The stupidest virus is cleverer than the cleverest immunologist.
—GEORGE KLEIN
In the past two decades it has become clear that infection with certain viruses—as well as certain bacteria—accounts for a substantial proportion of cancer worldwide. However, this knowledge has been achieved only with great difficulty and after pursuing many false leads. The question of the possible role of infection in the cancer process was first raised in 1911 when Peyton Rous of the Rockefeller Institute in New York succeeded in inducing tumors in healthy chickens by injecting them with chicken sarcoma virus from tumor-bearing chickens.
1 But owing to the difficulty of demonstrating the causal role of a virus in human cancer—a disease that often takes decades to develop—it would be another fifty years before there was compelling evidence for a direct link between a virus and a specific cancer in humans. When this evidence finally emerged, it came from a remote corner of the world and was due to the acuity and persistence of a general surgeon working in colonial Africa.
After the end of World War II the Irish physician-surgeon and Presbyterian missionary Denis Parsons Burkitt went to Uganda to minister to people with virtually no access to medical care. In 1957 he was asked to see a 5-year-old boy whose face was deformed by a swollen jaw. Soon after he saw a girl with identical swelling of the jaw and noticed that as the disease progressed she developed swelling in other organs as well.
2 This led him to undertake a meticulous search of district medical records, which revealed that rapidly growing tumors of the jaw were common in children in Uganda and were often associated with tumors in other parts of the body. Tumors of the jaw were, in fact, the most common childhood cancer in Africa as well as the fastest-growing childhood tumor. While the common occurrence of this tumor in African children had been noted early in the twentieth century, Burkitt was the first to posit that the apparently different childhood cancers were all manifestations of a “single distinctive tumour syndrome.”
3 Two years after his initial publication in 1958, two of his colleagues identified the tumor as a lymphoma. Although Burkitt himself continued to refer to the “African lymphoma,” the newly identified cancer came to be known as “Burkitt’s lymphoma.”
4To trace the contours of the disease, Burkitt first sent an illustrated leaflet and questionnaire to medical practitioners throughout Africa. (The printing and postage costs were covered by a £25 sterling grant from the British government.) To maximize the accuracy of reporting, he hit on the idea of using the distinctive swelling of the jaw in young children as an “index” of the disease, since any medical practitioner who had seen a case was unlikely to forget it or to mistake it for any other condition. The responses to his questionnaire indicated that the tumor occurred in a belt across Central Africa. To obtain a finer-grained picture of what distinguished areas where the tumor was prevalent from those where it was absent, Burkitt and two physician friends organized a “tumor safari,” whose purpose was, in his words, “to determine more accurately the limits of this ‘tumor belt’ and the physical and climatic conditions determining the boundaries.”
5 In an inspired formulation that likened the task of tracing the geographical boundaries of the occurrence of the tumor to the surgeon’s task of delineating the boundaries of a cancerous lesion, Burkitt referred to his journeys as “surgical biopsies” in which the pathologist attempts to define the “edges” that separate diseased from normal tissue.
In the course of a ten-week safari, the three friends covered ten thousand miles and visited fifty-six hospitals in nine countries. In a paper in the
British Medical Journal in 1962, Burkitt gave a methodical description of their findings and included four maps displaying the data in different ways. Their surveys showed that the “lymphoma belt” stretched across equatorial Africa extending roughly 15° on either side of the equator and continued south in a “tail” along the east coast. The tumor was prevalent in areas at lower altitude and with greater rainfall and was often concentrated along rivers. The greater the distance from the equator, the lower the upper limit on the altitude at which the tumor occurred. Burkitt noted that because of climatic and topographic differences, the tumor could be prevalent in one area and absent in another area less than one hundred miles away. As a number of his colleagues pointed out, the epidemiological maps of malaria and the tumor belt overlapped.
Based on his findings, Burkitt concluded that the effect of altitude must reflect the requirement of a minimum temperature of 60 degrees Fahrenheit. And he went on to speculate that “the fact that this unusual tumor is temperature dependent, implies that some vector may be involved in its transmission. This in turn suggests the possibility that a virus is implicated.”
6On a visit to London in March 1961, Burkitt presented the results of his research in a lecture entitled “The Commonest Children’s Cancer in Tropical Africa—A Hitherto Unrecognised Syndrome.” At the back of the lecture hall was a young pathologist named Anthony Epstein of the Bland Sutton Institute. Epstein was working on the role of viruses in carcinogenesis, and he had been drawn by the title of the lecture. He later recalled that he was riveted by Burkitt’s description of the tumor’s geographical distribution as well as by its association with different organs within the body: “I could hardly sit still because it was immediately clear that anything which had its distribution determined by temperature and rainfall had a biological cause. And of course for me working with the Rous sarcoma virus, a tumor virus of animals, it had to be that it was a virus induced tumor in humans, and that so far as I was concerned was it.”
7 The two men spoke after the lecture, and Burkitt agreed to send Epstein frozen specimens of tumor tissue taken from affected children in Uganda.
8For two years Epstein and his coworkers’ attempts to isolate a virus using standard techniques and electron microscopy were unsuccessful. They then decided to grow Burkitt lymphoma cells in vitro, away from host defenses to allow the hypothesized oncogenic virus to replicate. Though previous attempts to establish cells of a lymphoid tumor in culture had failed, they were able to establish the first Burkitt lymphoma–derived cell line. Under the electron microscope, a cell in the first grid square showed particles that were recognizable as having a herpes virus morphology. Their results were published in the
Lancet in 1964 in what has become a citation classic (“This Week’s Citation Classic,” Apr. 2, 1979).
9 The virus, which was named Epstein-Barr virus (EBV), was the first cancer-causing virus to be identified in humans. In the late 1970s the results of a large epidemiologic study in which blood samples were collected from forty-two thousand children between the ages of 4 and 8 years in the West Nile district of Uganda showed that children with high levels of antibodies to EBV, indicating past infection with the virus, were at high risk of developing Burkitt’s lymphoma.
10 These findings strongly supported a causal relationship between EBV infection and the disease but suggested that the oncogenic potential of the virus is realized only in exceptional circumstances. Evidence later emerged that infection with malaria, which impaired resistance to the Epstein-Barr virus, was a necessary cofactor.
11Although Burkitt’s lymphoma is the most common childhood cancer in areas where malaria is endemic (equatorial Africa, Brazil, and Papua New Guinea), it is extremely rare in other parts of the world, and the profound implications of the discovery of Epstein-Barr virus in Burkitt’s lymphoma cells were not immediately apparent. Nevertheless, the work of Burkitt, Epstein, and colleagues spurred further research into the biological mechanism whereby the virus transformed lymphatic cells as well as the role of cofactors, such as malaria infection and immune status.
12 The Epstein-Barr virus is widespread, causing silent infections and infectious mononucleosis, and strikingly is associated with two human cancers, nasopharyngeal carcinoma as well as Burkitt’s lymphoma.
