This post is an edited extract from my unpublished book on scientific error and fraud; it is one of many case-studies of individual fraudulent scientists from different areas of the biological sciences, psychology and medicine. In this case study, error and fraud both contributed to the belief that high dose chemotherapy was a more effective therapy for breast cancer than conventional dose chemotherapy.
Until he was formally dismissed in March 2001, Dr Werner Bezwoda was head of the Department of Haematology and Oncology at the University of Witwatersrand in Johannesburg, South Africa. He was an oncologist or cancer specialist who published fraudulent clinical trial data which suggested that use of a very aggressive chemotherapy regimen improved survival time in women with advanced metastatic breast cancer i.e. cancer which had spread to sites outside of the breast. This is a disturbing case because tens of thousands of women, particularly in the US, underwent this very expensive and distressing treatment for no discernible benefit partly as a result of his fraudulent data. To commit a fraud which gives seriously ill and desperate people false hope that an expensive, very unpleasant and dangerous treatment will help them seems a particularly callous deception.
In normal cancer chemotherapy (or radiotherapy) the dose of the treatment that can be used is limited because of the harm caused to other cells and organs of the body. These treatments are particularly damaging to normal cells which, like cancer cells, are rapidly dividing e.g. the gut lining and bone marrow. If the bone marrow is severely affected by the therapy then this will depress formation of new blood cells leading to anaemia and an inability to fight infections. The basic reasoning behind Bezwoda’s favoured treatment was to harvest bone marrow or peripheral stem cells so that these could be given back to the patient after high dose chemotherapy to replace the damaged bone marrow. This mode of treatment had been successfully used to treat certain blood cancers like leukaemia and lymphoma. The idea of using this treatment for solid tumours and particularly advanced breast cancer had been discussed for some time prior to Bezwoda’s publications and some preliminary (Phase II trials) studies had been performed. These early studies gave promising results but were difficult to interpret because they were not randomised controlled trials and so bias in the way patients were selected for the high dose treatment could explain the apparent increase in their survival time. Bezwoda published the first, and probably the only highly positive randomised controlled trial of this therapy.
High dose chemotherapy with bone marrow transplant (HDC)
HDC is the abbreviation used to describe the type of aggressive treatment for advance breast cancer favoured by Bezwoda in which high doses of multiple chemotherapeutic agents are used followed by transplants of the patient’s own previously harvested bone marrow to replace the destroyed bone marrow. In later years it became possible to use stem cells from peripheral blood rather than always needing to transplant bone marrow per se. These stem cells whether from bone marrow or peripheral blood are injected into the patient and they migrate to the bone marrow and re-establish functioning bone marrow which produces new blood cells. The abbreviation HDC will be used throughout to describe this procedure whether bone marrow transplants or stem cells were used to effect the so-called haematopoietic rescue. In this section, I have overviewed the history and development of this treatment strategy largely leaving out Bezwoda’s involvement at this stage. I think that had Bezwoda’s fraudulent data not existed, this example could still have justifiably been included in this book in the section dealing with major errors.
In so-called Phase II clinical trials, small groups of patients (say 20-30) are treated with likely therapeutic doses of the drug to see if patients respond to it and the severity of side-effects produced. If the drug has apparent anti-cancer activity then this may warrant initiating Phase III trials. In phase II trials, the outcome is often compared with the expected outcome when using the standard best treatment obtained from previous studies i.e. historical controls. In phase III trials, patients with matched disease severity (e.g. stage of cancer) are randomly allocated to receive either the new treatment or the existing (best) treatment. Good phase III trials should give a clear indication of the relative benefits and hazards (e.g. side-effects) of the two treatments. It may take many patients monitored over some years to get a clear indication of which is the better of the two treatments. This may mean that patients need to be recruited and treated at several different centres in order to get sufficient numbers. For the phase III trials, patients should be randomly allocated to the two treatment groups but because the regimens are so different double blinding would not be possible when comparing conventional and HDC treatment.
