Framingham Heart Study Director Daniel Levy

In 1948, as death rates from cardiovascular disease continued their gradual yet steadfast rise from decades prior and were more than a little worrisome, the National Heart Institute established a novel study in the historic town of Framingham, Mass.

Little was known then about the general causes of heart disease and stroke. The Framingham Heart Study, a longitudinal study originally composed of 5,209 Framingham adult residents, ultimately established most of the risk factors that are now embedded in the vernacular:  high blood pressure, high blood cholesterol, smoking, obesity, diabetes, and physical inactivity.

Associate Director Christopher O'Donnell

To embark on a long-term study of thousands of people who had no overt symptoms of cardiovascular disease was an ambitious project not without its share of criticism. Early peer-reviewed papers from the Framingham research team mostly justified the epidemiological proach to studying heart disease. 

But by the late 1950s, the study was bearing fruit, as researchers began to see the effects of high blood pressure and cigarette smoking on the hearts of their subjects, who returned every two years for a detailed examination. 

Dozens of landmark papers followed. Framingham, the city, soon became synonymous with the study. In 1971, the study enrolled a second-generation cohort—5,124 of the original participants’ adult children and their spouses. Then the grandchildren joined in 2002.  Throughout this period the study enjoyed the continuous support of NIH extramural funding.

Beating Stronger Every Day

Sixty years and nearly 2,000 journal articles later, the NHLBI’s Framingham research team has joined the NIH intramural program. The study is still within NHLBI, only now it has reinvented itself once again. Indeed, the Framingham Heart Study remains as innovative and promising as it was a half-century ago.

The impetus for the move was to “leverage 60 years of data collected at Framingham” with intramural resources, such as in gene-expression profiling and bioinformatics,” in order “to make the investment all the more cost effective,” said Daniel Levy, who joined the study in 1984 and became its director in 1995.

Levy has numerous projects under consideration that can best be done in-house. Entering a feasibility stage this June is a study to correlate gene expression with phenotypes and 500,000 genotyped SNPs in Framingham participants, using microarrays (perhaps better described as “macroarrays”)—a 96-sample peg plate instead of the standard chip. NHLBI has the core facility to undertake this project in the Clinical Center.

Modern probes of the heart: The NHLBI Genomics Core can study the gene expression profile of 96 samples at once on one plate. Facility director Nalina Raghavachari holds one such plate

“We’re looking for new biology,” said Peter Munson, head of the Mathematical and Statistical Computing Laboratory within CIT’s Division of Computational Bioscience, who shifted his attention to the Framingham project a year ago to develop data-analysis software. Munson describes Framingham’s new focus as “a melding of good, old-fashioned science and people with lots of technical know-how.”

Genome of the Heart

Framingham hinted at its new direction in recent years with several large-scale genotyping projects, such as a genome-wide scan of nearly 100,000 SNPs from 1,345 study subjects, using the Affymetrix 100K Genechip. That study, published as a series of 17 papers in September 2007, included one paper exploring associations between the SNPs and four major cardiovascular disease outcomes: major atherosclerotic CVD, major coronary heart disease, atrial fibrillation, and congestive heart failure. Although there were no blockbuster results for those traits, intriguing findings emerged, most notably the replicated associations of chromosome 9p21 with major CVD.

Digging deeper, NHLBI launched the Framingham SNP Health Association Resource (SHARe) project in early 2007, under the leadership of Christopher O’Donnell, Framingham’s associate director and SHARe’s scientific director.

They upped the ante, too, with the goal of genotyping approximately 550,000 SNPs in more than 9,000 participants from  three generations, encompassing more than 900 families.

Stored within NCBI’s dbGaP and, as its acronym signals, open to scientists worldwide (beginning in October 2008), the SHARe database will contain all previous Framingham SNP and microsatellite genotyping, as well as extensive phenotype information from the three-generation cohort:  quantitative measures such as systolic blood pressure, total and HDL cholesterol, fasting glucose, and cigarette use; anthropomorphic measures such as body mass index; biomarkers such as fibrinogen and C-reactive proteins; and electrocardiography measures such as the QT interval. 

SHARe contains the 500K-SNP data and may possibly house much more.

Core scientists: (left to right) facility director Nalina Raghavachari, technologist Kimberly Woodhouse, and research biologist Poching Liu

Using the SHARe resource, O’Donnell hopes to discover new genes underlying coronary heart disease and subclinical atherosclerosis detected by computed tomography and other imaging measures. Levy hopes to discover genes involved in hypertension and altered vascular function.  The project might relate common genetic variation to alterations in gene expression as a means to get one step closer to understanding functional changes in human DNA.

“The focus is on discovering new genomic and genetic risk factors to identify the specific genetic sequences underlying associations seen previously  and to test how these new genetic risk factors might be used to predict and prevent cardiovascular disease,” said O’Donnell, who joined the Framingham Heart Study in 1996 and, like Levy, joined the intramural program in 2007 as a tenured investigator in NHLBI who maintains his base in Massachusetts.

“[Framingham] is a study that reflects the real world,” he said, and SHARe brings that study to the world by allowing scientists to compare genes within the Framingham study and between similar heart studies.

New Expressions

The next level, as Levy sees it, is discovering biomarkers and related therapeutics via a “phenomic” analysis of gene expression—to link proteins and metabolites to risk factors. 

Enter the Genomics Core, a facility on the 8th floor of the CC to study gene expression, initiated by Eric Billings, head of Bioinformatics and Systems Biology in NHLBI’s Intramural Research Program, and now under the direction of Nalini Raghavachari. 

This facility has automated the sample-preparation protocol, allowing for a tremendous increase in capacity while reducing noise to a 15 percent coefficient of variation.  A robot can manipulate an “array of arrays,” processing 96 samples at once, reducing batch effects, and exerting exogenous controls.

Beginning this summer for about two months, the Genomics Core will become an assembly line to study Framingham samples. The feasibility aspect is to assess which types of biological samples would be most useful in this mRNA analysis. Handling the sheer number of samples once the project moves forward—estimated to be at least 7,000 samples—is less of a concern, although this is the largest project by far that the Genomics Core has undertaken. 

The Genomics Core processes about 1,000 samples a year; its largest single project has been a heart study for NHLBI Director Elizabeth Nabel involving 200 samples. “This blows away anything we’ve done before,” said Mark Gladwin, former chief of NHLBI’s Vascular Medicine Branch, who coordinated this and other projects between NHLBI and the CC.

Levy speaks eagerly of the enormous research potential for Framingham within the intramural program, anticipating biomarker discoveries that are proteomic, metabolomic, and lipomic. He’s enthused about combining genetic and genomic information and applying systems biology approaches on a population level, as well as about the possibility of internal collaborations and recruiting high-quality fellows.

Genome-wide association studies have proven themselves successful for diabetes and several inflammatory diseases,“ but the Framingham study can go far beyond that,” said O’Donnell. “I’m optimistic this will yield a lot of fruit.” 

The Gene Chip Array Station robot can process and analyze 96 RNA samples at once within a few hours	Affymetrix fluidic stations, which can wash and stain gene chips in an automated manner, are used in the Genomics Core to maximize efficiency	The 96-sample plate that has been hybridized to 96 different RNA samples, washed and stained with fluorescent dye is scanned by the laser scanner to detect expressed genes in the samples

Return to Table of Contents