Spectacular advances in the scientific understanding of life processes since World War II have vastly increased the ability of physicians to diagnose, treat, and prevent disease. Nevertheless, even today there is much about the cause of most common diseases and their prevention or treatment that remains unknown. However, the huge and constantly expanding store of information in the medical literature is still mainly the province of physicians, who make most of the decisions about the use of health care resources. As always, patients depend largely on their physicians for counsel and treatment.
A new book by Dr. Eric Topol predicts that this is all about to change. He sees an exciting new era created by the increasing application of digital technology and computers to the study of human biology and health care, which he predicts will cause the “creative destruction” of medicine. Medical care will no longer be controlled by physicians. Neither will medical information be based on the results of studies of large populations, which Topol says give less precise information than do studies of individual patients. Instead, patients will largely manage their own care, based on their access to detailed and digitized information in the world’s medical literature and to information about their own bodily functions and individual genetic makeup.
Thus, Topol heralds the coming of “personalized” medicine, in which the convergence of digitized information from all these sources will enable patients to make most of the decisions now reserved to physicians. There will be a parity of knowledge between patient and doctor, which he calls the “democratization” of medical information.
According to Topol, patients will receive, in real time, continuous information about their own physiology, biochemistry, and general health generated by microchip wireless sensors embedded in, or attached to, their body. They will know about their genetically determined diseases, their risks of contracting diseases, and their probable response to therapeutic drugs, as determined by the variations in their fully sequenced genome. All of this vast store of information will be made instantly available and interpreted by their smart phone. Many health care decisions would be made or suggested by computers, some by patients themselves, and far fewer by physicians.
Physicians would continue to manage surgical procedures and the treatment of serious injuries. Even here, however, their direct hands-on activities would be limited to their control of robots, aided by electronic imaging. The result of all these changes, Topol says, will be better and probably less expensive clinical care. There will be much less use of hospitals and physicians’ offices, more rational, safe, and effective use of drugs and medical devices, and greater success in the prevention of disease and disability.
How much credence should be given to this description of a coming medical revolution? Although Topol’s account sounds like hyperbole, it cannot be easily dismissed, for he bears impressive credentials. He has had a distinguished academic career in cardiology, genetics, and medical education, and he currently is the director of the Scripps Translational Science Institute in La Jolla, California. So, when Topol tells us that patients’ biological information about themselves, monitored by digital wireless detectors, is going to produce a new kind of “personalized” medicine, his message deserves careful attention.
He has written his book for a wide public, so it is full of interesting information about new technologies. It describes many new types of micro-sensors and his experience with some of them. There is a helpful primer on genomics (“Genomics 101”), which alone makes his book worthwhile. His language is plain; nevertheless, some of his exposition will be tough going for scientifically uninformed lay readers. Topol and his contemporaries have lived with computers for their entire professional lives. I entered medicine a generation before him, and this may explain some of my skepticism about Topol’s prediction of a medical care future so dependent on, and regulated by, computers and digitized wireless devices. Still, I think he stretches current medical science too far. He is much too sanguine about the potential of new technology for providing clinical care and not sufficiently concerned about its limitations.
A particularly valuable section of the book describes recent advances in human genetics, including the sequencing of the total human genome. That sequencing is the process of determining the order of the four nucleic acids in a person’s DNA. Thanks to new technology, this stunning achievement is becoming faster and cheaper, and therefore more generally available. However, this instructive description is accompanied by predictions that invite disbelief. Topol envisions a time, not too distant, when every patient’s total genetic structure will be known and will be digitally accessible to him or her. He believes the unique variations and mutations in each genome will help to foretell susceptibility to diseases and will guide preventive and therapeutic interventions, including what drugs to use and at what dosage. As will be seen, I regard his predictions about genome sequencing skeptically.
Topol has an interesting chapter on the remarkable advances in digitized imaging of the internal structures of the body (e.g., pocket-sized echocardiographs, CT and PET scanning, and MRIs), which are replacing physical examination and old-fashioned X-rays. And finally, he discusses the application of electronic digital technology to medical records and doctor–patient communication, which is already beginning to replace handwritten records and personal encounters between doctor and patient. These technologies enable the creation of digitized records that at least in theory can be transferred between medical institutions and providers, and collected anonymously for analysis by public agencies, and would also be available to each patient on his or her smart phone.
Simply as descriptions of advances in medical technology, however glowing they may be, most of this material is not controversial. But Topol’s expectation that patients will want to attend constantly to the recordings of their bodily physiology and chemistry is, to say the least, questionable. And so is his belief that most people will pay close attention to the innumerable variations in their genetic code that might have a relatively weak statistical association with common diseases.