The ability to successfully transplant bone marrow in people had been clearly demonstrated by the 1960s. Deliberate or accidental exposure to lethal doses of radiation or cytotoxic drugs will destroy a patient’s bone marrow. Transplantation of bone marrow from a donor represents the only long term solution for these patients. Such transplants from unrelated donors often result in a potentially fatal problem, graft versus host disease, in which the newly grafted bone marrow recognises the recipient’s tissues as foreign and so starts to destroy them. This is analogous to the rejection of a transplanted kidney or heart by the recipient’s immune system (host versus graft disease). Both organ rejection and graft versus host disease can be minimised if there is some matching of the tissue types of the donor and recipient e.g. there is no rejection if donor and recipient are identical twins. Total body irradiation or high dose chemotherapy followed by bone marrow transplant had been shown to be an effective cure for some patients with cancers like the leukaemia and lymphomas (e.g. Hodgkin’s disease) that involve very rapidly dividing cancer cells. Some cancer specialists had advocated the potential value of this approach in also treating solid tumours like breast cancer. In theory enough chemotherapy or radiotherapy is used to destroy all of the cancer cells so that it does not recur after the treatment has finished. In conditions, like solid tumours, where the patient’s bone marrow is not diseased then the patient’s own bone marrow can be harvested prior to treatment and then re-implanted once the radiotherapy or chemotherapy is finished; there would thus be no graft versus host disease.
There were a number of potential problems and hazards with using this procedure for solid tumours like breast cancer such as those listed below.
- Breast cancer cells grow much less quickly than those that cause lymphomas and leukaemias and so it is more difficult to kill all of the cancer cells and prevent cancer recurrence. This also means that the speed of recurrence of breast cancer is likely to be slower than in leukaemia or lymphoma thus sometimes raising false hopes that the treatment had brought about a cure.
- The treatment itself has significant mortality; this was estimated at 10-20% in the early years. This was reduced later when so-called haematopoietic growth factors reduced the time it took for new bone marrow to start to function after drug treatment. These factors also made it possible to use as stem cells from peripheral blood as an alternative to harvested bone marrow.
- The treatment was very gruelling for the patient. Bone marrow was harvested by insertion of a large biopsy needle into the pelvis. There is considerable potentially life threatening damage to other tissues like lungs, heart and kidneys by the large doses of chemotherapeutic drugs used. Most patients undergoing this treatment would, for example, experience severe nausea, vomiting, diarrhoea, and mucositis a painful condition where the mucous membranes lining the mouth and gastrointestinal tract become inflamed and ulcerated making it impossible to eat.
- After treatment and destruction of existing bone marrow there is a period before the newly transplanted bone marrow has become established when the patient is very susceptible to infection and is at risk of dying from an uncontrollable infection.
- The treatment was very expensive because of: the high cost of the drugs, the need to keep patients isolated and not exposed to infection (barrier nursed) whilst the new bone marrow became established etc. In 1995 it was estimated that the cost of standard therapy for stage IV (advanced) breast cancer in the USA was $15,000-$40,000 whereas the HDC cost $70,000-150,000.
Some early phase II trials suggested considerable improvements in survival times for HDC treated patients compared to that expected for conventionally treated advanced breast cancer patients; some recorded 2 year disease-free survival rates of 20% for HDC patients with advanced breast cancer compared to just 2% for those treated conventionally. These results were certainly promising enough to justify conducting phase III trials but the methodology of phase II trials was necessarily inadequate to show conclusively that HDC was a better treatment that should be used routinely outside of clinical trials. One of the most important criticisms of these phase II trials was that they did not use a representative sample of advanced breast cancer patients and that the results were thus biased towards a positive outcome for the HDC. In order to be selected for a phase II trials women had to satisfy certain selection criteria e.g. they had to be healthy enough to withstand the gruelling treatment and they had to have shown evidence that they had responded well to chemotherapy in the past. In 1997, a group at the University of Texas in Houston (Rahman et al, 1997) set out to test whether this selection of patients considered suitable for inclusion in a phase II HDC trial was biasing the results. They identified over 1500 patients with advanced breast cancer who were being treated with a standard protocol for chemotherapy. They then classified these patients into two groups; those who met the general eligibility criteria for entry into a phase II HDC trial and those who did not. They found that by several measures those who were HDC eligible generally fared much better than those who were not. For example, 2 year survival rate was 21% for the HDC compatible but only 7% for those who were not and 5 year survival rate was 6% for the compatible but only 2% for the not compatible. This suggests that all or most of the apparent improvement in survival time in phase II trials for HDC could be due to bias in the selection of subjects.
