Harvard Reports on Cancer Prevention
Volume I: Human Causes of Cancer
Cancer Causes & Control:
An International Journal of Studies of Cancer in Human Populations
Official Journal of the International Association of Cancer Registries
Volume 7 Supplement November 1996 ISSN 0957-5243

Summary

Introduction
The purpose of the Harvard Report on Cancer Prevention is to summarize what we know from epidemiologic research on what does and does not cause cancer in the United States. Each chapter reviews the recent literature on several potential sources of cancer risk (e.g., smoking, diet, obesity, occupational exposure, environmental pollutants) and offers a consensus statement on what the evidence tells us at this time.

Cancer is a preventable illness
One of the most important conclusions to be drawn from this report is that cancer is indeed a preventable illness. As shown in the table below, nearly two-thirds of cancer deaths in the US can be linked to tobacco use, diet, obesity, and lack of exercise - all of which can be modified through action, both at the individual and societal level.

By addressing these risk factors, we believe that cancer mortality in the US can be reduced substantially, perhaps by as much as one-third. Progress is likely to be incremental, however, and not the result of major breakthroughs. Ultimately, the prevention of cancer will depend on changes in individual lifestyles, development and implementation of government regulations, societal change, and further research.

Table 1: Causes of cancer: a review of the evidence
Causes of cancer in the United States:

Estimated percentage of total cancer deaths
attributable to established causes of cancer

Widespread recognition that cancer can be prevented would represent a major shift in how the public thinks about this disease. Avoiding or stopping tobacco use is broadly known to lower cancer risk, but beyond that, cancer is commonly viewed as a mystery whose cause eludes our intellectual grasp and therefore our ability to prevent it. Instead, the public's focus has been on cancer treatment. This is due, in part, to rhetoric about finding a 'cure for cancer.' This focus also results from the fact that the diagnosis of cancer in a friend or family member is a landmark event whose successful treatment is celebrated, while the results of effective prevention cannot be individually recognized and therefore go unnoticed without painstaking research. Newer research showing an association between cancer and diet, obesity, exercise, and alcohol use are not well known. Our hope is that once public attention is brought to these findings, the idea that cancer is a preventable illness will take hold and the power of that idea will drive both individual and societal change.

Public concern about environmental carcinogens is out of proportion with the true risk
A primary focus of public concern has been potential environmental carcinogens, especially chemicals, electro-magnetic fields, and other products of our technological age. Recent years have seen increases in some types of cancer rates, which the public thinks can be attributed to environmental pollution or occupational exposures. Continued vigilance on these fronts is necessary, of course, but in fact these appear to be far less important sources of cancer risk than most people assume.

Stating that the public has been led to worry excessively about environmental carcinogens is likely to be controversial, given the conventional wisdom that industrialization has resulted in increasing rates of certain types of cancers. We do not diminish the importance of these sources of cancer, nor do we stand as apologists for industrial polluters.

Our point is that, with widespread news coverage of a variety of suspected carcinogens, public attention is drawn away from the most important causal factors - tobacco use, diet, obesity, and lack of exercise. Ironically, it is not uncommon to meet heavy smokers who are genuinely concerned about the possible health effects of magnetic fields or `environmental estrogens' while denying or choosing to ignore the health impact of their smoking habit.

Tobacco
With a barrage of new research findings constantly assaulting the public, it is easy for this critical fact to get lost:

Stemming the epidemic of tobacco smoking is our most effective means for preventing cancer.

An estimated 30 percent of all US cancer deaths can be attributed to tobacco use. Over the past decade, scientific studies have shown that the involuntary exposure of nonsmokers to smoke from other people's tobacco products - environmental tobacco smoke - poses a health risk for nonsmokers, including increased risks of lung cancer and other diseases. In 1993, the US Environmental Protection Agency designated tobacco smoke as a Group A carcinogen, for which there is no known safe level of exposure.

