In my long academic and professional career, I have often found myself arguing on the other side of the table from practicing oncologists who are serving as expert witnesses, members of task forces on prevention, consultants, or on panels and round tables. In my opinion, it is important to understand that oncologists are essential professionals in a difficult and sometimes brutal clinical specialty but that neither their specialty training nor their experience renders them experts on how cancer is caused. Diagnosis, yes. Treatment, yes. Management of complications of the cancer and its treatment, yes. Screening and secondary prevention, sometimes. Causation and forensics, no, unless the oncologist has received special training, and not many do.
Cancer experts (oncologists) are usually not experts in the causation of disease. They are experts in diagnosis and treatment. Unless an oncologist is trained as a cancer causation epidemiologist (clinical trials for cancer agents are something else again) or in carcinogenesis, their opinions on causation are just that, the opinions of practitioners who have read what someone else has written. When you have a disease, you want one of these well-trained specialists. When you need an expert to opine on causation, not so much.
Cancer risk is a probability, not a certainty, but it is not entirely random. Not at all. Various exposure risk factors increase the chance of cancer but rarely determine it for sure (there are exceptions). There is randomness about who (specifically) has the bad “luck” (yes, luck) to get the disease compared to others in similar circumstances, but not about the elevation in risk among people (in general, as a group) in those circumstances. This randomness acts within the bounds of increased or decreased risk conferred by a person’s exposure to external causes (such as carcinogens in the environment or personal habits such as smoking) and their constitutional makeup. This constitutional factor (variously called predisposition or susceptibility) reflects genetics (family history) and also “acquired risk” due to other health conditions (for example obesity, drugs, and earlier cancers or cancer treatment). On the other hand, exposure even at low levels can measurably raise the chance of developing cancer. In my opinion, a person who got the disease usually got it because of some intrinsic factor or risk factor in their past that did not act in another person or that they did not have, and upon which random chance operated.
Some cases, not many in the total population of course but some, arise from even modest or low exposures. You can see that in exposure-response curves, where the incidence of cancer increases, at first slightly, even at relatively low levels of exposure. The typical (a priori) risk estimates derived from epidemiological studies and used for decision making (including legal determination) look at the risk of cancer in a real or imagined future. They do not look at the probability of having been exposed to a risk factor in the past among people who got the disease later. (For my colleagues: Case-referent studies do measure prior, not posterior probabilities but the design is based on a sampled notional cohort. Too complicated for this blog.) Among the people who get the disease, the proportion of people who were exposed or shared a particular risk factor is higher than among those who did not, because their ad hoc probability of having a risk factor was higher. This is not circular reasoning. This is Bayesian probability. People who don’t understand this are overly impressed with risk estimates going forward in time (frequentist) and overlook the contribution to risk in the past within the small group that actually got the disease.
The probability of a disease outcome that actually happened having been associated with a risk factor in the past is greater than the probability going forward of a health outcome given the presence of the risk factor. In other words, one should not focus too much on the “future” risks calculated from epidemiological studies or you will be misled, because the event being predicted has already happened. The issue is whether in a particular case the risk factor played a role, and in the case of cancer whether it played a role in increasing the probability of cancer, not as the sole determining factor. After all, even a low elevation is an elevation and contributes some additional risk which translates into cases that would not otherwise have happened.
Oncologists sometimes believe that their training in clinical trials and in screening for cancer qualifies them to opine on epidemiology in general. It does not. Oncology training does not go into depth on etiological epidemiology or the molecular biology of cancer causation. Any oncologist interested in these topics gets special training if they want to work in those fields. Clinical trials are about treatments, identifying subgroups with better outcomes, and side effects (including elevated rates of problems and complications), which is reflected in the learning objectives and outlines of their training programs. Oncology training is not about risk factors for cancer causation in the community or what happened in the past. In fact, most oncologists are not particularly interested in a typical patient’s family history because it usually does not help them manage the patient. (The exception is patients with certain genetic conditions that present management problems.)
Too many experts, in my opinion, are overly impressed with the family history. Almost all common genetic predispositions to cancer that have been studied on a molecular level (and there are many) are defects in the mechanism of repair of damaged DNA. In other words, they interfere with a person’s ability to stop a cancer from developing after a mutation has occurred that could lead to cancer. But something has to start the sequence of events leading to cancer in the first place. A genetic defect in cancer repair (and most conditions reflected in a family history of cancer) simply allows an incipient cancer to make it to a full malignant tumor. They actually make cancer following exposure to a carcinogen more likely. Hence, they are not competing risk factors, but actually amplify the risk of an external (occupational or environmental cause). Thus, it is a fallacy to argue against an occupational cause for a cancer just because a person has a family history of cancer.
Occupational physicians assess causation for a living and it is an essential part of prevention, which is their mission. They are trained in toxicology and epidemiology with an emphasis on etiology (causation, and particularly general causation), not clinical trials or clinical outcomes, which have little or nothing to do with causation. They also have training in clinical medicine, which is invaluable in interpreting the risk of the individual claimant, plaintiff, injured worker, or patient (special causation).
Match the right medical expert to the right problem.
© Tee L. Guidotti