The apparently favourable results of phase II trials led to a number of phase III trials of HDC being initiated in the USA. However, because the results of phase II trials appeared to be so positive (supported by the fake favourable results of Bezwoda’s controlled trial) many women were reluctant to take part in double-blind trials where they had a 50% chance of being allocated to receive the standard treatment. Many women paid for this expensive treatment or persuaded their health insurers to fund this expensive treatment and only a tiny fraction of those being treated in this way were part of a clinical trial. These trials therefore took much longer than expected to be completed.
Some health insurers in the USA were reluctant to fund this expensive and unproven HDC treatment outside of clinical trials. Some patients who were denied financial support for this treatment successfully sued their insurers either to force them to fund it or to pay compensation to families of patients who died after being denied this treatment. The most famous case was that of a California schoolteacher called Nelene Fox whose health insurer refused in 1991 to fund this experimental treatment. Ms Fox did eventually raise enough money to fund this treatment herself but she died in March 1993. Her family successfully sued the health insurer for delaying her access to this treatment which they argued contributed to her death. The initial award of damages by the Californian jury was $89 million.
A number of other successful court cases followed which forced many health insurers to fund this treatment and in some US states it was made mandatory for health insurers to cover this treatment. In Discover magazine in a report by Shannon Brownlee and Dan Winters in August 2002 it was estimated that over a period of 20 years around 30,000 breast cancer patients underwent this treatment with 4,000-9,000 dying as a direct consequence of the treatment. Different sources offer different estimates of the total number of patients treated with HDC and the number of treatment associated deaths; in an article published in Health Affairs in September 2001 Michelle Mello and Troyen Brennan estimate that 41000 women underwent HDC. If one thinks in terms of approximately $100000 to treat each patient and then allows for 15+ years of inflation then somewhere near $6billion at 2014 prices was apparently wasted on this ineffective therapy along with all the personal suffering of the patients and their families. Even if one subtracts the cost of conventional treatment this still leaves a sum of several billion dollars.
At a meeting in of the American Society of Clinical Oncology in Atlanta, Georgia in May 1999, a summary of the results of five randomised controlled trials of HDC were presented. Four of these trials (2 from the USA and 2 from Europe) indicated no significant survival benefits for women receiving the HDC treatment; only the South African study of Werner Bezwoda’s group reported highly positive benefits for HDC. This large discrepancy between the results of Bezwoda and the rest of the world turned out to be the trigger for an audit of his research that led to his being unmasked as a fraudster and his dismissal from his academic and clinical research post.
In 2005, Cindy Farquhar and several colleagues based in New Zealand and Australia conducted a meta-analysis of all the randomised controlled trials RCTs) that had compared HDC with conventional chemotherapy in women with metastatic breast cancer. They found six RCTs that met their inclusion criteria and in total these studies reported results from well over 800 women. The HDC resulted in 15 treatment deaths compared to 2 in the conventionally treated group. There were no statistically significant differences in survival rate at three years or five years between the two treatments. There was some limited evidence of increased event-free survival in those treated with HDC at 5 years but not 3 years (i.e. survival with no indication of cancer recurrence). There was limited evidence of an adverse effect on quality of life in the HDC group but this was not always reported and different criteria were used in different studies making amalgamation of results not possible. These authors concluded that HDC should not be given to women with metastatic breast cancer except as part of a clinical trial. This review was published in 2005 but assessed as still up to date in 2010 without addition of any further data. This would suggest that trials of HDC treatment for advanced breast cancer had largely ceased after 2005 and it is now little used as a treatment for breast cancer.
As already stated, this example could have been used as a case-study in the section of this book dealing with major scientific mistakes. It could be used to reinforce the message in chapter 2 that extrapolation of limited evidence at a relatively low level on the hierarchy of evidence can lead to faulty or exaggerated conclusions about the likely benefits of any treatment or intervention. In this case, results from uncontrolled phase II trials seems to have greatly overstated the benefits of HDC as compared to conventional therapy; this gruelling, dangerous and expensive treatment at best produced a marginal benefit on some limited outcome measures which would certainly not warrant its general use. This case-study could also be used to illustrate to lawyers and law students how a jury’s natural sympathy for victims of some personal tragedy can lead them to falsely attributing blame for the plaintiff’s circumstances to a defendant’s actions especially if the defendant is a major commercial enterprise.
