of ophthalmology, especially retina. The interaction with the FDA on finding and testing, validating, and implementing an endpoint for Stargardt disease goes back to the year 2011 when, initiated by the Foundation Fighting Blindness, a group of investigators met with the FDA in Bethesda in person. It was before all these online meetings, and we discussed about what would be needed to develop therapy for Stargardt in an arena where traditionally only visual acuity was an accepted endpoint. It was about that time when it was found that in GA, you can actually use atrophy image on autofluorescence imaging that correlates strongly with sensitivity loss in the retina to be used as a primary endpoint in GA.

We had a precedence there in an important other disease and discussed with the FDA what would be needed, and they recommended a natural history study conducted as if it was a trial and to specifically look into imaging technologies such as autofluorescence and OCT to allow outcome measures to be validated. This is exactly what we did that led to the birth of the so-called Progression of Stargardt Disease study called ProgStar, and that was a worldwide natural history multi-center study looking into the progression of Stargardt disease in 6-monthly intervals prospectively and investigating a large database retrospectively. That allowed to establish autofluorescence imaging as an outcome measure and specifically the definition of lesions called DDAF, definite decreased autofluorescence, as a primary endpoint.

ProgStar has been instrumental for the design of the DRAGON trial and instrumental for regulatory, especially FDA, how they understand Stargardt disease and why they are willing to accept an imaging endpoint as a real approval endpoint.

Yeah. It's important that you mentioned supplemental data because obviously we seek approval based on one trial, and it is not a small ask with the FDA to approve Tinlarebant based on a single trial. But the trial data speak for itself, and we have a large package of so-called confirmatory data, and this is what is needed as a guidance of the FDA, what is being regarded as supporting evidence for a single trial, and we have that. First of all, in the trial itself, we have a key secondary endpoint that we call DAF, decreased autofluorescence, meaning all areas that show a decrease in signal when we measure autofluorescence imaging.

That goes beyond just areas that are completely atrophic. When we look at DDAF, definitely decreased autofluorescence, and we look at histology, we find there's no photoreceptor left and there's no sensitivity left when we look at microperimetry. So these areas are completely atrophic. We have shown in the trial that Tinlarebant has a very significant efficacy signal in slowing down the growth of these lesions. That's clinically meaningful because a patient wouldn't see in these areas, and we can slow the growth of these lesions by 36%. When we speak about DAF, that's a more comprehensive measure of health or sickness of the RPE. DAF is the sum of DDAF, obviously, but also QDAF, questionably decreased autofluorescence, meaning these areas show sick RPE but are not as sick as the area of DDAF.

What we find for DAF is that we find essentially the same efficacy signal, and that means because DAF could grow slower than DDAF because when DDAF is expand, then QDAF could become smaller on the expense of DDAF, but it is not what we find. We find that in the trial, DAF grows as fast as DDAF, meaning that not only allow us Tinlarebant to slow down the growth of these completely atrophic lesions, it appears to also slow down the transition from healthy RPE into sick RPE, meaning that likely that would allow to prevent incident lesions in the future. So this is trial data. I think this is confirmatory evidence from the trial, is the fellow eye data. We are not used to that any longer in ophthalmology because over the last 20 years, essentially every innovation was eye specific.

If it was a device, if it was gene therapy, surgery, what have you, all of that was eye specific. We have an oral therapy. This has a systemic effect on both eyes. We looked into the fellow eye data that was available, and we find essentially the same effect size. We find statistical significance, although the trial could not optimize for the fellow eye. The fellow eye was all commers. There could be no lesion in the fellow eye, there could be a very large lesion in the fellow eye, and still we reached a significance level when we analyzed the fellow eye. So obviously, Tinlarebant works on both eyes together. When it comes to confirmatory evidence, we have access to this very large natural history database of the ProgStar study.

We use that database to supplement our placebo arm in the DRAGON trial by a very large number of natural history controls. This has a very positive impact on the result that is already very positive, but it bumps the efficacy signal even further up and it bumps the P value even further down if we use that natural history controls out of ProgStar.