Design of clinical trials in advanced prostate cancer: avoiding the dead ends
Frans M.J. Debruyne, Department of Urology, Radboud University Nijmegen Medical Centre, PO Box 9101, 6500 HB Nijmegen, the Netherlands.
Despite more than 30 years of clinical trials, investigations in prostate cancer have not succeeded in making advances comparable to those in other branches of research, such as breast cancer. Indeed, prostate cancer trials have repeatedly run into a series of ‘dead ends’, as investigators face the problems of inadequate funding for research, treatments that result in only minimal improvements in survival, and lack of treatment options that have sufficient prospects for success. This article briefly reviews the strategies behind clinical investigations into prostate cancer over the last three decades, evaluates the pitfalls that have hindered research, and makes suggestions for the appropriate design of clinical trials that are safe and beneficial to patients while maintaining cost-effectiveness and accountability to patients and society.
androgen-independent prostate cancer
hormone-resistant prostate cancer
National Institutes of Health
- EORTC (GU)
European Organization for Research and Treatment of Cancer (Genito-Urinary)
European Association of Urology
Good Clinical Practice
Food and Drug Administration
European Medicines Evaluation Agency
National Cancer Institute
Despite more than 30 years of concentrated effort, clinical investigation in prostate cancer has run into many ‘dead ends.’ Traditionally, most of the clinical trials in this field have focused on hormonal therapy, which has not yielded a significant improvement in survival. Long after initial trials showed that diethylstilbestrol (DES) worsened survival because of its cardiovascular complications, studies of DES continued to be conducted, in part due to the politics of research funding and a lack of stringent requirements for patient safety, which resulted in priorities that placed human considerations behind scientific inquiry. More than 10 years after the initial publications on the lethal toxicity of DES, clinical research on it was finally halted. However, the DES period was followed by that of LHRH agonists, and it was more than another decade before it became clear that the combination of LHRH analogues and nonsteroidal antiandrogens, the so-called standard of care for patients with metastatic prostate cancer, was no better than castration therapy alone. Although millions of dollars had been spent and thousands of patients enrolled in trials, the gain in real research terms remains limited.
Current clinical research in prostate cancer is now focused on both earlier and more advanced stages of the disease. Rising levels of PSA after initial curative treatment is now a major clinical issue, and the question of whether to treat early or late with hormones is at the centre of a clinical research debate. Likewise the treatment of hormone-resistant prostate cancer (HRPC) with numerous new drugs and approaches is being examined in many new clinical trials. But real breakthroughs are lacking. Are we once again on the wrong track with respect to new compounds? Novel approaches, such as gene therapy, immunotherapy and monoclonal antibodies, are still in the laboratory stage of investigation and not yet part of the treatment strategies investigated in clinical research.
At the same time, there are obstacles to conducting clinical trials in a manner most likely to benefit patients. Breast cancer researchers, for example, have had a great degree of success in recruiting patients for very large trials in prevention and in treatment. Prostate cancer investigators, in comparison, have failed to achieve parity in the design of their trials. Thus, the purpose of this article is to explore the question of how prostate cancer researchers can improve their investigations in a manner most beneficial to patients, clinicians, and society, while maintaining an awareness of the restrictions that are an inherent part of research funding. Several pitfalls in designing and performing clinical trials are discussed, and some suggestions for avoiding them are offered.
ECONOMIC ASPECTS OF CLINICAL TRIALS: FUNDING CHALLENGES AND SCIENTIFIC PRIORITIES
Phase I trials are typically small enough to be conducted in one centre. For phase II and higher trials, multicentre involvement is needed to recruit enough patients in a reasonable period. These trials require a great deal of time and money. They require large international clinical research organizations with sufficient resources to conduct the trials. Government agencies and healthcare providers (private and/or public) obviously have different priorities. Research institutions that specialize in cancer and concentrate on fundamental research must compete for the same limited funding. Thus, it seems that only the pharmaceutical industry has the resources to fund larger trials, and because these companies must balance support of research against fiduciary responsibilities to shareholders, each drug company must make determinations of the clinical relevance of any given study based on a host of factors other than strict scientific investigation. The result is that overall, the pharmaceutical industry provides little funding for trials that are unlikely to lead to product registration or are otherwise incompatible with development strategies. Moreover, because of the huge investments they have made, they have to look for ‘blockbuster’ drugs, which have a high probability of return on investment.
