From The Sciences (New York Academy of Sciences), Nov/Dec 1985. Reprinted with permission.

Birth of an Errant Cell

A New Theory About the Cause of Cancer

by Jorge J. Yunis and William R. Hoffman
Related Websites and Articles

There comes a time in every life when cell division is supposed to slow down. In human skin, mucous membranes, and blood-forming lymph nodes and bone marrow, the pace is rapid throughout life, even after the slackening that occurs at adulthood to make the shift from growth to replacement. But in nerve, bone, and muscle, cell division grinds virtually to a halt. The cellular population becomes relatively static. There is no growth in these tissues, not even much turnover; and cells do not constantly die and get replaced. Of all the biological rites of passage into adulthood, this slowing of cell division and cessation of tissue growth are perhaps the most crucial, for they help keep the cells in a differentiated state, concentrating on their specialized functions instead of on multiplying. This quality of differentiation is basically all that distinguishes a normal cells from a cancer cell.

It takes no more than a single cell gone awry to start a tumor. Consider the simplified case described by Christian de Duve, of Rockefeller University: One cell somehow escapes the normal checks on growth and divides, on average, once every hundred days. (This seemingly moderate pace may come as a surprise, but cancer cells do not, in fact, divide all the time or perform the process any faster than do normal ones.) The result, a simple exercise in exponential growth, is chilling nonetheless. The errant cell may take eight years to form a pea-sized lump; just two years later, the tumor weighs a pound.

Where does such a process begin? What makes a healthy cell turn malignant in the first place? Biologists have suspected since the turn of the century that the answers would be found within the cell's genetic machinery, but only during the past few decades has there been any real progress toward linking specific genetic defects with particular types of cancer. Perhaps one of the most important findings in recent months is that during cell division, normal human chromosomes--the gene-bearing threads of DNA within the cell nucleus--break easily at certain weak points, under certain conditions. Everyone has such fragile sites, and breaks at some of them are indistinguishable from chromosome breaks found in various types of malignancies; their role in the development of cancer, though not yet fully defined, may well be critical.

These findings have profound implications for the diagnosis, treatment, and prevention of cancer. In leukemia, for instance, analysis of chromosome breaks and rearrangements can help doctors determine how severe the disease is and whether a patient is heading for remission or deterioration; thus it can aid in tailoring treatment. In addition, because fragile sites appear to be weaker in some people than in others, it may be possible to develop a blood test to detect a predisposition to certain cancers. And because the B vitamin folic acid appears to enable fragile sites to resist breakage, and least in cell culture, it may turn out to play a role in the prevention of cancer. Although this work is preliminary, it provides a framework for understanding how diet and oncogenes--the cancer-causing genes that normally lie dormant within all cells--may interact at fragile sites in the initiation of cancer.

Biologists have studied the chromosomes of other organisms since early in this century, but little headway was made with human chromosomes before the 1950s. Until then, in fact, the analysis of human chromosomes was considered a maddeningly difficult, somewhat esoteric pursuit. The chromosomes are so scrunched up within the cell nucleus that it seemed virtually impossible to disentangle them and stretch them out for a look. (It is now known that the DNA in the largest human chromosome, number 1, is five inches long.) The first technical advance, in the early 1950s, resulted from a mistake in the laboratory: T.C. Hsu, of the University of Texas, in Galveston, inadvertently immersed human cells in a dilute salt solution prepared by a technician who had misread the instructions for making the mixture; the cells swelled osmotically, scattering the chromosomes. Thus, they could be analyzed individually for the first time.

In 1956, Joe Hin Tjio and Albert Levan, at the Institute of Genetics, in Lund, Sweden, studied human cells with the aid of both dilute salt solutions and the drug colchicine, which freezes cells division when the chromosomes have separated into clearly visible bodies. To their amazement, Tjio and Levan counted forty-six chromosomes (twenty-three pairs) rather than forty-eight, which had been the accepted tally for thirty-three years. Just three years later, in 1959, the French geneticist Jerome Lejeune, of the Universite de Paris, presented the first evidence linking a human disease with an abnormal chromosome; each of his patients with Down's syndrome, which causes mental retardation and various physical defects, had an extra chromosome 21.

