The Natural History Of The Epidemiological Identification Of A Cancer Cause

The formulation of aetiological hypotheses is usually based on the examination of existing data. These data may represent the results of studies in experimental animals, e.g. the occurrence of papillary carcinoma in the bladder of mice after exposure to tobacco tar encouraged investigators to examine whether an association between tobacco smoking and bladder cancer also existed in humans. In other instances, the data may refer to 'unusual' or 'interesting' cases reported in the clinical literature, e.g. the hypothesis linking inorganic trivalent arsenic compounds to skin cancer (IARC, 1980a) and phenacetin-containing analgesics to renal pelvic carcinoma (IARC, 1980b) have been based, to a large extent, on clinical observations and pathophysiological considerations. There have also been situations where hypotheses were developed and subsequently tested on the basis of biological and theoretical arguments. For example, it was hypothesized that passive smoking may cause lung cancer, because sidestream smoke is not qualitatively different from mainstream smoke, and there is no threshold in the dose-dependent relation between active smoking and risk of lung cancer (Trichopoulos, 1994).

In most instances, however, aetiological hypotheses are developed on the basis of statistical associations between the cancer under consideration on the one hand, and various personal characteristics of the affected individuals, in addition to the time and place occurrence pattern of their disease, on the other. By revealing who, when and where are affected by a particular cancer, one has already gone a long way towards discovering why the particular cancer has occurred. Such observations, collectively considered under the term 'descriptive epidemiology,' represent either the products of routinely recorded information or the byproducts of analytical epidemiological studies designed to address other specific aetiological hypotheses.

An aetiological relation presupposes the existence of a statistical association, but for diseases such as cancer, which are defined according to criteria at the histological, cytological or subcellular level, a cause does not have to be, and usually is not, either necessary or sufficient. For example, not all hepatitis B virus carriers develop hepatocellular carcinoma, and this cancer can develop without the presence of, or even exposure to, the hepatitis B virus (Stuver, 1998). Furthermore, the existence of a statistical association between a particular agent and a particular form of cancer does not necessarily imply the existence of an underlying causal relation. It is possible, in fact common, that the association reflects coexistence of the particular agent with another agent (the confounding factor) which represents the real cause of the particular cancer. Thus, lung cancer patients may report excessive use of alcoholic beverages, simply because in several cultures tobacco smoking and alcohol drinking tend to be positively correlated (Figure 1). Even in the absence of confounding, a statistical association is not an infallible indication of a causal relation. Non-smoking lung cancer patients, for example, may report higher alcohol intakes compared with healthy individuals, because they provide more truthful and accurate histories of habits for which there is real or perceived social disapproval (information bias).

Analytical epidemiological studies are designed to explore whether an association between a particular agent or characteristic and a particular cancer actually exists, and what is its real strength after eliminating, as far as possible, all recognizable effects of confounding and bias. The most commonly used measures of strength of the association between a particular agent or characteristic and a particular cancer is the relative risk; this generic term covers the rate ratio, the risk ratio and the odds ratio. The relative risk indicates how many times higher (or lower) is the frequency of occurrence of the particular cancer among individuals exposed to the agent (or possessing the characteristic), compared with individuals not exposed to the agent (or not possessing the characteristic).

Figure 1 Confounding of the association between consumption of alcoholic beverages and lung cancer by tobacco smoking.

Analytical epidemiological studies have been tradi- under investigation in the two (or more, if several exposure tionally designated as cohort or case control. In cohort levels can be ascertained) groups is calculated, allowing studies, exposed and nonexposed individuals are followed the direct estimation of the incidence rate ratio, which is over time and the frequency of occurrence of the cancer a variant of the relative risk (Figure 2). In case-control

Quit Smoking Today

Quit Smoking Today

Quit smoking for good! Stop your bad habits for good, learn to cope with the addiction of cigarettes and how to curb cravings and begin a new life. You will never again have to leave a meeting and find a place outside to smoke, losing valuable time. This is the key to your freedom from addiction, take the first step!

Get My Free Ebook

Post a comment