Specialist Medical Review Council
Reasons for Decision
|
Part 1 |..| Part 2 |..| Part 3 |..| Part 4 |..| Part 5 |..| Part 6 |..| Part 7 |Part 6
Submissions from Repatriation Commission
Both the first written submission and the supplementary written submission of the Repatriation Commission supported maintaining the current contents of both Statements of Principles without amendment. At the meeting, Dr Horsley submitted that the available information does not support smoking as a cause of prostate cancer but the available information does support exposure to phenoxy-herbicides as a cause.
Dr Horsley stated that the evidence that smoking causes prostate cancer is largely confined to a small number of North American studies. He stated that the increased risks that most of these studies have demonstrated are small - usually a relative risk less than two and often barely above one. He noted that, against these few studies, are many others that do not show a significantly increased risk, and publication bias probably gives a distorted picture of the overall research effort.
Dr Horsley submitted that the positive studies had not taken confounding factors into account. He stated:
Smokers eat less vegetable fibre, they eat more animal fat. Smokers have diets with different amounts of vitamins. Smokers drink more alcohol than non-smokers. Smokers exercise less than non-smokers. Former smokers are more likely to be obese. Smokers are more likely to be sexually active at a younger age. They have more sexual partners. They are more likely to have a history of sexually transmitted disease. All of these are factors which ... are likely to confound studies which try to demonstrate a causal link between prostate cancer and smoking and most in a manner which is likely to give a false positive association. For diseases like bronchogenic carcinoma, similar such confounders may exists, but the relative risks for smoking are far, far greater in the case of bronchogenic carcinoma than they are for prostate cancer.
Dr Horsley also stated that the few studies that have shown an association are mainly mortality studies and that morbidity studies are nearly universally negative. He suggested that a simple treatment effect could explain these very small elevations in relative risk for death due to prostate cancer, in that smokers often represent an unacceptable surgical risk.
In relation to herbicides, Dr Horsley made the following submission:
In ... a meta analysis, the effects of confounders tend to be eliminated in their meta analysis of the size performed by Blair, which we submitted to the Authority and to the Council involving 21 different studies with the mortality cancer - of a total prostate cancer mortality of 7500 cases. A risk of 1.08 is a finding of far greater importance than a finding of the same size in a single study. This is even more the case when it is remembered that our knowledge of farmers suggest that there are some factors in this group which may give rise to lower levels of prostate cancer, they have different diets, they have higher levels of exercise and they have higher UV exposure than the general population.
Secondly, when this general finding of a higher rate of prostate cancer in farmers is hypothesis-tested against various exposures, the Commission is aware of only one positive association which has been found, that is, exposure to phenoxy-herbicides. While farmers have exposures to a wide range of chemicals, animals and potential carcinogens, phenoxy-herbicides, alone, as far as we can tell, show a positive and significant relationship. When the smoking prostate cancer hypothesis is tested in purpose designed studies, it usually fails.
Thirdly, when we test for dose response relationship, we find it inconsistently for smokers, and with paradoxes such as former smokers having a rate as high as current smokers. These point away from a true cause or link. With phenoxy-herbicides, Morrison, et al, have demonstrated a good dose relationship. Further, the more carefully this group looked at their data, looking at those who were more likely to spray herbicides, the more pronounced this effect became.
Fourthly, whereas there are abundant confounders which could explain the very slight positive associations found between smokers and prostate cancer, dietary fat, obesity, lack of exercise, male hormones, and these may also explain the slight dose response effects that are sometimes seen, there are no logical confounders that can explain a dose response for herbicide exposure. Do you eat more fat the more you spray? Surely, the more you spray, the more you exercise and therefore should be less likely to get prostate cancer.
Fifthly, the coherence and the consistency of the material that points to a causal chain between prostate cancer and herbicides is demonstrably better than the material for smoking. Most of the herbicide studies are positive, some significantly so. The great majority of smoking studies are not positive and the totality of the material straddle zero. The farming studies are joined by studies of production workers and people accidentally exposed to herbicides. With appropriate latencies, the same small, but consistent elevations are shown. Again, no obvious confounder is present. In short, the most obvious explanation for the association between smoking and prostate cancer is confounder.
