The scientific facts and the unfortunate experiment
The Cervical Cancer Inquiry of 1987–88 was commissioned by the Minister of Health. In summarizing Professor Linda Bryder’s critique of the Inquiry and its subsequent Report in the Listener (August 15–21), Joanne Black did not make any independent evaluation of the claims made, nor seek alternative views. We are writing now, as two of the specialist advisers to the Inquiry, to provide readers with a fuller account of the medical and scientific matters covered by the above article and Linda Bryder’s book (A History of the “Unfortunate Experiment” at National Women’s Hospital).
This response is focused on a sample of the misunderstandings and mistakes evident in Dr Bryder’s book, some of which were repeated in the Listener article. To assist readers, we have also suggested to the Ministry of Health that the Cartwright Report (as it is known) is placed on the Ministry’s website.
The Inquiry was led by Judge Silvia Cartwright. She had two legal counsel assisting and three medical advisors: Eric McKay, a professor of obstetrics and gynaecology from Queensland, and ourselves, a pathologist and a medical epidemiologist. There were four months of public hearings. Seven gynaecologists from overseas gave evidence and reviewed patient files. Not one of these gynaecologists, including the one who was called by the counsel for Dr Herbert Green and Professor Denis Bonham, agreed with Green’s management of patients with carcinoma in situ during the time of his clinical study.
Carcinoma in situ (CIS) of the cervix is an abnormality in which the cells look like cancer but only the surface layer is affected. There is no invasion of these cells into the underlying tissue (stroma). Treatment of this condition has always meant cutting out (excising) or otherwise destroying this abnormal tissue. Conservative treatment aims to do this while conserving the uterus. Adequate treatment is aimed at removing the abnormal area, with follow-up and further treatment if there is a persistent or recurrent abnormality, as identified by abnormal smears.
Dr Bryder has taken a selective approach in her use of evidence. The consequence is a series of regrettable (and confusing) misinterpretations.
First, Bryder claims that Dr Green did not conduct an experiment. Judge Cartwright considered this question and concluded that he did. He had a research aim, he withheld conventional treatment from a group of patients with CIS, and he systematically recorded the outcome in a series of publications. To do an experiment in itself is nothing terrible: the most powerful form, the randomized controlled trial, has been exceptionally useful and can be conducted in an ethical manner. This was not, however, what happened at National Women’s Hospital, over the period considered by the Inquiry.
Second, Bryder seems to accept there was a research aim, but not that it was: “To attempt to prove that carcinoma in situ is not a premalignant disease”. That aim was recorded in the Minutes of the Senior Medical Staff meeting at National Women’s Hospital in 1966, at which the study was approved (on the basis of a proposal from Green). At the Inquiry, Green made the claim that the word ‘invariably’ had been accidentally omitted from the minuted description of the research aim. The Report (p.33) considers this suggestion, and rejects it. One of the experts pointed out that the addition of ‘invariably’ made the proposal worse: “in fact to prove that it does not invariably progress, one would have to go on until every single women developed invasive cancer”.
Third, the Report concluded that some women had not received adequate treatment during the period they were in the care of Dr Green. This judgment was based on evidence from expert gynaecologists from around the world who reviewed the patients’ files. It was not based on an interpretation of the 1984 paper by McIndoe and colleagues, as Bryder seems to believe. As noted above, all the gynaecologists concluded that the treatment was inadequate. Professor Per Kolstad, the eminent Norwegian gynaecologist, indeed concluded that some women suffered “severe” and “terrifying mismanagement”.
In one specific instance, Bryder claims that the patient, Clare Matheson, was discharged from National Women’s Hospital following “five negative smears” and “one equivocal report of probably carcinoma in situ in the four years preceding her discharge”. This is wrong. At the time of Clare Matheson’s discharge from the hospital, she had an unequivocal histological report of “fragments of carcinoma devoid of underlying stroma, probably carcinoma in situ” on a tissue specimen. The pathologist was certain about the presence of carcinoma; he was equivocating between CIS and invasive cervical cancer; it was not possible to tell, because there was no “stroma” (underlying tissue) to show invasion into. What is more, multiple biopsies had left her cervix scarred and narrowed, affecting the reliability of the smear test. Amazingly, Dr Green wrote in the discharge letter to the GP: “she has no more chance than the next person of developing any carcinoma of the cervix”. Unsurprisingly, she went on to develop invasive cancer of the cervix and required even more extensive treatment.
Fourth, Bryder does not understand the 1984 paper. She seems to think it somehow underpins the Cartwright Report, but it needs to be restated that Judge Cartwright’s conclusion that women had had inadequate treatment did not rely on any information in this paper by McIndoe and colleagues. In particular, Bryder is confused about the question of ‘two groups’ of patients described in the 1984 paper. It needs to be stated clearly that groups of patients were also identified by Green in his research publications.