So the pharmaceutical companies are the largest source of financing for clinical research in prostate cancer, and they indeed have invested many millions of dollars or euros. Recent examples are the Early Prostate Cancer Program funded by AstraZeneca (>8000 patients), the atrasentan trial financed by Abbott, the zoledronic acid study supported by Novartis, and the docetaxel trial organized and financed by sanofi-aventis. Between 1985 and 1995 all studies of combination therapy (LHRH analogues and antiandrogens) were sponsored by the pharmaceutical industry, and almost all of these studies have lead to the registration of new drugs, which has allowed the industry to recoup the huge initial investments.
However, there is concern in the medical community about whether drug companies will ultimately direct clinical research and perhaps restrict scientific inquiry. How can we ensure that research continues to be originated and driven by urologists, who confront and are familiar with important clinical questions in prostate cancer? Realistically, we must acknowledge that the priorities of medicine and business may sometimes be at odds.
Some articles in the non-medical press have made the claim that pharmaceutical firms first develop a drug, and then search for a disease against which the agent can be targeted . While this charge may be extreme, we must realise that the economic pressure on the pharmaceutical industry is enormous, with fewer new drugs reaching registration and fewer new treatment approaches being evaluated. Moreover, clinical research with a new drug may be halted as soon as drug registration is acquired, causing important clinical questions to go unexamined. For example, some years ago we were involved in the design of a phase III trial in which zoledronic acid was to be investigated in the management of patients with advanced prostate cancer. The objective of the study was to evaluate whether the combination of hormonal treatment + zoledronic acid could significantly prevent the development of (symptomatic) bone metastasis; an important clinical question. The trial was immediately suspended when Novartis obtained registration in Europe of zoledronic acid for all patients with advanced prostate cancer, including patients with asymptomatic metastatic prostate cancer, the group of patients for whom the trial was specifically designed. So this important clinical question may remain unanswered, at least in Europe.
In the USA, where zoledronic acid is approved only for symptomatic metastatic prostate cancer, a similar trial is now underway. Despite a broad registration, Novartis agreed to sponsor a European study in patients with high-risk, non-metastatic local disease, investigating whether zoledronic acid (vs placebo) can prevent the onset of metastases. However, because this was not a registration trial, funding was limited and trial support was relegated to national Novartis subsidiaries.
Insufficient funding is not a charge that can be laid entirely at the doorstep of pharmaceutical corporations. What about public and governmental support? Although no figures exist for individual European countries, it can be concluded from individual experience that governmental support of prostate cancer research in Europe remains minimal. In the Netherlands some infrastructural facilities, more specifically for data management, are available through a network of integrated cancer centres. Similar facilities are present in some other (Western) European countries, such as Scandinavian countries and the UK (through the Medical Research Council). Direct public support has been insufficient, but there have been some notable changes. The European Union is supporting a randomized trial to screen for prostate cancer, the ‘European Randomised Study of Screening for Prostate Cancer’, and the European Union recently approved a grant to investigate new markers for diagnosis and progression of this disease. In the USA public funding for prostate cancer research also remains limited. However, through the lobbying efforts of the AUA some positive changes, especially for fundamental prostate cancer funding, have been initiated. The National Institutes of Health (NIH) has sponsored some important studies in prostate cancer, such as the Prostate Cancer Prevention Trial, the outcome of which was recently published .
