NEUROPROTECTION

In January 2008 glaucoma neuroprotection development suffered a setback. Two large phase III clinical trials of oral memantine failed to reach their primary endpoint of slowing glaucoma progression compared with placebo – even though progression was significantly slower in the higher dose group than in the lower dose group.

“Many companies that had been carefully monitoring the emerging activity in glaucoma neuroprotection lost their appetite for what appeared to be a high risk venture, and others that had already made a commitment to develop a neuroprotective drug reallocated their resources from glaucoma into areas they thought might be more productive, including retina,” Robert N Weinreb MD told Glaucoma Day at the 2014 American Academy of Ophthalmology annual meeting in Chicago.

But recent advances may make new neuroprotection trials viable, said Dr Weinreb, of University of California, San Diego, US. “With new technologies, new methods of assessment and our better understanding of clinical trials, I wonder whether it’s time to reassess our ability to do these trials.”

What went wrong? The failed memantine glaucoma trial involved two parallel arms of 1,100 patients each followed for four years. Although the total cost for these trials has not been made public, it is likely that it was sizeable and a huge risk for any investor.

Given memantine’s success in animal models, including monkeys, the failure was especially disappointing, Dr Weinreb said. But animal glaucoma models are difficult to translate to human disease due to differences in anatomy and mechanisms of ganglion cell damage.

Differences in design between preclinical and clinical trials include dosing, timing and intervention duration. Subject age and previous treatment with intraocular pressure-lowering agents are also significant differences.

Detecting progression and defining study endpoints have been especially problematic, Dr Weinreb noted. Regulatory agencies equate glaucoma progression with achromatic visual field loss. However, visual fields are not sensitive in early disease and are highly variable in late disease.

Neuroprotective agents also must be evaluated in glaucoma patients already treated with IOP-lowering drugs, so trials are necessarily longer than required to establish IOP reduction. The lack of endpoint precision combined with slow and uneven disease progress means large numbers of patients must be followed for more than a year to generate meaningful data.

Better endpoints

Still, many glaucoma specialists are unwilling to give up on neuroprotection. Many glaucoma patients progress despite lowering their IOP. It is possible that other factors besides IOP contribute to optic nerve injury in glaucoma including ischaemia, failure of trophic support and others.

New trials could be feasible with changes that address previous weaknesses, Dr Weinreb said. For example, opening participation to patients other than those with advanced disease would make recruiting easier. Including patients at high risk for progression such as those with disc haemorrhage increases the difficulty of recruitment but would reduce the sample size.

Adopting structural as well as functional endpoints might also help. Measures such as retinal nerve fibre layer thickness on OCT are more sensitive for assessing early progression, with changes often appearing long before any measurable vision loss, Dr Weinreb noted. Such testing also does not require pupil dilation and so are less burdensome for patients, enabling quicker data collection and fewer patients lost to follow-up.

Better still may be a combined structure and function index for estimating retinal ganglion cell numbers. One such test, the retinal ganglion cell index developed by Felipe Medeiros MD, PhD, has proven more sensitive than visual function, OCT retinal thickness or mean deviation for detecting both early and late glaucoma progression, Dr Weinreb said (Medeiros FA et al. Am J Ophthalmol 2012;154:814-24).

Dr Weinreb suggests that it may be possible to design a neuroprotection trial with fewer than 500 patients. “In fact if you look at some of the assumptions and tweak them a bit you can get samples even smaller and durations less than one year,” Dr Weinreb said.

He believes a new trial with an appropriate drug and testing parameters has a good chance of succeeding at a reasonable cost.

Robert N Weinreb: rweinreb@ucsd.edu