Dr Green’s 1966 research proposal described selecting groups of patients to be given no further treatment after a small diagnostic biopsy of CIS, or after positive smear test results following treatment. In his subsequent papers these patients were referred to as one or more “special series” of cases and as “untreated or incompletely treated”. In his last paper, in 1974, he described “following of 96 of them with positive cytology for at least two years….”
In the 1984 paper, McIndoe and his co-authors reported on their review of the files of patients with CIS first diagnosed at the Hospital between 1955 and 1976. In order to investigate the natural history of CIS by comparing patients with no evidence of persisting disease (negative smears) to those with evidence of persisting disease (positive smears), they identified two groups. “Group 1” had negative smears at two years after diagnosis. “Group 2” were those with positive smears at two years after diagnosis. Group 2 is basically a retrospective reconstruction of Dr Green’s group described above: “96 of them with positive cytology for at least two years”. The 96 were those diagnosed up to 1971. The McIndoe paper included women diagnosed up to 1976 and included 131 women. This extension of the period by five years explains the difference in number of cases.
Bryder also claims that “there was no difference in treatment between group 1 and group 2”. This is obviously not the case. Adequate treatment is not just about the type of surgery. Adequacy depends on removing the lesion; it must be gauged by the examination of the tissue removed (to check whether disease has been left at the margins) and by follow-up smears. Positive smears indicate continuing disease. In the normal course of events, a positive smear would be an indication for further surgical excision. The two groups had different profiles in these key respects: the timing of surgical excision; the proportion of biopsies which showed “incomplete excision”; and smear results. Bryder’s claim that “there was no difference in treatment” between groups 1 and 2 also leaves her with no explanation for the fact that the rate at which cancer appeared in group 2 was 25 times higher than in group 1. As Judge Cartwright concluded, the group with ‘limited treatment’ (i.e. group 2) did much worse.
Fifth, Bryder asserts that Green made rational judgments aimed at protecting his patients from unnecessary surgery. On the contrary, the women in Green’s study were subjected to repeated biopsies to check for invasive cancer. The biopsies were intended to leave some of the abnormal area so that the disease process could continue to be observed. The consequence of this approach was that many women had more biopsies than if they had had conventional treatment. The Report states (p.42): “A review of the patients’ notes demonstrates that many women with continuing positive cytology have had more than one cone or ring biopsy and in some cases up to six…” And “In some patients the practice of multiple cone biopsies or conizations resulted in unpleasant and painful side effects.”
This means the women suffered in two ways from having conventional treatment withheld. Many had multiple biopsies with the attendant repeat general anaesthetics and painful side effects. Many also developed invasive cancer: this required even more extensive surgery plus radiation, and put them at risk of death.
Sixth, Bryder misunderstands the results of the re-examination of the natural history of CIS, drawing on material from National Women’s Hospital during 1955 to 1976. This was undertaken by Margaret McCredie and colleagues (including Charlotte Paul), and published in 2008 in Lancet Oncology (9:425-34). One of the tables in the Lancet Oncology paper shows the direct effect of Green’s clinical study on the 143 women who received only a small diagnostic biopsy (as opposed to treatment by a cone biopsy) within six months of diagnosis. Over the next 30 years, 34 of these women went on to develop invasive cancer. Even within five years, 11 percent had developed cancer, compared to 1 percent among women who had received a cone biopsy initially. The implication of these figures should be obvious.
The conclusion of the Lancet Oncology paper was that 31 percent of women who received only a small diagnostic biopsy and no subsequent adequate treatment developed cancer after 30 years. Bryder says about this finding: “while an aggressive approach to CIS could be shown after 30 years to be a better approach than the more conservative one, this knowledge was not available at the time” (History, p43).
This statement cuts to the heart of the difficulties that Bryder has in her interpretation of scientific material. First, the women in this group received no treatment, not conservative treatment. Second, there was already evidence in 1966 that a substantial proportion of cases would progress to invasion if untreated. For instance, in 1961 Professor Kolstad compiled a table titled “Development of invasive carcinoma in untreated cases of CIS”: the average proportion of cases progressing was 28 percent over a range of 1-23 years of observation. This table appears on p.23 of the Report of the Cervical Cancer Inquiry. Third, Bryder assumes that the outcomes of Green’s study at National Women’s Hospital were only available last year (2008)! As the data given in the Lancet Oncology paper show, the differences between the initially untreated and the treated women should have been obvious to Green well within five years of the beginning of his study (1966).