Diet and obesity
An estimated 30 percent of all US cancer deaths can be attributed to diet in adult life, including its effect on obesity. Evidence to date indicates that a diet that reduces cancer risk should be:

Obesity in adult life, either due to excess energy intake or reduced expenditure, is an important cause of endometrial cancer and an established cause of post menopausal breast cancer. Obesity also is associated with cancers of the colon, rectum, and kidney. Obesity can be avoided through a combination of diet modification and increased physical activity.

Sedentary lifestyle
Higher levels of physical activity can reduce the incidence of colorectal cancer and may help reduce cancers of the breast and prostate. Regular physical activity during childhood and adolescence may prevent excessive weight gain and delay onset of menstruation (early age at menarche is a major risk factor for breast cancer).

Occupational factors
Control of occupational carcinogens in the US and other industrialized countries represents an important but insufficiently recognized triumph for primary cancer prevention. Collectively, occupational factors are thought to cause about five percent of all fatal cancers, mostly in the lung, bladder, and bone marrow.

Continuing progress in reducing cancers due to these causes can be expected in the US because of technological advances and a continuing shift away from industrial employment. The public must continue to demand that governmental regulations be enforced to minimize occupational exposure to carcinogens.

Family history of cancer
Certain individuals have susceptibility to cancer due to genetic mutations running in families. Fortunately, the occurrence of such mutations is not common, resulting in about two percent of all fatal cancers. Preventive measures that can be taken include genetic counseling, through which couples may decide not to have children, and more frequent screening among those at high risk.

Cancer incidence also depends on genetic polymorphisms that affect the absorption, transport, metabolic activation, or detoxification of environmental carcinogens. Genetic susceptibility due to these cancer-facilitating processes could play an interactive role in the majority of cases of cancer.

Viruses and other biologic agents
Infectious agents, overlooked as causes of human cancer only 30 years ago, are now considered to be a factor in five percent or more of all fatal cancers in the US. Among the more significant infectious agents are human papilloma viruses (HPV) types 16 and 18, which are implicated in cancer of the uterine cervix, and hepatitis B virus (HBV), which is implicated in liver carcinoma.

Hepatitis B and C viruses are responsible for a minority of cases of liver carcinoma in the US. The proportion of such cases is likely to decrease following the availability of anti-HBV vaccine, improved screening of blood and blood products, and more frequent use of disposable syringes and needles by injection drug users. Measures directed against HIV transmission, coupled with declining mortality from cervical cancer due to effective Pap screening programs, suggest that human papilloma virus-related cancer mortality is also likely to decline.

Perinatal growth factors
Excess energy intake early in life is probably responsible for the positive association between height and the risk of breast cancer and possibly other cancers. Evidence is also emerging that larger birth weight is associated positively with some cancer types, notably breast cancer and possibly prostate cancer.

Reproductive factors
Among physiologic processes, reproductive factors are the most closely linked to human cancer. Early age at menarche, late age at first birth, and late age at menopause tend to increase the risk for breast cancer, while parity is associated inversely with cancers of the endometrium and ovaries.

Several other associations have been noted, but they have not been established conclusively, are of marginal importance, or are thought to be surrogates for other recognized causal factors. For example, having multiple sexual partners, an established risk factor for cancer of the cervix, reflects likelihood of exposure to sexually transmitted viruses that are carcinogenic.

Alcohol
The use of alcoholic beverages interacts with tobacco smoking in the causation of cancers of the upper respiratory and gastrointestinal tracts. Moreover, alcohol alone is implicated in cirrhosis-mediated liver cancer and also may cause a proportion of cancer of the breast and the large bowel.

Because the epidemiologic findings here are complicated, the advice to minimize cancer risk is also complex. Most importantly, heavy alcohol consumption should be avoided. Current guidelines define moderate drinking as two or fewer drinks per day for men and no more than one drink per day for women. Women in particular should be cautious about their level of consumption since alcohol may be involved in the etiology of breast cancer. Indeed, women who are at high risk for breast cancer (e.g., because of family history) should avoid drinking alcohol. Women who are at low risk for breast cancer but high risk for cardiovascular disease (against which moderate alcohol consumption is protective) might be counseled to drink alcohol in moderation (i.e., one drink per day). Before taking that step, however, consultation with a physician is strongly encouraged.

