Preventive Medicine 76 (2015) 129–131
Contents lists available at ScienceDirect
Preventive Medicine journal homepage: www.elsevier.com/locate/ypmed
Editorial
Sobering realizations in cancer prevention and screening and their lessons
The year in the world of cancer prevention began on a gloomy note with the publication of a paper showing that risk of cancer among different anatomical sites is strongly correlated with the lifetime number of stem cell divisions in the underlying tissues (Tomasetti and Vogelstein, 2015). Tomasetti and Vogelstein elegantly elucidated long-standing enigmas such as, why cancer is rare in certain organs (e.g., heart) or disproportionately more common in the large rather than small intestine, despite the fact that the latter is much longer than the former. The most prominent message from their paper (Tomasetti and Vogelstein, 2015), however, was in interpreting the coefficient of correlation for their analysis (0.804) as indicating that some 65% (i.e., 0.804 squared and expressed as a proportion) of the variation in cancer risk among tissues could be explained by the number of stem cell divisions. This left only some 35% of the variation to be attributable to exposure circumstances that we could potentially control or modify via preventive strategies. Sadly, non-modifiable risk determinants, such as low-penetrance genetic mutations and endocrine hormonal influences, are part of the 35%, thus leaving very little for public health practitioners to set targets for cancer prevention and for epidemiologists to make bold claims of cancer causes yet to be discovered. Having appeared in the journal Science and owing to the authors' solid reputation in the cancer research community, Tomasetti and Vogelstein's paper received unparalleled attention in the lay and professional media. Yet, their findings and methods came into dispute, most notably in a series of letters to the editor of Science that was followed by their response (Sills, 2015). Although Tomasetti and Vogelstein made many simplifying generalizations and assumptions their primary finding remains, i.e., that replicative mutations, accumulating over a lifetime, explain a substantial proportion of the variation in cancer incidence among organ sites. The proverbial devil, however, is in the details. That the variation in risk within a country and among anatomical sites is explained by these mutations does not imply that there cannot be further variation among individuals or populations. Cancer is a group of diseases of multi-factorial etiology and long latency. For instance, today's incidence of lung cancer in the US reflects the prevalence of tobacco smoking and occupational risk factors as of some 20–30 years ago in that country, which were already declining. Further reductions in lung cancer incidence in the future will be dependent upon actions to curb smoking and other risk factors today. The same can be said for all other cancers with modifiable risk factors that are clearly identified. Calculating the proportion of cancers that are attributable to specific external causes is an extremely important goal in public health and must take into account the added complexity of latency (i.e., the past gave us today's cancer burden but it is the present dynamics of risk factors that will shape future burden). With empirically valid estimates of population attributable risk (PAR), pairing specific risk factors to their corresponding cancer, policymakers can have targets for prevention and
http://dx.doi.org/10.1016/j.ypmed.2015.04.014 0091-7435/© 2015 Elsevier Inc. All rights reserved.
benchmarks for measuring success in deploying interventions. In this issue of Preventive Medicine, King et al. (2015-in this issue) interpret the message from the Tomasetti and Vogelstein paper (Tomasetti and Vogelstein, 2015) in the context of estimating PARs for a country as large and diverse as Canada. King et al. explain the public health value of PARs and in taking advantage of the granularity in knowledge that comes from estimating disease- and cause-specific PARs via comprehensive literature reviews and country- or region-specific prevalence data on the exposures of interest (King et al., 2015-in this issue). Elimination or inhibition of risk factors over time is a laudable goal for society. At a minimum, this can be accomplished via robust tobacco control policies, public education regarding healthy diets and physical activity, vaccination to prevent infections with cancer-causing viruses, such as human papillomavirus (HPV) and hepatitis B virus, and reduction of environmental and occupational exposures to chemical and physical agents that are known carcinogens. We can perhaps accomplish further gains via chemoprevention, such as tamoxifen or aromatase inhibitors for reducing risk of breast cancer (Sestak and Cuzick, 2015), or daily aspirin for preventing mortality from several cancers (Cuzick et al., 2015). If in combination these measures can be successfully implemented in Canada, a country that places cancer prevention high in the public health agenda, we would be able to verify if the unexplained (regression) variation in Tomasetti and Vogelstein's correlation analysis (Tomasetti and Vogelstein, 2015) would be