In Europe, the organization of clinical research in prostate cancer continues to pose problems and challenges. Most of the large phase III prostate cancer studies that were not sponsored by the pharmaceutical industry were organized through the European Organization for Research and Treatment of Cancer (EORTC) Genito-Urinary (GU) Group. In the 1980s and 1990s, this group performed significant clinical research, primarily in the hormonal treatment of metastatic prostate cancer. For various reasons this excellent research group has deteriorated from a proactive organization to a ‘club’ with a lot of ideas, embedded in rather rigid, bureaucratic structures, and significant financial restrictions.
Nevertheless, the future of clinical research in prostate cancer lies in the hands of the professionals who know and can define the important clinical questions. The EORTC GU Group can overcome its current problems and shortcomings by collaborating with other professional organizations, such as the European Association of Urology (EAU), which has created an office that has specific responsibilities for European clinical urological research. The EORTC and its highly developed data centre can join forces with the EAU Clinical Research Office (together with the GU Group) and the pharmaceutical industry to set up the organizations and secure the necessary funds to initiate trials addressing all the important issues in the management of prostate cancer. In this way non-registration trials will be possible. An example of such a trial is the comparison between open and laparoscopic radical prostatectomy, a much-needed study that has not yet been realized due to lack of organizational structure and insufficient funding.
CHALLENGES IN THE DEVELOPMENT OF CLINICAL TRIAL DESIGN
Contemporary researchers now use the Good Clinical Practice (GCP) guidelines developed by the International Conference on Harmonization for designing, conducting, recording, and reporting clinical trials. The GCP is accepted by the European Union, Japan and the USA to facilitate mutual acceptance of clinical data by regulatory authorities in those areas . Other groups, such as the EORTC, also make recommendations about trial design . Standardized designs typically range from phase I to IV studies, with phase II and III studies usually prospective, randomized, and often double-blind. Phase IV studies are less interesting from a clinical and scientific perspective, but are of value as postmarketing studies of general questions such as quality of life, patient compliance, and general outcome research. The pharmaceutical industry also uses phase IV studies to create awareness of the treatment among doctors and patients.
There are at least 12 key factors that merit consideration when seeking to conduct a well-designed clinical trial:
- • Objectives
- • Design
- • Ethical considerations
- • Identification of study end points
- • Patient selection and eligibility
- • Recruitment of patients and study or treatment centres
- • Statistical analysis
- • Quality control
- • Data collection and analysis
- • Trial organization
- • Publication and reports of trial results
- • Creating public awareness
A focused approach and a clear and efficient design at the onset of a trial offer the potential to improve the quality and reduce the cost of research.
STUDY OBJECTIVE AND DESIGN
The study objective and target population should be clearly defined in a way that leaves no room for misinterpretation. For example, the term ‘advanced prostate cancer’ may have several possible definitions, each of which would involve many types of patients, approaches, and trials. This also holds true for hormone-resistant (HRPC) and androgen-independent prostate cancer (AIPC).
Possible definitions of advanced prostate cancer:
- • Hormone resistant
- • Locally advanced
- • Metastatic
- • Metastatic symptomatic
- • Metastatic asymptomatic
- • Rising PSA level after local treatment
- • Rising PSA level during hormone treatment
The categories of patients eligible for the trial should be clarified before the trial begins. This will prevent the inclusion of subjects in trials from which they are unlikely to benefit, and in which they would be unnecessarily exposed to the toxicity of the investigational drug or treatment. Patients should never be misled, and the strict observation of all GCP rules will obviate such events. Ethical committees that oversee and approve the design of the trial will also help prevent possible mistakes.
A detailed trial design will also minimize the number of patients who are dropped from the trial in its early stages, and may help to ensure a better fit between the goals of the scientific investigators and the subjects studied.
Importantly, precisely defined categories would prevent a trial outcome that may be difficult to translate to guidelines and recommendations for use in routine clinical urological prostate cancer practice. More patients recruited for the trial will consequently reach the trial endpoint, and hence the power of the trial will increase.
