- Email: [email protected]
The efficacy of surgical treatment of cancer
IMedical
Hypotheses
Medical Hypothesis (1993) 40.129-138 Q Lm-qu~ UK Ltd 1993
Group
The Efficacy of Surgical Treatment of Cancer D. J. BENJAMIN 23/453 Bourke St, Surty Hills, NSW, Australia,
2010
Abstract-This paper is the result of a study to provide a scientific basis for the claim that surgery is effective in extending the life of cancer patients. The study failed to locate any scientifically valid evidence for the claim. The paper includes an outline of some of the factors which led to surgery becoming an ‘accepted’ treatment for cancer and a discussion of how the problem of ethics continues to prevent proper trials being carried out to prove the efficacy of surgery. Alternative ways of obtaining evidence for the efficacy of surgery are discussed including the Graphical Method, Comparative Studies, Survival Rates, Comparison of Incidence and Mortality, Epidemiological studies and Comparison of Survival after treatments based on two different hypotheses about what cancer is. Each analysis showed no reliable evidence of surgery’s efficacy or, where evidence for efficacy could be inferred from the evidence, an alternative conclusion was also possible. Poor methodology, invalid assumptions and definitions, and poor logic are some of the characteristics of this field of medicine. This paper brings together results from scientific papers published over the past 35 years which, taken together, suggest that the reason for the lack of evidence for surgery’s efficacy might be that it is not effective because cancer is a systemic rather than a localised disease when first diagnosed.
Until about 200 years ago there was a widespread belief that cancer was a systemic disease (1). Treatment was largely unsuccessful except for remissions which probably had little relation to trcatmcnt. As a result of studies by pathologist G.B. Morgagni around 1750 and French rcscarchcr Xavier Bichat around 1790, work by R. LaCnncc in the 1820s on the classification of malignant tumours, the development of the achromatic microscope and the study by Rudolph Virchow on cell multiplication in the 1WOs, the medical profession became convinced that cancer was a localised disease (1). At the same time anaesthcsia and ascpsis influenced the transition of surgery from quackery to re-
spectability (2). By the end of the early 1900s surgery had become an ‘accepted’ treatment for most types of cancer. Over the past 35 years there have been isolated claims that surgery has no effect on cancer patient survival, except whcrc it is used to remove tumours threatening life e.g. by obstructing the bowel (3-8). In contrast to this there have been continued claims by organisations such as the American Cancer Society that life expectancy is improving in most types of cancer where surgery is the main form of treatment (9). Both of these perceptions cannot be true. There is therefore a need to collect evidence for the opposing claims to see how it stands up to analysis.
Date received 11 June 1992 Date accepted 27 August 1992
129
130 A definition
MEDICAL HYPOTHESES
of effective
In order to analyse whether or not a therapy is effective it is necessary to define what is meant by ‘effective’. The following definition will be used: A therapy is effective ifit has an impact on the course or outcome of the disease by extending a patient’s life.
This is an unambiguous definition because it is based on the objective measurement of the time of death. Therefore, if it is possible to deline the date of commencement of the disease or its first diagnosis, the length of time between diagnosis and death becomes an objectively measurable quantity. If this time interval is extended by treatment the treatment is ‘effective’. Other definitions involving the presence or absence of tumours are of limited use because they are based on the assumption that the tumour is the disease and ignore the possibility that the tumour might be merely a symptom or element of a systemic disease. In this case the physical removal of the tumour (symptom) might not affect the underlying disease. To demonstrate the efficacy of surgery it is therefore necessary, using statistics, to show that those cancer patients who were trealcd with surgery lived longer than those who were not. If this had been done 100 years ago, before surgery became an ‘accepted’ treatment for cancer, by comparing the survival of two randomly selected groups, one treated and the other untreated, there would not bc a dispute now about the efficacy of surgery. But this has never been done (3, 10, 11). Instead over the past 65 years attempts have been made to prove that surgery is effective in extending survival using, for the untreated control group, patients who remained untreated for a variety of reasons. These were quite unrepresentative of the total group. So it would not be surprising to find that retrospective analysis of these early studies has shown them to be quite invalid as an objective measure of the efficacy of surgery (3, 10).
the basis of treatment of cancer for the past 100
years.) Since proper randomised clinical trials have never been carried out to prove the efficacy of surgical treatment of cancer, and such trials are now considered unethical, and are politically impossible, what other scientific methods can be used to throw some light on the impact of surgery on cancer? One answer has been-provided by a small group of medical scientists who have looked at the statistics of disease in general rather than confine their attention to cancer in particular The graphical method
In 1825 B. Gompertz (12) established that, as a person ages from birth to death, the age-specific mortality rate doubles about every 8.5 years. A typical curve for humans is one with a formula such as y = eu.u*x where x is the chronological age in years (Fig. 1). He described this relationship as a uniform logarithmic worsening of the tendency to die with increasing age. When plotted on semi-logarithmic graph paper this gives a straight line sloping up to the right (Fig. 2). Such a curve is sometimes referred to as representing a ‘constant’ mortality rate. In this context there are fundamentally three types of mortality increasing with the duration of the disease; a straight line, representing constant mortality, unaffected by the duration of the disease; and concave downwards, representing decreasing mortality with duration of the disease (4) (Fig. 3). 1500r
t
The problem of ethics
In order to carry out a randomiscd trial now one would have to withhold from a randomly selected group of cancer patients an ‘accepted’ technique which, it is argued, might save their lives. So these patients would have to be ‘sacrificed’ to prove the efficacy of surgery, a measure which is considered unethical. (Although this reasoning is quite illogical and contrary to scientific principles it is sufficiently widely accepted to make such a trial politically impossible. The mere undertaking of such a trial would be an admission that an unproven method had been
I
0
IO
20
30
40 AGE
50
60
I
I
I
70
80
90
YEARS
Fig. 1 The variation in Mortality Rate with Age for humans. Gcmpertz in 1825 described this relationship as a uniform logarithmic worsening of the tendency to die with increasing age. The formula y = eO.08~ is only approximate. When plotted on semi-logarithmic graph paper this becomes a straight line sloping up to the right and is sometimes referred to as representing a ‘constant’ mortality rate.
131
THE EFFICACY OF SURGICAL TREATMENT OF CANCER
CHRONOLOGICAL
AGE,
AGE (on linear scale)
YEARS
Fig. 2 Sub-populations of people with specific degenerative disease have a constant-slope mortality rate curve, their mortality doubling about every 8.5 years (the same as healthy people). Those with a particular type of cancer have a characteristically high mortality rate. The line is displaced in time as if they had aged by 10-15 years.
‘Commonsense’ among medical workers would lead to an expectation that most diseases will follow either the first or third type of curvature. For example it appears from conventional wisdom that disseminated cancer is a disease whose mortality rate increases with duration of the disease, i.e. the patient’s health deteriorates with time; whereas myocardial infarction (a heart attack) is associated with a mortality rate which decreases with time, i.e. the patients get over their disease. In facl these suppositions are both incorrect, as Jones (3) had noted in 1956 and Zumoff and his colleagues confirmed 10 years later (4). Jones re-analysed the statistics on mortality rates for different diseases from studies throughout the world, including those on cancer from the previous 30 years. He found that for chronic diseases, including both heart disease and cancer, the sub-populations of people with these diseases lie on a constant slope mortality rate curve. Contrary to conventional wisdom, they neither get over their disease (as is assumed with heart disease) nor get worse with time (as is assumed with cancer). Instead people with particular types of cancer have a characteristically high mortality rate; but the apparently increasing mortality rate with cancer is due simply to increasing age, the rate continuing to double ever 8.5 years. He found the same for
Fig. 3 There are fundamentally three types of mortality curves. ‘Conventional wisdom’ says that cancer shows increasing mortality and heart disease patients get over their disease after a heart attack. Both these suppositions are incorrect. Both diseases are associated with a constant slope mortality rate.