diminished, thus leaving only the influence of replicative mutations to explain cancer risk. Although this may be useful as a thought experiment intended to validate their argument, attaining complete elimination of a cancer ‘exposome’ is unrealistic. Many people will continue to smoke, eat poorly, and eschew exercising. There will always be noise in the correlation but reducing the impact of modifiable risk factors should lead to an improvement in the goodness of fit of Tomasetti and Vogelstein's correlation. Still on the topic of sobering realizations in cancer control, we now turn our attention to the limits of what can be achieved by cancer screening. Much like the newfound modesty in cancer prevention targets communicated by the aforementioned January Science paper (Tomasetti and Vogelstein, 2015), guidelines in cancer screening are frequently revised to reflect the uncertainty of the evidence. It used to be simple to grade the evidence for or against a given cancer screening technology. There were precious few randomized controlled trials of screening techniques, and their efficacy in reducing cause-specific mortality was the primary driver for country-wide policy decisions or clinical guideline recommendations. Finding evidence that the screening intervention merely leads to improved detection of precancerous lesions or early cancers, as it happens with many technologies proposed in the last decades for screening for cervical, breast, colorectal, ovarian, lung, and prostate cancers, is not enough. To be sure, accuracy in lesion detection is important, but we must be able to show also
130
Editorial
that ablation or excision of the lesion and its surrounding tissue will subsequently lead to cure or an improvement in survival that will translate into a reduced mortality from the cancer in question. While it is hard to argue with the cogency of this definition, we also have to take into account that the mortality reduction benefit must outweigh the potential harms that come from screening. Employing any given screening technology in a large enough group of individuals will inevitably lead to untoward clinical outcomes. These can originate from unnecessary procedures done on people erroneously presumed to have the disease because of a false-positive test result. They can also result from local damage to tissue (e.g., intestinal perforation in colorectal cancer screening) during a follow-up diagnostic procedure on a screen-positive individual, irrespective of whether the procedure was based on best practices or not. Harm can also happen in the form of late sequelae of a successful screening intervention (e.g., miscarriages or premature births following cervical lesion treatment). In the past 15 years, health technology assessment (HTA) has become eclectic enough to incorporate the entire evidence base concerning benefits and harms when passing judgment regarding cancer screening technologies and their practice recommendations (Franco, 2012). The latter is not a trivial part of the judgment, i.e., at which ages one should start or stop screening, and how frequently one should be screened within that span of life when the benefit-toharm ratio is considered maximal. HTA agencies or task forces, such as the US National Institutes of Health's Physician Desk Query program or the US and Canadian Preventive Services Task Forces, employ indepth systematic reviews of the literature, health economic modeling, and expert panels to arrive at consensual and graded recommendations. Expectedly, these recommendations end up influencing professional societies when issuing their own clinical guidelines. Not infrequently, recommendations become quite controversial, such as the US Preventive Services Task Force's assignment of a grade D recommendation (i.e., against) regarding prostate-specific antigen (PSA) based screening for prostate cancer (Moyer, 2012). The public and healthcare providers are beginning to realize that the mantra of ‘early detection saves lives’ is not as simple an argument as it used to be. The bewildering complexity of multi-disciplinary concerns that come into play for HTA recommendations is the new framework for cancer screening. Notwithstanding the above new world of increased complexity for screening assessment, we knew at least that one classic criterion for deciding on screening stood tall, i.e., that the disease should be an important public health problem (Wilson and Jungner, 1968). Although there are no benchmarks for deciding what is important enough, investigator-initiated studies on candidate screening technologies have traditionally justified their pursuits by stating that the screeningtargeted disease was appreciably common and/or had poor prognosis. Defining the importance of a given cancer as a public health problem is era-dependent. The onset of cervical cancer screening in much of the Western world happened at a time (the 1960s) when incidence rates, albeit declining, were high enough (above 20 per 100,000 women per year) and treatment outcomes were dismal enough (about 50% survival at 5 years) for this disease to be considered an important public health problem. Most conveniently, however, there was a simple screening test (Pap cytology), and an easily accessible