Perspectives On Medical Research


Volume 5, 1995

Aping Science


A Critical Analysis of Research at the Yerkes
Regional Primate Research Center

Public Health Research for the Future





A. The Importance of Understanding Disease Causation

Historically, medical researchers have sought to identify disease causes--those environmental agents that change a person from a healthy state to one of disease. Obviously, such information is vital in preventing disease and developing rational treatment approaches. Nevertheless, over the past 20 years, medical researchers, teachers, and practitioners have all revealed a reduced interest in causation.1 Rather than look for specific causative agents, the tendency today is to identify a multitude of "risk factors" associated with disease. For example, at least 246 separate risk factors for coronary artery disease have been identified, including cigarette smoking, elevated plasma cholesterol, obesity, type A personality, Jewish religion, baldness, sense of exhaustion in college, slow beard growth, and psychiatric disorders.2 Cardiovascular researchers are recognizing the limits of this approach, because many "risk factors" do not cause cardiovascular disease but, rather, represent mere associations,3 and many "risk factors" are unavoidable and therefore not amenable to medical or social intervention.

Another problem with contemporary research strategies is that scientists have focussed primarily on risk factors related to lifestyle, tending to overlook specific causative agents in the environment. For example, researchers have generally seen coronary artery disease as a consequence of elevated blood levels of certain fats and fat-protein compounds, usually a consequence of poor diet. Based on this theory, cardiologists have tried to remove, displace, or bypass plaques of presumed fatty deposits. However, clinicians have been frustrated by a persistent problem: dividing cells in the arterial wall that repeatedly narrow the arteries after interventions to widen them. For much of the 20th century, cardiovascular researchers had interpreted the smooth-muscle cell masses bulging into patients' arteries as inflammatory scar tissue. However, growing evidence has pointed to benign smooth muscle cell tumors as the main cause of renewed blockage of angioplastied arteries.4-6 It appears that human atherosclerotic plaques are composed primarily of benign tumors that may, secondarily, accumulate fatty deposits.

The group of scientists who had shown clearly by 1968 that chemicals in cigarette smoke cause both lung tumors and atherosclerosis7 subsequently showed that diagnostic X-rays also cause coronary artery disease,8 indicating that coronary artery disease resulted from DNA damage. Indeed, when scientists began to look more closely at the composition of human atherosclerotic plaques lining the arterial wall, they found benign tumors comprised of cells of monoclonal origin.4,5 Increasingly, researchers have found that the chronic, degenerative diseases responsible for most morbidity and mortality in Western nations result from environmental mutagens (that damage DNA).9 Nevertheless, the careful retrospective analyses needed to elucidate the sources of DNA damage from environmental mutagens has rarely been funded by the NIH. The NIH primarily funds human, animal, and cell experiments to determine the relative importance of different "risk factors." However, the very nature of experimentation involves inducing a condition on an artificial, experimental system. While certain experiments can yield some kinds of information about anatomy, physiology, and pathology, they cannot determine the underlying causes of naturally occurring diseases through a process that merely simulates the disease in humans or animals. Experimental conditions may resemble naturally occurring human diseases, but invariably they exhibit fundamental biological differences in etiology, clinical presentation, and natural history. In contrast, human population studies are devoid of artifact, because they study actual human disease in human beings. Researchers can identify disease causation by studying disease incidence in populations at high and low risk for suspected disease-causing agents. While such research is often difficult and time-consuming, it is the only way to reliably determine human disease causation.


B. Problems With Current NIH Research Resources


Former lipid researcher Edward Ahrens has decried the "crisis in clinical research" --the shifting of funding priorities from learning about human diseases by direct observation of human patients to experiments on animals, cells, and human volunteer subjects.10 Only 7.4% of new NIH grants in 1987 were directed to patient-oriented research; a survey of 1990-1991 grants revealed that only 4.5% were basic research involving human subjects.11 Many of these grants were experimental studies, indicating that the federal investment in direct human clinical observational study is minimal.