* * *
During the precise years when Burkitt and his colleagues were documenting the extent of the occurrence of childhood lymphoma in equatorial Africa and developing chemotherapeutic treatments, a young German virologist, Harald zur Hausen, had completed his training and in 1966 had taken a postdoctoral position with the virologists Werner and Gertrude Henle at the Children’s Hospital of Philadelphia. The Henles’ laboratory was focused on the recently discovered Epstein-Barr virus. They had received EBV cell lines from Epstein’s lab and were working to develop serologic markers of infection (that is, to detect antibodies to the virus in blood) for use in epidemiologic studies. In Philadelphia, zur Hausen requested to work on another virus (adenovirus) in order to familiarize himself with the techniques of molecular biology. To please his mentor, however, using the electron microscope, he demonstrated the presence of EBV particles directly in Burkitt’s lymphoma cells that showed serologic evidence of infection, confirming the usefulness of the Henles’ antibody test.
In 1969, having received an offer to set up his own research group at the University of Würzburg, zur Hausen returned to Germany and immediately decided to shift his focus entirely to EBV. His objective was to prove that EBV DNA persists in every tumor cell of Burkitt’s lymphoma but does not establish a persistent infection there, as the Henles and others had assumed. In contrast to his mentors, who believed that only a minority of lymphoma cells harbored persistent infection, zur Hausen posited that EBV is present in all Burkitt’s lymphoma cells and might be spontaneously reactivated. He obtained a large number of Burkitt’s lymphoma cell lines and tumor biopsies, as well as material from nasopharyngeal carcinomas, which also appeared to be associated with EBV infection. Quickly overcoming the major obstacle, namely, the purification of adequate quantities of EBV DNA from a small number of virus-producing cells, he was able to show that a particular Burkitt’s lymphoma cell line that was not actively producing EBV nevertheless contained multiple copies of EBV DNA in each cell. Soon after, using the new technique of in situ hybridization, zur Hausen was able to identify EBV genetic material in all Burkitt’s lymphoma samples, as well as in samples of nasopharyngeal carcinoma, but not in any of the controls. He went on to demonstrate that the amount of viral material remained constant over time, suggesting an “intimate interaction of viral DNA with most, if not all, tumor cells.” As he later wrote in his Nobel biography, “It seems that this was the first demonstration of persistent tumour virus DNA in human malignancies.”
13During this period the Henles also demonstrated the causative role of the virus in infectious mononucleosis as well as the presence of EBV in nasopharyngeal carcinoma (the latter work involved zur Hausen). A summary of this work contained in the Henles’ papers at the National Library of Medicine describes its significance as follows: “This significant relationship between Epstein-Barr virus and cancer demonstrated that the presence of certain viruses in the nucleus of a cell could transform a healthy cell into a malignant one.”
14Zur Hausen was to maintain an interest in EBV and Burkitt’s lymphoma throughout his career. In 1972, however, he was appointed chairman of the newly established Institute of Clinical Virology in Erlangen-Nűrnberg. In his new position he decided to turn his attention to another cancer—cancer of the cervix—and a particular family of viruses—the papillomaviruses—which was to become the main focus of his energies. His work over the next eleven to twelve years was to radically transform our understanding of the role of viruses in human cancer, shedding light on this large and varied class of viruses, with momentous implications for the prevention and control of a major cancer in women worldwide, work that would eventually earn him a Nobel Prize.
The human papillomaviruses (HPVs) are small, double-stranded DNA viruses wrapped in a protein shell that have coexisted with the human species for hundreds of thousands of years, undergoing relatively few changes in their genetic makeup. Among the more than one hundred different HPV types that have been identified, some have an affinity for colonizing the skin, where they can produce warts, while others are adapted to the lining of the genital tract and other internal tissues. A third group is equally at home in either environment.
15Cervical cancer is a malignant neoplasm arising from the cells lining the cervix. Most cervical cancers are squamous cell carcinomas arising in the squamous, or flattened, epithelial cells lining the outer surface of the cervix. Adenocarcinoma arising in the glandular epithelial cells is the second most common type.
The earliest observations linking cervical cancer to a sexually transmitted agent date from the nineteenth century. In 1842 the Italian physician Domenico Antonio Rigoni-Stern had observed that cervical cancer rarely occurred in unmarried women and was virtually nonexistent among nuns, in contrast to its occurrence in married women, widows, and, particularly, prostitutes.
16 Later studies indicated that the disease was much rarer among women in certain religious groups, including Jews, the Amish, Mormons, and Seventh Day Adventists.
17 In the 1970s cervical cancer was the most common cancer occurring in women worldwide—today it is the fourth most common, surpassed only by cancers of the breast, intestines, and lung. Its highest rates were—and still are—seen in developing countries, particularly in East Africa, Central America, and the Pacific Islands.
18 Both internationally and within countries, cervical cancer incidence was associated with lower socioeconomic status. In the United States, cervical cancer rates are 45 percent higher among black women and 65 percent higher among Hispanic women compared to white women.
19Epidemiologic studies had identified a number of risk factors for cervical cancer in addition to socioeconomic status, including religion, having a larger number of children, use of oral contraceptives, smoking, and possibly nutrition. The two strongest and most consistent risk factors, however, were an early onset of sexual activity and a history of multiple sexual partners.
20These findings pointed to an infectious agent as a cause of the disease, though by the end of the 1960s there was no solid evidence implicating a particular agent. Around this time the first reports appeared suggesting infection with Herpes simplex virus type 2 (HSV-2) as an agent in cervical cancer etiology, and HSV-2 became the prime suspect.
21 But over the next decade, attempts to isolate HSV-2 particles in cells from cervical cancers were uniformly unsuccessful, and large-scale epidemiologic studies failed to support a role of the virus. Zur Hausen, together with a coworker, had also failed in his efforts to find evidence of HSV-2 infection in cervical carcinoma.
22 This is where things stood when he returned to Germany.
In addition to the epidemiologic evidence implicating a transmissible agent in the development of cervical cancer, another body of work had contributed to zur Hausen’s thinking. Looking back in a recent interview on how he came to focus on HPV and cervical cancer, he explained, “I was only interested in cervical cancer.”
23 His attention had been directed to papillomaviruses by going back to the literature from the 1930s, when the researchers Richard Shope and Peyton Rous had observed lentil-like structures in wild U.S. cottontail rabbits. By taking cell-free extracts from these lesions and injecting them into domestic rabbits, they were able to produce similar warts that eventually became malignant. Zur Hausen had also come across anecdotal reports regarding genital warts that had occasionally converted gradually to malignant tumors. “These findings triggered the idea that there may be an agent in the genital lesions that could also cause cervical cancer.”
24After the failure to find a link between HSV-2 and cervical cancer, in 1972 zur Hausen and his group started to work experimentally on HPV. He was convinced that genital warts were caused by a virus. He had observed HPV particles in genital warts and felt that HPV would be a “good candidate” for the infectious cause of cervical carcinoma.
25 In 1973 at a meeting in Key Biscayne, Florida, he proposed that HPV was the cause of cervical cancer, but, given the prevailing consensus favoring HSV-2, his proposal met with little interest. Zur Hausen had collected a few hundred warts from individual patients and had isolated wart viruses from the skin of the hands and feet. But, to his disappointment, the wart virus could not be detected in cervical cancer biopsies or in genital warts. This was the first hint that there must be different types of human papillomavirus. Owing to the small number of viral particles in genital warts, it took zur Hausen and his colleagues several years to characterize and isolate HPV type 6 from a genital wart—this was achieved in 1977.