anatomical site (the uterine cervix). Cervical cancer screening with the Pap test thus became the paradigm against which all other cancer screening interventions had to be judged. The burden of cervical cancer in Western countries, such as Canada and the US, is much less today than it was in the 1960s. Prostate, colorectal, lung, and breast cancers have become far more common relative to cervical cancer. Yet, as societies, we do not screen for these cancers with the same unequivocal conviction that we screen for cervical cancer. A compelling line of reasoning for gauging the importance of the cancer burden that justifies screening was advanced in this issue of Preventive Medicine by Whitham and Kulasingam (2015-in this issue). These authors used data from the Surveillance, Epidemiology, and End
Results (SEER) program maintained by the US National Cancer Institute to compare the risk of breast, colon, and cervical cancer at and after the recommended ages for screening. Presenting the risk of each cancer relative to the others in this way is very insightful and (to our knowledge) has not been done before. The authors' contention hinges on the fact that the risk of cervical cancer is currently much lower than that of breast and colon cancer across the age span, but particularly at the ages when guidelines define risk to be unacceptably high to justify continued screening. Whitham and Kulasingam found that based on precancerous lesions alone (which is where the bulk of mortality risk reduction lies), US guidelines recommend to start screening for colorectal adenomas at an age (50 years) when these lesions are 45 times more common than high-grade cervical intraepithelial neoplasia at the age (21 years) US providers judge appropriate to initiate cervical cancer screening. In other words, there is a disproportionately high tolerance of risk for colorectal cancer than is practiced for cervical cancer. The authors also demonstrate that the same asymmetry of risk tolerance exists when considering someone old enough to exit screening. Breast and colorectal cancers are more than 10 times more common than cervical cancer at the prescribed ages to exit screening (Whitham and Kulasingam, 2015-in this issue). With HPV vaccination expected to further reduce the incidence of cervical pre-invasive lesions and cancer, the high intensity with which US practitioners screen for cervical cancer will become a glaring anomaly in preventive medicine when judged against the backdrop of conservatism that prevails for colorectal and breast cancer screening. Is it surprising that we are far less tolerant of risk for cervical cancer than we are for cancers that are much more common than that of the uterine cervix? Why is it that we do far more (and have done for so long) for cervical cancer than for other screening-detectable cancers? First, we have had Pap tests since the 1940s, a simple technique that requires relatively little technology and no complex equipment or procedures. Second, the cervix is far more accessible via direct visualization than the breast or the colon/rectum. Third, the decision context for the onset of wide-scale cervical cancer screening was at a time (late 50s, early 60s) when this disease was as much of a killer of women as are breast and colorectal cancers. Fourth, Pap tests are recognized as an important acquired right of women and a key component of well-women care by providers. Fifth, entire subspecialties of medicine were created and come to depend upon a continued flow of patients in cervical cancer screening. Attempts to scale down screening in the US have been met with resistance by women and providers. With the escalating costs of healthcare delivery in Western countries, the lucidly explained framework by Whitham and Kulasingam (2015-in this issue) should form the basis for defining epidemiologic benchmarks for risk thresholds to begin and end screening, and for the intensity of screening during that time span. Clearly, the evidence that motivates these decisions must be bolstered by strong science on the natural history of cancers that can be prevented by screening, such as cervical and colorectal cancers, or those whose mortality can be reduced via screening (cervical, colorectal, breast, and lung). Preventing cervical cancer is akin to an embarrassment of riches. It has a single distal cause: HPV infection, which for the majority of women begins in the mid-to-late adolescent years, and can be prevented by vaccination. Even without vaccination, continued screening for cervical cancer over the reproductive years tends to capture the vast majority of precancers that would have eventually developed into invasive disease. The same cannot be said for colorectal and breast cancers, diseases whose onset is throughout the life span and carcinogenesis is mediated by genetic susceptibility or age-related changes in endocrine hormonal influences. While proper understanding of natural history is essential, it behooves us also to make better use of analogy to bring symmetry and fairness in policy decisions concerning cancer prevention and control. Conflict of interest statement The authors declare that they have no conflicts of interest with the content and message of this editorial.