patients with tuberculosis, cirrhosis of the liver and cerebral vascular disease. Each disease had a different age-specific mortality rate line but all lines had the same slope. Using the same graphical technique, Zumoff confirmed Jones’ findings for cirrhosis of the liver, disseminated breast cancer, chronic lymphatic leukemia and myocardial infarction (4). Like Jones he concluded that such diseases have in common an alteration of undefined physiological systems that influence susceptibility to aging and dying, producing an elevated and constant increase in this susceptibility. The power mtiel Some authors such as Armitage and Do11(13) noted that the shape of the cancer mortality curve differed markedly from that of other chronic diseases and they developed a ‘power’ model of carcinogencsis to cxplain the curve’s shape. However Zajicek (14) concluded that the gradual departure from the constant slope curve observed with some cancers after the age of 50 years could be due to inaccurate diagnosis. Autopsies on pcoplc over 65 years have shown cancer incidence of about 40%, which is much higher than
132 is seen in mortality reports. This shows that cancer often co-exists with other fatal diseases and is therefore likely to be inaccurately diagnosed. By correcting for this factor Zajicek was able to show that cancer does in fact have the same constant slope curve as the other degenerative diseases mentioned. One of the features of this graphical technique is that it is sensitive to the effects of any therapy which can affect the mortality rate. Reduction in the mortality rate would result in patients dropping onto a lower line. In this way Jones was able to demonstrate that, after correcting for poor methodology, all patients with a particular form of cancer experienced the same mortality whether treated or not, once the cancer was of established malignancy. Further confirmation of this phenomenon has since come from Green (5). More recently Fox (15) applied the graphical method to breast cancer. He took data that Bloom (16) had reported on the survival of untreated breast cancer patients at Middlesex Hospital at the turn of the century. This group of patients had received no surgery, radiation or chemotherapy. The death rate for this sub-population of cancer patients is 25% per year with few survivors after 7 or 8 years. Fox then took survival data of patients treated by surgery, radiation and chemotherapy during the period 1950-1973, and plotted it in the same manner as before. This is shown as the black dots in Figure 4. He found that the total population could be broken down into two sub-populations each with a completely different prognosis. One group, representing 40% of the total, had a poor prognosis and, like the untreated sub-population from Middlesex Hospital, few had survived 7 or 8 years. These are shown as the open circles in Figure 4. The remaining 60% exhibited a survival characteristic only slightly lower than that of women of similar ages without evidence of cancer. These are shown by the other line in Figure 4. The conclusion that has been drawn from this data is that breast cancer can be cured if detected early enough. (The Middlesex Hospital group would have consisted mainly of advanccd cases.) However a second possible conclusion is that there arc two groups of patients with breast cancer: 40% with a suppressed immune system for whom the cancer grows and spreads rapidly; and 60% whose immune defences are still capable of keeping the cancer under control. If the latter conclusion is valid it is possible that treatment had no effect in either of the groups. Thus the Graphical Method does not support the claim that surgery has an impact on the course of cancer. Rather it is consistent with the claim that surgery is ineffective.
MEDICAL HYPOTHESES
100
I
I
I
I
I
90 80 70 60 50 z? m.? ? 2 .?Q)
40
30
I B
20
10
\ Year After Diagnosis
Fig. 4 Survival data of breast cancer patients treated by surgery, radiation and chemotherapy from 1950-1973. It can be considered as two sub-populations: one (lower left) identical to that of untreated patients with advanced disease at the turn of the century, and the aher (upper right) with survival only slightly less than that of healthy wanen of similar ages. From this data it is possible to conclude either that surgev reduces mortality if the cancer is detected early (the normal conclusion) or that there are women with two different levels of immune status: a) low, unable to control the disease and b) normal, able to keep the disease under control. If the latter conclusion is correct it is possible that treatment had no effect.
Comparative
studies
Another useful source of evidence for the efficacy of surgery is the comparative study. By comparing the mortality of different groups of cancer patients who have undergone different degrees of surgery it is possible to evaluate the efficacy of surgery itself. According to the rationale for cancer surgery it is important to remove all malignant cells; otherwise the remaining cells will continue to grow. Therefore the more extensive the surgery, the less the likelihood of any malignant cells remaining and the lower the mortality rate should be. If this rationale is invalid,
THE EFFICACY
OF SURGICAL
TREATMENT
133
OF CANCER
and tumours are only local, late-stage symptoms of a systemic disease, there would be no difference in mortality between groups receiving different degrees of excision. The comparative trial avoids the ethical problem because it does not require the withholding of an ‘accepted’ treatment in order to have an untreated control group. In recent years several randomized clinical trials have been carried out to compare the survival of breast cancer patients after different degrees of excision for tumours up to 4 cm in diameter. There was no difference in survival between radical mastectomy, total (simple) mastectomy, quadrantectomy, segmental mastectomy (lumpectomy) and excisional biopsy (17-20). These results show the relevance of the comment made in 1963 by Shimkin that ‘when many forms of treatment appear to yield the same results or lack thereof suspicion should arise that none is really effective, and a no-treatment group in subsequent comparisons may be acceptable’ (10). Or as Lewison commented in the same year in relation to breast cancer surgery: ‘In recording our surgical triumphs are we merely measuring the natural history of this malignancy?’ (21). That is, the results of these comparative trials suggest that surgery has no impact on the course of cancer. Claims of increased survival rates If, as the above evidence suggests, surgery is ineffective, what is the explanation for the apparent improvement in the percentage 5-year survival rates for all cancer sites between 1960 and 1975 as claimed by the American Cancer Society (9)? Iwo possible explanations have been offered: 1. These figures are unreliable for reasons of poor methodology. One main example of poor methodology is that increased 5-year survival can result from death happening later (real progress) or by making an earlier diagnosis, as has happened in more recent times. (There is no progress here. Death still occurs at the same time but the existence of the cancer has been known for a longer time, leading to an ‘apparent’ increase in survival.) As Pete and Easton stated in relation lo the effect of earlier diagnosis on staging, ‘Stage III in 1990 is therefore not the same thing as Stage III in 1970, and has a better prognosis irrespective of the effects of therapy’ (22). Other factors include comparison between unmatched groups. For reasons such as these EnStrom and Austin concluded that ‘survival rates should not be used as a sole or primary measure of progress in cancer control because factors un-
related to the efficacy of treatment play an important role in the determination of those rates and their trends’ (23). 2. Earlier figures with lower survival applied when more aggressive treatments were being used and were reducing survival (24). Whatever the reason, survival figures arc unreliable as a measure of the efficacy of surgery as a treatment for cancer.
Comparison of incidence and mortality Enstrom and Austin concluded that a better measure of progress in cancer treatment is to compare the incidence of each type of cancer with the mortality rate over the time interval in question. For, so long as incidence and mortality remain unchanged, or change proportionately, no genuine change in survival can occur. Progress in cancer control requires that the mortality rate dcclinc mom rapidly or rise more slowly than the incidence for the particular type of cancer (23), (see Fig. 5). If incidence is compared with mortality over the period from 1950 to 1970 it is found that there have been large changes in incidence and mortality over this time with several types of cancer (see Table). For most types the incidence and mortality have changed by the same amount within the limits of uncertainty of the statistics. Only in the case of prostate cancer can some claim be made for improved survival due to treatment. Yet even here there is another possible explanation. Most prostatic cancer is ‘occult’ and relatively benign in its course. The increasing rate of autopsies and operative procedures for other causes makes for an increase in the number of occult tumours found, leading to an increasing recorded incidence. This increasing rate of discovery of occult prostatic cancer has caused the stage distribution to change (23, ref 28 on p851), i.e. there is a higher proportion of early stage tumours in the more recent samples than in those recorded 20 years previously, one of the factors rcfcrred to above in relation to poor methodology which has resulted in an inflated value of the percentage 5-year survival rate. The apparent survival rate between 1950 and 1970 closely follows the increased incidence of this type of cancer. As Cairns (25) has pointed out, the incidence of prostatic cancer discovered by autopsies on 70-year old men who had died of other causes was found to be up to 100 times as great as the incidence based on diagnosis of such patients presenting with symptoms. So the incidence figures for these slow-growing tumours are unreliable for the purpose of comparison with mortality rates as a means of assessing progress in cancer control.