26 This type also could not be detected in carcinoma specimens. The researchers persisted, however, and a year later they found a related virus—HPV-11—in genital warts. Using HPV-11 as a probe and relaxing the stringency of the assay, they finally managed to isolate the distantly related HPV-16 and HPV-18 and, eleven years after embarking on this effort, to “link them convincingly to cancer.” Reflecting, in an interview with the journal
Nature (2012), on the painstaking path that led to the discovery, zur Hausen commented, “It was not a Eureka moment.”
27As often happens with radically new ideas, the scientific community treated zur Hausen’s findings dismissively. Infectious disease researchers, who had devoted years to investigating the role of HSV-2 and other sexually transmitted infections in cervical cancer, were unpersuaded; tumor virologists were skeptical in the absence of evidence demonstrating how the virus initiated cancer; and epidemiologists wanted to see data from carefully designed studies of human populations demonstrating a convincing association. While epidemiological and serologic studies had quickly linked hepatitis B virus with liver cancer, EBV with B-cell lymphomas, and
Helicobacter pylori infections with gastric cancer, it took longer to work out and validate the serologic markers of HPV infection and to scale up the HPV molecular hybridization assays from the basic research laboratory to the clinical research laboratory to enable processing of the thousands of specimens needed in epidemiologic studies.
28 It took nearly another decade of intensive research into the natural history of HPV infection and the epidemiology of cervical cancer before the cancer community accepted the evidence that HPV was the cause of the disease.
The bafflement regarding the causes of cervical cancer just before the identification of HPV-16 and HPV-18 is captured in a review article by Barbara Hulka, an epidemiologist at the University of North Carolina, summarizing the state of knowledge regarding the etiology of cervical cancer in the early 1980s:
Despite a long history of research into the epidemiology and biology of cervical carcinoma, a definitive statement about its probable causes still remains elusive…. Although vigorously pursued, an increased risk from oral contraceptives has not been convincingly demonstrated. A variety of venereally transmitted organisms appear to be frequent cohabitants with cervical neoplastic cells. Herpesvirus type 2 still remains the prime suspect in the complex pathogenesis of cervical neoplasia. Clinical findings, biological characteristics of the virus, serological studies and interactions of host cells and viral particles continue to stimulate the most intensive investigative efforts.
29Only in the final paragraph did Hulka mention the link between genital warts and cervical dysplasia and HPV, stating, “A role of this virus in the development of cervical cancer has not yet been demonstrated.” The next year zur Hausen would publish his definitive results implicating the high-risk HPV-16 as a cause of cervical cancer.
30A number of factors contributed to the difficulty of establishing that HPV played a causative role in cervical cancer. First, papillomaviruses are virtually ubiquitous on human skin and are widespread in the epithelial cells that form the lining of anogenital tissues, and it is difficult to determine whether evidence of viral infection in tumor cells points to causation or whether the virus is merely a “passenger” or “bystander,” “cohabiting,” as Hulka had put it, in cervical cancer cells.
Second, as research into HPV progressed, the number of specific types continued to grow, and at present well over one hundred types have been identified. Only a minority of HPV types are associated with cancer.
31Third, several features of the virus made it difficult to study. It is not easily cultured in the laboratory, and infected individuals do not mount a consistent antibody response, limiting the use of antibody levels in the blood for identifying viral types. This has meant that historically the only way to identify HPV has been from biopsies of warts or lesions. Furthermore, the molecular techniques for detecting viral particles and for amplifying viral DNA in the 1970s and 1980s were cumbersome and insensitive. These were to undergo dramatic improvements with in situ hybridization and culminating in the development of polymerase chain reaction, or PCR, in 1983. PCR made it possible to amplify a single or a few copies of a piece of DNA by orders of magnitude, generating thousands to millions of copies of a particular DNA sequence.
A final factor that complicated the analysis and interpretation of the role of HPV infection in the cancer process was the lack of understanding of the “natural history” of cervical cancer, that is, the fact that cervical malignancy is the end result of a gradual progression from normal tissue to mild abnormalities to increasingly distorted behavior of the cells lining the cervix and finally to invasive cancer.
32 The process culminating in malignancy typically unfolds over a period of fifteen to twenty years. Further obscuring the role of HPV in the carcinogenic process was the fact that, in most cases, early and even intermediate stages of dysplasia undergo spontaneous regression to normal epithelium (referred to as “clearance” of the infection).
33 It took years of research examining the acquisition of HPV infections in young women and then following them over a period of years to appreciate that most HPV infections resolved on their own owing to the body’s immune defenses, and that persistence of the virus leading to advanced dysplasia and ultimately to invasive cancer was relatively rare.
34 Robert Burk, pediatrician at the Albert Einstein College of Medicine, who got involved in studying HPV in the early 1980s, did early work that contributed to clarifying the natural history of the virus. He carried out one of the first studies showing that the prevalence of infection was strongly age-dependent—rates of infection were highest in young women and then declined, reflecting clearance of the virus.
35Until the natural history of the progression to cervical cancer was understood, there seemed to be a disconnect between the high incidence of HPV infections in the general population and the comparatively much lower incidence of invasive cervical cancer.
36* * *
HPV is the most common sexually transmitted infection worldwide. It infects roughly 10 percent of women at any given point in time, making it a “universal and pandemic” infection, as one researcher put it to me. Most women in the world will be infected by the virus at some point in their lives.
37 The highest rates of infection are seen in women in their teens and early twenties—those most sexually active—after which the rates decline with age, although there is a second lower peak around the age of menopause. In young women in the United States the prevalence of infection is about 25 percent.
38 These infections are, for the most part, transient and regress spontaneously. Ninety-five percent of lesions disappear within two years on their own and are of no clinical significance. However, in 5 to 10 percent of infected women, the infection persists and can result in precancerous lesions, which, if not removed, can progress to invasive cervical cancer. Thus cervical cancer is the end result of gradual progression of cellular pathology from normal tissue to increasingly abnormal features, and finally to invasive cancer.
39 The progression to invasive cancer normally takes decades, and this accounts for the success in the second half of the twentieth century in preventing cervical cancer in developed countries by widespread Papanicolaou (Pap) screening in which cells from the cervix are sampled and examined under a microscope. Regular Pap screening is effective because the cervix is accessible to clinical examination and because, once detected, precancerous lesions can be excised.
Of the more than one hundred HPV types, about fifteen are high-risk types for cervical cancer. Chief among these are, first and foremost, HPV-16 and, secondarily, HPV-18. The predominant role of HPV-16 is underscored by the fact that it is present in less than 3 percent of normal cervical tissues but is present in 20 percent of low-grade dysplasia, 45 percent of high-grade dysplasia, and 50 percent of invasive cancers. Together HPV-16 and 18 are responsible for roughly 70 percent of cervical cancer globally.
40Researchers at the National Cancer Institute recently summarized the significance of the natural history and the multistage process of HPV infection for the development of cervical cancer as follows: “The stages of cervical carcinogenesis include HPV infection; persistence, rather than clearance of the virus, linked to the development of a high-grade precursor lesion or ‘precancer’; and invasion.