The next year brought the first report of an association between a chromosomal defect and cancer. Peter Nowell, of the University of Pennsylvania, and David Hungerford, of the Institute for Cancer Research, both in Philadelphia, observed that chromosome 22 in patients with the chronic myelogenous form of leukemia was abnormally short in the malignant cells. The researchers suggested that the Philadelphia chromosome, as it is now called, somehow caused the disease, and their work came to be regarded as a landmark in the study of cancer in human beings.

But Lejeune and Nowell and Hungerford had identified relatively gross abnormalities. Subtler defects eluded detection, for under the microscope the chromosomes showed up as just shadowy threadlike figures of slightly different lengths. It was recognized that each chromosome has a long arm and a short arm that join at a constriction called the centromere and that the chromosomes reproduce themselves before cells division, so each daughter cell receives a set identical to the parent's. But little was known of the chromosomes' structure. Then, in the late 1960s, a staining technique known as chromosome banding brought the shadowy figures into focus.

In essence, banding marks the chromosomes in a way that makes it possible to detect chemical and physical changes in them through a light microscope. Thus, it offers a means of comparing chromosomes and charting structural differences between them. The chromosomes are often stained with a dye called Giemsa, which is taken up in differing degrees by the four base molecules that make up DNA. What results are distinctive and reproducible patterns of dark and light bands that change noticeably if the chromosomes undergo some alteration, such as a rearrangement or the loss of a segment. Not only does banding permit the detection of changes, it also makes it possible to pinpoint their locations and thus to home in on specific defects that crop up again and again in association with particular diseases. The more stretched a chromosome, and the more bands it carries, the more accurately its structure can be mapped. The original techniques produced between 250 and 320 bands on a set of twenty-three chromosomes--half the human complement.

Naturally enough, researchers eagerly sought to apply these findings to cancer research. But the results were perplexing: only about half the cancers studied seemed to involve an abnormal chromosome. So it was presumed that most chromosome defects were random, probably unrelated to the disease. The presence of the Philadelphia chromosome in patients with chronic myelogenous leukemia, and of specific chromosome defects in a handful of other cancers, was regarded as an exception. But the explanation for this paradox lay in the early banding techniques, which in tumor cells often produced only 150 to 250 bands for each set of twenty-three chromosomes.

During the past ten years, the perspective has changed dramatically. High-resolution banding techniques developed in our laboratory have unraveled the chromosomes to unprecedented lengths and have increased the number of bands produced in tumor cells to anywhere from 320 to 1,200. This extraordinary degree of resolution is the result of adding drugs to the cells, which causes them to divide in synchrony, then stopping the process at a certain point to yield as many chromosomes as possible in earlier stages of cell division.

Since 1980, we have analyzed tumor cells from more than five hundred patients and found that most have a specific chromosome defect--involving breaks at places where we have also discovered fragile sites--that is characteristic of a given disease. For example, in studying the bone marrow of patients with acute nonlymphocytic leukemia and the lymph nodes of victims of non-Hodgkin's lymphoma, we found that nearly all the patients had defective chromosomes and that in most the defect was specific and recurrent. More important, we showed that acute nonlymphocytic leukemia is actually seventeen diseases, in terms of chromosome defects, and that the type of defect found indicates the severity of the disease. A patient who has a defect associated with the best outlook would have time to try for a cure by undergoing bone marrow transplantation, whereas one with a worse prognosis could be offered aggressive chemotherapy or experimental treatments. The vast majority of solid tumors--those affecting bone, muscle, connective tissue, skin, and mucous membrane--have not been studied extensively, but investigators have found recurring chromosomal defects in several, including those of the lung, the kidney, and the ovary.