134
MEDICAL IlYPOTHESES
Fig. 5 So long as incidence and mortality remain unchanged or change proportionately, no genuine change in survival can occur. The apparent increase in the percentage 5-year survival rate is due to factors unrelated to the efficacy of treatment, such as the increasing proportion of localised cases in the more recent samples. This graph (taken from a paper by G. Linden) is used by Enstrom and Austin to &monstrate this effect.
From a comparison of the incidence and mortality rates over this 20-year period there are therefore no clear cases where survival could have improved as a result of surgery. This is further evidence that surgery has not had a proven impact on the course of cancer. In April 1987 the US General Accounting Office issued a report in which it stated that gains in treating cancer over the last 3 decades have been small and overstated by Federal health officials. For a majority of the most common tumours there was little or no improvement from 1950 to 1982 in the rate at which patients survived their disease (26). Epidemiological studies Another useful type of information for testing the efficacy of surgery is the epidemiological tables of mortality which enable international mortality trends to be compared and correlated with social changes and the introduction of new diagnostic techniques. There is some evidence from this source that surgery may be effective in invasive cervical turnours. For example, the decline in mortality from cervical cancer which has been observed in many countries throughout the world is claimed to have accelerated in those countries at the time PAP smear programs were intro-
duced, when it can be assumed that more women were diagnosed with, and therefore treated for, invasive cervical cancer. However there are counter claims that this decline occurred prior to the introduction of the PAP smear (27) (see Fig. 6). Nothing that has been said is intended to imply that surgery has no place in cancer treatment; only that, with few exceptional conditions, there is no reliable evidence that it extends life. Surgery can of course extend the life of a cancer patient by removing a lifethreatening obstruction in a vital organ or pressure on the brain. Surgery is also useful for reducing pain, removal of a source of toxins from the body, removal of unsightly growths and reducing the load on the immune system. But these benefits can be achieved without affecting the disease itself. An increasing number of cancer specialists, particularly surgeons, are recognizing that surgery has limited effect on cancer. For example, in 1970 Sir John Bruce, Regius Professor of Clinical Surgery at Edinburgh University and past President of the Royal College of Surgeons, stated that ‘the future (of cancer treatment) lies elscwhcre than in the operating room; but when the answer is eventually found, the surgeon will have no cause to be ashamed of his attempts to relieve suffering and not infrequently avert the arrows
THE EFFICACY
OF SURGICAL
TREATMENT
135
OF CANCER
Table There have heen large changes in incidence and mortality between 1950 and 1970. In most types of cancer the incidence and mortality have changed by the same amount, within limits of uncertainty. So there cannot have been an improvement in survival rates over this period for these cancer types. The apparent improvement is due to poor methodology. For prostate cancer the apparent improvement is due to increasing discovery of prostate cancers during operations and autopsies Site
Death rate/lo0 1950
Lung Breast Colon Rectum Prostate ovary Pancreas uterus,
1970
Change %
corpus Stomach Bladder Esophagus Kidney All sites Ref. source
CKKI
Percentage
1950
1970
Change %
1950-54
1965-69
5.year survival rate
Change %
1960-63
1973-80
Change %
8s 63 41
12a 74 50
+5Oa +17 +22
50 32
68 37
+36 +16
58
68
+17
12 13 15.5
30 13.5 15
+150 +4 -3
17 38 25
36 39 27
+112 +3 +8
7 59 42
9 64 44
+29 +8 +5
7 7.5 4 6 5.5
4.5 7 4.5 7.5 3
-36 -7 +13 t2.5 45
17 17 8 7
12 21 7 8
-29 +24 -13 t14
41 44 28 1 57
40 56 32 1 56
-2 I +27 +14 0 -1
32
20
-38 71
74
+4
73
88
+21
tY
11 53 4 37
14 73 5 50
+27 +38 +25 +35
cervixh Uterus,
Incidence/100
OOU
4.5 15 4 2.5 2.5 135
2
-56
6 3.5 2.5 3
-60 -13 0 +20
24 13 4 5
9 12 3 6
-63 -8 -25 +20
11 54 4 32
12 60 3 41
+I1 -25 +28
+4
290
275
-5
39
39
0
140 (23)
(23)
(23)
Notes: aThese figures for lung include bronchus, bfigures for uterine cervix exclude Leukemia have heen omitted as surgery is not applicable.
of death in one of the greatest scourges of mankind’ (6). In 1973, in a collection of essays celebrating the 75th anniversary of the Rosewell Park Memorial Institute, Thomas L. Dao of the Institute’s Department of Breast Cancer wrote: ‘Despite improved surgical techniques, advanced methods on radiotherapies, and widespread use of chemotherapies, breast cancer mortality has not changed in the last 70 years’ (28). In 1974, addressing a cancer conference at Sydney University, Miles Little, Professor of Surgery, stated that: Despite refinements in surgical technique and management, and increasingly radical surgery, there is comiderable doubt about fhe impact of surgery on the natural history of most malignancies. The apparently logical hypothesis that earlier diagnosis and more radical excision would lead to more cures, has not been borne out in practice. Surgery brings a mechanical approach to a biological problem (7).
In 1986 Bailar and Smith pointed out that the death rate from all cancers other than lung, stomach and cervix (sites which have shown marked changes in in-
‘in
(9) situ’,
Cligures for Hodgkin’s
Disease and
cidence in recent decades) had not altcrcd appreciably since 1950 (8). These arc implied admissions by the more honest medical scientists that there has never been any reliable scientific evidence to show that surgery affects the course of cancer. It became an ‘accepted’ method for cancer treatment but was never a ‘proven’ method. Peto and Easton state that in the many situations where it is not known whether treatment is effective: many clinicians
respond by developing a set of fwmly held but unsupported beliefs in the merits of particular regimens. The primary treatment of advanced nonmetaslatic laryngeal cancer, for example, will usually be by surgery at certain treatment centres and by radiotherapy at others. Whether chemotherapy is given as well and, if so. what form it will take, are also determined more by the idiosyncrasies and outpatient arrangements of the particular treatment centre than by objective evidence of long-term efficacy. Similar examples could be taken from most areas of cancer therapy (22).
Stephanie Short, a sociologist, has carried out a study of cancer treatment in Australia and discovered that if a process is technically feasible, and cancer
136
MEDICAL HYPOTHESES
Death
$
0;
cl
1930
Rate
for
Cancer of
the Uterus,
I
U.S.
I
7
I
1935
1940
1945
1950
1955
1960
1965
19’10
1930-1979
1975
1980
YEAR Fig. 6 Earlier surgical treatment of uterine cancer following the widespread introduction of the PAP test has been claimed to be the cause of the observed fall in the mortality rate. However there have been other claims that the fall began before the introdcmion of the PAP smear. Source: Cancer Facts and Figures, 1984, American Cancer Society.
patientsexpectsomethingto be done for their disease, it will be done, even if there is no proven efficacy for the treatment. She described this as the ‘technological imperative’ (29). This growing realisation of the inefficacy of surgery in extending life should lead to a reappraisal of the underlying assumptions of the nature of the disease. For if tumours are in fact only local, late-stage symptoms or elements of a systemic disease, removal of such symptoms would not be expected to have any impact on the natural history of the disease. Comparison of survival after treatments based on different hypotheses There is growing evidence from a wide variety of sources that cancer is a systemic disease, even when first diagnosed. For example in the comparative trial of breast cancer in the US mentioned above (19), of those patients who had received radiation in addition to segmental mastectomy, about 90% were free of tumours (‘disease free’) after 5 years compared to about 70% of those who had received no radiation.