Bezwoda’s role in the HDC controversy
Bezwoda’s fraudulent data played a significant role in exaggerating the benefits of HDC and increasing its use. In 1995 Bezwoda and two colleagues from the University of Witwatersrand published the results of what purported to be the first randomised controlled trial of HDC versus conventional dose chemotherapy as a primary treatment for metastatic breast cancer. Ninety women were recruited onto the study and they were randomly allocated to receive either 2 cycles of high doses of three cytotoxic anti-cancer drugs (cyclophosphamide, mitoxantrone and etoposide) or six to eight cycles of more conventional doses of three drugs (cyclophosphamide, mitoxantrone and vincristine). All of those receiving the high dose therapy were given either transplants of their own bone marrow or stem cells derived from their peripheral blood.
The results reported for the two groups were strikingly different with 23 out of 45 of the HDC group achieving complete response but this was only achieved in 2 out of 45 of the conventional dose group. All of the HDC patients required haematopoietic rescue but otherwise the side effects were described as moderate with mucositis listed specifically as a universal side effect of this therapy. The length of response and survival time were significantly larger in the HDC group and so this paper, published in a major international cancer journal, appeared to be a clear evidence-based endorsement of HDC as an effective treatment for advanced breast cancer and that their particular regimen:
“…appears to be a promising schedule that results in a significant proportion of complete response and increased survival in patients with metastatic breast cancer”
Up until February 2001 this paper was cited no less than 354 times and so it was clearly attracting the attention of oncologists around the world and thus indirectly their patients. The results of this trial were also recounted and/or updated in several other publications by Bezwoda over the next 4 years including 3 review articles. One of these follow-up articles was published in Haematology and Cell Therapy in April 1999, just before suspicions about the integrity of Bezwoda’s work were first made public. In this 1999 article there is an updated graph of patient survival from the 1995 study. This graph is shown in figure 1 and this indicates that after two years none of the patients treated with conventional doses of the chemotherapy drugs were still alive but around 40% of the HDC patients were alive – a dramatic positive result. After 5 years, approximately 30% of the HDC patients were still alive and this remained largely unchanged at 400 weeks (over 7 years) which suggests that in almost a third of patients HDC brought about an effective long term cure for metastatic breast cancer; no further deaths occurred between 300 and 400 weeks. There is also no indication of any extra treatment deaths in the HDC group.
Figure 1 Updated survival data purportedly for patients involved in Bezwoda’s 1995 trial comparing HDC with conventional chemotherapy. Source: Bezwoda (1999)
At a meeting of the American Society of Clinical Oncology in Atlanta, Georgia in May 1999, dramatic differences between the results of Bezwoda’s group and four other randomised controlled trials became very apparent. Bezwoda reported a clear advantage of the HDC therapy in terms of survival time and unlike the other trials he reported no treatment-associated deaths. In a report on the conference by Katherine Arnold published in 1999 in the Journal of the National Cancer Institute, the views of some of the participants are summarised and discussed. Dr Gabriel Hortobagyi one of those in the Texas group that had earlier published the data which suggested that positive results from phase II trials were probably due to biased selection of subject is quoted as saying that:
“The therapy is not useful if toxicity-related deaths increase”
Dr William Peters, an ardent longstanding advocate of HDC, highlighted the better survival rates in both arms of the clinical trials (HDC and conventional dose therapy) compared to women who were treated outside of clinical trials. This is an important point to remember if results from uncontrolled clinical trials are compared to historical outcomes for the condition as phase II trials often were; people treated as part of a clinical trial in specialist units tend to have better survival rates than women treated elsewhere.
The five studies reporting summary results at the conference had variations in their methodology and the most significant of these was that Bezwoda was using HDC as a first treatment for metastatic breast cancer whereas all of the others used it after patients had already been treated with conventional doses of chemotherapy. The difference between the South African results of Bezwoda and the four other groups were discussed and one contributor highlighted the fact that Bezwoda was using HDC as a primary treatment and suggested that the difference in his results might suggest that standard treatment made cells resistant to chemotherapy and added that:
“We will probably see a national study developed in the United States to test high-dose therapy alone”
If such a study had been initiated it would have been largely because of Bezwoda’s apparent success in using this mode of treatment because as he put it:
“We were giving the most effective treatment first”
In this conference report in an academic journal there was no hint of any suspicions about the results of Bezwoda.