However, my major concerns are with Topol’s conclusions that digitized data recorded from each patient, together with information about his or her fully sequenced genome, will transform the practice of medicine, and will “reboot” the pharmaceutical industry. I also question Topol’s speculation about the rapidly approaching parity of knowledge between the public and the medical profession. He thinks that patients, aided by computerized access to the world’s medical literature and to computer-interpreted information about their own body and its unique genomal structure, will be almost as knowledgeable about their medical conditions as any physician, and therefore will be an equal participant with their physicians in the management of their medical care.
I doubt that most people will be able to analyze or fully comprehend the avalanche of information that would be available to them, even if recorded and interpreted by a computer. And this would be particularly true when the information is most important to them, i.e., when they are very sick, seriously injured, or simply terrified by the possibility that they might be dangerously ill. Not many people will be obsessive enough to keep up with all the available information about their own body, its vulnerability to disease and likely response to treatment. They will continue to rely on physicians for interpretation of the data and for objective, professional advice, as well as for help with any needed medical procedures. In modern society, with few exceptions, people turn to experts for help, and they probably always will—especially when threatened by a disease or injury they do not fully understand.
Topol also is more optimistic about the clinical benefits of genetic information than many experts would consider justified. It is true that advances in molecular biology have identified a few dozen relatively uncommon inherited diseases that are attributable to specific mutations in the structure of a single gene. Knowledge of the proteins encoded by those genes has helped to pinpoint and understand the molecular events causing the disease. Examples of such diseases include familial occurrence of very high blood cholesterol, cystic fibrosis, sickle-cell disease, Gaucher’s disease (fatty deposits in the liver, bone marrow, and other organs), Huntington’s disease, and a certain type of breast cancer. They usually follow Mendelian laws of inheritance. Knowledge of the genetic defect, the mode of inheritance, and the causative mechanisms of these diseases has often led to improved medical treatment.
But most diseases, including the most common ones, such as diabetes, asthma, hypertension, coronary artery disease, Parkinson’s disease, late-onset Alzheimer’s disease, and most cancers, are associated with many variations in the genome—not with just a single mutation. The identification of these multiple genetic variations has often established a statistical association with the disease—usually a relatively weak association—but not the cause.
Neither has genetic information about most of these common diseases led to specific and definitive treatment. This is not only because of the multiplicity of the genetic variations found in patients with these diseases, but also because most of the diseases have so-called “multifactorial” causes. That is, they are the net result not only of multiple inherited and acquired variations in the genome itself, but of many factors outside the genome, elsewhere in the body or in the environment.
These factors may affect the way the genome works, or may themselves be on the molecular pathway directly leading to the disease. As a result, although they may run in families, the development of one of these diseases in a particular individual cannot be reliably predicted by Mendelian laws. Their association with genetic variations is purely statistical and not causal. It will probably be a long time before genomal sequencing results in the conquest of one of the common diseases. Far more likely, and much sooner, we will make progress through increased understanding of the molecular events directly responsible for the disease rather than from the sequencing of everyone’s whole genome.1
Topol is an expert in medical genetics, and he must know the arguments against his optimistic predictions; in fact, occasional brief comments in his book suggest that he is aware of the limited use of genomic sequencing in most patients with common diseases. For example, after describing the initial excitement in the year 2000 over the first complete sequencing of the human genome, he acknowledges that “we have yet to gain real knowledge about what was discovered….” Later, when talking about the widespread application of genomic sequencing, he says: “It’s not immediately clear what the clinical importance of any of this information is.” But the burden and tone of most of his argument leave a different impression. He repeatedly implies that we are on the verge of a great revolution based on understanding how genes determine our health.
Recent studies of the genomic DNA that is not a part of any of the known 21,000 genes have confirmed what most geneticists have assumed: this so-called “dark” DNA controls the activity of the genes. The new work finds that many individual variations in the regulatory DNA have loose statistical associations with common diseases that cluster in families, but cannot be predicted from Mendelian laws of inheritance. Whether these associations will lead to advances in the prevention and treatment of disease remains to be seen.
If genome sequencing ever leads to worthwhile advances in the treatment of disease, as Topol predicts, cancer will probably be the disease for which such advances are first achieved. Topol devotes much space to the genetics of cancer and its potential clinical applications. He explains: “Cancer is a genomic disease. It can’t develop without a change in the DNA sequence of the cancer cell genome.” (Strictly speaking, the change could be either in the DNA sequence itself or in its expression, as influenced by many factors outside the DNA genome.)