Yet these two groups showed the same survival. This shows that the presence or absence of tumours after surgery is not a reliable measure of the existence of the disease. (Hence the definition of ‘effective’ used at the beginning of this paper.) A second source of evidence is survival statistics. The best 5-year survival statistics have been produced using therapies based on the hypothesis that cancer is a systemic disease, and tumours are ony local symptoms. Therefore the cancer patient should be treated using a therapy designed to restore the body’s own natural immune system. After undergoing such a therapy tcrminal cancer patients showed a 16.6% 5-year survival compared with about 2% expected with conventional therapies and a 15% 15-year survival compared with less than 1%. With pre-terminal patients there was an 85% 5-year survival compared with about 50% (30). These results prompted John Anderson, Professor of Physical Medicine at King’s College Hospital in London to state that ‘the Isscls approach to the treatment of cancer is an unique and pioneering solution to a very difficult problem. He is undoubtedly producing
137
THE EFFICACY OF SURGICAL TREATMFiNT OF CANCER
clinical remission in patients who have been regarded as hopeless and left to fall back on their own resources’ (3 1). Similarly a therapy designed to augment the body’s own immune system is reported to have produced 50% 5-year survival with 11 cases of peritoneal mesothelioma, a malignancy with an expected prognosis of about 12 months (32). Conclusions In summary, surgery became an ‘accepted’ treatment for cancer as a result of several factors. Because of the ethical problems of providing a proper control group of untreated patients, no scientifically acceptable trials have ever been carried out to prove that surgery is ‘effective’ in extending life. A graphical method used for analysing mortality rates suggests that there is no difference in survival between treated and untreated patients. Comparative studies of surgical techniques using differing degrees of excision show no difference in survival. This would be expected if tumours were only symptoms of a systemic disease. Percentage 5-year survival statistics offered to show that surgery is an increasingly effective method of cancer treatment are unreliable because of several methodological problems. The difference between incidence and mortality, which is a more reliable measure of the effects of surgery on cancer control, suggests that no improvement in survival has occurred as a result of surgery. The best, properly documented results of survival with terminal cancer patients have been achieved using therapies based on the hypothesis that cancer is a systemic disease and therefore should be treated by methods designed to restore the body’s own natural immune system. Surgery has therefore not been shown to be an effective or proven method for the treatment of cancer. These findings add weight to the hypothesis that cancer is a systemic disease when first diagnosed.
2.
3. 4. 5.
6. 7.
8.
9. 10. 11. 12.
13. 14. 1.5. 16.
17.
18.
19.
20.
Acknowledgements 21. I would like to acknowledge the encouragement and help given by Alec Forbes, former Consultant Physician at Plvmouth General Hospital, Advisor to the WHO on Traditional Medicine. and cofounder and original Medical Director of the Cancer Help Centre in Bristol, UK. I would also like to acknowledge the help of John McAllan. a CSIRO colleague who has since died of mesothelioma.
22.
23. 24.
References 1. Issels J. The Classic Methods, pp 32-36 in Cancer, A Second
25. 26.
Opinion. Hodder & Stoughton, London, 1975. Moss R W. Surgery pp37Xl. and The Crisis of Credibility ~~22-23 in The Cancer Syndrome. Grove Press, New York, 1980. Jones H B. Demographic Considerations of the Cancer Problem. Trans N Y Acad Sci 18 (4): 298, 1956. Zumoff B, Hart H, Hehnan L. Considerations of Mortality in Certain Chronic Diseases. Annals Intern Med 64: 595, 1966. Green G II. Duration of Symptoms and Survival Rates for Invasive Cervical Cancer, Aust & N I, Journal of Ohstet and Gynec 10: 238, 1970. BNCC Sir J. Quoted by Issels J (1) at ~36. Little M. Quoted by Jacka J in Statistics relating to cancer, pl in Cancer-A Physical and Psychic Profile. Currency, Melbourne, 1977. Bailar J C, Smith E M. Progress against cancer? N Engl J Med 314: 1226, 1986 and following correspondence on June 5 and July 15, 1986. American Cancer Society. Cancer Facts & Figures 1984. American Cancer Society, New York, 1983. Shimkin M B. The Numerical Method in I’herapeutic Medicine, Public Iicalth Reports 79 (1): 5, 1964. Phillips S. Dr Joseph Issels and the Ringberg Klinik. Clinical Trials Journal, 35.5, August 1970. Gompertz B. On the nature of the function expressive of the law of human mortality, and on a new mode of determining value of life contingencies. Phil Trans Roy Sot 115: 513, 1825. hrmitage P, Doll K. The age distribution of cancer and a multistage theory of carcinogenesis. Brit J Cant 8: 1, 1954. hjicck G. Cancer as a Systemic Disease. Medical Hypotheses 4 (3): 193, 1087. Fox M S. On the Diagnosis and Treatment of Breast Cancer. JAMA 241 (5): (Feh 2). 1979. Bloom 11 J G. Survival of women with untrcatcd breast cancer+st and present. pp 3-19 in Prognosis Factors in Breast Cancer (A P M Forrest and P B Kunkler, eds) II & S Livingstone Ltd. Edinburgh 1968. Fisher B, Redmond C. Fisher E et al. Ten-year results of randomired clinical trial comparing radical mastectcmy and total mastectomy with or without radiation. N Engi J Med 312 (11): 614, 1985. Vcroncsi U, Saccozzi R. Del Vccchio M ct al. Comparing Radical Mastectomy with Quadrantectomy, Axillary Dissection and Radiotherapy in patients with small cancers of the hrcast. N Engl J Med 305 (1): 6, 1981. Fisher B, Bauer M, Margoiese R et al. Five-year results of a randomimd clinical trial comparing Total Mastectomy and Scgmcntal LMastcctomy with or without radiation in the trcatment of Breast Cancer. N Engl J Med 312 (11): 665, 1985. Sarrazin I,. IL M. Rouesse J et al. Conservative Treatment versus Mastectomy in Breast Cancer Tumors with Macroscopic Diamctcr of 20 Millimeters or Less. Cancer 53 (5): 1200-1213, 1984. lewison E F. An Appraisal of Long-Term Results in Surgical Treatment of Breast Cancer. JAMA 186 (11): 975-978, 1963. Peto J, liaston D, Cancer treatment trials-past failures, current progress and future prospects. Cancer Surveys 8: 513-533, 1989. Enstrom J Ii, Austin D I- Interpreting Cancer Survival Rates. Science 195: 847-851, 1977. Greenhcrg D S. A critical look at cancer coverage. Columbia Journalism Rcvicw. 4044 Jan&b, 1975. Cairns J. The Cancer Problem. Sci Am 233 (5): 64-78, 1975. US Congress General Accounting Office. Cancer patients survival: what progress has hcen made? PEMD-87-13, (3/31/87).
138 27. 28.
29. 30.
MEDICAL HYPOTHESES
Editorial. Uncertainties of cervical cytology. Br Med J 5891: 501-502, 1973. Dao T L. Quoted by Greenberg D S, in ‘Progress’ in Cancer Research-Don’t Say It Isn’t So. N Engl J Med 292 (13): 707-708, 1975. Short S. The War Against Cancer. A sociological study of cancer treatment. New Doctor 35: 25-28, 1985. Issels J. Immunotherapy in Progressive Metastatic Cancer-A
31.
32.
Fifteen-Year Follow-Up. Clinical Trials Journal, 357-365, August 1970. Anderson J. Quoted by Thomas G. in Issels--The biography of a doctor. London, Hodder and Stoughton, (in Introduction, ~16). 1974. Clement R J, Button L. Lampe G N. Peritoneal Mesotltelioma, Quantum Medicine 1 (1 & 2): 6&73, 1988.