C. Prototypes of Scientific Research Resources for the Future


1. A National Numbering System

In order to advance public health effectively, Americans must recognize that we are all subjects of ongoing, unplanned "experiments." These "experiments" reveal the real human health implications of contemporary life, with all its sociological, psychological, toxicological, and physiological stressors. For example, careful population studies can determine the dangers of radiation and other mutagens (that initiate the genetic damage responsible for tumor formation and birth defects). Over the past few decades, Americans have been exposed to diagnostic X-rays and to a wide range of chemical pollutants that damage cells' genetic material. Identifying who gets cancer, birth defects, and other manifestations of genetic damage, and then analyzing cellular and molecular characteristics of those individuals is the best (perhaps the only) way to determine how these mutagens affect humans.12 The question for Americans is not whether they are experimental subjects, but whether they wish to know the results of the unplanned experiments in which they participate.

Recognizing this, Sweden has initiated a massive public health research project. The Swedish government assigns each newborn infant a tracking number for recording all medically relevant life events of that person and his/her parents. For example, in studies to investigate hazards to the developing fetus, researchers can access information about the parents’ diseases, occupations, and exposures to mutagens (such as cigarette smoke). The primary concern of this numbering system is its potential invasion of privacy. American health officials would, therefore, need to find a balance between individual rights to privacy and improved public health. Without such a tracking system, American public health research will remain incomplete, often uninterpretable, and consequently frequently erroneous.13


2. Deep Mathematical Models

A team of biostatisticians at Roswell Park Memorial Institute for Cancer Research (RPMI) pioneered new ways of modelling human diseases in the 1960s and 1970s. Irwin Bross, Leslie Blumenson, and colleagues developed the first "deep mathematical models" of human disease,14 which describe the complex chain of events that occur in the human body, from initiating molecular events to full-blown diseases. Unlike laboratory-based models, deep models utilize data from human patients during life and at autopsy. For example, Bross and Blumenson's deep mathematical model of breast cancer used data from two large-scale human breast cancer chemotherapy trials involving 2,000 women at RPMI.14 In attempting to explain the data in terms of the growth of a clone of tumor cells in the breast and their eventual spread through the body, Bross and Blumenson identified at least two distinct forms of breast cancer. One is a fast-growing, fast-spreading variety that tends to occur in women under 50; the other is a relatively slow-growing tumor of older women.15,16 While these two tumor types have different clinical features, they look identical microscopically. Treatments and screening methods that construe all breast tumors as identical cannot possibly lower mortality significantly, because many tumors will be treated inappropriately. Only by tailoring treatments to the specific subtypes of breast cancer can substantial progress be made. Bross and Blumenson’s model, therefore, had profound implications for breast cancer treatment and prevention strategies.


3. Medical Technology Assessment Research: Number-Crunching Versus "Metatechnology"

Health services research includes assessing cost effectiveness of medical technologies and healthcare delivery systems. Many argue that the efficacy of all medical services should be assessed through randomized controlled clinical trials (RCTs). Such trials have been useful in determining whether surgical interventions are more effective than medical therapy in lowering mortality from chronic diseases. However, RCTs have not resolved controversies about services for which risk/benefit considerations are complex. Screening mammography for women under age 50 is a case in point. In the 1970s, widespread mammography screening of asymptomatic women under age 50 was sponsored by the NCI based on RCTs showing mortality reduction in women over 50. While RCTs have consistently failed to show any benefit (and some have shown danger) from screening mammography for women under age 50,17,18 the NCI still encourages routine mammograms in these women.

The entire Breast Cancer Detection Demonstration Project of the NCI had been based on an assumption that the number of breast cancers detected by mammography and thereby cured far outnumbered those induced by X-ray.19 This may well have been unfounded. Bross and Blumenson20 undertook a thorough analysis of the screening program. Interlinking a mathematical model of the screening process with their "deep mathematical model" of breast cancer, Blumenson and Bross20 estimated that the actual benefit/risk ratio was at best 1/1: (1 cured breast cancer for each induced by the mammography), and possibly much worse. Mammography screening for women under 50 continues to have its advocates, despite the prediction of the deep model of breast cancer that mammography screening could not lower mortality from breast cancer in women under age 50.