In a front page story in The New York Times of July 8, the science reporter Gina Kolata tells about the discovery of a mutated overactive gene in the white blood cells of a patient dying of acute lymphoblastic leukemia (a cancer of white blood cells in the bone marrow), and his dramatic, positive response to a drug that stopped the activity of the abnormal gene. Whether his response will be sustained and whether the mutation in this patient will be found in other patients with this disease are still unclear. Another story by Kolata (July 19) reports the discovery of many mutations in the genes of colon cancer cells, which raises the hope that it may be possible to block their effect with drugs and slow or stop the cancer’s growth. The key word here is “hope.” Although genomic abnormalities that account for their rapid growth will probably be found in other cancers, and drugs will be found that specifically target the responsible gene, it is impossible to predict how clinically useful this information will prove to be. Topol’s optimism seems at best premature.2
Many experts, particularly those in epidemiology and biostatistics, will also question Topol’s seemingly boundless enthusiasm for “pharmacogenomics,” i.e., the application of knowledge about each person’s genome to the design, testing, and clinical use of drugs. His argument starts with the recognized fact that in clinical trials the response of a large number of patients to a given dose of a particular drug is always variable, with individual results falling above and below the average. Topol evidently believes that much of this variation in responses among a large population of patients is caused by individual variations in the genome, which determine how a drug is metabolized and therefore what kind of effect it will have. He cites a few examples from among the hundreds of drugs now in use, in which genomic variations have been shown to result in different responses to a drug, ranging from great sensitivity to no response at all. He notes, for example, that a single small variation in a gene greatly reduces the clinical response to Plavix, a patented drug that prevents clotting of blood in coronary artery stents.
Assuming that these examples will be typical of most drugs, he concludes that drug treatment will henceforth have to be guided by knowledge of each patient’s genome and therefore will be “personalized.” He further predicts that future trials of drugs will need much smaller numbers of patients, grouped according to their genome, thus changing the way pharmaceutical companies develop new drugs. (He does not discuss what the business model would be for drug companies in fragmenting their markets in this way.)
Topol admits that post-marketing information about adverse side effects of drugs will continue to require data from large numbers of patients because side effects usually occur infrequently and would be missed if only small numbers of patients were studied. But epidemiological science tells us that statistical analysis of large trials is also needed for the evaluation of the clinical effectiveness of new drugs, and not simply for discovery of their side effects.
Epidemiologists know that much of the variation in the response of patients to a drug results from random errors, imprecise measurement, or a host of unknown events outside the genome, either in the body or in the environment surrounding the trial. Some of the largest variations may be genetically determined or might even be due to the effects of the “microbiome” (i.e., the recently discovered universe of bacteria that live symbiotically in and on the human body) on the absorption and metabolism of drugs; but epidemiological science teaches that most of the variation is simply random and not a fixed characteristic of each patient. This means that repeating the same treatment of the same patients would probably change the results for particular patients, though the average would stay the same. That is why we need statistical summaries of the results from clinical trials on large groups of patients, and why genomic information is not likely to determine pharmacotherapy anywhere near as frequently as Topol suggests.
There is a curious disconnect between Topol’s conviction that medical care of the future will be based on the unique, largely genetically determined characteristics of each patient (as particularly emphasized in his enthusiasm for “pharmacogenomics”) and his comments on the medical wisdom of a “community,” which are scattered throughout his book. Topol evidently believes that the widespread sharing of information through Facebook and other such social networks will provide a useful source of information about diseases and their treatments. It will lead to the “democratization” of medicine and the empowerment of patients—results that he strongly believes will lead to better medical care.
More understanding of their illnesses and greater participation by patients in decisions about their personal medical care are certainly desirable, but I do not believe the opinions of a community can reliably inform personal medical decisions or determine which drugs are useful. It is precisely because there are so many unique variables (genetic and other) that influence the clinical condition of each patient and his or her response to treatment, and because the assessment of the relevance of these variable factors to the treatment of each patient requires the judgment and experience of competent physicians, that the opinions of a broad constituency of people cannot be trusted.
Facebook provides advertisers with personalized information that can focus their marketing efforts. Genomic information might similarly provide personalized information for the marketing of medical products (it already has), but genomic variations (and non-genetic variables) make the medical experience and opinions of the social community of dubious relevance to a particular patient’s medical problems.
Digital communication has undoubtedly made it much easier for physicians and patients to access the world’s store of medical information. And the adoption of electronic technology (which is being subsidized and promoted by federal legislation) has facilitated the keeping and sharing of medical records and has probably reduced errors in ordering medications and treatments in hospitals.