Hypotheses
Medical Hypothesis (1993) 40.129-138 Q Lm-qu~ UK Ltd 1993
Group
The Efficacy of Surgical Treatment of Cancer D. J. BENJAMIN 23/453 Bourke St, Surty Hills, NSW, Australia,
2010
Abstract-This paper is the result of a study to provide a scientific basis for the claim that surgery is effective in extending the life of cancer patients. The study failed to locate any scientifically valid evidence for the claim. The paper includes an outline of some of the factors which led to surgery becoming an ‘accepted’ treatment for cancer and a discussion of how the problem of ethics continues to prevent proper trials being carried out to prove the efficacy of surgery. Alternative ways of obtaining evidence for the efficacy of surgery are discussed including the Graphical Method, Comparative Studies, Survival Rates, Comparison of Incidence and Mortality, Epidemiological studies and Comparison of Survival after treatments based on two different hypotheses about what cancer is. Each analysis showed no reliable evidence of surgery’s efficacy or, where evidence for efficacy could be inferred from the evidence, an alternative conclusion was also possible. Poor methodology, invalid assumptions and definitions, and poor logic are some of the characteristics of this field of medicine. This paper brings together results from scientific papers published over the past 35 years which, taken together, suggest that the reason for the lack of evidence for surgery’s efficacy might be that it is not effective because cancer is a systemic rather than a localised disease when first diagnosed.
Until about 200 years ago there was a widespread belief that cancer was a systemic disease (1). Treatment was largely unsuccessful except for remissions which probably had little relation to trcatmcnt. As a result of studies by pathologist G.B. Morgagni around 1750 and French rcscarchcr Xavier Bichat around 1790, work by R. LaCnncc in the 1820s on the classification of malignant tumours, the development of the achromatic microscope and the study by Rudolph Virchow on cell multiplication in the 1WOs, the medical profession became convinced that cancer was a localised disease (1). At the same time anaesthcsia and ascpsis influenced the transition of surgery from quackery to re-
spectability (2). By the end of the early 1900s surgery had become an ‘accepted’ treatment for most types of cancer. Over the past 35 years there have been isolated claims that surgery has no effect on cancer patient survival, except whcrc it is used to remove tumours threatening life e.g. by obstructing the bowel (3-8). In contrast to this there have been continued claims by organisations such as the American Cancer Society that life expectancy is improving in most types of cancer where surgery is the main form of treatment (9). Both of these perceptions cannot be true. There is therefore a need to collect evidence for the opposing claims to see how it stands up to analysis.
Date received 11 June 1992 Date accepted 27 August 1992
129
130 A definition
MEDICAL HYPOTHESES
of effective
In order to analyse whether or not a therapy is effective it is necessary to define what is meant by ‘effective’. The following definition will be used: A therapy is effective ifit has an impact on the course or outcome of the disease by extending a patient’s life.
This is an unambiguous definition because it is based on the objective measurement of the time of death. Therefore, if it is possible to deline the date of commencement of the disease or its first diagnosis, the length of time between diagnosis and death becomes an objectively measurable quantity. If this time interval is extended by treatment the treatment is ‘effective’. Other definitions involving the presence or absence of tumours are of limited use because they are based on the assumption that the tumour is the disease and ignore the possibility that the tumour might be merely a symptom or element of a systemic disease. In this case the physical removal of the tumour (symptom) might not affect the underlying disease. To demonstrate the efficacy of surgery it is therefore necessary, using statistics, to show that those cancer patients who were trealcd with surgery lived longer than those who were not. If this had been done 100 years ago, before surgery became an ‘accepted’ treatment for cancer, by comparing the survival of two randomly selected groups, one treated and the other untreated, there would not bc a dispute now about the efficacy of surgery. But this has never been done (3, 10, 11). Instead over the past 65 years attempts have been made to prove that surgery is effective in extending survival using, for the untreated control group, patients who remained untreated for a variety of reasons. These were quite unrepresentative of the total group. So it would not be surprising to find that retrospective analysis of these early studies has shown them to be quite invalid as an objective measure of the efficacy of surgery (3, 10).
the basis of treatment of cancer for the past 100
years.) Since proper randomised clinical trials have never been carried out to prove the efficacy of surgical treatment of cancer, and such trials are now considered unethical, and are politically impossible, what other scientific methods can be used to throw some light on the impact of surgery on cancer? One answer has been-provided by a small group of medical scientists who have looked at the statistics of disease in general rather than confine their attention to cancer in particular The graphical method
In 1825 B. Gompertz (12) established that, as a person ages from birth to death, the age-specific mortality rate doubles about every 8.5 years. A typical curve for humans is one with a formula such as y = eu.u*x where x is the chronological age in years (Fig. 1). He described this relationship as a uniform logarithmic worsening of the tendency to die with increasing age. When plotted on semi-logarithmic graph paper this gives a straight line sloping up to the right (Fig. 2). Such a curve is sometimes referred to as representing a ‘constant’ mortality rate. In this context there are fundamentally three types of mortality increasing with the duration of the disease; a straight line, representing constant mortality, unaffected by the duration of the disease; and concave downwards, representing decreasing mortality with duration of the disease (4) (Fig. 3). 1500r
t
The problem of ethics
In order to carry out a randomiscd trial now one would have to withhold from a randomly selected group of cancer patients an ‘accepted’ technique which, it is argued, might save their lives. So these patients would have to be ‘sacrificed’ to prove the efficacy of surgery, a measure which is considered unethical. (Although this reasoning is quite illogical and contrary to scientific principles it is sufficiently widely accepted to make such a trial politically impossible. The mere undertaking of such a trial would be an admission that an unproven method had been
I
0
IO
20
30
40 AGE
50
60
I
I
I
70
80
90
YEARS
Fig. 1 The variation in Mortality Rate with Age for humans. Gcmpertz in 1825 described this relationship as a uniform logarithmic worsening of the tendency to die with increasing age. The formula y = eO.08~ is only approximate. When plotted on semi-logarithmic graph paper this becomes a straight line sloping up to the right and is sometimes referred to as representing a ‘constant’ mortality rate.
131
THE EFFICACY OF SURGICAL TREATMENT OF CANCER
CHRONOLOGICAL
AGE,
AGE (on linear scale)
YEARS
Fig. 2 Sub-populations of people with specific degenerative disease have a constant-slope mortality rate curve, their mortality doubling about every 8.5 years (the same as healthy people). Those with a particular type of cancer have a characteristically high mortality rate. The line is displaced in time as if they had aged by 10-15 years.
‘Commonsense’ among medical workers would lead to an expectation that most diseases will follow either the first or third type of curvature. For example it appears from conventional wisdom that disseminated cancer is a disease whose mortality rate increases with duration of the disease, i.e. the patient’s health deteriorates with time; whereas myocardial infarction (a heart attack) is associated with a mortality rate which decreases with time, i.e. the patients get over their disease. In facl these suppositions are both incorrect, as Jones (3) had noted in 1956 and Zumoff and his colleagues confirmed 10 years later (4). Jones re-analysed the statistics on mortality rates for different diseases from studies throughout the world, including those on cancer from the previous 30 years. He found that for chronic diseases, including both heart disease and cancer, the sub-populations of people with these diseases lie on a constant slope mortality rate curve. Contrary to conventional wisdom, they neither get over their disease (as is assumed with heart disease) nor get worse with time (as is assumed with cancer). Instead people with particular types of cancer have a characteristically high mortality rate; but the apparently increasing mortality rate with cancer is due simply to increasing age, the rate continuing to double ever 8.5 years. He found the same for
Fig. 3 There are fundamentally three types of mortality curves. ‘Conventional wisdom’ says that cancer shows increasing mortality and heart disease patients get over their disease after a heart attack. Both these suppositions are incorrect. Both diseases are associated with a constant slope mortality rate.