These are the necessary stages; cervical cancer is virtually impossible in the absence of sexually transmitted HPV infection and in the absence of intermediate progression to precancer” (emphasis added).
41With the identification of specific HPV types associated with developing genital warts and cervical cancer, research into human papillomavirus and its role in cervical cancer intensified starting in the 1980s. The rate of acceleration is conveyed by the number of PubMed citations containing the terms “human papillomavirus” and “cervical cancer” over the past forty years: 1 in 1974, 5 in 1980, 81 in 1985, 221 in 1990, 432 in 2000, and 988 in 2010. Studies addressed different aspects of HPV infection with specific types and its role in cervical carcinogenesis. As the results of these studies appeared, they helped resolve poorly understood issues, fill in important gaps, and raise new questions. The HPV story is a powerful example of the painstaking, incremental progress between many groups working independently and collaboratively. Inevitably, there were missteps, misconceptions, and cumbersome and insensitive laboratory methods. But there was also a fortunate confluence of enormous energy, new insights, sharing of samples and results, and improved laboratory methods, which helped move the field forward.
42 Major results were replicated and extended, and concepts were refined and reformulated, contributing to filling out the picture of a major scientific and public health problem.
The identification of carcinogenic HPV genotypes in the mid-1980s spurred epidemiologists to undertake population-based studies to characterize the association of HPV infection with cervical cancer and precancer, and to determine whether HPV infection—the hypothesized cause—was consistent with the already identified risk factors, principally, age at first intercourse and number of lifetime sexual partners. The results of early epidemiologic studies were inconsistent, however, and suggested only a weak association of HPV exposure with disease. The uncertainty surrounding these studies reflected the facts that the definition of a precancerous lesion is somewhat subjective and that the different methods used to test for HPV seropositivity had variable sensitivity and accuracy.
43It took carefully designed studies carried out over more than a decade to improve the accuracy of these tests and establish reliable criteria for identifying women at high risk of developing cervical cancer. Researchers at the National Cancer Institute, led by Mark Schiffman, have been in the forefront of these efforts. As the criteria for defining high-risk cytology improved and the sensitivity of methods for detecting high-risk HPV DNA increased, the strength of the observed associations increased dramatically. The fact that this sharpening of measures of both exposure and disease outcome led to a strengthening of the association provided compelling support for a causal relationship.
Broadly speaking, three distinct types of studies contributed to elucidating major aspects of the HPV-cervical cancer relationship, and their results converged to provide overwhelming evidence that HPV was the obligatory cause of the disease. First, studies of the natural history of HPV infection, as described earlier, showed that persistent infection with high-risk types was necessary to cause cervical cancer. Second, large coordinated surveys documenting the prevalence of HPV infection with different HPV types in women with normal cytology were carried out in different countries throughout the world under the aegis of the International Agency for Research on Cancer.
44 These studies made it possible to compare the rates of infection in different regions with greatly differing socioeconomic conditions, health care systems, sexual mores, and rates of cervical cancer. Third, case-control and prospective studies conducted in different countries made it possible to gauge the magnitude of the association of HPV infection (any HPV infection, as well as infection with specific high-risk types) and the risk of cervical cancer. These studies took into account other risk factors for cervical cancer in addition to HPV infection, thereby enabling researchers to assess the relative importance of different risk factors. The results of these different types of studies complemented and buttressed one another and led to a three-dimensional and therefore much more convincing picture of the relationship of the virus to cervical cancer.
It was not until the early and mid-1990s that the results of large, population-based investigations and epidemiologic studies of HPV infection and cervical cancer began to appear. Under the auspices of IARC, an international survey of the prevalence of HPV infection among cervical cancer cases was carried out. Over a thousand frozen biopsy specimens were collected from twenty-two countries around the world. In each center, collaborators recruited fifty consecutive cases of invasive cervical cancer. A major strength of this study was that all specimens underwent centralized pathology review and centralized testing for HPV DNA in the tumor by PCR.
45 Initially, HPV DNA was detected in 93 percent of the specimens using PCR. However, when the 7 percent of specimens that initially tested negative were retested using a more sensitive technique, and when specimens were limited to those with clear evidence of malignancy, the prevalence of HPV DNA was 99.7 percent.
46 This provided strong evidence that HPV was a necessary cause of cervical cancer. In other words, in the absence of HPV infection there would be virtually no cervical cancer. As the McGill epidemiologist Eduardo Franco has pointed out, “This is the first instance in which a necessary cause has been demonstrated in cancer epidemiology.”
47In the 1980s and 1990s epidemiologic studies—both case-control and prospective studies—were carried out in different countries to determine the magnitude of the risk of developing cervical cancer associated with HPV exposure. In addition to information on infection with HPV, studies typically gathered information on other exposures that might be independent risk factors or might modify the effect of HPV exposure. By the mid-1990s the results of these studies had demonstrated clearly and consistently that, after other factors were taken into account, HPV exposure (defined as the presence of HPV DNA in tumor specimens) was associated with a dramatically increased risk. In fact, the risk estimates from these studies are the highest found in epidemiologic studies of any cancer. A woman with evidence of infection with any HPV type has a relative risk of developing cervical cancer ranging from 50 to 100 (i.e., a fifty- to one hundred-fold increased risk). Women with evidence of infection with HPV-16 and HPV-18 have relative risks ranging from 100 to 500. And in some studies, the risk estimates reach values of between 500 and 1,000.
48 For comparison, compared to someone who has never smoked, a heavy smoker may have a twenty- to fiftyfold increased risk of developing lung cancer, depending on how many cigarettes he or she usually smoked per day.
Work in the early 1990s seemed to point to a mechanism by which HPV-16 initiates the carcinogenic process. HPV DNA becomes integrated into the host cell’s genome—a process believed to be irreversible—leading to the inactivation of tumor suppressor genes and the immortalization of transformed cells.
49 Two critical viral proteins, known as E6 and E7, appear to interfere with cellular proteins involved in the regulation of the cell cycle, leading to uncontrolled cell growth and transformation to malignancy.
50The accumulation of scientific evidence from virology, molecular biology, and clinical and epidemiologic studies in many different countries provided strong support for a causal association of infection with specific HPV genotypes with risk of cervical cancer. This evidence can be summarized in terms of the “criteria for judging causal associations” discussed in
chapter 2. As noted above, the association of HPV infection with risk of developing cervical cancer is the strongest in the field of cancer epidemiology. The association is consistent across studies carried out in different populations. There is a dose-response relationship between markers of persistent infection and risk (i.e., risk increases dramatically with markers of viral persistence). Prospective studies of young women demonstrate that infection with HPV precedes the development of disease by several decades. Epidemiologic evidence coheres with molecular pathologic evidence. Finally, evidence from molecular biology that HPV DNA is integrated into the host genome and is carcinogenic provides a mechanism, thereby satisfying the criterion of biological plausibility.