Editorial
Acknowledgments ELF has served as occasional consultant to pharmaceutical (GSK, Merck) and biotechnology (Roche, Gen-Probe, BD, Qiagen, Ikonisys) companies involved with HPV vaccination, HPV diagnostics, and cervical cytology screening. Research on cancer prevention and screening by the authors has been funded by grants from the Canadian Institutes of Health Research, National Institutes of Health, Canadian Cancer Society Research Institute, and Cancer Research Society of Canada. References Cuzick, J., Thorat, M.A., Bosetti, C., et al., 2015. Estimates of benefits and harms of prophylactic use of aspirin in the general population. Ann. Oncol. 26 (1), 47–57. Franco, E.L., 2012. Towards more eclectic evidence-based medicine in cancer prevention and control. Prev. Med. 55, 552–553. King, W.D., Friedenreich, C.M., Brenner, D., et al., 2015. The contribution of lifestyle, environment, genetics and chance to cancer risk in individuals and populations. Prev. Med. 76, 132–134 (in this issue). Moyer, V.A., 2012. on behalf of the U.S. Preventive Services Task Force, Screening for prostate cancer: U.S. Preventive Services Task Force recommendation statement. Ann. Intern. Med. 157, 120–134. Sestak, I., Cuzick, J., 2015. Update on breast cancer risk prediction and prevention. Curr. Opin. Obstet. Gynecol. 27 (1), 92–97. Sills, J. (Ed.), 2015. Letters in response to ‘Cancer Risk’. Science 347, pp. 727–731. Tomasetti, C., Vogelstein, B., 2015. Variation in cancer risk among tissues can be explained by the number of stem cell divisions. Science 347, 78–81. Whitham, H.K., Kulasingam, S.L., 2015. The significantly lower risk of cervical cancer at and after the recommended age to begin and end screening compared to breast and colorectal cancer. Prev. Med. 76, 135–140 (in this issue).
131
Wilson, J.M.G., Jungner, G., 1968. Principles and practice of screening for disease. Public Health Papers, No. 34. World Health Organization.
Eduardo L. Franco⁎ Department of Oncology, McGill University, Montreal, Canada Department of Epidemiology, Biostatistics, and Occupational Health, McGill University, Montreal, Canada ⁎ Corresponding author at: Division of Cancer Epidemiology McGill University 546 Pine Avenue West Montreal, QC, H2W 1S6 Canada. E-mail address: [email protected] (E.L. Franco). Gayle A. Shinder Department of Oncology, McGill University, Montreal, Canada Joseph E. Tota Division of Cancer Epidemiology and Genetics, National Cancer Institute, Rockville, USA Sandra D. Isidean Department of Oncology, McGill University, Montreal, Canada Department of Epidemiology, Biostatistics, and Occupational Health, McGill University, Montreal, Canada
Contents lists available at ScienceDirect
Preventive Medicine journal homepage: www.elsevier.com/locate/ypmed
Editorial
Sobering realizations in cancer prevention and screening and their lessons
The year in the world of cancer prevention began on a gloomy note with the publication of a paper showing that risk of cancer among different anatomical sites is strongly correlated with the lifetime number of stem cell divisions in the underlying tissues (Tomasetti and Vogelstein, 2015). Tomasetti and Vogelstein elegantly elucidated long-standing enigmas such as, why cancer is rare in certain organs (e.g., heart) or disproportionately more common in the large rather than small intestine, despite the fact that the latter is much longer than the former. The most prominent message from their paper (Tomasetti and Vogelstein, 2015), however, was in interpreting the coefficient of correlation for their analysis (0.804) as indicating that some 65% (i.e., 0.804 squared and expressed as a proportion) of the variation in cancer risk among tissues could be explained by the number of stem cell divisions. This left only some 35% of the variation to be attributable to exposure circumstances that we could potentially control or modify via preventive strategies. Sadly, non-modifiable risk determinants, such as low-penetrance genetic mutations and endocrine hormonal influences, are part of the 35%, thus leaving very little for public health practitioners to set targets for cancer prevention and for epidemiologists to make bold claims of cancer causes yet to be discovered. Having appeared in the journal Science and owing to the authors' solid reputation in the cancer research community, Tomasetti and Vogelstein's paper received unparalleled attention in the lay and professional media. Yet, their findings and methods came into dispute, most notably in a series of letters to the editor of Science that was followed by their response (Sills, 2015). Although Tomasetti and Vogelstein made many simplifying generalizations and assumptions their primary finding remains, i.e., that replicative mutations, accumulating over a lifetime, explain a substantial proportion of the variation in cancer incidence among organ sites. The proverbial devil, however, is in the details. That the variation in risk within a country and among anatomical sites is explained by these mutations does not imply that there cannot be further variation among individuals or populations. Cancer is a group of diseases of multi-factorial etiology and long latency. For instance, today's incidence of lung cancer in the US reflects the prevalence of tobacco smoking and occupational risk factors as of some 20–30 years ago in that country, which were already declining. Further reductions in lung cancer incidence in the future will be dependent upon actions to curb smoking and other risk factors today. The same can be said for all other cancers with modifiable risk factors that are clearly identified. Calculating the proportion of cancers that are attributable to specific external causes is an extremely important goal in public health and must take into account the added complexity of latency (i.e., the past gave us today's cancer burden but it is the present dynamics of risk factors that will shape future burden). With empirically valid estimates of population attributable risk (PAR), pairing specific risk factors to their corresponding cancer, policymakers can have targets for prevention and
http://dx.doi.org/10.1016/j.ypmed.2015.04.014 0091-7435/© 2015 Elsevier Inc. All rights reserved.