patients with tuberculosis, cirrhosis of the liver and cerebral vascular disease. Each disease had a different age-specific mortality rate line but all lines had the same slope. Using the same graphical technique, Zumoff confirmed Jones’ findings for cirrhosis of the liver, disseminated breast cancer, chronic lymphatic leukemia and myocardial infarction (4). Like Jones he concluded that such diseases have in common an alteration of undefined physiological systems that influence susceptibility to aging and dying, producing an elevated and constant increase in this susceptibility. The power mtiel Some authors such as Armitage and Do11(13) noted that the shape of the cancer mortality curve differed markedly from that of other chronic diseases and they developed a ‘power’ model of carcinogencsis to cxplain the curve’s shape. However Zajicek (14) concluded that the gradual departure from the constant slope curve observed with some cancers after the age of 50 years could be due to inaccurate diagnosis. Autopsies on pcoplc over 65 years have shown cancer incidence of about 40%, which is much higher than
132 is seen in mortality reports. This shows that cancer often co-exists with other fatal diseases and is therefore likely to be inaccurately diagnosed. By correcting for this factor Zajicek was able to show that cancer does in fact have the same constant slope curve as the other degenerative diseases mentioned. One of the features of this graphical technique is that it is sensitive to the effects of any therapy which can affect the mortality rate. Reduction in the mortality rate would result in patients dropping onto a lower line. In this way Jones was able to demonstrate that, after correcting for poor methodology, all patients with a particular form of cancer experienced the same mortality whether treated or not, once the cancer was of established malignancy. Further confirmation of this phenomenon has since come from Green (5). More recently Fox (15) applied the graphical method to breast cancer. He took data that Bloom (16) had reported on the survival of untreated breast cancer patients at Middlesex Hospital at the turn of the century. This group of patients had received no surgery, radiation or chemotherapy. The death rate for this sub-population of cancer patients is 25% per year with few survivors after 7 or 8 years. Fox then took survival data of patients treated by surgery, radiation and chemotherapy during the period 1950-1973, and plotted it in the same manner as before. This is shown as the black dots in Figure 4. He found that the total population could be broken down into two sub-populations each with a completely different prognosis. One group, representing 40% of the total, had a poor prognosis and, like the untreated sub-population from Middlesex Hospital, few had survived 7 or 8 years. These are shown as the open circles in Figure 4. The remaining 60% exhibited a survival characteristic only slightly lower than that of women of similar ages without evidence of cancer. These are shown by the other line in Figure 4. The conclusion that has been drawn from this data is that breast cancer can be cured if detected early enough. (The Middlesex Hospital group would have consisted mainly of advanccd cases.) However a second possible conclusion is that there arc two groups of patients with breast cancer: 40% with a suppressed immune system for whom the cancer grows and spreads rapidly; and 60% whose immune defences are still capable of keeping the cancer under control. If the latter conclusion is valid it is possible that treatment had no effect in either of the groups. Thus the Graphical Method does not support the claim that surgery has an impact on the course of cancer. Rather it is consistent with the claim that surgery is ineffective.
MEDICAL HYPOTHESES
100
I
I
I
I
I
90 80 70 60 50 z? m.? ? 2 .?Q)
40
30
I B
20
10
\ Year After Diagnosis
Fig. 4 Survival data of breast cancer patients treated by surgery, radiation and chemotherapy from 1950-1973. It can be considered as two sub-populations: one (lower left) identical to that of untreated patients with advanced disease at the turn of the century, and the aher (upper right) with survival only slightly less than that of healthy wanen of similar ages. From this data it is possible to conclude either that surgev reduces mortality if the cancer is detected early (the normal conclusion) or that there are women with two different levels of immune status: a) low, unable to control the disease and b) normal, able to keep the disease under control. If the latter conclusion is correct it is possible that treatment had no effect.
Comparative
studies
Another useful source of evidence for the efficacy of surgery is the comparative study. By comparing the mortality of different groups of cancer patients who have undergone different degrees of surgery it is possible to evaluate the efficacy of surgery itself. According to the rationale for cancer surgery it is important to remove all malignant cells; otherwise the remaining cells will continue to grow. Therefore the more extensive the surgery, the less the likelihood of any malignant cells remaining and the lower the mortality rate should be. If this rationale is invalid,
THE EFFICACY
OF SURGICAL
TREATMENT
133
OF CANCER
and tumours are only local, late-stage symptoms of a systemic disease, there would be no difference in mortality between groups receiving different degrees of excision. The comparative trial avoids the ethical problem because it does not require the withholding of an ‘accepted’ treatment in order to have an untreated control group. In recent years several randomized clinical trials have been carried out to compare the survival of breast cancer patients after different degrees of excision for tumours up to 4 cm in diameter. There was no difference in survival between radical mastectomy, total (simple) mastectomy, quadrantectomy, segmental mastectomy (lumpectomy) and excisional biopsy (17-20). These results show the relevance of the comment made in 1963 by Shimkin that ‘when many forms of treatment appear to yield the same results or lack thereof suspicion should arise that none is really effective, and a no-treatment group in subsequent comparisons may be acceptable’ (10). Or as Lewison commented in the same year in relation to breast cancer surgery: ‘In recording our surgical triumphs are we merely measuring the natural history of this malignancy?’ (21). That is, the results of these comparative trials suggest that surgery has no impact on the course of cancer. Claims of increased survival rates If, as the above evidence suggests, surgery is ineffective, what is the explanation for the apparent improvement in the percentage 5-year survival rates for all cancer sites between 1960 and 1975 as claimed by the American Cancer Society (9)? Iwo possible explanations have been offered: 1. These figures are unreliable for reasons of poor methodology. One main example of poor methodology is that increased 5-year survival can result from death happening later (real progress) or by making an earlier diagnosis, as has happened in more recent times. (There is no progress here. Death still occurs at the same time but the existence of the cancer has been known for a longer time, leading to an ‘apparent’ increase in survival.) As Pete and Easton stated in relation lo the effect of earlier diagnosis on staging, ‘Stage III in 1990 is therefore not the same thing as Stage III in 1970, and has a better prognosis irrespective of the effects of therapy’ (22). Other factors include comparison between unmatched groups. For reasons such as these EnStrom and Austin concluded that ‘survival rates should not be used as a sole or primary measure of progress in cancer control because factors un-
related to the efficacy of treatment play an important role in the determination of those rates and their trends’ (23). 2. Earlier figures with lower survival applied when more aggressive treatments were being used and were reducing survival (24). Whatever the reason, survival figures arc unreliable as a measure of the efficacy of surgery as a treatment for cancer.
Comparison of incidence and mortality Enstrom and Austin concluded that a better measure of progress in cancer treatment is to compare the incidence of each type of cancer with the mortality rate over the time interval in question. For, so long as incidence and mortality remain unchanged, or change proportionately, no genuine change in survival can occur. Progress in cancer control requires that the mortality rate dcclinc mom rapidly or rise more slowly than the incidence for the particular type of cancer (23), (see Fig. 5). If incidence is compared with mortality over the period from 1950 to 1970 it is found that there have been large changes in incidence and mortality over this time with several types of cancer (see Table). For most types the incidence and mortality have changed by the same amount within the limits of uncertainty of the statistics. Only in the case of prostate cancer can some claim be made for improved survival due to treatment. Yet even here there is another possible explanation. Most prostatic cancer is ‘occult’ and relatively benign in its course. The increasing rate of autopsies and operative procedures for other causes makes for an increase in the number of occult tumours found, leading to an increasing recorded incidence. This increasing rate of discovery of occult prostatic cancer has caused the stage distribution to change (23, ref 28 on p851), i.e. there is a higher proportion of early stage tumours in the more recent samples than in those recorded 20 years previously, one of the factors rcfcrred to above in relation to poor methodology which has resulted in an inflated value of the percentage 5-year survival rate. The apparent survival rate between 1950 and 1970 closely follows the increased incidence of this type of cancer. As Cairns (25) has pointed out, the incidence of prostatic cancer discovered by autopsies on 70-year old men who had died of other causes was found to be up to 100 times as great as the incidence based on diagnosis of such patients presenting with symptoms. So the incidence figures for these slow-growing tumours are unreliable for the purpose of comparison with mortality rates as a means of assessing progress in cancer control.