51The demonstration that persistent infection with HPV is the necessary cause of cervical cancer informs the interpretation of other potential risk factors, such as smoking, oral contraceptive use, number of live births, diet, and infection with other transmissible agents. The contribution of these other cofactors is modest compared to that of infection with a high-risk HPV type, as reflected in relative risks of between two- and threefold. As the sensitivity of assays for HPV increased, the relative risk estimates for these cofactors decreased; hence the elevated risk estimates may partially reflect residual confounding, since most of these variables are markers of sexual activity. Any role of these factors must now be understood in the context of their ability to modify (either enhancing or inhibiting) the process of HPV-initiated carcinogenesis. As researchers at the National Cancer Institute wrote, “Because HPV infection is a necessary cause of cervical cancer worldwide, no other risk factors are important in the absence of HPV, a somewhat startling conclusion that greatly affects usual epidemiologic approaches to effect modification and confounding.”
52* * *
Although the cervix is the leading cancer site associated with HPV infection, the virus can also replicate and take hold in the cells that line the anogenital tissues generally and the throat of both sexes. Shortly after the identification of HPV-16 and 18 DNA in cervical cancer biopsies in 1983 and 1984, respectively, these HPV types, as well as several others, were found in other anogenital cancers.
53 Sites associated with HPV infection now include the vulva, vagina, penis, anus, and oropharynx. The proportion of cancer attributable to HPV is roughly 50 percent for vulvar cancer, 30–50 percent for penile cancers, and 60–90 percent for cancer of the vagina and anal and perianal cancers.
54 However, these cancers are extremely rare compared to cervical cancer.
A possible role of an infectious agent in the development of oropharyngeal cancer was suggested by a study in 1975 indicating that women with cervical cancer had a five- to sixfold increased risk of going on to develop oral cancer.
55 But it was not until 1985 that direct evidence established the presence of specific HPV types (principally HPV-16) in squamous cell carcinomas of the tongue and other subsites within the oropharynx. Between one-quarter and one-third of these cancers are now thought to be caused by anogenital high-risk HPV infections. In the past, most oropharyngeal cancers, like all head and neck cancers, had been strongly associated with two “traditional” risk factors, namely, smoking and heavy alcohol consumption. In spite of recent decreases in rates of other head and neck cancers, however, incidence rates of oropharyngeal cancer have increased dramatically since the 1980s, particularly in heterosexual, middle-aged men, mostly nonsmokers and nondrinkers. This change in the pattern of occurrence of oral cancer has been ascribed to changes in sexual behavior, including an increased number of sexual partners and an increase in oral-genital sexual practices predominantly among younger people starting in the 1970s.
Oropharyngeal cancers associated with HPV and those associated with smoking and drinking differ in a number of ways. HPV-associated oropharyngeal cancers tend to occur at the base of the tongue, in the tonsils, and in the back of the throat, whereas smoking- and alcohol-associated oral cancer has a wider distribution within the oral cavity. In addition, the prognosis of HPV-positive oropharyngeal cancer is much more favorable than that of the HPV-negative type—a difference that may partly be due to the absence of mutations in p53, a major tumor suppressor gene, which are a hallmark of the traditional oropharyngeal cancer. These differences, together with the contrasting profiles of those who develop the two types of oral cancer, have led researchers to conclude that HPV-negative and HPV-positive oral cancers represent two distinct diseases.
56Whereas the incidence of cervical cancer declined markedly in the United States from 1985 to 2005 and is expected to decline further by 2025, the incidence of oropharyngeal cancer has increased, and, strikingly, the proportion of these squamous cell carcinomas attributable to HPV infection has surged—from an estimated 16 percent in 1985 to an estimated 72 percent in 2005. And the proportion is projected to reach 90 percent by 2025.
57The role of HPV infection as a novel risk factor for oropharyngeal carcinoma received widespread media attention in July 2013, when the actor Michael Douglas announced that his throat cancer was caused by his having engaged in oral sex.
58 However, there is little public awareness of HPV infection as a risk factor for oral cancer. And, in contrast to the use of Pap testing for lesions of the cervix, no premalignant lesion has been identified for HPV-induced oropharyngeal cancer that could be used for screening.
Based on combining the number of cases at all anatomic sites, the total number of cancers that are associated with HPV infection in the United States in a given year is approximately thirty-one thousand.
59* * *
In the first half of the twentieth century, cervical cancer was one of the most common cancers among women and a leading cause of cancer death among women in the United States. With the incorporation of Papanicolaou testing into gynecologic practice starting in the 1950s, cervical cancer mortality in the United States has declined by roughly 75 percent, and incidence has declined by half during the same period. The American Cancer Society estimated that there would be 12,900 new cases in 2015 and 4,100 deaths due to cervical cancer. Today cervical cancer now ranks fourteenth among female cancers in terms of incidence. Similar declines have taken place in other advanced industrial societies, and the widespread use of Pap testing has been credited with averting an epidemic of cervical cancer in the United Kingdom over the past several decades.
60Although Pap screening is credited with preventing many deaths from cervical cancer, it has serious limitations. Evaluation of Pap smears under the microscope is somewhat subjective, and it can miss precancerous lesions. Follow-up and treatment of women with ambiguous cytology findings entails an enormous burden on the health care system and on women. Because Pap testing has a substantial false positive rate, many women undergo colposcopy (examination of the cervix with a magnifying device) and repeated testing when, in fact, their risk of developing cervical cancer is low. The annual cost of these procedures has been estimated at four billion dollars. Finally, Pap testing cannot detect adenocarcinoma of the cervix, which for this reason has been increasing in incidence, while the more common squamous cell carcinoma has declined. Adenocarcinoma is caused principally by HPV-18.
In a landmark decision, in March 2014 an FDA panel unanimously recommended that the Pap test be replaced with HPV-DNA testing. The new test detects HPV-16 and HPV-18, which account for 70 percent of cervical cancer cases. Testing for high-risk HPV DNA in cervical tissues represents an enormous advance in the ability to identify women who are truly at high risk and to reduce overtreatment of women at low risk. HPV DNA testing makes it possible to classify women very finely as to their risk. The test may reveal that one woman has a 60 percent chance of developing in situ cancer of the cervix within five years, whereas another woman has virtually no chance of developing cancer.
61An added benefit is that, owing to the greater accuracy of HPV DNA testing, women who test negative for high-risk types will be able to go without screening for three to five years. This is referred to as the test’s “negative predictive value.” According to Burk, “That’s a really big deal. The negative predictive value of a negative HPV DNA test is phenomenal.” To Burk, who was a member of the FDA panel, the decision represents “a historic moment.” “So you have the vaccine and you have the transformation of evidence-based medicine, where we really can now put into practice what we’ve learned about the epidemiology of HPV.”
62* * *
By the early 1990s new insight into the natural history and epidemiology of HPV infection with specific high-risk genotypes opened up the possibility of developing a vaccine against HPV infection that would prevent cervical cancer. A highly effective vaccine against hepatitis B virus (HBV), a major cause of primary liver cancer in Asia and southern Africa, had been in use since the 1980s.
Soon after identifying HPV-16 and 18 in the mid-1980s, zur Hausen had tried to interest the pharmaceutical industry in the prospect of developing a vaccine against HPV infection. But his overtures met with little interest. Then, in the early 1990s, researchers in Australia and at the National Institutes of Health perfected the synthesis of empty virus-like particles (VLPs) from the HPV-16 protein shell, which triggers immunity to the virus.
63 Shortly thereafter, vaccine trials demonstrated that VLPs for species-specific papillomavirus prevented infections and tumors in cows, rabbits, and dogs. This work prompted pharmaceutical companies to pursue the technology to develop HPV vaccines.