Tumors in young women simply grow and spread too quickly to be detected early enough to affect cure. By the time a young woman's tumor is identified on X-ray film, it has very likely already spread to a distant organ (metastasized), and prognosis at this point is poor. In terms of providing direction for future research, the deep model showed that an effective diagnostic tool for younger women would need to detect breast tumors at a microscopic stage, and that it should not utilize carcinogenic ionizing radiation. Bross refers to deep mathematical models that evaluate risks and benefits of medical technology as "metatechnology."21


4. Medical Informatics

The overlapping concerns about the shortcomings of healthcare and health research have spurred development of a new discipline called "medical informatics," which utilizes sophisticated computer analyses to enhance the value of patient-derived data.22 For instance, most patient records are written and filed manually. Consequently, any scientist interested in assessing the quality of patient care or exploring a medical hypothesis must sort through reams of hand-written patient records. Computer-based, multimedia records that include free text, high-resolution images, and sounds (e.g., auscultation) could significantly expedite what is now a daunting task. While no fully developed computerized information systems are currently operative, there are a few prototypes.23

Such a system would greatly facilitate important clinical research projects, as illustrated by the work of John Spratt and co-workers to optimize breast cancer treatment.24 They found that survival was greater when the surgery was performed between days 7-20 of the menstrual cycle. A more accessible database would greatly assist further studies following up theft research into how the timing of breast cancer surgery affects prognosis.


5. Specialized Clinical and Autopsy Data Banks

Currently, the primary data collected by the National Center for Health Statistics (NCHS) consists of mortality rates for the major diseases and injuries. In order to discharge its Congressional mandate, however, the NCHS should be collecting and analyzing a wide variety of clinical and autopsy data. Currently, the NCHS obtains cause of death data from physicians’ death certificates, and the accuracy of such certificates is questionable because physicians base their diagnoses on clinical findings, not on autopsy findings.25 In order to increase the accuracy of national mortality data, several clinical investigators have advocated that current cause-of-death certificates be based on autopsy diagnoses which are computerized and collected in a centralized database at the NCHS.26

In 1977, the Technical Consultant Panel of the United States National Committee on Vital and Health Statistics issued recommendations for broadening the data-collection activities of the NCHS.27 Those suggestions, never followed up, mostly concerned the need for more careful monitoring of environmental exposures. The World Health Organization has also emphasized the importance of public health research on pollution hazards, requiring computerized data banks.28 Such a system benefitted Rosalie Bertell, who accessed voluminous clinical data to investigate the capacity of low-level radiation from X-rays to accelerate the aging process.29,30 Based on her findings, Bertell has contended that the minimum human database necessary to effectively audit the health of a population includes: 1) decreasing fertility rate, 2) increasing infant death and birth defect incidence rate, 3) increasing rates of severe asthmatic or allergic reactions, 4) declining scholastic ability of children, 5) decreasing avenge age of diagnosis of chronic, life-threatening diseases such as hypertension, diabetes, and cancer.31 Such use of computerized clinical and autopsy data, which can be modelled on the Swedish numbering system, is critical for gathering the public health information needed to identify causes and means of prevention of disease in America.


1. Krieger N. Epidemiology and the web of causation: Has anyone seen the spider? Social Science and Medicine l994;39: 887-903.

2. Hopkins PN, Wiffiams RR. A survey of 246 suggested coronary risk factors. Atherosclerosis 1981:40:1-52.

3. McCormick I. The multifactorial aetiology of coronary heart disease: A dangerous delusion. Perspectives in Biology and Medicine 1988;32: 103-108.

4. Benditt EP, Gown AM. Atheroma: The artery wall and the environment. International Review of Experimental Pathology 1980;21:55-1 18.

5. Benditt EP. Origins of human atherosclerotic plaques: The role of altered gene expression. Archives of Pathology and Laboratory Medicine 1988;112:997-1001.

6. Speir E, Modali It, Huang E-S. Potential role of human cytomegalovirus p53 interaction in coronary restenosis. Science 1994;265:391-395.

