Determining the Role of Subclinical Disease Testing in Patients at Intermediate Risk

July 13 - 14, 2004
Bethesda, MD


The objective of the Working Group was to provide advice to the National Heart, Lung, and Blood Institute about research that is currently needed to inform clinical guidelines for the use of subclinical cardiovascular disease testing to identify persons at high risk for CVD and to target intervention. The group met in Bethesda for a day and a half, July 13-14, 2004, followed by a one-hour conference call on July 23 among a subset of members and NHLBI staff. A list of the Working Group members is at the end of this report.


Review of Previous Recommendations

Drs. Greenland and Taylor noted that approximately 650,000 persons in the United States have first coronary heart disease events annually, and that 25% of all coronary heart disease patients present with sudden death. Numerous studies have documented the effectiveness of targeting and treating high risk patients. However, a large proportion of cases arise from persons at intermediate risk. The question now is whether risk prediction can be refined and treatment targeted more effectively among those at intermediate risk. One challenge is to identify tests that are powerful enough to discriminate ? particularly beyond our current ability to discriminate ? between persons who will and will not develop CVD.

Drs. Greenland and Taylor reviewed existing guidelines for risk reduction of CHD aimed at those at high and intermediate risk, including the following:

  • American Heart Association Statement on Electron Beam Computed Tomography (2000)
  • AHA Prevention V Conference (2000)
  • National Cholesterol Education Program (2001)
  • Joint National Committee on the Detection, Evaluation, and Treatment of High Blood Pressure (2001)
  • Addendum to AHA Prevention V conference (2001)
  • American College of Cardiology Bethesda Conference #33 (2002)
  • American College of Cardiology Bethesda Conference #34 (2003)
  • AHA/CDC Statement on CRP (2003)
  • US Preventive Health Services Task Force Clinical Guidelines on Screening for Coronary Heart Disease (2004)

It was also noted that Blue Cross/Blue Shield issued a medical policy in 2000 that did not recommend use of coronary calcium imaging. A group promoting the concept of the ?vulnerable patient? was also acknowledged, but this approach appears to be aimed at extremely high risk patients.

Conclusions from review of these recommendations and discussion were:

  • Persons at intermediate risk have the greatest likelihood of having their estimated risk altered by subclinical disease testing.
  • Tests should be chosen for clinical utilization based on availability, reproducibility, evidence-base, and incremental value over current office-based risk assessments.
  • A clinical trial is needed to determine the clinical utility of the tests available. A negative trial might serve to stem over-utilization of unproven technology.
  • No test has been definitively recommended for screening, because there is no evidence that such testing improves clinical outcomes. A low ABI (<0.90), however, has been recommended by National Cholesterol Education Program as a ?CHD equivalent.?

Subclinical CVD and the Framingham Risk Score

Dr. Kronmal reviewed data from the Multi-Ethnic Study of Atherosclerosis (MESA), an ongoing multicenter cohort study examining the relationship of subclinical CVD measures to each other, progression of subclinical CVD, and the relationship of subclinical CVD to clinical CVD. The Framingham risk score was most strongly related to coronary calcium and carotid IMT measures in both men and women, in the whole cohort and in those at intermediate Framingham risk. He also presented data on ankle-brachial index, CRP, and left ventricular mass from MRI.

Subclinical Disease Testing: Study Design

Dr. Califf noted that the rate of growth of heart disease and expenditures for treatment during the next several decades will be considerable, due to aging in the population. The number of Medicare enrollees is expected to increase from 40 million in 2002 to 77 million in 2030. The Center for Medicare and Medicaid Services (CMS) is beginning to put more emphasis on cost-effectiveness. Without evidence-based guidelines, marketing of subclinical disease testing is untethered, and there is the potential for enormous gaps to be created in utilization: third party payers will not pay for unproven procedures, which will be obtained only by the economically advantaged.

The following hypothesis was submitted as the basis for a clinical trial: Subclinical disease testing will improve patient outcome compared with current standard screening for CAD risk factors. He proposed a design that randomized patients into testing vs. no testing, following a clinical screen. Those who tested positive would be further randomized into aggressive vs. conservative care, and the no test group would receive usual care. All groups would be followed for events.

subclinical chart

Several issues were discussed:

  • Feasibility of blinding of participants to test results
  • The need to tailor the intervention to the test results.
  • Feasibility of including one test vs. a battery of tests.
  • Potential intensification of therapy in the control group.
  • Additional comparison of aggressive vs. usual care among the Test Positive group
  • The need for intervention to be powerful enough to have an effect, suggesting that treatment should be prescribed within the trial.