64By 2006 the FDA had approved two vaccines for prevention of HPV infection. Merck’s Gardasil is a quadrivalent vaccine, which targets HPV-16 and 18 as well as HPV-6 and 11, the most common types causing genital warts. Glaxo-Smith-Kline’s Cervarix is a bivalent vaccine targeting HPV-16 and 18. Since it would take decades to determine whether the vaccines protected against cervical cancer, the FDA and other agencies decided to judge the efficacy of the vaccine based on how effective vaccination was in preventing precancerous cervical lesions. Both vaccines have been shown to be highly efficacious, conferring high levels of protection (greater than 90 percent) against persistent infection with HPV-16 and 18. Based on the studies carried out to date, this protection lasts undiminished for nearly a decade, and studies are currently in progress to assess protection afforded by the vaccines through at least fourteen years.
65 For the vaccine to be effective, it must be administered before the onset of sexual activity. For this reason, the Centers for Disease Control and Prevention (CDC) recommends that all girls be given either vaccine at 11 or 12 years of age and that boys be vaccinated with Gardasil at 11 or 12 years old. Both vaccines require three doses. So far, however, the response to the FDA recommendation has been disappointing. In spite of the vaccine’s safety and efficacy, according to the CDC, as of 2014 only 40 percent of girls and only 22 percent of boys had received all three doses of the vaccine.
66 The much poorer compliance with HPV vaccination compared to other childhood vaccinations appears to be largely due to parents’ and physicians’ reluctance to confront the topic of sex, even though behavioral research demonstrates that teens who receive the HPV vaccine are no more likely to engage in casual or unsafe sex than those who do not.
67The most important outstanding question regarding HPV immunization is whether the vaccine will provide lifelong immunity or whether one or more booster shots will be required later in life. A second question is whether vaccination protects against infection at other sites, such as the oropharynx. In addition, for use in developing countries there is an urgent need for alternatives to the current vaccines, which are type-specific and expensive and require cold chain transportation.
Since HPV-16 and 18 infections account for approximately 70 percent of all cervical cancers, second-generation vaccines are being developed that would cover over 90 percent of the cancer-causing HPV genotypes, including, in addition to HPV-16 and 18, HPV-45, 31, 33, 52, 35, 58, 39 (highly prevalent in Latin America), and 51 (highly prevalent in Africa). The results of a clinical trial comparing a new “nonavalent” vaccine (i.e., targeting nine different HPV types) to the quadrivalent vaccine have demonstrated protection against five additional HPV types.
68* * *
Although its incidence has been dramatically reduced in developed countries due to Pap testing, cervical cancer is the third most common cancer among women worldwide and causes the largest number of cancer-related deaths among women in developing countries. According to estimates from IARC’s GLOBOCAN program, each year there are 530,000 new cases of cervical cancer worldwide and 275,000 deaths from the disease.
69 The vast majority of the burden of cervical cancer—more than 85 percent of new cases and 88 percent of deaths—occurs in the developing world, giving it the most inequitable burden of any cancer. Without changes in prevention and control, due solely to population growth and aging of the population, cervical cancer deaths are projected to reach 430,000 annually by 2030, virtually all in developing countries.
Mortality from cervical cancer varies widely internationally (
fig. 7.2). Age-standardized mortality rates are highest in East and West Africa, intermediate in Southern Africa, South-Central Asia, South America, and Central Africa, and lowest in West Asia, North America, and Australia/New Zealand. There is a tenfold difference between the rates in the highest versus the lowest mortality regions of the world. Although the highest rates are seen in East Africa, it is India, with an intermediate mortality rate and a population of 1.3 billion, that had the largest number of deaths from cervical cancer (72,824) and accounts for roughly a quarter of the worldwide burden of cervical cancer.
70 The death rate from cervical cancer in less developed countries is on average roughly three times that in more developed countries. In the developing world cervical cancer kills women in their prime who are often the sole support of young children. Thus it exacts an enormous toll in terms of premature death, years of life lost, and family and societal impacts.
71
The substantial variation in cervical cancer rates largely reflects the pattern of screening availability in different parts of the world. The low mortality areas of the map in
figure 7.2 are those in which Pap screening is available. If one had drawn such a map a century ago, before the introduction of Pap testing, mortality rates from cervical cancer would be much more uniform throughout the world, reflecting the universal distribution of HPV.
The global prevalence of HPV infection in the cervix in women with normal cytology at any given point in time is about 10 percent. As shown in
figure 7.3, HPV prevalence does vary by region, although not as much as incidence or, especially, mortality from cervical cancer. Rates are generally higher in low-income countries compared to those in more developed regions. Women in Africa, and in particular in East Africa, have the highest HPV prevalence rates (32 percent), while the lowest estimates are seen in Southeast Asia (6 percent). These differences in the prevalence of HPV infection in the general population likely reflect cultural norms affecting sexual behavior, such as age at first marriage, marriage to older men or to men who have several contemporaneous partners, and poor hygiene.
72
The recognition that infection with high-risk HPV types is the necessary cause of virtually all cervical cancers has created an unprecedented opportunity—and implicitly, an ethical obligation—with regard to a major cause of cancer deaths in the developing world. That is to say, through a combination of screening and vaccination, there is the potential to drastically reduce the number of deaths from this disease. With the exception of liver cancer, this cannot be said of any other major cancer, such as those of the breast, colon, lung, prostate, endometrium, ovary, or leukemia or lymphoma.
* * *
As a result of painstaking work over the past thirty years, the strategies and tools available for reducing morbidity and mortality from cervical cancer have been radically transformed.
73 Depending on the target age-group, one of two strategies is available. These are referred to as “primary prevention” and “secondary prevention.”
Primary prevention refers to preventing infection with high-risk HPV genotypes in preteens—before exposure to the virus. In theory, currently available prophylactic vaccines could prevent approximately 70 percent of cervical cancer in the future in girls vaccinated before the age of 12.
Secondary prevention refers to screening of women who are already sexually active to identify and surgically remove precancerous lesions or cancers before they become life-threatening. Novel methods, referred to as “screen and treat,” have been developed for use in low-resource countries with limited health care infrastructure. However, implementing these strategies in the places where they are most needed—in developing countries like Uganda and India—is anything but simple, and realizing the promise of HPV prevention research will require overcoming formidable obstacles.
These obstacles pervade all sectors of low-income societies, from the level of the household to that of government policy and international aid. The most immediate problems are a lack of health services and qualified staff, but at another level endemic corruption, government bureaucracy, and weak rule of law stand in the way of adopting the needed programs.
74 Furthermore, while preventing deaths from cervical cancer is an urgent need, developing countries face many other urgent needs competing for attention and scarce resources, including malnutrition, high infant mortality and maternal childbirth deaths, and lack of clean water and basic sanitation. At the same time, the incidence of chronic diseases, such as cancer, diabetes, and cardiovascular disease, is increasing owing to changes in society and lifestyle, principally, increasing tobacco use and obesity. Moreover, in some countries a disease that affects women only may not command as much support as other health problems. Beyond the question of material resources, attitudes toward vaccination can pose a serious obstacle in regions where there is mistrust among groups that may view government efforts to vaccinate young girls as part of a birth control program or, worse, as attempts to spread AIDS.