134
MEDICAL IlYPOTHESES
Fig. 5 So long as incidence and mortality remain unchanged or change proportionately, no genuine change in survival can occur. The apparent increase in the percentage 5-year survival rate is due to factors unrelated to the efficacy of treatment, such as the increasing proportion of localised cases in the more recent samples. This graph (taken from a paper by G. Linden) is used by Enstrom and Austin to &monstrate this effect.
From a comparison of the incidence and mortality rates over this 20-year period there are therefore no clear cases where survival could have improved as a result of surgery. This is further evidence that surgery has not had a proven impact on the course of cancer. In April 1987 the US General Accounting Office issued a report in which it stated that gains in treating cancer over the last 3 decades have been small and overstated by Federal health officials. For a majority of the most common tumours there was little or no improvement from 1950 to 1982 in the rate at which patients survived their disease (26). Epidemiological studies Another useful type of information for testing the efficacy of surgery is the epidemiological tables of mortality which enable international mortality trends to be compared and correlated with social changes and the introduction of new diagnostic techniques. There is some evidence from this source that surgery may be effective in invasive cervical turnours. For example, the decline in mortality from cervical cancer which has been observed in many countries throughout the world is claimed to have accelerated in those countries at the time PAP smear programs were intro-
duced, when it can be assumed that more women were diagnosed with, and therefore treated for, invasive cervical cancer. However there are counter claims that this decline occurred prior to the introduction of the PAP smear (27) (see Fig. 6). Nothing that has been said is intended to imply that surgery has no place in cancer treatment; only that, with few exceptional conditions, there is no reliable evidence that it extends life. Surgery can of course extend the life of a cancer patient by removing a lifethreatening obstruction in a vital organ or pressure on the brain. Surgery is also useful for reducing pain, removal of a source of toxins from the body, removal of unsightly growths and reducing the load on the immune system. But these benefits can be achieved without affecting the disease itself. An increasing number of cancer specialists, particularly surgeons, are recognizing that surgery has limited effect on cancer. For example, in 1970 Sir John Bruce, Regius Professor of Clinical Surgery at Edinburgh University and past President of the Royal College of Surgeons, stated that ‘the future (of cancer treatment) lies elscwhcre than in the operating room; but when the answer is eventually found, the surgeon will have no cause to be ashamed of his attempts to relieve suffering and not infrequently avert the arrows
THE EFFICACY
OF SURGICAL
TREATMENT
135
OF CANCER
Table There have heen large changes in incidence and mortality between 1950 and 1970. In most types of cancer the incidence and mortality have changed by the same amount, within limits of uncertainty. So there cannot have been an improvement in survival rates over this period for these cancer types. The apparent improvement is due to poor methodology. For prostate cancer the apparent improvement is due to increasing discovery of prostate cancers during operations and autopsies Site
Death rate/lo0 1950
Lung Breast Colon Rectum Prostate ovary Pancreas uterus,
1970
Change %
corpus Stomach Bladder Esophagus Kidney All sites Ref. source
CKKI
Percentage
1950
1970
Change %
1950-54
1965-69
5.year survival rate
Change %
1960-63
1973-80
Change %
8s 63 41
12a 74 50
+5Oa +17 +22
50 32
68 37
+36 +16
58
68
+17
12 13 15.5
30 13.5 15
+150 +4 -3
17 38 25
36 39 27
+112 +3 +8
7 59 42
9 64 44
+29 +8 +5
7 7.5 4 6 5.5
4.5 7 4.5 7.5 3
-36 -7 +13 t2.5 45
17 17 8 7
12 21 7 8
-29 +24 -13 t14
41 44 28 1 57
40 56 32 1 56
-2 I +27 +14 0 -1
32
20
-38 71
74
+4
73
88
+21
tY
11 53 4 37
14 73 5 50
+27 +38 +25 +35
cervixh Uterus,
Incidence/100
OOU
4.5 15 4 2.5 2.5 135
2
-56
6 3.5 2.5 3
-60 -13 0 +20
24 13 4 5
9 12 3 6
-63 -8 -25 +20
11 54 4 32
12 60 3 41
+I1 -25 +28
+4
290
275
-5
39
39
0
140 (23)
(23)
(23)
Notes: aThese figures for lung include bronchus, bfigures for uterine cervix exclude Leukemia have heen omitted as surgery is not applicable.
of death in one of the greatest scourges of mankind’ (6). In 1973, in a collection of essays celebrating the 75th anniversary of the Rosewell Park Memorial Institute, Thomas L. Dao of the Institute’s Department of Breast Cancer wrote: ‘Despite improved surgical techniques, advanced methods on radiotherapies, and widespread use of chemotherapies, breast cancer mortality has not changed in the last 70 years’ (28). In 1974, addressing a cancer conference at Sydney University, Miles Little, Professor of Surgery, stated that: Despite refinements in surgical technique and management, and increasingly radical surgery, there is comiderable doubt about fhe impact of surgery on the natural history of most malignancies. The apparently logical hypothesis that earlier diagnosis and more radical excision would lead to more cures, has not been borne out in practice. Surgery brings a mechanical approach to a biological problem (7).
In 1986 Bailar and Smith pointed out that the death rate from all cancers other than lung, stomach and cervix (sites which have shown marked changes in in-
‘in
(9) situ’,
Cligures for Hodgkin’s
Disease and
cidence in recent decades) had not altcrcd appreciably since 1950 (8). These arc implied admissions by the more honest medical scientists that there has never been any reliable scientific evidence to show that surgery affects the course of cancer. It became an ‘accepted’ method for cancer treatment but was never a ‘proven’ method. Peto and Easton state that in the many situations where it is not known whether treatment is effective: many clinicians
respond by developing a set of fwmly held but unsupported beliefs in the merits of particular regimens. The primary treatment of advanced nonmetaslatic laryngeal cancer, for example, will usually be by surgery at certain treatment centres and by radiotherapy at others. Whether chemotherapy is given as well and, if so. what form it will take, are also determined more by the idiosyncrasies and outpatient arrangements of the particular treatment centre than by objective evidence of long-term efficacy. Similar examples could be taken from most areas of cancer therapy (22).
Stephanie Short, a sociologist, has carried out a study of cancer treatment in Australia and discovered that if a process is technically feasible, and cancer
136
MEDICAL HYPOTHESES
Death
$
0;
cl
1930
Rate
for
Cancer of
the Uterus,
I
U.S.
I
7
I
1935
1940
1945
1950
1955
1960
1965
19’10
1930-1979
1975
1980
YEAR Fig. 6 Earlier surgical treatment of uterine cancer following the widespread introduction of the PAP test has been claimed to be the cause of the observed fall in the mortality rate. However there have been other claims that the fall began before the introdcmion of the PAP smear. Source: Cancer Facts and Figures, 1984, American Cancer Society.
patientsexpectsomethingto be done for their disease, it will be done, even if there is no proven efficacy for the treatment. She described this as the ‘technological imperative’ (29). This growing realisation of the inefficacy of surgery in extending life should lead to a reappraisal of the underlying assumptions of the nature of the disease. For if tumours are in fact only local, late-stage symptoms or elements of a systemic disease, removal of such symptoms would not be expected to have any impact on the natural history of the disease. Comparison of survival after treatments based on different hypotheses There is growing evidence from a wide variety of sources that cancer is a systemic disease, even when first diagnosed. For example in the comparative trial of breast cancer in the US mentioned above (19), of those patients who had received radiation in addition to segmental mastectomy, about 90% were free of tumours (‘disease free’) after 5 years compared to about 70% of those who had received no radiation.