An article by Denise Grady in the New York Times reporting on this Atlanta conference published on May 18th 1999 again highlighted the difference between the results of Bezwoda’s study and the four other controlled studies. Once again experts attending the conference were quoted voicing a range of different conclusions about the usefulness of HDC as a result of the conference presentations. Again the possibility that Bezwoda got better results because he gave patients HDC without any previous use of lower doses of chemotherapy was given as one possible reason for his success. However right at the end of the piece there is a statement that indicates that some delegates were suspicious of Bezwoda’s results and it was stated that some of them had asked:
“The National Cancer Center to send a scientist to South Africa to re-examine the study data”
In fact two audits of Bezwoda’s results took place: one relating to the data presented at the Atlanta meeting (Weiss et al, 2000) and published in the Lancet in March 2000 and one relating to his earlier highly cited 1995 trial and related articles Weiss et al (2001).
In the Lancet article reviewing the data presented at the 1999 Atlanta conference a number of serious problems with the work were found and some of these are summarised below.
- The conventional drug regimen that was used was not the same as that described at the meeting; a different drug combination was used.
- There was a small discrepancy between the number of patients listed on the enrolment log and those reported in Atlanta.
- Only 58 patient records were made available (out of 154 claimed to be enrolled on the study). These were generally very short and many patients did not meet the eligibility criteria for enrolment on the study. There were discrepancies between the doses of drugs recorded in the notes and those in the study protocol and other discrepancies between patient characteristics recorded and those presented in Atlanta.
- The on-site review noted 8 deaths which was the same as in the meeting abstract but records for many patients were not available and the panel identified a further 7 patients who had almost certainly died.
- No signed consent forms were present in the patient records and there was no record of ethical approval having been granted by the University of Witwatersrand.
- The protocol which was purported to have been written in 1990 referred to a 1997 paper in the bibliography and made reference to a substance that only became available in South Africa in March 1992.
Bezwoda did admit to misreporting the drug regimen for the conventional dose group:
“Out of a foolish desire to make the presentation more acceptable to an audience which I believed would have regarded CAF as more a familiar and more standard control arm”
The review team concluded that their audit did not confirm the validity of what was done and what was reported and that:
“The multiple discrepancies were of sufficient magnitude to invalidate the study results.”
This review team went on to recommend that the South African results should not be used to justify the conduct of further clinical trials.
The conclusions of the second audit (Weiss et al, 2001) into the 1995 trial was published in the Journal of Clinical Oncology and many of the findings are similar to those noted above, some of these are listed below.
- Again no ethical approval or signed patient consent forms could be traced.
- Bezwoda admitted that the protocol was actually written in January 2000 i.e. nine years after the study was supposed to have started.
- Again the drug regimen used was at variance with that in the paper.
- The stated process of randomisation of patients to control and HDC groups was not actually used.
- Many of the patients did not meet inclusion criteria for participating in the study.
- Records for only 61 of the 90 patients were found and they could only verify that 25 had actually received the treatment recorded on the enrolment log. Only 3 of these had received the conventional dose treatment.
- Of the 61 patients actually reviewed only 7 could be determined to be alive after 1995.
There were numerous other inconsistencies between what was reported and what was found by the audit team. Given that figure 1 was published in 1999 as an updated survival log of this study then it is inconceivable that if it is correct then the reviewers would not have found evidence in early 2001 of substantial numbers of patients still alive after 1995.
Taken together, these audits suggest that Bezwoda wrote protocols for the two studies only after he knew his work was to be audited, that he failed to obtain ethical approval or patient consent and that the published results bear no resemblance to what was actually done or what results (if any) were actually obtained.
Legal, political and economic aspects of the HDC affair
In a September 2001 article by Michelle Mello and Troyen Brennan published in the journal Health Affairs there is an extended discussion of some of the legal, political and economic aspects of the HDC controversy and some of the lessons that should be learned from the case study and some of these are briefly outlined below.
- This case highlights how political pressure and single issue pressure groups can overwhelm scientific arguments in deciding whether a new treatment should be implemented and funded.