75In spite of these considerable obstacles, international agencies, nongovernmental organizations, pharmaceutical companies, donors, and ministries of health are working together to address the prevention of cervical cancer. A variety of demonstration projects are underway in developing countries to determine the most effective way to deliver the vaccine to preadolescent girls. In 2012 Uganda’s Ministry of Health, in collaboration with Merck, Sharpe, and Dohm, launched a program to vaccinate 140,000 girls aged 9–12 with Gardasil. The program included twelve out of one hundred districts in the country. The GAVI Alliance, a public-private partnership including international agencies, pharmaceutical companies, donors, and governments, is supporting the introduction of HPV vaccination demonstration programs targeting 180,000 girls in eight developing countries, mainly in sub-Saharan Africa. These programs will provide valuable experience and preliminary data for the design of effective national vaccination programs. By 2015 GAVI planned to extend its pilot projects to reach approximately one million girls in twenty countries; and by 2020 the goal is to have vaccinated more than thirty million girls in over forty countries. In countries where the average per capita income is less than two dollars a day, cost is a serious issue, and GAVI is making the vaccine available in poor countries for $4.50 per dose.
76If these programs deliver on their promise, they will succeed in reaching and vaccinating preteen girls and will then be expanded to cover the entire populations of these countries. But sustained funding will be needed to support vaccination beyond the first wave as a permanent component of health care services. Once primary prevention through vaccination is assured, there will still be a vital need for screening programs for the female population above the age of thirteen who are still at risk of developing cervical cancer. Screening will also be essential to monitor the effectiveness of the vaccine as well as to detect cancers caused by types other than HPV-16 and HPV-18, which are included in currently available vaccines.
77One observer summed up what is at stake in developing countries as follows: “With the availability of an effective, safe vaccine, there is real hope for reducing the global burden of cervical cancer. Although achieving broad coverage of young adolescents, negotiated tiered pricing, and securing financing will be challenging, it is sobering to realize that with every 5-year delay in bringing vaccination to developing countries, 1.5 million to 2 million more women will die.”
78Given what has been learned in the past thirty years, we are in the unprecedented situation of having the ability to virtually eradicate this type of cancer. All that stands in the way are practical issues of resources and strategies and the political leadership to realize the potential solution. This would be the first time in history that a type of cancer was eradicated. In speaking of the potential for eradicating cervical cancer, Burk pointed to the global eradication of smallpox in 1977, which was the culmination of a campaign by the World Health Organization, and lamented the lack of drive and energy to implement the required program. “We have the means to eliminate cervical cancer, if we had the energy that he had [D.A. Henderson—who led the smallpox eradication campaign]. Of course, smallpox is a different kind of disease—it is more immediate—but still, we have the means to do that now.”
79The fact that HPV types have changed little over the past 200,000 years suggests that cervical cancer has been around at least since Homo sapiens diverged from other hominids. Yet only in the past thirty years has HPV been demonstrated to cause cancer in humans. The finding that infection with high-risk HPV types is estimated to cause 5 percent of all cancers worldwide underscores the momentousness of the present juncture.
* * *
If the HPV story ended here, it would qualify as one of the great achievements in cancer research, epidemiology, and global public health. But there is more to the story. The effort to identify high-risk carcinogenic HPV types and to understand how they induce cancer has led to a comprehensive cataloguing of the genetic differences both between and within viral types at a minute level. And an appreciation of the tremendous genetic variation within the Papillomaviradae family is yielding insights in several disparate fields. First, as the number of HPV types has grown, there is an increasing appreciation for the high degree of specialization and adaptation of the virus to its animal and human hosts. Second, a key question is why a minority of forms of the virus but not others are highly carcinogenic. Answering this question will contribute to understanding the carcinogenic process. Finally, comparison of HPV types and variants from cervical smears from different populations throughout the world has shown potential to shed new light on human evolution and human migrations.
Since zur Hausen isolated the first HPV types in the 1980s, the number of HPV types has steadily grown. At present 170 human and 20 animal papillomavirus genotypes have been fully characterized, and it is expected that more human and, particularly, animal genotypes will be identified in the future. Zur Hausen’s wife Ethel-Michelle de Villiers maintains a reference center in Heidelberg, Germany, for the confirmation and cataloguing of new HPV types. A distinct genotype, or species, of papillomavirus is one that differs from other identified types by more than 10 percent in a key segment of the HPV genome. Each of the more than one hundred distinct types has been isolated either from abnormal growths in different tissues (skin, epithelial tissues of the cervix, vagina, vulva, anus, penis, or the oropharynx), including warts, precancer, or invasive cancer, or from normal tissues.
De Villiers has constructed a phylogenetic tree depicting known HPV types in terms of the degree of relatedness of different genotypes (
fig. 7.4). The different types fall into three major groupings, or “genera,” denoted by Greek letters alpha, beta, and gamma. The alpha types tend to colonize the mucous membranes, whereas the beta and gamma types are adapted more to the skin. All the types found today, which are denoted by the extremities of the branches, evolved from a common ancestor that existed hundreds of millions of years ago. Unlike the family trees in which time is represented along the horizontal or vertical axis, here as we move inward toward the center, where the “trunks” of the three major genera meet, we are moving back in time. The presumed common ancestor of all HPV types is located in the center of the diagram.
Looking at another phylogenetic tree, this one limited to HPV types in the alpha papillomavirus genus—which includes HPV-16—Burk stressed that the key thing is that every one of the types that causes cancer has a common origin. However, not all the viruses that have evolved from this common ancestor cause cancer. This suggests that the ability to adapt to an ecologic niche that causes cancer has been lost in some of the types. For example, the alpha 6 grouping, which is associated with genital warts, represents a completely different ecological niche from the alpha 9 and 7 groupings, which cause cancer. And other groupings are more or less benign. Most alpha HPV types coexist with the host and don’t cause any pathology.
Burk told me that his real interest is in understanding the genetic basis of cervical cancer:
The key observation is that HPV causes cervical cancer as collateral damage, not as part of its natural life cycle. In fact, it’s the ones that cause cancer that have evolved into a certain ecological niche, where their survival is a little better and, unfortunately, there’s an overlap between those traits and the dysregulation of a particular cell population—that is, cells of the squamo-columnar junction of the female cervix. The female has these specific cells at the squamo-columnar junction that the virus has adapted to—only certain HPV types—and, unfortunately, as part of their evolution and adaptation, they also cause cervical cancer.
80Differences in the genetic sequence of HPV types are believed to determine two independent traits, which are necessary for the virus to cause cancer—the ability to persist (i.e., to maintain infection) and to progress on the path to cancer. According to Burk, something about HPV-16 makes it “uniquely likely both to persist and to cause neoplastic progression when it persisted, making it a remarkably powerful human carcinogen.” Other carcinogenic types, many related to HPV-16, were not particularly persistent but could cause neoplastic progression, at lower rates than HPV-16 if they did persist.