The dilemma of eligibility is particularly essential in studies addressing HRPC. There is a significant difference between symptomatic and asymptomatic patients. Most current trials primarily address asymptomatic patients, which could lead, for example, to a use of cytotoxic chemotherapy in earlier stages of prostate cancer. The recent publication of a docetaxel study (Tax 327)  already addresses these issues.
Perhaps as a consequence of that study, phase II cytotoxic chemotherapy studies in patients with high-risk localized disease have been initiated in the USA and are under discussion by medical oncologists in Europe. The future will tell whether studies of drugs that present a risk of cytotoxicity in usually fully asymptomatic patients will be harmful or beneficial. The risk that they are harmful is indeed present, because the design was developed as a consequence of a phase III trial that may not have been optimally designed .
The commonly accepted ideal endpoint in prostate cancer trials is overall survival, especially in patients with HRPC, where survival is on average ≈ 15 months. This allows for the detection of a survival advantage within a reasonable study time, as was recently shown in two large docetaxel studies [5,6].
However, this is a difficult and in some situations an impossible goal, especially in trials evaluating earlier stages of prostate cancer. Studies evaluating early vs late hormonal treatment in patients with asymptomatic, nonmetastatic node-positive disease need a much longer follow-up to mature (up to 8 years). As many patients will die from diseases other than prostate cancer, overall survival becomes a weaker endpoint. Cancer-specific survival may be a useful endpoint but may lead to inexplicable results, for example, the EORTC 30891 study, in which patients with early hormonal treatment showed a better overall survival but no difference in prostate cancer-specific survival (Studer, personal communication 2005) . This was another large trial (>1000 patients) that sought to investigate a very important clinical question, but because of the survival endpoint the results did not provide a clinically useful answer.
It is also necessary to consider alternative or secondary endpoints, such as objective and subjective response and quality of life, which are always measured by validated instruments and are of clinical significance. If a drug does not improve survival but significantly improves quality of life it is of value for patients. The US Food and Drug Administration (FDA) approved mitoxantrone/prednisolone therapy for HRPC because of the significant improvement in quality of life, as shown in a large Canadian phase III trial . Shortly afterwards, this combination was also approved by the European regulatory authority, the European Medicines Evaluation Agency (EMEA). However, caution is necessary, especially in evaluating objective response in metastatic prostate cancer. An objective response as defined by bone-scan findings is not a viable endpoint in advanced cancer because of the poor quality of bone scans, as shown by a review of EORTC trials . Soft-tissue metastasis is an appropriate endpoint in this patient category, but it occurs in < 20% of patients with metastatic disease.
Is the PSA level acceptable as a surrogate endpoint? The FDA does not accept PSA levels, but the EMEA does. The Liarozole Study Group showed that PSA level can be used as an endpoint in HRPC . In a study of liarozole for HRPC, patients with a ≥ 50% decrease in PSA from baseline had a 10-month significant survival advantage (hazard ratio 0.43, P = 0.0018). Does this mean that the PSA level can be used as an endpoint in other categories of patients with prostate cancer? The subject is under discussion, but if the answer is yes, it will make possible the design of studies in early-stage disease (e.g. rising PSA level after radical prostatectomy or radiotherapy). Indeed, prostate cancer is the only malignancy for which there is such a reliable marker. It is unfortunate that the FDA does not accept it as an endpoint in clinical trials. The EMEA has done so, and recent discussions in the FDA's Office of Cellular, Tissue, and Gene Therapies offer some hope that the FDA will eventually follow its example .
Ideally, patients should be stratified prospectively on the basis of prognostic factors, but this presents a problem in prostate cancer because prognostic factors are not specific enough to stratify patients according to risk. Even if patients are stratified, the weakness of prognostic factors in advanced prostate cancer can somewhat blur the outcome of the study. Many trials use retrospective stratification analysis, but the field of prostate cancer research is rife with arguments about the value of findings from retrospective stratification analysis. Ultimately, retrospective analyses waste both time and money, whereas well-designed prospective stratification can avoid such problems.