These defects, which prevail throughout the lifetime of a malignancy, take different forms. The two most common are an exchange, or rearrangement--in which segments from two chromosomes break off during cell division and trade places--and a deletion, or gap--in which a piece of a chromosome breaks off and gets lost. But why should such rearrangements and gaps lead to cancer? Because, apparently, they disrupt the mechanisms that normally control the activity of oncogenes. Twenty oncogenes have been found thus far in human cells, and seventeen have been mapped to specific bands on the chromosomes and linked to several forms of cancer, including chronic myelogenous leukemia and Burkitt's lymphoma. Oncogenes may start out as good, indeed, essential segments of DNA; they likely play a crucial part in the frenetic cell division, growth, and differentiation that characterize early life, by directing the synthesis of enzymes essential for these processes. But when growth is finished, the oncogenes are shut down or kept at low levels of activity.

Two widely discussed theories of cancer hold that a cell becomes malignant either after the DNA of certain viruses become integrated into the cell's genes or after the cell's DNA has changed chemically--undergone some mutation--perhaps because of an attack by a carcinogen on an oncogene. But our recent analysis of twenty-nine types of leukemia and lymphoma in humans suggests that the critical step in cancer often may be the break of a fragile site near an oncogene, without either mutations or viral involvement. The rearrangements that result from such breaks may bring the oncogene on one chromosome under the influence of another type of gene on another chromosome, a gene with the power to activate the oncogene.

Several chromosomal rearrangements and gaps involving oncogenes have been associated with different types of cancer. One is Burkitt's lymphoma, a disease affecting B lymphocytes, the cells of the immune system that normally produce disease-fighting antibodies. In Burkitt's, and oncogene designated myc shifts from its normal location on chromosome 8 to a region on chromosome 14 near the highly active genes that direct antibody synthesis. These gene, in turn, may activate the myc oncogene, which somehow waives the biological laws that would normally tell the B lymphocyte when to stop multiplying. When breaks at fragile sites produce gaps rather than rearrangements in the chromosomes, solid tumors are more likely to develop than are leukemia or lymphoma. These gaps may include the loss of a gene that normally keeps an oncogene in check. This happens in two cancers that strike mainly in children: Wilms's tumor, a kidney cancer; and retinoblastoma, a tumor of the eye.

The study of normal cells has shed a surprising amount of light on the role of fragile sites in cancer. It has been known since the mid-1960s that chromosomes break in a small percentage of healthy cells during routine culturing. The cause remained elusive until the next decade, when the Australian geneticist Grant Sutherland found that he could reproduce one class of these breaks by using a culture medium that was low in the B vitamin folic acid, which is essential for the synthesis of thymidine--a substance need to produce thymine, one of the four base molecules that make up DNA. This culture method, widely adopted in other laboratories, has revealed sixteen fragile sites, which are now known to be inherited by about two people in a thousand and which appear to be DNA sequences rich in thymine. Until 1983, most of these heritable sites were not linked to any disease except for one apparently associated with a form of mental retardation. But that year, we found that several leukemia and lymphoma patients who had had an exchange in their tumor cells involving chromosomes 11 and 12 also had heritable fragile sites at the same spots in their normal blood cells when these cells were cultured in a medium lacking folic acid. In each case, the rearranged chromosomes had broken at a fragile site, suggesting that the site had acted as a factor predisposing them to the disease. During the following months, the correlation was noted in more patients; it turned out that eight of the sixteen known heritable fragile sites were located at or near the spots where chromosomes broke in several types of leukemia and lymphoma.