The “remarkable” pattern of differences in natural history between the types is prompting more detailed investigations, whose goal is to figure out what genetic variant—what specific piece of the HPV genome—makes it so carcinogenic. What is the difference between HPV-16 and its closest sister types, HPV-31 and HPV-35, that accounts for the former’s much greater carcinogenic potential? Understanding this would provide an unprecedented insight into the mechanism of HPV carcinogenesis.
Using “whole genome sequencing,” researchers are now in a position to examine the entire HPV genome, which is small enough to permit a comprehensive analysis of all its components and functions on a population level. This is the goal of a project that Burk and his colleagues have called the HPV Human Genome Project, which includes researchers at Einstein, the National Cancer Institute, IARC, and BGI, a high-powered genome sequencing company in Shenzhen, China. This approach has the potential to reveal genetic differences in the papillomavirus that, depending on its interaction with its human host, can have very large effects. For example, the different variant lineages of HPV-16 have associations with risk of cervical cancer on the order of sixfold. However, comparing HPV-16 to related types in the alpha-9 species group yields huge differences—perhaps fiftyfold. Given this powerful phenotypic difference (that is, whether one develops cervical precancer or cancer or not, depending on the viral genome), these new genome-phenotype correlation studies have a much better chance of yielding results than the human genomics studies have.
As Burk put it:
The key to understanding the molecular basis of HPV carcinogenicity is realizing that the biological driving force has been the evolution of specific HPV’s into discrete host ecosystems, such as the epithelium from the cervix, vagina, external genital skin, or skin covering other anatomic surfaces. Each bodily ecosystem has characteristics that allow adapted HPVs some type of competitive advantage to infect, replicate, and transmit. Nevertheless… HPV-16 stands out as having the most pathogenic phenotype (e.g., HPV-16 causes both cervix and oropharyngeal cancer).
81* * *
The concerted effort to understand the human papillomavirus has been motivated by the virus’s ability to induce cancer. But, as an unanticipated by-product of medically motivated research, studies of the variation in the virus have yielded insights in a totally different area. Because the papillomavirus phylogenetic tree includes genomes isolated from cervical smears from all over the world, this molecular variation simultaneously reflects geographic variation and differences in population groups.
The existence of stored samples of cervical smears from populations around the world has enabled researchers to use fine variation in the HPV genome as a means of reconstructing prehistoric viral spread and the movement of ancient populations. This is possible due to the tremendous diversity of HPV types and the fact that HPV variants show the greatest divergence when they are obtained from ethnic groups that evolved for a long time without contact, such as Africans and American Indians.
82 The finer (and more recent) variation within types and the greater variation distinguishing types can be used as an evolutionary clock to trace HPV speciation going back millions of years.
Using this logic, Hans-Ulrich Bernard and colleagues at the University of California at Irvine examined worldwide variation within HPV-18 in samples obtained from population groups in different parts of the world. They concluded that diversity within the HPV-18 genome correlates with patterns of evolution and the spread of
Homo sapiens out of Africa. HPV-18 variants from Amazonian Indians were most closely related to those from Japanese and Chinese patients, in conformity with the posited dating of the migration of Asian peoples across the Bering Strait and down into the Americas approximately twelve thousand years ago. Bernard and colleagues speculated that the split between two closely related genotypes HPV-18 and HPV-45 occurred more than half a million years ago, and that speciation events between less closely related viral genotypes may have occurred “several million years ago, i.e., before the evolution of humans.”
83Bernard and colleagues have carried out similar studies of HPV-16. The strongest pattern among HPV-16 variants pointed to the independent evolution of the virus among Africans, Caucasians, and East Asians and reflected colonization of the Americas by Europeans and Africans. As in the analysis of HPV-18, HPV-16 appears to have evolved over more than 200,000 years from a precursor genome that may have originated in Africa. The authors concluded that “the identification of molecular variants is a powerful epidemiological and phylogenetic tool for revealing the ancient spread of papillomaviruses, whose trace through the world has not yet been completely lost.”
84The prevalence of different HPV types and variants within types in different populations worldwide reflects the history of the virus-host relationship at the most intimate level, which, in turn, is the outcome of innumerable interactions over time between different groups, including conquest, intermarriage, and migration. For genital HPVs, the fundamental interaction is the sexual encounter between groups carrying different HPV types. In the stark picture painted by Burk, the fundamental interaction comes down to an essential feature of human behavior:
If you imagine the history of man, people are nasty, they go around and rape one group, and if the virus is eliminated before they found the next person, it would be gone—there would be no more papillomavirus. In another instance, one tribe will go and rape all the women in another tribe and just kill everybody, so their victims’ genes wouldn’t be perpetuated, but their HPV would be! So those types that we see today have persisted over time and allowed replication. Because it’s evolution—whatever replicates wins.
85It turns out that the project of mapping the genome of the virus and understanding its genetic variation is inextricably intertwined with understanding the geographic distribution of different HPV types and variants and their evolution in tandem with the migrations of human and hominid populations, as well as the distribution of associated disease.
* * *
Looking back over the past three decades, in 2009 Burk summarized the fortunate “confluence” of a number of different factors that made possible the tremendous progress in the field encompassing basic research on HPV and its clinical implications. First, advances in technology (recombinant DNA, cloning of HPV genomes, and the use of molecular hybridization) represented a quantum advance over the standard virologic methods (i.e., serology), and these new methods were used in epidemiologic studies of disease. Second, the free and widespread distribution of cloned HPV genomes by the Heidelberg group and the inauguration of an annual international papillomavirus conference “fostered a collaborative culture within the PV community.” “From a public health viewpoint, HPV has become the model for molecular medicine and how technology can be readily applied to global health problems.”
86Others in the field of cancer epidemiology have argued that the HPV story provides a model for how basic science findings can be applied to real-world problems and make an enormous difference in reducing morbidity and mortality. Rather than a narrow focus on etiology, epidemiologists are encouraged to adopt a broader ecologic model of population health and to tackle issues of health care and survivorship.
87* * *
The enormous distance traversed and the astonishing progress over the past fifty years in understanding the role of viruses in the development of cancer and in the prevention and control of fatal cancers is highlighted by a story from Burkitt’s later life. In 1964 Burkitt resigned his post at Mulago Hospital in Kampala and two years later returned to England to work for the Medical Research Council in London. After having spent eighteen years in Uganda, he was left “at sea” by the move back to England and the lack of clinical involvement with patients, and he had no idea whether he would find any sort of occupation. As happened in his early career, however, another totally unanticipated opportunity was placed in his path. The eminent epidemiologist Sir Richard Doll introduced him to Peter Cleave, and he was galvanized by Cleave’s idea that many of the diseases of Western civilization could be ascribed to diet, and specifically to a high consumption of refined carbohydrates and a lack of dietary fiber.
88 This new interest was to preoccupy him for the rest of his life. In 1968, when Clifford Nelson, one of his two companions on the tumor safari, visited him and asked him about the latest on Burkitt’s lymphoma, he replied, “Cliff, it’s all out of my hands now. All the really clever chaps in epidemiology, virology, immunology, and biochemistry have left me in the dust.”
89 This was only ten years after the publication of his initial paper on Burkitt’s lymphoma. Burkitt died in 1993. One can only guess what he would have made of the decades of research on HPV—with its impressive advances—that, in an important way, grew out of his groundbreaking work on childhood lymphoma in Africa.