A current trial evaluating use of local treatment with zoledronic acid to prevent metastases has been designed with immediate stratification based on risk factors, PSA levels, Gleason scores, and positive surgical margins . The trial stratification is defined in detail in the study protocol but still presents some difficulties. One of the stratification items is the presence of positive margins, which requires a pathological evaluation. However, that is not feasible in this trial because funding for central review pathology is limited. Stratification thus remains a ‘dead end’ in these trials. However, it is important and in some cases essential for the application of trial outcomes to daily clinical practice.
The study investigators should clearly delineate parameters for determining when a trial should be abandoned because of toxicity. The EORTC has focused on this issue, at the suggestion of the medical oncologists participating in its trials. Ethics committees are important for monitoring compliance with GCP criteria, data control, independent data review, and implementation of previously defined rules for discontinuing the trial because of toxicity. However, ethics committees can sometimes create obstacles to clinical investigation. Ethics committees should not comment on the value of the therapy evaluated or other aspects of the trial that are outside their purview. However, they sometimes do, which can lead to undesirable restrictions, especially in international multicentre trials. It has indeed occurred that the same trial (e.g. photodynamic therapy before radical prostatectomy for localized prostate cancer) was accepted by the ethics committees in the UK and France but refused in the Netherlands, even though the evaluation involved the same patients and same treatment. A similar trial, of high-intensity, focused ultrasound before radical retropubic prostatectomy, had been clinically approved in the Netherlands some years earlier. Ethical approval of a trial is essential, necessary, and obligatory, but inconsistency and variable evaluation and judgement across centres should be avoided.
RECRUITMENT OF STUDY POPULATION AND STUDY CENTRES
An inability to recruit adequately for trials has been a problem for many years. For example, at the Department of Urology, Radboud University Nijmegen Medical Centre in the Netherlands, there are ≈ 15 physicians who see patients who might be eligible for a trial in which we are currently participating. Yet after 6 months, only one patient is enrolled. This is not an isolated case. Among the explanations for this problem are the following:
- • Lack of awareness and interest (sometimes due to time) in the trial
- • Physicians overestimate the number of patients who will meet selection criteria
- • Participating centres lack the commitment to enrol patients
- • Pharmaceutical sponsors recruit from a medical speciality that sees few patients appropriate for inclusion (e.g. from medical oncologists rather than urologists in continental Europe)
- • Patients refuse to be randomized
Researchers must develop strategies to address these issues effectively. First, we must develop stringent selection criteria and deadlines for recruiting patients and study centres. We should also consider closing trials that recruit poorly. When a trial recruits inadequately because few patients meet the selection criteria, this may be an indication that the trial has not been designed to answer a clinically relevant question. It quickly becomes obvious whether a trial will recruit well. Trials that remain open despite sparse recruitment waste considerable resources.
It also is important to mandate a minimum patient enrolment for centre participation and payment. For example, centres participating in an EAU intergroup trial must guarantee enrolment of ≥ 10 patients within 1 year to participate in the trial and receive funding. Such a strategy both improves patient recruitment and enhances the quality of data.
Independent statistical analysis and intent-to-treat analysis are integral to good clinical research. However, statistical demands can often present a burden in trial design. Sometimes the statisticians’ criteria for an adequately powered trial make it necessary to include large numbers of patients. For example, in the zoledronic acid trial one small change in the trial design meant that the number of patients required increased from 600 to 1300 . As discussed, this can present a major problem in recruiting sufficient numbers of patients for the study.
DATA COLLECTION AND DATA CONTROL
Data collection is usually complete in patients with cancer; once they consent to participate in the trial, patients adhere well to the scheduled appointments and are reliable, informed, and cognisant of their participation in the trial. Yet there remains room for improvement in the completeness and reliability of data that are gathered. Automation is expensive but would probably improve data collection and data control. Proper data control is best managed in registration trials, which can handle the cost and time required. Methods of internal and independent review of protocols, adherence to GCP or other standardized criteria, and other forms of data control, should be factored into the study.