The only obvious explanation for the association between prostate cancer and herbicide exposure is that of a causal relationship.
In the meeting, members of the Council asked a number of questions of Dr Horsley in connection with the Commission's submission.
DR GILES: I was interested in your comparison between the criteria of causality used and applied to tobacco and to herbicide, and you seem to come down in favour of one and not the other. But just going through the written submission, the strength of association on page 26, you note between herbicide exposure, you have got a range there from 0.8 to 2.2. Do I take it that you are saying that you think that that material gives strong evidence of association?
DR HORSLEY: I would not call it strong evidence. It is evidence of an association. It is - and those relative risks are relative risks that are similar to what have been found for the cigarette smoking - but you know, I would submit that we have submitted that strength of association is possibly weakly met for both, but we have to assess the evidence against all of the other criteria, and I think that cigarette smoking founders on some of the other criteria whereas the herbicide struggles through.
DR GILES: Okay. The next section on dose response effect, you state one study that showed dose response - a dose response, the Morrison study, which was largely an ecological study. Would you have a view as to the quality of evidence that you get from an ecological study versus an analytical study such as a cohort study?
DR HORSLEY: It is always poorer, but that was within that type of study, a particularly good one and they did have great care to try and get the right types of occupations listed, they took great care to try and show that the more you sprayed the more likely you were to develop prostate cancer. It has its flaws, but it is also true that it is not always possible to design studies to look at herbicide exposure. It is very difficult to get a dose response effect, for example, Seveso, because it is all or nothing in some respects, and those - the small group that were more highly exposed is so very small that it is difficult to know what is happening in that group.
...
DR McCREDIE: Can I bring you back to the point that you were making about confounders, and you have submitted in your written and oral presentations and you would like to submit some more about things that might confound in the smoking and prostate relation - prostate cancer relationship - yet you have not said that you think that many of those other factors would confound the relationship between the herbicides and prostate cancer, and why would you think that?
DR HORSLEY: Just from first principle reasoning, I mean, as far as I can see, sort of going through them, as we have listed them out on page 13 of our - I am sorry, paragraph 13, page 4 and 5 of our submission, looking through those things that have been postulated as possible causes, it is difficult to understand how any of them could be related to herbicide exposure.
DR McCREDIE: Well, if we come back to things like, you were saying that smoking - smokers have different sorts of diet, it is equally plausible and probably able to be referenced, that farmers have a different sort of diet, they have different sort of requirements in their life than other occupations, so it is possible that diet, for example, could confound the relationship that you are suggesting with herbicides too.
DR HORSLEY: It is possible, but I point to the meta-analysis by Blair which shows that the level of coronary artery disease mortality, for example, in farmers is substantially reduced. Now, he suggests that that is in part a reflection of the fact that they smoke a lot less, which of itself would argue against the cause, and is another piece of evidence, but also that it is a reflection of their diet; that they have a diet which he presumes is higher in vegetable fibre and lower in processed and refined foods. And I guess, from first principles, there is some evidence to suggest that.
When we actually searched the literature to try and prove that, we found a surprising lack of information that related to the amount of dietary fibre and animal fat that farmers, as a group, had. So in terms of whether or not diet can be a confounder for herbicide exposure, I expect that that would be a likely possibility, but when we go to look at the number of acres that you spray, I cannot see how diet could co-vary with the number of acres that you spray. Conversely, I can quite easily see - and there is ample and demonstrable literature which shows that the more you smoke, the more dietary fat you eat, the higher is your calorie intake, the lower is your consumption of fresh fruit and vegetables, and a beautiful dose response relationship is established.