benchmarks for measuring success in deploying interventions. In this issue of Preventive Medicine, King et al. (2015-in this issue) interpret the message from the Tomasetti and Vogelstein paper (Tomasetti and Vogelstein, 2015) in the context of estimating PARs for a country as large and diverse as Canada. King et al. explain the public health value of PARs and in taking advantage of the granularity in knowledge that comes from estimating disease- and cause-specific PARs via comprehensive literature reviews and country- or region-specific prevalence data on the exposures of interest (King et al., 2015-in this issue). Elimination or inhibition of risk factors over time is a laudable goal for society. At a minimum, this can be accomplished via robust tobacco control policies, public education regarding healthy diets and physical activity, vaccination to prevent infections with cancer-causing viruses, such as human papillomavirus (HPV) and hepatitis B virus, and reduction of environmental and occupational exposures to chemical and physical agents that are known carcinogens. We can perhaps accomplish further gains via chemoprevention, such as tamoxifen or aromatase inhibitors for reducing risk of breast cancer (Sestak and Cuzick, 2015), or daily aspirin for preventing mortality from several cancers (Cuzick et al., 2015). If in combination these measures can be successfully implemented in Canada, a country that places cancer prevention high in the public health agenda, we would be able to verify if the unexplained (regression) variation in Tomasetti and Vogelstein's correlation analysis (Tomasetti and Vogelstein, 2015) would be diminished, thus leaving only the influence of replicative mutations to explain cancer risk. Although this may be useful as a thought experiment intended to validate their argument, attaining complete elimination of a cancer ‘exposome’ is unrealistic. Many people will continue to smoke, eat poorly, and eschew exercising. There will always be noise in the correlation but reducing the impact of modifiable risk factors should lead to an improvement in the goodness of fit of Tomasetti and Vogelstein's correlation. Still on the topic of sobering realizations in cancer control, we now turn our attention to the limits of what can be achieved by cancer screening. Much like the newfound modesty in cancer prevention targets communicated by the aforementioned January Science paper (Tomasetti and Vogelstein, 2015), guidelines in cancer screening are frequently revised to reflect the uncertainty of the evidence. It used to be simple to grade the evidence for or against a given cancer screening technology. There were precious few randomized controlled trials of screening techniques, and their efficacy in reducing cause-specific mortality was the primary driver for country-wide policy decisions or clinical guideline recommendations. Finding evidence that the screening intervention merely leads to improved detection of precancerous lesions or early cancers, as it happens with many technologies proposed in the last decades for screening for cervical, breast, colorectal, ovarian, lung, and prostate cancers, is not enough. To be sure, accuracy in lesion detection is important, but we must be able to show also
130
Editorial
that ablation or excision of the lesion and its surrounding tissue will subsequently lead to cure or an improvement in survival that will translate into a reduced mortality from the cancer in question. While it is hard to argue with the cogency of this definition, we also have to take into account that the mortality reduction benefit must outweigh the potential harms that come from screening. Employing any given screening technology in a large enough group of individuals will inevitably lead to untoward clinical outcomes. These can originate from unnecessary procedures done on people erroneously presumed to have the disease because of a false-positive test result. They can also result from local damage to tissue (e.g., intestinal perforation in colorectal cancer screening) during a follow-up diagnostic procedure on a screen-positive individual, irrespective of whether the procedure was based on best practices or not. Harm can also happen in the form of late sequelae of a successful screening intervention (e.g., miscarriages or premature births following cervical lesion treatment). In the past 15 years, health technology assessment (HTA) has become eclectic enough to incorporate the entire evidence base concerning benefits and harms when passing judgment regarding cancer screening technologies and their practice recommendations (Franco, 2012). The latter is not a trivial part of the judgment, i.e., at which ages one should start or stop screening, and how frequently one should be screened within that span of life when the benefit-toharm ratio is considered maximal. HTA agencies or task forces, such as the US National Institutes of Health's Physician