Research nurses play an essential role in appropriate patient consulting and effective execution and follow-up. Prostate cancer trials, especially in advanced disease, should not be conducted without them. A good collaboration between physician (urologist) and research nurse will improve patient adherence to the protocol and thus the quality of the trial. When the trial is explained to the patients, research nurses will help to judge whether and which selection criteria are applicable to each patient. Data transfer should be done with the help of qualified data managers. The optimum situation is the use of a fully automated system, which will avoid most data transfer errors. However, the currently available systems are expensive and, for the most part, remain incomplete.
PUBLICATION OF TRIAL RESULTS
Once a trial is completed, findings should be disseminated quickly so as to benefit patients as soon as possible. In some cases trial results are completed but not published until the next major scientific meeting. This can delay publication by a year. Unpublished results then become a topic of discussion and speculation, and treatment choices may be based on rumours rather than data. Expediency is thus an important part of selecting a target scientific venue for publication. Both positive and negative findings should be published quickly so that physicians can prescribe the latest evidence-based treatments and avoid inappropriate therapies. Some journals now favour, and facilitate, rapid publication of trials data.
Perception of bias in the relationship between physicians and pharmaceuticals companies should be avoided. Results of research should be disclosed at scientific meetings or conferences. Any real or perceived conflict of interest on the part of presenters at scientific meetings and authors of published studies should be disclosed.
Once trial results are published and agreed upon in the scientific community, organizations and principal investigators should communicate findings to the general public in a simple, straightforward way. Knowledge of trials and their benefits may improve recruitment to new studies. Effective use of the media to create public awareness of major trials may explain, in part, why studies of breast cancer prevention and therapy recruit more patients than do such investigations in prostate cancer.
Clinical research in advanced prostate cancer will continue to challenge researchers, presenting problems and dead ends as well as opportunities and rewards. It is hoped that early-stage research will yield new approaches that may serve as topics for future large clinical trials. Ultimately, the goal should always be to approach the design of trials in a manner that is most beneficial to patients.
DISCUSSION AFTER PROF DEBRUYNE's PRESENTATION
Dr Kenneth Watson, BSc, MD, MBA (Yamanouchi Group Business, LLC, Paramus, New Jersey, USA): The new drugs for cancer have become increasingly specific. In recent papers, editors have indicated that the population recruited for trials may not reflect the properties of the drug in the results. I have asked to distribute this paper published in Fortune a month ago that alludes to this problem . How would you as investigators like to deal with this situation?
Prof Frans M. J. Debruyne, MD, PhD (University Medical Center, Nijmegen, Netherlands): That is certainly wrong. You should never in my view develop a drug and then find an indication. It should be the reverse. It is wrong to institute trials to find an indication. I see that happening a little bit with zoledronic acid. They explored experimentally all kinds of things and now they are trying to use it in RCC. But I understand also the drug companies’ viewpoint. You have invested many millions of dollars in a certain compound. If it doesn’t live up to expectations, you will try to find other uses for it.
Dr Daniel Petrylak, MD (Columbia-Presbyterian Medical Center, New York, New York, USA): One of the problems pointed out by the Fortune article was that we are content with survival improvements of 3–4 months. I don’t think any oncologist is happy with these numbers; this is a slow process where we build upon the results of other investigators. Dramatic impact on treatment, such as those seen with aspirin or penicillin, has yet to be seen in the treatment of advanced metastatic cancer. Targeted therapy is a rational step ahead. These treatments are far from panaceas. If we subscribe to the adage that two investigational drugs cannot be combined in a clinical trial, then we may be doomed to fail with many of these targeted agents. A cancer cell may ‘cross wires’, and it may be necessary for response/survival to turn off two pathways rather than one. So really the issue is how can we get people to collaborate, and how can we get investigators to accept basic science that supports the shutdown of multiple targets rather than just simply going after one target.