Outcomes to be considered should include both morbidity and mortality, procedures and interventions, psychological outcomes, including quality of life, costs, and changes in care in both intervention and control groups. It was noted that subclinical disease diagnosis may lead to surgical interventions as well. The issue of outcomes was discussed in more detail by Dr. Criqui (see below).

The consequences of not conducting a trial were considered: This would leave clinicians and patients with no evidence base for making rational decisions regarding newly proposed, and sometimes expensive, methods to identify and treat persons at high risk. The unrestrained growth of imaging services in the community is of considerable concern. If testing is not an effective tool, such clinical testing represents an enormous waste of expenditures. It was generally felt to be unlikely that the imaging industry would fund a diagnostic efficacy trial in this arena.

Dr. Califf suggested that the potential costs and benefits of screening could be modeled using existing data, an exercise that may be very informative although would be far less likely to be accepted by the medical community than actual clinical trial results.

Review of Subclinical Disease Measures

Coronary calcium. Dr. Raggi presented data on the association of coronary calcium, measured by computed tomography, with traditional CVD risk factors, and its incremental value in predicting CVD risk over that of traditional risk factors. Data from the South Bay Heart Watch, a cohort of adults who had at least one risk factor, showed that a score of 300 or over effectively ?moved? some persons at intermediate risk into a high risk category. Data recently published by Shaw et al in 10,377 subjects screened with EBCT showed that coronary artery calcium scores are predictive of all cause mortality and that they provide incremental prognostic information over traditional risk factors. It would be useful to further analyze these and similar data sets to determine the frequency of this occurrence. He recommended employing age-specific cut points to determine high risk. Approximately half of persons at intermediate risk may be expected to have scores of zero. This test is easy and well-tolerated, and readings are highly reproducible. The scanning does involve low-dose radiation, and during the discussion the problem of handling incidental findings, particularly lung nodules, was raised. Coronary calcium is considered a likely candidate as a test to identify those at high risk. During discussion, it was stated that consideration might also be given to including aortic calcium, which has a much higher prevalence. Previous research has shown that aortic calcium on plain abdominal films shows a strong independent association with CVD events.

C-reactive protein. Dr. Wilson reviewed considerations for a new risk factor, including the need for standardized measurement, low variability, low correlation with existing risk factors, observational data that demonstrate a relationship between the risk factor and outcomes, clinical trial data, improvement in overall risk prediction, and measurement costs. Among several markers of inflammation, CRP appears to meet more of these criteria than do other tests, such as fibrinogen. Overall, the association with CVD is modest, however, particularly in relation to other risk factors. The AHA/CDC Panel concluded that CRP is an independent marker of CVD risk and that in persons at intermediate risk (defined as 10-20% over 10 years), hsCRP measurement may help direct further evaluation and primary prevention. The JUPITER Trial is currently examining whether statins are beneficial in persons with low LDL and high CRP. He recommended that CRP not be used as the basis for a new clinical trial.

Microalbuminuria. Dr. Flack presented NHANES data showing that the prevalence of microalbuminuria is higher in women than men, is less than 10% among persons age 20-29 and increases to over 20% in persons aged 80 and older. The prevalence is higher in African Americans and Hispanics than Whites. It is associated with hypertension and diabetes and has been associated with all-cause and CVD mortality, independent of other risk factors. ACE inhibitors can reduce albumin excretion. He recommended that microalbuminuria be measured in a clinical trial but that it not be the basis for randomization to treatment.

Carotid intima-media wall thickness (IMT.) Dr. O?Leary discussed the basis for measurement of carotid IMT and showed the relationship of IMT to CVD. Among persons in the highest quintile of IMT, the age and gender-adjusted 7-year CVD rates in CHS were approximately 25%, compared to approximately 5% in the lowest quintile. IMT can be obtained in virtually all people, it is a continuous variable, and the procedure has no risk. Standardized measurement is difficult, although a technician using a relatively inexpensive machine in a physician?s office can probably categorize patients into low, medium and high. IMT is considered a good candidate for consideration as a test to identify persons at high risk.