Further, the direction that the confounders in farmers would tend to confound is towards a negative association. They exercise a lot, they are out and about, they eat those diets which we presume are protective - we know are protective against coronary artery disease, and we feel may be protective against other forms of cancer, such as colon, prostate; they exercise, they do not have a lot of sexual partners.
DR GILES: But notwithstanding all of the above, the range of relative risks you are talking about, from 0.8 to 2 or so, are very much within the ambit of the sorts of relative risks you are going to find due to uncontrolled or unconsidered confounding in epidemiological studies.
DR HORSLEY: Correct.
Issues raised by the submissions
Causation mechanism proposed by Dr King
The Council notes that Dr King did not point to any information that was available to the Repatriation Medical Authority to support the mechanism by which he thought that smoking might cause prostate cancer. The Council could not find any support for his theory in that information. Therefore, it is not sound medical-scientific evidence within the meaning of the Act and the Council cannot rely on it in making its decisions.
Different types of tobacco and cigarettes
There was no information available to the Repatriation Medical Authority concerning different types of tobacco or cigarettes, and no studies were referred to that have sought to differentiate between types of tobacco. The only information before the Council was the submission of Dr David Pope for the NSW Ex-Prisoners of War Association Inc, submitted by AVADSC, suggesting that Prisoners of War of the Japanese had a higher incidence of prostate cancer than Prisoners of War in the European theatre during World War 2. However, no information on that subject had been available to the Repatriation Medical Authority.
The fact that there may have been a higher incidence in one group compared to the other says nothing about the cause of that higher incidence. No details were provided as to how the study was constructed, the questions asked, how the populations were selected, whether or not or how cigarette smoking was considered to taken into account, or any other details by which the information could be adequately assessed. Additionally, no information was provided as to the statistical significance of the suggested higher incidence.
Therefore, there is no sound medical-scientific evidence available on which any distinction can be made concerning types of tobacco or cigarettes, or on which the Authority could have relied to make a decision relating to types of tobacco or cigarettes.
Cancers in other organs are associated with cigarette smoking
It was stated in one submission that all other cancers have been associated with smoking. There was no information available to the Repatriation Medical Authority that supported that statement, and the Council is aware of a number of cancers that have not been associated with cigarette smoking.
A number of submissions suggested that, because a large number of other cancers, including bladder cancer, have been accepted as being caused by cigarette smoking, then prostate cancer should be accepted as well, especially given the bladder's proximity to the prostate.
None of the submissions pointed to any sound medical-scientific evidence available to the Repatriation Medical Authority that supported the proposition that mere proximity to one organ could result in cancers from the same causes. Such submissions indicate a lack of knowledge of basic physiology and pathology.
It is true that cancers in many different organs and tissues have been associated with cigarette smoking, but that is because epidemiological and other studies have been conducted that have provided a proper epidemiological basis for accepting such an association for each of the particular cancer sites. Each organ of the body has a different function and metabolises different chemicals in different ways, depending on their functions. Some organs have similar functions and operate in similar ways. But the fact that some organs develop cancers due to a particular causal factor says nothing about the effect of that factor on another organ. All that can be said, without more information about the organ in question or the mechanism of causation, is that because the factor has been shown to be causal in one human organ, there is a possibility that it could cause cancer in other organs - but it is only a mere possibility, having nothing to support it other than the analogy.
Cadmium studies on animals
Both Mr McCombe and Commander Fahey suggested that studies in which animals were exposed to cadmium supported the proposition that cigarette smoking causes prostate cancer. Professor Kune and Mr Veigel also referred to studies relating to cadmium.
The information that was available to the Repatriation Medical Authority that related to cadmium was contained in five articles: items 7(a), 10(a), 22, 23, and 24 of the Appendix. Additionally, the submission to the Repatriation Medical Authority by the Department of Veterans' Affairs (item 1(a)) surveyed those and other studies relating to cadmium.
Item 7(a) related to high doses of cadmium being given to rats who subsequently developed prostate cancer.