It is a telling fact that in Dr Green’s publications up to 1970 he reported that none of his ‘special series’ had progressed to invasion. However, as is shown in the Report (p.54-55), there were cases originally included in Green’s special series which progressed to invasive cancer, but Green reclassified these cases and removed them from his series. Paul Gerber with Malcolm Coppleson, the gynaecologist who pioneered colposcopy in Australia, wrote about this in 1989 in the Medical Journal of Australia. They commented on his “highly unscientific approach”, in which he justified removing cases from his series by going back and changing the diagnosis. They also said: “Green should have concluded that if cases of invasion were capable of being missed at the outset, his protocol not only was unsafe in that some women would, without treatment, proceed to develop invasive cancer, but also was incapable of testing his hypothesis.” It is a signal failing that Green did not report his study accurately, and hence did not stop it when evidence of harm emerged.
Seventh, one of the most serious issues with Bryder’s book is her extensive quotation out of context. For example, on page 24 of her book, Bryder introduces Green’s 1966 proposal with a crucial section missing. The missing words are: “Including the four 1965 cases, there are at present under clinical, colposcopic, and cytological observation, 8 patients who have not had a cone or ring biopsy. All these continue to have positive smears at intervals.” The proposal continues: “It is suggested that this should be extended to include all women who…”
Bryder goes on to say “ this protocol would have met with the approval of Professor Per Kolstad”. She then quotes from a 1970 article by Kolstad: “By the complementary use of cytology, colposcopy and histopathology, the risks of overlooking invasive carcinoma of the cervix are minimal”. This sounds as if he is advocating following women without treatment, as Green did. What Bryder omits to quote from the article is: “Although we agree that it is not justified to allow a patient to go without treatment when a diagnosis of carcinoma in situ is made, we would like to plead for a conservative approach”. That approach, Kolstad explained, was conization (cone biopsy).
Similarly, on pp 12-13 of Bryder’s history, several references are made to descriptions of a conservative approach to treatment of CIS in a book by Coppleson and Reid published in 1967. The implication is that this conservative approach is like Green’s. On the contrary, the Coppleson and Reid book makes quite clear that the conservative approach is a colposcopically-directed cone biopsy. Despite Bryder’s contention to the contrary, her references provide no evidence that any authority (other than Green) at the time regarded CIS as not a cancer precursor or that no treatment was indicated. The authorities cited agreed that (a) not all CIS develops into invasive cancer, (b) conservative treatment (cone biopsy) is an appropriate treatment, (c) women who have had treatment still need careful follow-up to detect persistent or recurrent CIS, so that they can have further treatment if necessary. This reason for follow-up was different from Green’s reason – his was to wait until he detected invasive cancer.
Eighth, Bryder incorrectly calls into question the Medical Council of New Zealand’s finding of disgraceful conduct with respect to Professor Denis Bonham, head of the Postgraduate School of Obstetrics and Gynaecology at NWH (History, pp 167-8). As an example, she questions the Council’s conclusion that “the belief that patients with positive smears could be followed without risk was unjustified”. She quotes Kolstad’s advocacy of conservative treatment and follow-up, again omitting to note that Kolstad was advocating follow-up after treatment not without treatment.
Finally, the outcome of a similar experiment to that undertaken by Green, published in 1969, shows that proper consent procedures were used elsewhere at that time. From 1960 to 1966 Ralph Richart and Bruce Barron followed women with a cytological diagnosis of dysplasia (a less serious condition than CIS) without treatment in the US. In their paper Richart and Barron stated: “the patient was informed that she had a lesion which, based on medical knowledge at the time, could either regress or progress to a precancerous state. She was given the option of either being followed and basing further recommendations on her clinical course or of having a therapeutic conization”. A regime of close monitoring followed.
At the Inquiry, Professor Graham Liggins from National Women’s Hospital supported Green by likening his study to that of Richart. The difference was that Richart and Barron designed their study so that when women were found to have CIS (or in three cases invasive cancer), follow-up stopped and treatment was undertaken. Green’s study was designed to stop only at frankly invasive cancer. Moreover, Richart and Barron informed their patients of the research project, and gave them a choice as to whether to take part. Thus their patients also knew they remained at potential risk while they were being monitored, and understood why they needed to continue follow-up. This was not the case with Green’s patients.
Bryder’s book rehashes arguments that were carefully considered by the Judicial Inquiry, and re-examined by the Medical Council of New Zealand, and were found to be without merit.
Professor Linda Holloway
Bioethics Centre, Department of Medical and Surgical Sciences
Professor Charlotte Paul
Preventive and Social Medicine
Dunedin School of Medicine, University of Otago
The Cartwright Report is available online at www.nsu.govt.nz/Current-NSU-Programmes/3233.asp
—Linda Holloway and Charlotte Paul