Ankle-brachial index (ABI). Dr. Criqui stated that ABI is simple and inexpensive to measure in a reproducible manner. There is no risk to the procedure. The prevalence of an abnormal ABI, defined as <0.9, ranges from 3% among those aged 45-64 to 27% among elderly men and women, in different studies. Overall, the prevalence is likely to be <10% among persons at intermediate risk. Stress testing, such as toe rises, can increase the sensitivity of the test. ABI is strongly predictive of CVD death, and the current NCEP guidelines recommend that a low ABI be treated as a CHD risk equivalent. This would complicate its inclusion as a basis for randomization in a trial, unless a borderline level or a positive exercise test (e.g., toe rises) in persons with normal ABIs was included as a criterion for a positive test.

Functional capacity. Dr. Lauer discussed treadmill stress testing for assessment of functional capacity, which has been shown to discriminate well between persons at high and low risk for all-cause and cardiovascular mortality. Worse functional capacity is correlated with older age, female gender, and obesity, yet independently predicts death independent of the Framingham risk score. Furthermore, functional capacity has been shown to identify those subjects most likely to benefit from aggressive treatments like coronary bypass grafting. Testing must be symptom-limited, and age- and gender-specific cut points for abnormality should be used. One problem is linking an effective intervention with low functional capacity. While persons may improve functional capacity through exercise, this may be difficult to accomplish. Exercise testing was not considered a candidate as a measure to identify high risk patients in this trial. Nonetheless, it may have value as a baseline variable since, as with the case with CABG, it may identify those people most likely to benefit from a screening strategy.

Left ventricular hypertrophy (LVH). Dr. Flack discussed methods for measuring LVH. The electrocardiogram is insensitive and may overestimate the prevalence of LVH in African Americans. Echocardiogram may be difficult to obtain in a large proportion of people, particularly the elderly, and reproducibility is only fair. The prevalence of LVH increases with body mass index and blood pressure and is higher in African Americans than Whites. Prevalence estimates range from 1% of young White adults to 4% of young African American adults (CARDIA) to 28% of the middle-aged ARIC cohort. The relative risk of CVD for LVH in the Framingham Study was approximately 1.5. Magnetic resonance imaging provides much more reproducible measures of LV mass, but the cost is higher and many people cannot tolerate or are excluded from the procedure. A measure of ventricular size may also be made from cardiac CT scans. LVH was not considered a candidate test to identify high risk patients in this trial but should be considered as a possible effect modifier of intervention.

Other measures. Dr. Herrington discussed the potential for genetic variants or genetic or protein profiles to be used to identify high risk persons or to identify treatment responses. Individual variants have generally been found to have small effects, but multiple variants in combination (genomics) may have large effects. The costs of performing genetic scans and proteomics are currently high but are declining. Vascular MRI, SPEC imaging, flow-mediated vasodilation, arterial stiffness, and CT angiography may provide better opportunities for prediction in the future, but currently there is a lack of data on predictive ability, high costs, unknown test characteristics, and limited standardization.

During discussions, mild renal failure was noted to carry strong independent risk for CVD and may be synergistic with other risk factors. Cystatin C appears to be a promising risk factor. Another point o f discussion was the ?street credibility? of the chosen test ? that is, clinicians are more aware of and likely to adopt some tests over others. From this perspective, there is relatively more interest in coronary calcium, CRP, and carotid IMT than the other measures discussed.

Clinical Screening Algorithms

Dr. Newman discussed the main goal of screening ? to target more specific preventive treatment. The screening test should identify high risk individuals among those initially deemed to be at intermediate risk, and may be used to reinforce aggressive risk factor management and/or allow persons defer treatment if their risk is estimated to be lower (therapeutic avoidance). Knowledge of the test might enhance adherence to treatment.

The following considerations should be made in planning a screening strategy:

  • Screening guidelines should be simple, clear, and easy for people to understand. (The example of colonoscopy at age 50 was provided.)
  • The screening algorithm should be able to be easily incorporated into the health care system.
  • A two-stage approach may be needed ? first to establish ?intermediate? risk, then to refine with subclinical disease testing.
  • Consider the need for age- and gender-specific thresholds for testing and the need for repeat screening at some regular interval (e.g., every 5 years).
  • Screening tests should ideally have high sensitivity and specificity to identify high risk persons, and should be low cost and low risk.