In addition to the sixteen heritable fragile sites, human chromosomes have displayed seemingly spontaneous breaks at other sites. Generally, these breaks occurred so infrequently, even in the absence of folic acid, that they have been difficult to characterize. It occurred to us that perhaps the chromosomes were breaking but repairing themselves too fast for the breaks to be detected. A lucky accident helped confirm our hunch. In another series of experiments, we were using caffeine, which enhances the effects of carcinogens. When we inadvertently added caffeine to cultures of folic acid-deprived cells from healthy people, we were surprised to see that it caused at least a tenfold increase in chromosome breakage; perhaps the caffeine pushed the cells to finish dividing before they were able to repair the breaks. In any event, the process was reversible: adding folic acid or thymidine to the cultures dropped the breakage rate nearly to zero, despite the caffeine (Yunis and Soreng, Science, 1984).

Our experiments with caffeine and cells deprived of folic acid have revealed ninety-six fragile sites that occur in all human beings, as well as chimpanzees and gorillas. Of the forty-one specific chromosomal rearrangements that so far have been associated with various cancers in human beings, thirty-eight involve fragile sites, and nine of these lie near known oncogenes. The next step was to study healthy cells from people with various types of cancer and specific defects in their tumor cells to see if the chromosomes in their normal cells were more breakable than those of healthy people. Indeed, patients with five types of leukemia and lymphoma had an increased number of chromosome breaks in healthy cells that corresponded to the breaks in their cancer cells. Two of these patients had histories of heavy exposure to herbicides, pesticides, and organic solvents such as benzene, as well as a chromosome defect characteristic of leukemia victims with such exposure, suggesting that their fragile sites were especially vulnerable to attack by carcinogens.

These findings raise many intriguing possibilities. One of general interest is the question of whether caffeine is safe. Regrettably, there are no answers yet. Epidemiological studies have failed to find a connection between the drinking of coffee and tea and the incidence of cancer; moreover, the caffeine levels used to break chromosomes in the laboratory are now fairly high, the equivalent of more than twenty cups of coffee a day.

If confirmed, the work on fragile sites will suggest that some people are more susceptible to chromosome breaks that make them prone to certain types of cancer. It might thus be possible for them to lower their overall risk of the disease by avoiding exposure to known cancer-causing agents. But could those people also cut their risks by taking folic acid supplements? The question deserves study, because all cells need folic acid for DNA synthesis, and if folic acid is in short supply, the chromosomes may break at fragile sites. In addition, though the evidence is not clear, it appears that people with low levels of folic acid in their blood, such as chronic alcoholics, may be more prone than average to certain types of cancer. Nonetheless, we cannot recommend supplements.

Although folic acid, even in large doses, appears safe for most people, for some it may be dangerous. In patients with a vitamin B12 deficiency (a relatively uncommon problem), folic acid can mask the deficiency, which, untreated, can lead to serious neurological problems. What's more, most healthy people, with the notable exception of pregnant women, do not seem to need folic acid supplements. Basic human needs (the government's recommended daily allowance is 400 micrograms) are normally met by a balanced diet. Leafy and dark-green vegetables, asparagus, whole-grain breads and cereals, organ meats, nuts, and some fruits, particularly oranges, are rich in folic acid. As for the value of folic acid supplements in the prevention of cancer, it would take a ten-to-twenty-year study to decide.

What we have learned about fragile sites also raises questions from an evolutionary standpoint. If the sites are potentially dangerous, why has evolution preserved them? We suspect that fragile sites are rich in the DNA base thymine--hence their fragility and their need for folic acid and thymidine during replication--and we know that thymine-rich DNA unwinds faster than do other types. And a gene must unwind to become active. If the sites are regulatory sequences of DNA that control certain genes, such as oncogenes, there may be some adaptive advantage in the ability to unwind and act quickly. There may also be an advantage to the fragility itself; not every fragile site is associated with cancer, and some chromosomal changes, particularly those that occur in the germ cells to create genetic diversity in the offspring, may constitute a source of genetic flexibility and vitality. It is also possible that cancer, since it strikes most commonly in people who have reached the end of their reproductive lives, is probably not selected against by evolution. Perhaps, as the science historian June Goodfield has suggested, "cancer is a price that we as individuals pay for the life of man as a species."