Socioeconomic status
Cancers of the lung, stomach, and uterine cervix, and possibly other cancers as well, are particularly common among poor and underprivileged population groups. Poverty may be thought of as an important underlying cause for these cancers, since it is associated with increased exposure to tobacco smoke, alcoholism, poor nutrition, and certain infectious agents.

Environmental pollution
The search for carcinogenic agents from environmental pollution has been a very active area of research, yet few causal links have been firmly established. Investigations have focused on water fluoridation, chlorinated water byproducts, metabolites of organochlorine pesticides (e.g., DDT), and residential proximity to hazardous waste sites or contaminated wells.

Ecologic studies have indicated higher mortality from lung cancer in urban areas with high air pollution levels than in rural areas. On the other hand, epidemiologic studies have not documented an increased risk of lung cancer among nonsmokers living in urban rather than in rural areas but do suggest higher risks for urban smokers compared with rural smokers.

Ionizing and ultraviolet (UV) radiation
The ultraviolet (UV) part of the sunlight spectrum is responsible for over 90 percent of skin cancers, including skin melanomas. Both prolonged sun exposure and a history of severe sunburns have been implicated in skin cancer. Experience in Australia and other countries demonstrates that both higher public awareness of the cancer risk associated with prolonged sun exposure and greater use of sunscreens can be achieved, both resulting in reduced UV exposure.

Ionizing radiation is unquestionably carcinogenic, but the risk of cancer following exposure to low levels of radiation is generally overestimated. Even among Japanese atomic bomb survivors, only one percent have died from radiation-related cancers.

Extremely low-frequency magnetic fields have been intensively studied. The collective evidence suggests that if a risk exists at all, it is small. Radio-frequency-range electromagnetic radiation, as used in cellular telephones, is being studied currently for possible brain carcinogenicity, but at present there is no empirical evidence to substantiate this claim.

Prescription drugs and medical procedures
Several medical products and procedures can cause cancer, but when they are administered to patients suffering from serious disease, they result in exceedingly favorable benefit-to-risk ratios. The problem is more complicated when pharmaceutical agents or procedures are applied to healthy persons for preventive purposes, since the potential benefit is smaller.

Radiotherapy can cause cancer. It is also conceivable that medical diagnostic radiation is responsible for some cancer cases, although this has been documented only for intrauterine exposures in relation to childhood leukemia. It is now generally recognized that mammography conveys a negligible risk but a substantial benefit.

Several pharmaceutical agents are human carcinogens (at different sites): cancer chemotherapeutic and immunosuppressive drugs (bone marrow); menopausal estrogens (endometrium, possibly breast); anabolic steroids (liver); and phenacetin analgesics (kidney, pelvis). Other drugs with established carcinogenicity are not used anymore (e.g., DES).

Oral contraceptives, above and beyond their social benefits, prevent many more cancers of the ovary and endometrium than they cause in the liver, while their effect in causing breast cancer appears to be minimal.

Tamoxifen, a valuable drug for the treatment of breast cancer, is now being evaluated for the primary prevention of breast cancer among healthy women who are at high risk for the disease. There is considerable but not conclusive evidence, however, that tamoxifen can cause endometrial cancer.

Salt and other food additives and contaminants
As noted above, reduction of salt intake could reduce stomach cancer risk. No other food additive or contaminant has been linked conclusively to cancer.

Conclusion
We cannot pretend that preventing cancer deaths in the US will be easy, given the scope of individual and societal change that must occur. But change is possible. Studies of migrant populations show that they tend to adopt the cancer pattern of the host country within a period of time that varies from less than 10 years (e.g., cancer of the large bowel) to a few generations (e.g., cancer of the breast). Cancer rates are malleable, and with the proper commitment of resources, and with time, we believe that the United States can realistically reduce cancer mortality by as much as one-third.

Suggestions