Special consideration should be given to persons with chronic renal insufficiency, lupus, rheumatoid arthritis, polycystic ovaries, type 1 diabetes, and other low prevalence conditions that are associated with high risk. Age alone is a major factor; after age 65 in men and age 75 in women, the average risk for CHD is >20% in 10 years. Consideration should also be given to defining intermediate risk as 10-20% to assure a sufficient number of events in a trial. Dr. Newman recommended that for each test considered, the sensitivity, specificity, cost, availability, reproducibility and risk be examined, but that it appears that coronary artery calcium, IMT and perhaps CRP are the most likely candidates for study.

Population Considerations

Dr. Taylor addressed the question of treating global risk vs. risk based on individual factors. He considered the arguments for and against including low or high risk persons in a clinical trial; the intermediate risk approach appears to have the highest likelihood of clinical yield ? that is, changing the risk category of individuals based on test results. It is estimated that approximately 25% of persons at intermediate risk could be re-classified based on coronary calcium results; thus, there would be plenty of persons in this group for a study. The issue of studying whether therapeutic avoidance is beneficial in persons classified as low risk based on test results was discussed; this appears unfeasible as it would require abandoning current established guidelines. Also, according to the South Bay Heart Watch data, persons at high Framingham risk remain at high risk, even if their calcium scores are zero.

Consideration should be given to the gender composition of the population included in a trial, since women are at lower Framingham risk than men. African Americans have a lower prevalence of coronary calcium than Whites, and the risk associated with calcification in this group is not clear. However, they should not be excluded.


Dr. Greenland proposed using a ?polypill? ? a combination of medications to lower blood pressure and LDL-cholesterol, and possibly including aspirin ? as the intervention for persons identified to be at high risk. Aspirin does have a relatively high incidence of side effects, however. An alternative would be to create complicated treatment guidelines based on individual risk factors, and treat to specific blood pressure and LDL targets. Consideration should also be given to aggressive lifestyle modification, as several types of dietary and exercise interventions have been shown to be as effective as drug therapy in lowering blood pressure and modifying the lipid profile.


Dr. Criqui reviewed the possible outcomes that could be measured in a trial, either as primary or secondary endpoints, or even as ancillary studies. These include:

  • Risk factor change (obesity, physical activity, blood pressure, etc.).
  • Change in subclinical disease (coronary calcium, CRP, etc.).
  • Morbidity, including CHD, cerebrovascular disease, peripheral vascular disease, other CVD (CHF, in particular), and non-CVD morbidity.
  • Mortality, including fatal CHD, fatal cerebrovascular disease, fatal peripheral vascular disease, fatal other CVD, and non-CVD mortality.
  • Quality of life, functional capacity, activities of daily living, and psychosocial measures.
  • Costs; cost-effectiveness.

Combined outcomes should be considered as a primary endpoint; total mortality is always the most important hard clinical endpoint, however. Consider not only the number needed to treat to prevent one event, but also the number needed to treat to harm one person. Finally, cost-effectiveness may be a major trial result, particularly in the assessment of technology.

Based on the experience of other CVD intervention trials, this study is likely to require >10,000 subjects to test one type of screening test.

Recommendations and Remaining Issues

The following recommendations and issues emerged during the meeting and the post-meeting conference call regarding the motivation for and design of a clinical trial.

  1. Should the trial address whether testing reduces improves outcomes or testing + intervention in those who test positive improves outcomes?
    • Working Group members and meeting attendees pointed out the need to test both questions. Some proposed an ?encouragement? trial, whereby the test results would be used to inform and perhaps serve as an incentive for clinicians and patients to be more aggressive (or perhaps less aggressive, in the case of negative test results) in treatment to reduce CVD risk. Others stated that such a trial would likely be negative (i.e., find no benefit from screening), because there are few guidelines regarding the appropriate clinical response to positive tests, and clinical practice is highly variable. The result would likely be only modest differences in the actual treatment experienced by patients between those at high risk vs. not at high risk. Ideally, a trial would test both questions, including a randomization to screening, to test the effectiveness of screening, as well as randomization to different levels of intervention among those who test positive. It was also pointed out that it would be desirable to determine if more aggressive treatment without additional testing would be effective to improve outcomes. There would be some efficiency realized in testing more than one question in one study.

      Several Working Group members noted that more aggressive treatment will likely reduce CVD risk. The central question is at what cost this risk reduction would be achieved and whether a test and treat strategy would have a more acceptable cost-benefit ratio than alternative strategies.