Dr John T. Isaacs, PhD (Sidney Kimmel Comprehensive Cancer Center at Johns Hopkins, Baltimore, Maryland, USA): There is also a corporate problem with trying to affect apoptotic pathways. For example, if company A is working on the first target, and company B is focused on the second, then you need the two companies to do a cooperative trial with their respective agents, and that is almost unheard of. I don’t think the FDA is the obstacle to this problem. We are working with them and I have been very impressed. If two companies came in and said they were willing to collaborate on a trial of two drugs, then the FDA would be happy. The FDA does not require the drug to be tested as a single agent. If you have a compelling mechanism that says element A will affect element B, and C will affect D, and you need both effects to obtain the outcome, the FDA is saying, ‘Fine, but just do the appropriate toxicity study.’ It's the companies that are the obstacle to that sort of trial.
Dr Petrylak: There is a lack of consensus about how to identify and develop the proper marker to predict response. For example, gefitinib is thought to attack the epidermal growth factor receptor (EGFR), and for quite some time investigators were focused on whether, as with HER-2 and breast cancer responses, elevated EGFR levels predicted response. In fact, the response is probably not related to EGFR levels, and most responders have low-to-intermediate levels of EGFR expression. EGFR mutations rather than levels correlate with response. Selection of the proper markers is crucial to the right clinical trial results.
Dr Isaacs: The National Cancer Institute (NCI) tried to address this by forming a biomarkers network to fund research into developing new markers . The Cancer Biomarkers Research Group supports a variety of cancer research grants, including:
- • Development and validation of promising early cancer biomarkers for risk prediction and early detection of cancer
- • Development of databases and informatics systems to optimize tracking and assessment of biomarker utility and expression patterns
- • Development of new technologies or refinement of existing technologies to ascertain the molecular circuitry of preneoplastic cells
One problem with this system is that it separates aspects of bringing a new marker to clinical use.
There are three levels at which one can apply for funding: identifying a potential marker, developing an appropriate assay for a marker, and validating an assay. Different groups apply for each level. This replaces the former system in which a single group would conduct all research from identifying a potential marker, through completing trials, validating the assay for the marker.
This sets up additional hurdles to introducing a marker for clinical use. One group may perform and publish the basic science required to identify the marker; another must then apply for funding to create an assay for that marker. A third needs to seek funding to test an approach for validating the marker. If the process breaks down at any point, then the research may not proceed to the next level. This is why we may read about a possible marker such as pro-PSA or hepsin but never see it in clinical use. Typically, groups seek funding for the first two steps, but few seek to validate markers. This is where the process breaks down.
Many times the issue is not the assay but the standards for the assay. We need standards that can be shared globally. When antibodies became available to the oestrogen receptor, what held up development of a test for breast cancer was the decision about the standard among laboratories.
Prof John Fitzpatrick, MD (Mater Misericordiae Hospital, Dublin, Ireland): I have an example of how an inability to meet standards of sensitivity led to the death of a promising marker. When studying CD44 in breast cancer, my group examined the mRNA level, performed Western blots, did RTP share, then moved to clinical samples and used RT-PCR ELISA. We raised monoclonal antibodies and had the ELISAs for that. We did immunohistochemistry. A lot of papers went out about CD44 and cancer. When we took it down to RT-PCR ELISAs, it failed based on its sensitivity. If you reduced the sensitivity of the assays, you got rid of the cancers. If you increased the sensitivities, you had a sensitivity of 50–60%. In other words, it did not meet the ‘gold standards’ of PSA, carcinoembryonic antigen, or CA-125. So, the molecular biology was great but it didn’t work when it was applied to a practical setting. And that is why CD44 died a natural death, along with a lot of other tumour markers that we were developing.
Prof Paul Abrams (Bristol Urological Institute, Bristol, United Kingdom): I remember looking at the first EORTC trials in the 1970s and being amazed that very large centres recruited five patients and small centres recruited 25. This shows huge selection bias. The only way around that is to make it mandatory through ethics committees to keep a log of all suitable patients. Then you can begin to understand why patients are included and excluded.