- Courts adjudicated in favour of plaintiffs when they tried to force health insurers to pay for HDC treatment even though there was no persuasive scientific evidence to support its use. Different courts did not always come to the same conclusions.
- Billions of dollars of health insurer, government and private money was wasted in funding an experimental treatment that proved to offer no significant advantage to the patient. The treatment itself killed some patients and greatly increased the severity of side-effects in many others.
- Court decisions and state legislation forced many health insurers and public agencies to fund this treatment outside of clinical trials. This meant that it became difficult to recruit patients onto clinical trials. Less than 10% of patients who underwent HDC did so as part of a clinical trial and so clinical trials took more than twice as long to recruit the required number of patients which resulted in many years delay in producing clear indications that the treatment was in fact no more effective that conventional treatment.
Sad Footnote to the Bezwoda affair
As a result of the positive results reported by Bezwoda in his 1995 paper and at the meeting in Atlanta, a group at the University of Washington in Seattle designed a phase 2 clinical trial that aimed to replicate Bezwoda’s published regimen. Six patients started on this regimen before the trial was abandoned. Four patients experienced severe symptomatic cardiac toxicity. Two died of cardiomyopathy, the other two had suffered permanent cardiac impairment requiring intensive medical management (one had died of her cancer by September 2001 when this was reported in the American Journal of Clinical Oncology (Gralow and Livingston, 2001).
Arnold, K (1999) High-dose chemo and breast cancer: early results spark debate. Journal of the National Cancer Institute 91, 992-4.
Bezwoda, WR, Seymour, L and Dansey, RD (1995 RETRACTED) High-dose chemotherapy with hematopoietic rescue as primary treatment for metastatic breast cancer: a randomised trial. Journal of Clinical Oncology 13, 2483-9.
Bezwoda, WR (1999) High dose chemotherapy with haematopoietic rescue in breast cancer. Haematology and Cell Therapy 41, 58-65.
Brownlee, S and Winters, D (2002) Bad science and breast cancer. Discover magazine August 2002 issue.
Farquhar, C, Marjoribanks, J, Basser, R, Hetrick, SE and Lethaby, A (2005) High dose chemotherapy and autologous bone marrow or stem cell transplantation versus conventional chemotherapy for women with metastatic breast cancer. Cochrane Database of Systematic Reviews 2005 Issue 3 ART No:CD003142. DOI: 10.1002/14651858. CD003142. Pub2.
Grady, D (1999) Conference divided over high-dose breast cancer treatment. The New York Times May 18th 1999.
Gralow, JR and Livingston, RB (2001) University of Washington high-dose cyclophosphamide, mitoxantrone, and etoposide experience in metastatic breast cancer: unexpected cardiac toxicity. Journal of Clinical Oncology 19, 3903-4.
Mello, MM and Brennan, TB (2001) The controversy over high-dose chemotherapy with autologous bone marrow transplant for breast cancer. Health Affairs 20, 101-117.
Peters, WP, Ross, M, Vredenburgh, JJ, Karks, LB, Winer, E, Kurtzberg, J, Mast, RC Jr, Jones, E et al (1993) High-dose chemotherapy and autologous bone marrow support as consolidation after standard-dose adjuvant therapy for high risk primary breast cancer. Journal of Clinical Oncology 11, 1132-43.
Weiss, RB, Gill, GG and Hudis, CA (2001) An on-site audit of the South African trial of high-dose chemotherapy for metastatic breast cancer and associated publications. Journal of Clinical Oncology 19, 2271-7.
Weiss, RB, Rifkin, RM, Stewart, FM, Theriault, RL, Williams, LA, Herman, AA and Beveridge, RA (2000) High-dose chemotherapy for high-risk primary breast cancer: an on-site review of the Bezwoda study. Lancet 355, 999-1003.
Rahman, ZU, Frye, DK, Buzdar, AU, Smith, TL, Asmar, L, Champlin, RE and Hortobagyi, GN (1997) Impact of selection process on response rate and long-term survival of potential high-dose chemotherapy candidates treated with standard-dose doxorubicin-containing chemotherapy in patients with metastatic breast cancer. Journal of Clinical Oncology 15, 3171-7. http://ascopubs.org/doi/pdf/10.1200/JCO.19188.8.131.5271