However, there are significant downsides as well—some of which Topol clearly acknowledges. Digital electronic records may well result in serious problems with identity theft and violations of privacy and with the inappropriate use of personal information by government agencies and private insurers. In addition, unless government regulation mandates that all medical record information software be mutually compatible (not the present situation at all), problems with intercommunication between many different competing commercial systems will continue to recur.
Topol notes in passing that electronic records are potentially vulnerable to hackers. But he does not mention the recently expressed concern of the government that all US computerized information systems, particularly those related to national security, might be disabled by “cyberterrorists.” As patients’ records are increasingly digitized, theft and hacking problems have increased; an attack on national computer systems might also disable the health care system.
Topol also recognizes that computerized records, like computerized monitoring and imaging, draw physicians’ attention away from the real patient. Most patients are unhappy with physicians who pay more attention to their computer records than to them. They need and expect more human contact with their physician than is likely to occur in a future world of digitized information. Topol would have to admit that computers and wireless sensors cannot, for example, substitute for live physicians and nurses in hospital intensive care units (ICUs). Of course sophisticated monitoring instruments are essential in ICUs, but they cannot replace the judgment, support, and compassionate care provided by good physicians and nurses who use those instruments.
Neither can digital technology provide the medical care that terminally ill patients need. Computers have not even lived up to earlier expectations that they might largely replace physicians in diagnosing diseases and interpreting the significance of signs, symptoms, and laboratory findings in sick patients. Experience has shown that computers can be of great help but cannot fully substitute for the skills of good physicians. That is because they cannot have the breadth of view, the flexibility, or the intuitive sensitivity of the experienced physician when meeting a new patient with an undiagnosed disease.
Francis Peabody, a distinguished professor at Harvard Medical School in the 1920s, left a lasting mark on medical education by teaching that the treatment of a disease can be entirely impersonal, but the care of the patient must be completely personal.3 He was the author of an aphorism still widely cited to medical students: The secret of the care of patients is in caring for them. I doubt that any technological revolution, even one as profound as that envisioned by Topol, will ever change that fact. Good medical care, as is often said, is an art as well as a science. At best, computers can only deal with the science. Topol must know this, but he doesn’t say so.
No final assessment of his book can be made without brief mention of a problem that I have so far ignored. Topol has significant financial conflicts of interest. To his credit, he discloses them. In the acknowledgments at the end of the book he tells us that he serves on the boards of two companies that manufacture wireless sensing devices and consults for another, and he acknowledges that “these relationships are important for readers to know about.” The website of the Scripps Translational Science Institute also reveals that he and several colleagues there have recently founded a company that will help patients interpret and use information from their genome. Topol holds equity interests in this new business, and in the three he mentions in his book.
He writes that he has “no interest” in promoting the three companies on whose board he serves, although it is hard to believe he has no such interest in the genomic information company he and his Scripps colleagues have recently founded. In any case, few observers will give unquestioning credence to such declarations, however sincere these sentiments may be. His book promotes the products of an industry that includes businesses in which he has an equity interest, and that fact inevitably raises questions about a possible bias (perhaps unconscious) in his enthusiasm for the benefits of the industry’s products. That is the problem with all such financial conflicts of interest.
As an introduction for the lay reader to the sophisticated digital technology now being applied to medical care, and as an explanation of the rapidly advancing science of medical genetics and the new revelations from the sequencing of the human genome, Topol’s book is a tour de force. But as a credible prediction of how this technology and this information is about to revolutionize the practice of medicine and improve the health care system, it raises the serious doubts I have suggested. If major parts of Topol’s vision are eventually realized, as unlikely as that may seem to skeptical physicians like me, it will only be in the distant future. For now, I remain unpersuaded that the “creative destruction” of medicine is imminent.
Much of the foregoing skepticism about the clinical applications of whole-genome sequencing is supported by a recent, concise, and authoritative review by Liam R. Brunham and Michael R. Hayden, two medical geneticists at the University of British Columbia, in Science (“Whole-Genome Sequencing: The New Standard of Care?,” June 1, 2012).They say at the start that “the vast majority of genomic data is, at this time, not medically actionable.” Later, they add: “With few exceptions, there is a lack of data to suggest that genetic testing actually leads to improved health outcomes.” ↩
It is already known that another blood cell cancer, chronic myelogenous leukemia, is associated in the malignant cells with an abnormal gene that results from an acquired rearrangement of chromosomes. A drug that inhibits the expression of the abnormal gene induces remission in this otherwise fatal disease. However, the remissions usually last only a few years. ↩
F.W. Peabody, “The Care of the Patient,” Journal of the American Medical Association, Vol. 88 (1927), pp. 877–882. ↩