Could Metformin Fill the Bill for Geographic Atrophy Prevention?

STOCKHOLM — The risk of developing geographic atrophy (GA) declined significantly in patients who had a history of treatment with metformin, a case-control study involving claims data showed.

Overall, the prevalence of GA was 12% lower in patients with or without diabetes. The magnitude of reduction increased to almost 50% in patients without diabetes. Applying the findings only to patients without diabetes yielded a number needed to treat (NNT) of less than 100.

The findings add to other data showing an association between metformin use and a reduced risk of age-related macular degeneration (AMD), including GA, or dry AMD, said Dimitra Skondra, MD, PhD, of the University of Chicago, at the American Society of Retina Specialists meeting.

“We still have an unmet need for therapeutics to prevent geographic atrophy,” said Skondra. “Metformin use was associated with decreased risk of new-onset ICD coding for geographic atrophy of up to 47%. It has the potential to be repurposed to prevent GA, given how easy and accessible it is. Additional studies are needed to confirm these associations and to motivate a clinical trial.”

Following the talk, co-moderator Raj Maturi, MD, of the Midwest Eye Institute in Indianapolis, asked Skondra whether she currently uses metformin among patients with early and intermediate AMD and a high risk of GA, such as a family history of AMD.

“We don’t yet have the clinical data to say for sure that it works, only indirect evidence,” Skondra replied. “There is a strong body of evidence from preclinical data and big data. So what do I say when patients ask me what they should do? They have done genetic testing, they have a strong family history. They have noticed changes in the quality of their vision and color sensitivity.”

“I discuss what is out there and I let them decide. But I ask them to have a discussion with their family doctor first to see whether there might be some side effects or interactions with other medications. I think we need to do more work before we offer it to patients widely.”

Metformin has several desirable characteristics that make it an attractive option for the unmet therapeutic need for early intervention in GA, said Skondra. The drug is widely available and inexpensive, has an excellent side effect profile — including no risk of hypoglycemia — and an expanding off-label use for prediabetes, gestational diabetes, polycystic ovarian syndrome, and metabolic syndrome.

Several recent studies have suggested that metformin has potential to slow or prevent AMD, including GA. Three case-control studies showed associations between metformin use and a decreased risk of AMD, a reduced risk of new-onset dry AMD, and a decreased risk of AMD in nondiabetic patients. Skondra referenced a half dozen other studies that have added to the evidence that metformin might have a favorable effect on AMD.

Continuing to explore the relationship between metformin and AMD, Skondra and colleagues performed another case-control study using Medicare and commercial claims databases. The study included patients ≥60 with new-onset GA as identified by ICD-10 coding. Data analysis included 10,505 patients with GA and a control group of 10,502 patients without GA. Additionally, investigators matched a subgroup of 7,611 patients with GA but no diabetes to a control subgroup of 7,608 nondiabetic individuals without GA. The cases and controls were well balanced with respect to age, region, year, comorbidity score, and hypertension.

An evaluation of the entire study population (21,007 patients) produced an odds ratio of 0.88 for metformin and new-onset GA (95% CI 0.79-0.99). An analysis limited to the 15,219 patients without diabetes produced an odds ratio of 0.53 for metformin use and new-onset GA (95% CI 0.33-0.83).

Investigators then calculated the NNT, based an estimated 2.4% of patients with bilateral early or intermediate AMD that progresses to GA within 2 years (lower than recent actual estimates, said Skondra). The calculations showed an NNT of 355 across the total study population, decreasing to 90 when limited to patients without diabetes.

Skondra acknowledged limitations of the study: lack of clinical granularity in data from a claims database, uncertainty about how — or if — the findings would extend to publicly insured patients and the uninsured, and an inability to ensure the cases and controls were balanced with respect to race and ethnicity.

Please enable JavaScript to view the

comments powered by Disqus.