Desk Query program or the US and Canadian Preventive Services Task Forces, employ indepth systematic reviews of the literature, health economic modeling, and expert panels to arrive at consensual and graded recommendations. Expectedly, these recommendations end up influencing professional societies when issuing their own clinical guidelines. Not infrequently, recommendations become quite controversial, such as the US Preventive Services Task Force's assignment of a grade D recommendation (i.e., against) regarding prostate-specific antigen (PSA) based screening for prostate cancer (Moyer, 2012). The public and healthcare providers are beginning to realize that the mantra of ‘early detection saves lives’ is not as simple an argument as it used to be. The bewildering complexity of multi-disciplinary concerns that come into play for HTA recommendations is the new framework for cancer screening. Notwithstanding the above new world of increased complexity for screening assessment, we knew at least that one classic criterion for deciding on screening stood tall, i.e., that the disease should be an important public health problem (Wilson and Jungner, 1968). Although there are no benchmarks for deciding what is important enough, investigator-initiated studies on candidate screening technologies have traditionally justified their pursuits by stating that the screeningtargeted disease was appreciably common and/or had poor prognosis. Defining the importance of a given cancer as a public health problem is era-dependent. The onset of cervical cancer screening in much of the Western world happened at a time (the 1960s) when incidence rates, albeit declining, were high enough (above 20 per 100,000 women per year) and treatment outcomes were dismal enough (about 50% survival at 5 years) for this disease to be considered an important public health problem. Most conveniently, however, there was a simple screening test (Pap cytology), and an easily accessible anatomical site (the uterine cervix). Cervical cancer screening with the Pap test thus became the paradigm against which all other cancer screening interventions had to be judged. The burden of cervical cancer in Western countries, such as Canada and the US, is much less today than it was in the 1960s. Prostate, colorectal, lung, and breast cancers have become far more common relative to cervical cancer. Yet, as societies, we do not screen for these cancers with the same unequivocal conviction that we screen for cervical cancer. A compelling line of reasoning for gauging the importance of the cancer burden that justifies screening was advanced in this issue of Preventive Medicine by Whitham and Kulasingam (2015-in this issue). These authors used data from the Surveillance, Epidemiology, and End
Results (SEER) program maintained by the US National Cancer Institute to compare the risk of breast, colon, and cervical cancer at and after the recommended ages for screening. Presenting the risk of each cancer relative to the others in this way is very insightful and (to our knowledge) has not been done before. The authors' contention hinges on the fact that the risk of cervical cancer is currently much lower than that of breast and colon cancer across the age span, but particularly at the ages when guidelines define risk to be unacceptably high to justify continued screening. Whitham and Kulasingam found that based on precancerous lesions alone (which is where the bulk of mortality risk reduction lies), US guidelines recommend to start screening for colorectal adenomas at an age (50 years) when these lesions are 45 times more common than high-grade cervical intraepithelial neoplasia at the age (21 years) US providers judge appropriate to initiate cervical cancer screening. In other words, there is a disproportionately high tolerance of risk for colorectal cancer than is practiced for cervical cancer. The authors also demonstrate that the same asymmetry of risk tolerance exists when considering someone old enough to exit screening. Breast and colorectal cancers are more than 10 times more common than cervical cancer at the prescribed ages to exit screening (Whitham and Kulasingam, 2015-in this issue). With HPV vaccination expected to further reduce the incidence of cervical pre-invasive lesions and cancer, the high intensity with which US practitioners screen for cervical cancer will become a glaring anomaly in preventive medicine when judged against the backdrop of conservatism that prevails for colorectal and breast cancer screening. Is it surprising that we are far less tolerant of risk for cervical cancer than we are for cancers that are much more common than that of the uterine cervix? Why is it that we do far more (and have done for so long) for cervical cancer than for other screening-detectable cancers? First, we have had Pap tests since the 1940s, a simple technique that requires relatively little technology and no complex equipment