Item 10(a) gave no support whatsoever to the proposition. It studied, among other things, cadmium concentrations in benign prostatic hypertrophy tissue and found no correlation between cadmium concentrations and hypertrophy.
Item 22 gave minimal support for the proposition that occupational exposure to cadmium resulted in a very slight increased risk for prostate cancer. However, the study itself recognised that it was limited in its ability to quantify cadmium exposure, and that, in any case, the confidence interval included unity - meaning that the outcome of the study could have occurred by chance alone. Such a result is of no significance unless there are other studies also showing positive results and where the confidence interval excludes unity. No such studies are contained in the information that was available to the Repatriation Medical Authority
Item 23 gave no support to the proposition that cadmium can cause prostate cancer. It was stated that while an increased risk was seen only for occupational exposure, the results were not statistically significant. It was also noted that no synergistic effect between occupational exposure and cigarette smoking was observed.
Item 24 related to a study of some 248 workers exposed to cadmium oxide for at least one year. Four prostate cancer cases were observed. While this was significantly more than expected, the researchers stated that it would be manifestly unwise to infer the existence of an industrial hazard.
The Departmental submission referred to various other studies, concluding that while at heavy industrial levels cadmium might possibly cause the development of prostate cancer, there is no evidence of it doings so at lower levels.
From the information that was available to the Repatriation Medical Authority, there is nothing to suggest that the doses of cadmium to which a person would be exposed through cigarette smoking could cause prostate cancer. The studies have either not dealt with the issue of cadmium in cigarette smoke and prostate cancer , or have only related to occupational exposure levels which are orders of magnitude greater than obtained through cigarette smoking. Occupational studies have not shown significantly strong associations to suggest a causal relationship. .
The rat experiments merely show that extraordinarily high doses of cadmium, many degrees of magnitude greater than that obtained through smoking cigarettes, can result in prostate cancer in rats.
N-nitroso compound studies on animals
Both Mr McCombe and Professor Kune mentioned the possible role of N-nitroso compounds in the cause of prostate cancer. Professor Kune stated that N-nitroso compounds are potent carcinogens and have been shown to induce prostate cancer in laboratory animals. He then stated that if the association between smoking and prostate cancer is causal, then N-nitroso compounds might be one of the mediators.
There was no information available to the Repatriation Medical Authority concerning N-nitroso compounds and prostate cancer when it determined the Statements of Principles. There was no information before the Council by which it could assess Professor Kune's assertions, and therefore they cannot be said to be sound medical-scientific evidence within the meaning of the Act.
Suppression of the auto immune system by smoking
Commander Fahey suggested that smoking suppresses the auto immune system thus rendering smokers more susceptible to cancer. The only information relating to that proposition before the Repatriation Medical Authority was a 1983 article by Professor R Burton. However, that article does no more than put up a proposition for consideration. No studies or other information were available to test the proposition against the criteria for determining whether or not it is sound medical-scientific evidence.
In any event, the proposition is inconsistent with the fact that there are a number of cancers that show no association with smoking.
Synergistic effect of smoking
Mr McCombe submitted, on the basis of Mr Veigel's report, that smoking might operate synergistically with some other factor to cause prostate cancer. If that were the case, one would expect a relative risk of the order of 4 or more rather than a relative risk that is close to unity. When factors operate synergistically, relative risks are significantly greater than the relative risks for each of the co-factors independently. Relative risks of the order seen in the US veterans study would, in fact, point against synergistic effects being operative.
Pesticides
The Naval Association mentioned pesticides as being associated with prostate cancer. There was no information before either the Repatriation Medical Authority or the Council that would support such a proposition. Therefore, such a proposition cannot be said to be based upon sound medical-scientific evidence.
Studies concerning smoking and prostate cancer
The main study that has been pointed to as supporting the proposition that smoking can cause prostate cancer is the 26 year follow-up of US veterans. This study has been reported many times by different reporters in different journals, and a cursory examination of the list of reports relating to smoking and prostate cancer might lead one to think that there have been many studies resulting in similar findings. But that is not the case.