Prof Fitzpatrick: The number of patients screened should be put into the paper, along with what happened to the patients screened. That would be very interesting and has never been done.
Dr Petrylak: Another issue is that patients refuse randomization to trials if they can obtain the drugs off-protocol. Patients are extremely educated and can easily obtain information from the Internet about current trials and standards of care. For the most part, they want to do what is best for their own healthcare. If a patient is motivated, eventually he or she will find someone who will treat them with FDA-approved drugs off-protocol. The physician can bill for these services in the community. The South-west Oncology Group (SWOG) study comparing mitoxantrane/prednisolone to docetaxel/estramustine is a good example. Patients often asked, ‘Why should I go on a study if you can treat me with the “best” drug off study?’ Trials can be successfully performed only with drugs that are not yet FDA approved because they are less likely to be affected by crossover of the investigational drug. In the SWOG study, the survival difference could have been diluted by patients having received secondary treatment; 52% of patients who received docetaxel/estramustine or mitoxantrone/prednisolone received secondary chemotherapy afterwards. The same is true with the Aventis trial comparing docetaxel weekly or every 3 weeks to mitoxantrone and prednisolone. Secondary treatment affects patients but we cannot quantify its effect. Is it selection bias? Does the patient want to go on a phase I study? Are they at an academic centre? Will they be treated with whatever comes off the shelf? It is a big issue.
Another question that remains to be answered is why there have been more and larger clinical trials in breast cancer than in prostate cancer. Breast cancer physicians are used to putting patients on trials. We did not have many phase III trials open in the past for prostate cancer. Also, there is an issue of whether a physician ‘believes’ in a trial. Certainly, a physician's enthusiasm or lack thereof can easily sway a patient's opinion about entering a study. People have opinions, not necessarily based upon facts, that one chemotherapeutic drug or approach may be better than another.
I don’t think that there is a major difference in clinical trial reimbursement in different organ sites. I don’t think that is the issue at all. It is the perception about whether certain types of treatment had activity or not. I think that it is a matter of making people aware of what is out there. Take the SWOG trial 99–21 of hormones vs hormones plus mitoxantrone as adjuvant therapy in high-risk patients after prostatectomy . It nearly was closed twice for lack of accrual. Urologists say they are unaware that the trial is open. There was not enough publicity for it. This is partly because mitoxantrone is expected to go off patent within about 3 years.
In contrast, SWOG 99–16 moved rapidly because there was a press release when it opened . Publicity from the patient advocacy groups along with Aventis advertising helped spread the word that this trial was open. I think the people will enrol patients in these studies if they know about them. As to payment, there is remuneration from the NCI.
One other issue with dissemination of the results of clinical trials involves rules about presentation of abstracts at many meetings. It is often stipulated that an abstract can only be presented at one meeting. The number of urologists and oncologists who attend both ASCO and the AUA are probably few. The minimal crossover of audiences means that only half the intended target audience may hear an abstract presentation. There needs to be a way to overcome this problem and promote the multidisciplinary approach to prostate cancer.
Dr Isaacs: Releasing results in abstracts assumes that reviewers will uphold the authors’ conclusions. There are some pretty good examples of peer review changing papers substantially. When David Crawford first submitted his paper about complete androgen blockade to the New England Journal of Medicine, the conclusion was that it did not increase survival. The Journal said that this was such an important paper that we want you to reach 50% survival, not projected 50% survival. While waiting for this to happen, the conclusion changed. Now there was increased survival so that the paper was modified.
Prof Abrams: That raises the issue about collaboration between journals to set standards.
Prof Fitzpatrick: People say that all the time but haven’t defined what it means. I am the editor of BJU International, and I think it is natural that journal editors are going to try to define criteria for success in a trial, and to state that a given drug is a useful addition to the available options.