  2. Which test(s) should be included as the basis for identifying high risk people?
    • The main candidates were coronary calcium and carotid IMT, because these tests appear to be best at identifying persons at higher risk. Factors such as cost, reproducibility, and feasibility of measurement should be considered closely. A separate consideration was the need to study technology such as computed tomography in particular, because of its widespread usage without a complete evidence base.
  3. Can test results be blinded?
    • It was felt to be highly unlikely that test results could be blinded, due to public interest in test results and ethical considerations. Most other studies in the US have not been able to blind the test results.
  4. Should the population included be at intermediate risk? What other population characteristics should be considered?
    • There was consensus that an intermediate risk group should be the focus of a trial. The population should be broadly representative, including women and minorities.
  5. If a prescribed treatment approach is used, what type of intervention would be most appropriate?
    • If treatment is prescribed by the trial, particularly pharmacotherapy, consideration must be given whether to tailor treatment (i.e., treat to specific blood pressure and/or LDL-cholesterol goals) or use a global risk approach, such as the ?polypill.? Behavioral measures to lower risk factors, such as diet and physical activity, should also be considered. The possibility of providing intensive treatment guidelines for test-positive patients in the intervention group of a trial (an approach that is between that of an ?encouragement trial? and a prescribed intensive intervention) was also raised. The trial design must take into account current (at the time) clinical guidelines and the possibility of near-term changes in guidelines. The choice of interventions, particularly their effect sizes, will influence sample size -- and ultimately, feasibility.
  6. Which outcomes should be included?
    • Major morbid and mortal CVD outcomes and total mortality should be the primary outcomes. Quality of life and costs must be assessed. Cost-effectiveness assessment should be a major goal, and could be considered a primary outcome in some design scenarios. Physician and patient behavior should also be monitored, particularly if the intervention is not prescribed.
  7. What would be the feasibility of a trial to answer these questions?
    • It was noted that a trial to address these questions would likely be quite large and that delivery of intensive treatment would be more expensive than an ?encouragement? design. These issues would need to be considered in determining what design features would make a trial feasible.

      The Institute will work with these recommendations to explore possible trial designs. Ultimately, the decisions to conduct a trial and the type of trial to conduct are driven by factors of feasibility, including cost, as well as science and public health. Working Group members may be consulted in the future to provide more specific advice on trial design.


Previous Recommendations for Subclinical CVD Testing

Smith SC, Greenland P, Grundy SM. AHA conference proceedings. Prevention V: Beyond secondary prevention: Identifying the high-risk patient for primary prevention. Executive Summary. Circulation 2000;101:111-116.

Greenland P, Smith SC, Grundy S.M. Improving coronary heart disease risk assessment in asymptomatic people: role of traditional risk factors and noninvasive cardiovascular tests.Circulation. 2001;104:1863-1867.

Taylor AJ, Merz CN, Udelson JE. 34th Bethesda Conference: Executive summary--can atherosclerosis imaging techniques improve the detection of patients at risk for ischemic heart disease?J Am Coll Cardiol. 2003;41:1860-1862.

Greenland P, Abrams J, Aurigemma G, Bond M, Clark L, Criqui M, Crouse J, Friedman L, Fuster V, Herrington D, Kuller L, Ridker P, Roberts W, Stanford W, Stone N, Swam J, Taubert K, Wexler L. Prevention Conference V. Beyond secondary prevention: identifying the high-risk patient for primary prevention. Noninvasive tests of atherosclerotic burden.Circulation 101, 111-116. 1-4-2003

U.S. Preventive Services Task Force. Screening for Coronary Heart Disease: Recommendation Statement. Ann Intern Med. 2004;140:569-572.

Information on CVD Risk Distribution

Ford, ES, Giles WH, Mokdad AH. The distribution of 10-Year risk for coronary heart disease among U.S. adults. Findings from the National Health and Nutrition Examination Survey III.Journal of the American College of Cardiology 2004;43:1791-1796.

Greenland P, LaBree L, Azen SP et al. Coronary artery calcium score combined with Framingham Score for risk prediction in asymptomatic individuals.JAMA. 2004;291:210-215.