It is relevant to note, in examining the US veterans study, that the Hsing report of that study in 1991 noted that the positive association that was found between smoking and deaths from prostate cancer could be the result of a treatment effect. People who are smokers are more likely to have other significant diseases, such as chronic respiratory and cardiovascular diseases, that would be contraindicators for surgery, and would also increase post-operative and peri-operative mortality. Such smoking related diseases may also prevent administration of oestrogen therapy. Thus, the mortality from prostate cancer may well be higher in smokers than non-smokers, but not because prostate cancer is related to smoking.
In the 1995 report of the US veterans study by McLaughlin and others, there is a very slight elevation in skin cancer and brain cancer in various groups of smokers. These cancers are ones in which smoking has not been implicated, and the most obvious explanation is that given by Hsing, of a treatment effect.
It is also significant to note that the relative risk for prostate cancer in the US veterans group has been declining, and approaching unity. Indeed, in the most recent analysis the relative risk was only 1.1.
Mr McCombe submitted that, because the US veterans study was such a large study, this prevented chance playing a part in the outcome, and that the narrowing of the confidence interval relating to the relative risk made the results more certain.
However, the size of the studies says nothing about the effect of confounders or bias. With a relative risk of less than 2 it does not take very much at all for one or more confounders or a selection bias to account totally for the perceived increased risk. Indeed, given the lack of any association being found in many other studies, this is the most likely explanation. In her letter to Dr Grant, dated 21 September 1993, Professor Hsing stated:
It is possible that the observed effect of smoking and prostate cancer in the US veterans study and the LBS was confounded by other unknown or unmeasured factors. This is especially true for the US veterans study, since we did not have information on other related factors, such as diet, alcohol use, and hormones.
Professor Kune, in his report, stated that "confounding etiological factors have not been corrected for" in the US veterans study.
There is no consistent observation of an association between smoking and prostate cancer in the studies available to the Repatriation Medical Authority. Indeed, most of the studies showed no association, including the British doctors study. This too, was a very large study.
It is also relevant to note that broad population and time-trend data appear to point away from cigarette smoking as being implicated in prostate cancer. There has been a steady and significant decline and stabilisation of incidence of lung cancer and other known smoking related diseases that parallels the decline in the incidence of smoking in men. However, prostate cancer mortality in Australia has been stable for decades. This observation points against an association with smoking.
In its supplementary written submission, the Repatriation Commission stated:
The Commission is aware that a small number of studies of prostate cancer have tried to control for dietary differences between smokers and non-smokers, and that these studies were submitted by the Commission to the Authority. A few studies have indicated smoking increases risk after diet has been controlled. Many more show that diet increases risk and smoking is irrelevant. Many more studies (that do not refer to prostate cancer) show that smokers have diets that are high in fat. Much of this material was not submitted to the Authority by the Commission, nor is it contained in the list of material considered by the Authority. The Commission would submit that this material is relevant to the question of the relationship between prostate cancer and smoking.
... the Commission is now aware of a very large number of articles that it considers important to the relationship between smoking and prostate cancer which were not before the Authority. In the main, these articles go to the relationship between diet and cigarette smoking. These articles suggest that smokers have a diet rich in animal fat and high energy carbohydrates, and a diet low in vegetable fibre and in some vitamins. The Commission is of the view that there is a considerable body of evidence that suggests that these diets have a causal relationship to prostate cancer. ...
At the meeting, Dr Horsley tendered to the Council those studies referred to above that were not before the Repatriation Medical Authority. The Council has not taken those studies into account in this review because they were not available to the Authority when it made its determinations. However, the Council considers that as the material is said to support the submission that the results of most the positive studies are affected by confounding factors - a finding that the Council has made on the information that was available to the Authority - the further material would not have affected the Authority's decision on that issue. Nevertheless, the Council will forward all of the material submitted to it to the Authority so that it might be taken into account in any future considerations that the Authority might make concerning malignant neoplasm of the prostate.