7. Sackett DL, Bross IDJ, Gibson RW, Pickren JW. Relation between aortic atherosclerosis and the use of cigarettes and alcohol: An autopsy study. New England Journal of Medicine 1968;279:1413-1420.

8. Bross IDJ, Ball M, Rzepka T, Laws R. Preliminary report on radiation and heart disease, Journal of Medicine l978;9:3-15.

9. Lower GM. Human carcinogenesis: A disciplinary perspective. Medical Hypothesis 1990;33:1-6.

10. Ahrens EH. The Crisis in Clinical Research: Overcoming Inuitutional Obstacles New York, Oxford University Press, 1992.

11. Marshall E. Does NIH shortchange clinicians? Science 1994;265:20-21.

12. Lower GM. The mutation theory of chronic, noninfectious disease: relevance to epidemiologic theory. American Journal of Epidemiology 1982;115:803-817.

13, Erickson JD. Assessing occupational hazards to reproduction: Uses of existing data sets, in Reproduction: The New Frontier in Occupational and Environmental Health Research. New York, Alan R. Liss, 1984, pp 99-107.

14. Bross IDJ, Blumenson LE. Statistical testing of a deep mathematical model for human breast cancer. Journal of Chronic Diseases 1968;21:493-506.

15. Bross IDJ, Blumenson LE, Slack N, Priore R. A two-disease model for breast cancer, in Forrest APM, Kunkler PB (eds). Prognostic Factors in Breast Cancer. Edinburgh, Livingstone, 1968, pp 288-300.

16. Slack NH, Blumenson LE, Bross IDJ. Therapeutic implications from a mathematical model characterizing the course of breast cancer. Cancer 1969;24:960-971.

17. Anon. Breast cancer screening in women under 50. The Lancet 1991;337: 1575-1576.

18. Miller AB, Baines CJ, To T, Wall C. Canadian National Breast Screening Study: 1. Breast cancer detection and death rates among women aged 40 to 49 years. Canadian Medical Association Journal 1992;147:1459-1476.

19. Bross IDJ. Fifty Years of Folly and Fraud "in the Name of Science." Buffalo, Biomedical Technology Press, 1994.

20. Bross IDJ, Blumenson LE. Screening random asymptomatic women under 50 by annual mammographies: Does it may sense? Journal of Surgical Oncology 1976;8:437-445.

21. Bross IDJ. Metatechnology: A technology for the safe, effective and economic use of technology. Metamedicine 1981;2:145-153.

22. American Association of Medical Colleges. Physicians for the 21st Century: Report of the Panel on the General Education of the Physician and College Preparation for Medicine. Journal of Medical Education 1984;59: 1-208.

23. Carter JR, Nash NP, Cechner RL, Plan RD. Proposal for a national autopsy data bank: a potential major contribution of pathologists to the health care of the nation. American Journal of Clinical Pathology 1981;76(suppl):597-617.

24. Spratt JS, Zirnheld J, Yancey JM. Breast cancer detection demonstration project data can determine whether the prognosis of breast cancer is affected by the time of surgery during the menstrual cycle. Journal of Surgical Oncology 1993;3:4-9.

25. Hill RB, Anderson RE. The Autopsy--Medical Practice and Public Policy. Boston, Butterworth Pub, 1988.

26. Kircher T, Carter JR, Sinton E. The National Autopsy Data Bank. Pathologist Nov 1985, pp 22-27.

27. Technical Consultant Panel to the United States National Committee on Vital and Health Statistics. Statistics Needed for Determining the Effects of the Environment on Health. US Department of Health, Education and Welfare, DHEW Publication No. (HRA) 77-1457, 1977.

28. WHO European Office, Priority Research for Health for All. Copenhagen, WHO, 1988.

29. Bertell R. X-ray exposure and premature aging. Journal of Surgical Oncology 1977;9: 379-39 1.

30. Bertell R. The nuclear worker and ionizing radiation. American Industrial Hygiene Association Journal 1979;40:395-401.

31. Bertell R. No Immediate Danger: Prognosis for a Radioactive Earth. Toronto, Women’s Publishing, 1985.