Working Group Roster

Working Group on Determining the Role of Subclinical Cardiovascular Disease
Testing in Patients at Intermediate Risk

Robert M. Califf, MD (Co-Chair)* 
Professor of Medicine 
Associate Vice Chancellor for Clinical Research 
Duke Clinical Research Institute 
Trent Drive, Box 3850 
Durham, NC 27710 
Phone: 919-668-8820 
Fax: 919-668-7103

Allen Taylor, MD (Co-Chair)* 
Walter Reed Army Medical Center 
6900 Georgia Avenue, NW 
Building 2, Room 4A 
Washington, DC 20307-5001 
Phone: 202-782-2887 
Fax: 202-782-7063

Michael Criqui, MD, MPH 
University of California, San Diego 
Stein Clinical Research Building 
9500 Gilman Drive 
Room 349La Jolla, CA 92093-5004 
Site Phone: 858-534-3722 
Office Phone: 858-534-3722 
Alternate Phone: 858-534-8625 
Fax: 858-534-8625

John Flack, MD, MPH 
Professor and Associate Chair 
Wayne State University 
Department of Medicine 
2E 4201 St. Antoine 
Detroit, MI 48201 
Phone: 313-955-0635 
Fax: 313-993-0645

Philip Greenland, MD* 
Feinberg School of Medicine 
Northwestern University 
680 North Lake Shore Dirve 
Suite 1102 
Chicago, IL 60611 
Office Phone: 312-908-1723 
Fax: 312-908-9588

David Herrington, MD, MHS 
Wake Forest University 
School of Medicine 
Department of Internal 
Medical Center Blvd. 
Winston-Salem, NC 27157 
Site Phone: 336-716-4950 
Office Phone: 336-716-4950 
Alternate Phone: 336-716-9124 
Fax: 336-716-9188

Richard Kronnal, PhD 
Principal Investigator 
University of Washington 
Collaborative Health Studies 
Coordinationg Center 
Box 354922 
Building 29, Suite 310 
6200 NE 74th Street 
Seattle, WA 98115-8160 
Office Phone: 206-685-7123 
Fax: 206-616-4075

Michael S. Lauer, MD, MPH 
Director of Clinical Research and 
Director of Stress 
Laboratory, Desk F25 
Cleveland Clinic Foundation 
9500 Euclid Avenue 
Cleveland, Ohio 44195 
Phone: 216-444-3407 
Fax: 216-444-6798

Anne B. Newman, MD, MPH 
Associate Professor of Medicine 
and Epidemiology 
Healthy Aging Research Program 
Bellefield Professional Building 
130 North Bellefield Avenue-Room 532 
Pittsburgh, PA 15213 
Phone: 412-383-1871 
Fax: 412-383-1308

Daniel O'Leary, MD 
tufts-New England Medical Center 
Department of Radiology 
750 Washington Street, Box 380 
Boston, MA 02111 
Site Phone: 617-636-8050 
Office Phone: 617-636-8050 
Alternate Phone: 617-636-0041 
Fax: 617-636-0041

Paolo Raggi, MD 
Tulane University 
1430 Tulane Avenue SL-48 
New Orleans, LA 70112 
Phone: 504-588-5492 
Charleston, SC 29425 
Fax: 504-587-4237

Peter W.F.Wilson, MD 
Medical University of South Carolina 
Medicine Division of Endocrinology 
PO Box 250624 
96 Jonathan Lucas Street 
Charleston, SC 29425 
Phone: 843-792-6245 
Fax: 843-792-2601

*participant on 7/23 conference call

Planning Group Roster

Working Group on Determining the Role of Subclinical Cardiovascular Disease 
Testing in Patients at Intermediate Risk 

Diane Bild, MD, MPH* 
Telephone: 301-435-0457 
Lawton Cooper, MD, MPH 
Telephone: 301-435-0419 

Jeffrey Cutler, MD* 
Telephone: 301-435-0413 
Michael Domanski, MD 
Telephone: 301-435-0396 

Teri Manolio, MD, Ph.D 
Telephone: 301-435-0708 
Merle Myerson, MD, EdD. 
Telephone: 301-435-1290 

Christopher O'Donnell. MD 
Telephone: 508-935-3435 

Peter Savage, MD 
Telephone: 301-435-0421 

Denise Simons-Morton, MD, Ph.D* 
Telephone: 301-435-0384 

Eser Tolunay, Ph.D. 
Telephone: 301-435-0557 

Momtaz Wassef, MD 
Telephone: 301-435-0550 

*participant on 7/23 conference call