or procedures. Second, the cervix is far more accessible via direct visualization than the breast or the colon/rectum. Third, the decision context for the onset of wide-scale cervical cancer screening was at a time (late 50s, early 60s) when this disease was as much of a killer of women as are breast and colorectal cancers. Fourth, Pap tests are recognized as an important acquired right of women and a key component of well-women care by providers. Fifth, entire subspecialties of medicine were created and come to depend upon a continued flow of patients in cervical cancer screening. Attempts to scale down screening in the US have been met with resistance by women and providers. With the escalating costs of healthcare delivery in Western countries, the lucidly explained framework by Whitham and Kulasingam (2015-in this issue) should form the basis for defining epidemiologic benchmarks for risk thresholds to begin and end screening, and for the intensity of screening during that time span. Clearly, the evidence that motivates these decisions must be bolstered by strong science on the natural history of cancers that can be prevented by screening, such as cervical and colorectal cancers, or those whose mortality can be reduced via screening (cervical, colorectal, breast, and lung). Preventing cervical cancer is akin to an embarrassment of riches. It has a single distal cause: HPV infection, which for the majority of women begins in the mid-to-late adolescent years, and can be prevented by vaccination. Even without vaccination, continued screening for cervical cancer over the reproductive years tends to capture the vast majority of precancers that would have eventually developed into invasive disease. The same cannot be said for colorectal and breast cancers, diseases whose onset is throughout the life span and carcinogenesis is mediated by genetic susceptibility or age-related changes in endocrine hormonal influences. While proper understanding of natural history is essential, it behooves us also to make better use of analogy to bring symmetry and fairness in policy decisions concerning cancer prevention and control. Conflict of interest statement The authors declare that they have no conflicts of interest with the content and message of this editorial.
Editorial
Acknowledgments ELF has served as occasional consultant to pharmaceutical (GSK, Merck) and biotechnology (Roche, Gen-Probe, BD, Qiagen, Ikonisys) companies involved with HPV vaccination, HPV diagnostics, and cervical cytology screening. Research on cancer prevention and screening by the authors has been funded by grants from the Canadian Institutes of Health Research, National Institutes of Health, Canadian Cancer Society Research Institute, and Cancer Research Society of Canada. References Cuzick, J., Thorat, M.A., Bosetti, C., et al., 2015. Estimates of benefits and harms of prophylactic use of aspirin in the general population. Ann. Oncol. 26 (1), 47–57. Franco, E.L., 2012. Towards more eclectic evidence-based medicine in cancer prevention and control. Prev. Med. 55, 552–553. King, W.D., Friedenreich, C.M., Brenner, D., et al., 2015. The contribution of lifestyle, environment, genetics and chance to cancer risk in individuals and populations. Prev. Med. 76, 132–134 (in this issue). Moyer, V.A., 2012. on behalf of the U.S. Preventive Services Task Force, Screening for prostate cancer: U.S. Preventive Services Task Force recommendation statement. Ann. Intern. Med. 157, 120–134. Sestak, I., Cuzick, J., 2015. Update on breast cancer risk prediction and prevention. Curr. Opin. Obstet. Gynecol. 27 (1), 92–97. Sills, J. (Ed.), 2015. Letters in response to ‘Cancer Risk’. Science 347, pp. 727–731. Tomasetti, C., Vogelstein, B., 2015. Variation in cancer risk among tissues can be explained by the number of stem cell divisions. Science 347, 78–81. Whitham, H.K., Kulasingam, S.L., 2015. The significantly lower risk of cervical cancer at and after the recommended age to begin and end screening compared to breast and colorectal cancer. Prev. Med. 76, 135–140 (in this issue).
131
Wilson, J.M.G., Jungner, G., 1968. Principles and practice of screening for disease. Public Health Papers, No. 34. World Health Organization.
Eduardo L. Franco⁎ Department of Oncology, McGill University, Montreal, Canada Department of Epidemiology, Biostatistics, and Occupational Health, McGill University, Montreal, Canada ⁎ Corresponding author at: Division of Cancer Epidemiology McGill University 546 Pine Avenue West Montreal, QC, H2W 1S6 Canada. E-mail address: [email protected] (E.L. Franco). Gayle A. Shinder Department of Oncology, McGill University, Montreal, Canada Joseph E. Tota Division of Cancer Epidemiology and Genetics, National Cancer Institute, Rockville, USA Sandra D. Isidean Department of Oncology, McGill University, Montreal, Canada Department of Epidemiology, Biostatistics, and Occupational Health, McGill University, Montreal, Canada
