Prediction model for late AMD includes genetic, environmental, clinical risk factors

A new model may be more accurate than current prediction models based on case-control studies, which tend to overestimate risks.

Issue: February 10, 2014

A new prediction model for late age-related macular degeneration using data from three large population-based studies found that the most potent predictive combination of risk factors included age, sex, 26 genetic variants, two environmental variables — current smoking and body mass index — and early AMD phenotype.

Perspective from Timothy W. Olsen, MD

The gray zone

“The prediction of AMD in the literature is based on case-control studies, which represent the black and white of the disease spectrum. Most people are in the gray zone, and risk prediction for these people was not possible,” co-author Caroline C.W. Klaver, MD, PhD, said about the inspiration for developing the new model. Case-control studies tend to overestimate the risk of disease.

Klaver, a professor of ophthalmology and epidemiology at Erasmus Medical Center in Rotterdam, Netherlands, leads the ophthalmic division of the ongoing Rotterdam Study, one of three studies in the Three Continent AMD Consortium (3CC) from which participants were recruited. The other two studies were the Beaver Dam Eye Study and the Blue Mountains Eye Study. A fourth study in the 3CC, the Los Angeles Latino Eye Study, was excluded because of the lack of genotype and follow-up data.

“We included all known and speculative genetic and environmental risk factors in a model to evaluate their independent predictive ability,” Klaver told Ocular Surgery News. “The risk score was the additive sum of betas (measure of association) of all predictors (risk factors) in the model.”

The prediction model, published in Ophthalmology, provided 87% accuracy, compared with 73% accuracy for the current commercially available prediction tests, Klaver said.

“We think there is much to gain by using a comprehensive test like ours,” Klaver said.

Some factors had a stronger predictive effect, such as the complement factor H (CFH) and age-related maculopathy susceptibility 2 (ARMS2) genes, she said.

A total of 10,106 participants were included in the study, all with gradable fundus photographs, genotype data and follow-up data without late AMD at baseline. The incidence of late AMD was determined during four to five examinations over a median follow-up time of 11.1 years.

Generalizing results

Incident late AMD — the study’s outcome variable — developed in 363 people, whereas early AMD developed in 3,378, and 6,365 remained free of any AMD.

“The final model performed very well when used in study populations from the U.S. and Australia, so results are very generalizable,” Klaver said. “Also, incorporating the patient’s retinal abnormalities in the model was better for prediction than [using] only a minimal set of genetic factors.”

Klaver said that “quick and dirty” prediction testing for AMD is not very accurate.

Limitations of the new prediction model include the relatively low number of late AMD cases and the exclusion of several risk factors, including dietary factors, biomarkers and rare genetic variants.

Klaver and colleagues are now investigating anatomic markers visible on optical coherence tomography as predictors for the course of AMD.

“I also look forward to the development of therapies that target high-risk individuals and are capable of lowering their risk before it is too late,” she said. – by Bob Kronemyer

Reference:

Buitendijk GH, et al. Ophthalmology. 2013;doi:10.1016/j.ophtha.2013.07.053.

For more information:

Caroline C.W. Klaver, MD, PhD, can be reached at Department of Ophthalmology, Erasmus Medical Center, P.O. Box 2040, NL-3000 CA Rotterdam, Netherlands; 31-10-7033691; email: c.c.w.klaver@erasmusmc.nl.
Disclosure: Klaver has no relevant financial disclosures.

Perspective

Timothy W. Olsen, MD

What is AMD? What is the accepted definition used to describe this clinical condition, especially to a patient? What are the risk factors for predicting progression? There are currently many phenotypic descriptions for the entity referred to as AMD. In fact, patients are often confused and frustrated because one physician will indicate that they have AMD while another will say otherwise. They really are asking the physician to assess their risk.

Buitendijk and colleagues assess risk in a meta-analysis of three large population-based studies from three continents to determine the incidence of late AMD using the best predictive factors, including age, sex, 26 genetic polymorphisms, smoking, body mass index and baseline phenotype. Without any information, other than age, one’s risk for progression to late AMD is 17% by age 90 years (three late AMD cases occur per 1,000 person-years). Other than age, fundus phenotype is the most important predictor of late AMD. Smoking and major genetic variants add to the predictive value and determine the highest predictive risk strata (most risk factors), increasing one’s risk to late AMD to 66% by age 90 years.

Timothy W. Olsen, MD

OSN Retina/Vitreous Board Member

Disclosures: Olsen has no relevant financial disclosures. He has received grants from the National Institutes of Health/National Eye Institute and from the following foundations: R. Howard Dobbs Jr. Foundation, Minnesota Lions, Research to Prevent Blindness, Abraham J. and Phyllis Katz Foundation, Emtech Biotechnology Grant, Georgia Research Alliance and The Fraser Parker Foundation. He also holds patents for Scleral Depressor 8,083,751 B2 (US, 12-27-2011), Tissue Support Structure: 13/511,690 (US Pending) #1 986 581 (Europe Oct-3-2012) and Postop Pain Formulation: 14011 Prov 09-27-2013.