Bernard Gilly, co-founder and CEO of Gensight Biologics SA, was upbeat about the readout of Gensight's phase III Rescue study testing its gene therapy, GS-010, in patients with 11778-ND4 Leber hereditary optic neuropathy (LHON).

"These results increase our confidence level in the efficacy of GS-010," Gilly told investors on a conference call.

Investors didn't share that opinion, sending shares of Paris-based Gensight (Paris:SIGHT) down €0.61, or 21.7 percent, to close at €2.20 on Monday, after the 39-patient Rescue study missed its primary endpoint of a +15-letter difference in visual acuity improvement for GS-010-treated eyes compared to sham-treated eyes at 48 weeks.

GS-010 is designed to treat LHON patients with mutations in NADH dehydrogenase 4 (ND4), which are maternally inherited through mitochondria. Using adeno-associated virus (type 2), the gene therapy delivers an ND4 gene tagged with Gensight's mitochondrial targeting sequence, which shuttles the mRNA from the nucleus into the mitochondria, where the protein is expressed and integrated into complex 1 of the respiratory chain.

GS-010 wasn't even close to hitting the 15-letter threshold of the primary endpoint. At 48 weeks, the change from baseline for GS-010-treated eyes was -19 ETDRS letters equivalent, while sham-treated eyes had a -20 ETDRS letters equivalent, a difference of just one letter.

Rescue enrolled patients who only needed to have the disease in one eye, which the company believes could have contributed to the disappointing data readout at 48 weeks. "We're enrolling and catching these patients early – in the more acute phase of the disease," said Barrett Katz, Gensight's chief medical officer.

Eyesight of LHON patients tend to reach a low point, or nadir, in three to five months, before stabilizing, which was seen in both treated and untreated eyes in the Rescue study.

Looking from nadir, patients had an improvement of 13 ETDRS letters equivalent in GS-010-treated eyes, compared to an 11-letter improvement from nadir in the sham-treated eyes.

"These changes of visual acuity from nadir are counter to what we would expect from what we know of the natural history of the disease," Katz said.

Gensight also saw an improvement in both GS-010-treated and sham-treated eyes in another phase III study, called Reverse, which enrolled 37 patients who had visual loss due to LHON between six months and 12 months prior to study treatment. In Reverse, both groups improved 11 ETDRS letters from baseline at 48 weeks. By 72 weeks, the improvement was 15 letters for the GS-010-treated eyes and 12 letters for the sham-treated eyes.

Beyond the post-nadir improvement, there were other efficacy readings that suggest GS-010 is having an effect. For instance, GS-010-treated eyes were 2.9 times more likely to have 20/200 or better vision, the threshold for legal blindness, than sham-treated eyes (p=0.0347).

"This is what the patients are looking for. This is what their families are looking for. This is what the community of clinicians is looking for," Katz noted of the ability to not go legally blind.

In a responder analysis of high-contrast visual acuity, 24 percent of patients had at least 0.3 LogMAR (15 ETDRS letters) better response in their GS-010-treated eyes compared to their sham-treated eyes. Similarly, a responder analysis of low-contrast acuity measured on the Pelli-Robson scale showed 24 percent of patients had GS-010-treated eyes that were at least 0.3 LogCS better than sham-treated eyes.

Anatomic measures of the eye trended in the right direction, but weren't statistically significant, including the change from baseline of temporal retinal nerve fiber layer (p=0.0513) and changes from baseline of papillomacular bundle thickness and ganglion cell volume (both p>0.05).

"When you look at this from afar and make best sense out of it, our thoughts are that a 48-week readout may be a premature look to recognize the true efficacy and power of the intervention we've offered," Katz concluded.

Reverse and Reflect

Like in the Reverse study, the Rescue study will evaluate patients at 72 weeks and 96 weeks, the data for which are scheduled to be available later this year. Data from the Reverse study at 96 weeks are due in the second quarter of 2019.

Gensight is running a third phase III study, called Reflect, testing the effect of GS-010 in 90 later-stage patients. The patients with LHON due to an ND4 mutation can have vision loss up to one year in duration.

Reflect, which treated the first patient in March 2018, has an active arm where both eyes of patients will be treated with GS-010 and a control arm where the first affected eye will be treated with GS-010 and the second eye will get a sham injection. The primary endpoint of the study will measure the LogMAR at one-year post-treatment in the second-affected/not-yet-affected eyes.

If it makes it to market, GS-010 will have limited competition. Santhera Pharmaceuticals AG, of Pratteln, Switzerland, markets Raxone (idebenone), an oral small-molecule analogue of co-enzyme Q10, for LHON in more than 20 countries, although the U.S. isn't one of them. Through the first nine months of 2018, sales of Raxone came in at CHF 23.6 million (US$23.65 million).