Yet these two groups showed the same survival. This shows that the presence or absence of tumours after surgery is not a reliable measure of the existence of the disease. (Hence the definition of ‘effective’ used at the beginning of this paper.) A second source of evidence is survival statistics. The best 5-year survival statistics have been produced using therapies based on the hypothesis that cancer is a systemic disease, and tumours are ony local symptoms. Therefore the cancer patient should be treated using a therapy designed to restore the body’s own natural immune system. After undergoing such a therapy tcrminal cancer patients showed a 16.6% 5-year survival compared with about 2% expected with conventional therapies and a 15% 15-year survival compared with less than 1%. With pre-terminal patients there was an 85% 5-year survival compared with about 50% (30). These results prompted John Anderson, Professor of Physical Medicine at King’s College Hospital in London to state that ‘the Isscls approach to the treatment of cancer is an unique and pioneering solution to a very difficult problem. He is undoubtedly producing
137
THE EFFICACY OF SURGICAL TREATMFiNT OF CANCER
clinical remission in patients who have been regarded as hopeless and left to fall back on their own resources’ (3 1). Similarly a therapy designed to augment the body’s own immune system is reported to have produced 50% 5-year survival with 11 cases of peritoneal mesothelioma, a malignancy with an expected prognosis of about 12 months (32). Conclusions In summary, surgery became an ‘accepted’ treatment for cancer as a result of several factors. Because of the ethical problems of providing a proper control group of untreated patients, no scientifically acceptable trials have ever been carried out to prove that surgery is ‘effective’ in extending life. A graphical method used for analysing mortality rates suggests that there is no difference in survival between treated and untreated patients. Comparative studies of surgical techniques using differing degrees of excision show no difference in survival. This would be expected if tumours were only symptoms of a systemic disease. Percentage 5-year survival statistics offered to show that surgery is an increasingly effective method of cancer treatment are unreliable because of several methodological problems. The difference between incidence and mortality, which is a more reliable measure of the effects of surgery on cancer control, suggests that no improvement in survival has occurred as a result of surgery. The best, properly documented results of survival with terminal cancer patients have been achieved using therapies based on the hypothesis that cancer is a systemic disease and therefore should be treated by methods designed to restore the body’s own natural immune system. Surgery has therefore not been shown to be an effective or proven method for the treatment of cancer. These findings add weight to the hypothesis that cancer is a systemic disease when first diagnosed.
2.
3. 4. 5.
6. 7.
8.
9. 10. 11. 12.
13. 14. 1.5. 16.
17.
18.
19.
20.
Acknowledgements 21. I would like to acknowledge the encouragement and help given by Alec Forbes, former Consultant Physician at Plvmouth General Hospital, Advisor to the WHO on Traditional Medicine. and cofounder and original Medical Director of the Cancer Help Centre in Bristol, UK. I would also like to acknowledge the help of John McAllan. a CSIRO colleague who has since died of mesothelioma.
22.
23. 24.
References 1. Issels J. The Classic Methods, pp 32-36 in Cancer, A Second
25. 26.
Opinion. Hodder & Stoughton, London, 1975. Moss R W. Surgery pp37Xl. and The Crisis of Credibility ~~22-23 in The Cancer Syndrome. Grove Press, New York, 1980. Jones H B. Demographic Considerations of the Cancer Problem. Trans N Y Acad Sci 18 (4): 298, 1956. Zumoff B, Hart H, Hehnan L. Considerations of Mortality in Certain Chronic Diseases. Annals Intern Med 64: 595, 1966. Green G II. Duration of Symptoms and Survival Rates for Invasive Cervical Cancer, Aust & N I, Journal of Ohstet and Gynec 10: 238, 1970. BNCC Sir J. Quoted by Issels J (1) at ~36. Little M. Quoted by Jacka J in Statistics relating to cancer, pl in Cancer-A Physical and Psychic Profile. Currency, Melbourne, 1977. Bailar J C, Smith E M. Progress against cancer? N Engl J Med 314: 1226, 1986 and following correspondence on June 5 and July 15, 1986. American Cancer Society. Cancer Facts & Figures 1984. American Cancer Society, New York, 1983. Shimkin M B. The Numerical Method in I’herapeutic Medicine, Public Iicalth Reports 79 (1): 5, 1964. Phillips S. Dr Joseph Issels and the Ringberg Klinik. Clinical Trials Journal, 35.5, August 1970. Gompertz B. On the nature of the function expressive of the law of human mortality, and on a new mode of determining value of life contingencies. Phil Trans Roy Sot 115: 513, 1825. hrmitage P, Doll K. The age distribution of cancer and a multistage theory of carcinogenesis. Brit J Cant 8: 1, 1954. hjicck G. Cancer as a Systemic Disease. Medical Hypotheses 4 (3): 193, 1087. Fox M S. On the Diagnosis and Treatment of Breast Cancer. JAMA 241 (5): (Feh 2). 1979. Bloom 11 J G. Survival of women with untrcatcd breast cancer+st and present. pp 3-19 in Prognosis Factors in Breast Cancer (A P M Forrest and P B Kunkler, eds) II & S Livingstone Ltd. Edinburgh 1968. Fisher B, Redmond C. Fisher E et al. Ten-year results of randomired clinical trial comparing radical mastectcmy and total mastectomy with or without radiation. N Engi J Med 312 (11): 614, 1985. Vcroncsi U, Saccozzi R. Del Vccchio M ct al. Comparing Radical Mastectomy with Quadrantectomy, Axillary Dissection and Radiotherapy in patients with small cancers of the hrcast. N Engl J Med 305 (1): 6, 1981. Fisher B, Bauer M, Margoiese R et al. Five-year results of a randomimd clinical trial comparing Total Mastectomy and Scgmcntal LMastcctomy with or without radiation in the trcatment of Breast Cancer. N Engl J Med 312 (11): 665, 1985. Sarrazin I,. IL M. Rouesse J et al. Conservative Treatment versus Mastectomy in Breast Cancer Tumors with Macroscopic Diamctcr of 20 Millimeters or Less. Cancer 53 (5): 1200-1213, 1984. lewison E F. An Appraisal of Long-Term Results in Surgical Treatment of Breast Cancer. JAMA 186 (11): 975-978, 1963. Peto J, liaston D, Cancer treatment trials-past failures, current progress and future prospects. Cancer Surveys 8: 513-533, 1989. Enstrom J Ii, Austin D I- Interpreting Cancer Survival Rates. Science 195: 847-851, 1977. Greenhcrg D S. A critical look at cancer coverage. Columbia Journalism Rcvicw. 4044 Jan&b, 1975. Cairns J. The Cancer Problem. Sci Am 233 (5): 64-78, 1975. US Congress General Accounting Office. Cancer patients survival: what progress has hcen made? PEMD-87-13, (3/31/87).
138 27. 28.
29. 30.
MEDICAL HYPOTHESES
Editorial. Uncertainties of cervical cytology. Br Med J 5891: 501-502, 1973. Dao T L. Quoted by Greenberg D S, in ‘Progress’ in Cancer Research-Don’t Say It Isn’t So. N Engl J Med 292 (13): 707-708, 1975. Short S. The War Against Cancer. A sociological study of cancer treatment. New Doctor 35: 25-28, 1985. Issels J. Immunotherapy in Progressive Metastatic Cancer-A
31.
32.
Fifteen-Year Follow-Up. Clinical Trials Journal, 357-365, August 1970. Anderson J. Quoted by Thomas G. in Issels--The biography of a doctor. London, Hodder and Stoughton, (in Introduction, ~16). 1974. Clement R J, Button L. Lampe G N. Peritoneal Mesotltelioma, Quantum Medicine 1 (1 & 2): 6&73, 1988.