Japanese Journal of Clinical Oncology

INTRODUCTION

Surgical trials are difficult to conduct, but not impossible. Since the mid-1980s our department has run several major randomized trials within the framework of the Dutch Gastric Cancer Group and the Dutch ColoRectal Cancer Group (3-5). It is the purpose of this article to present practical information on how we, in The Netherlands, do succeed in performing trials in surgical oncology. This will be illustrated on the basis of a gastric and rectal cancer trial.

DEVELOPMENT OF A TRIAL DESIGN

Japanese gastric cancer patients turned out to have a much better survival than Western patients. One of the reasons could be the more extensive surgery (6). A similar problem occurred in the field of rectal cancer: extended surgery (7) and Total Mesorectal Excision (8,9) lead to strikingly lower local recurrence rates. This has led us to design two randomized surgical trials: a pure surgical trial of D1 dissection versus D2 dissection in gastric cancer and a trial in rectal cancer of which the genesis will be explained below.

In The Netherlands, several population-based studies observed local recurrence rates of about 20% (10-12).The goal of a new trial would be to achieve standardization of rectal surgery leading to lower local recurrence rates. Since 1994 the idea was further developed by some enthusiastic surgeons throughout The Netherlands. Meetings had been organized and two newsletters had been produced. The idea for a phase III rectal cancer trial at that moment was to randomize patients for standard D1/D2 dissection plus preoperative radiotherapy versus extended D3 dissection alone. A pilot study had been carried out from November 1994 to March 1995, with professor Moriya from the National Cancer Center in Tokyo who visited 24 hospitals throughout The Netherlands as instructor surgeon. Many Dutch surgeons, however, feared a considerable morbidity with the Japanese D3 technique in Dutch patients, as experienced with the D2 dissection in the Dutch Gastric Cancer Trial as well as the MRC trial in which postoperative mortality after D2 dissection was twice as high as after D1 dissection (13,14). Moriya therefore performed a more or less `modified D3 dissection' (15). This changed the views to a TME approach, as advocated by Heald and Enker (8,9).

A second and third proposal was to compare `conventional' surgery with TME surgery or compare in a two by two factorial design yes/no preoperative radiotherapy and conventional versus TME surgery. Both designs, however, would allocate 25-50% of the patients to the inferior arm of conventional surgery without preoperative radiotherapy. Literature data were so convincing with regard to the superiority of the TME technique over conventional surgery, that a majority of the Dutch surgeons had the opinion that it would be unethical to randomize patients in such a design. It is difficult to conduct a randomized trial in an evolutionary phase of a new operation (16,17). A prerequisite for a trial is that the participating surgeons are equally conversant with both techniques. An alternative could be the use of a randomized-surgeon design in which groups of surgeons would perform conventional surgery and other groups of surgeons would perform TME surgery (17,18). Finally, the last proposal was made: compare TME surgery with or without preoperative radiotherapy.

Financial support is crucial, especially in multicenter trials (19,20). Approval by the Dutch Cancer Society was obtained in August 1995, merely for on-site and central datamanagement support. The omittance of a pure surgical trial of conventional surgery versus TME surgery, was a major concern for the Dutch Health Council for three years. Final approval and financial support for a surgical co-ordinator, pathology co-ordinator, quality of life co-ordinator, central datamanager, and also compensation for instructor surgeons and pathologists was only just obtained in April 1998.

From the early moments it was stated that the study should be a quality controlled study. This meant quality control in surgery, radiotherapy and pathology. Results from a questionnaire which was mailed to all 21 Dutch radiotherapy departments showed that the use of the 5×5 Gy scheme, as used in Sweden (21), was accepted by most institutes.

INSTRUCTION AND QUALITY CONTROL

The TME trial aims at lowering the percentage locoregional failures with nerve preservation by TME surgery itself and perhaps even more in combination with preoperative radiotherapy. The TME procedure provides an excellent specimen and therefore the pathologist is able to check if the procedure is performed according to the protocol using the transverse slicing method of Quirke which is highly predictive for the development of a local recurrence (22).

Thus, surgeons as well as pathologists had to learn new skills very different from their daily practice. How did we facilitate this learning process? In addition to the TME protocol, a special pathology protocol was written which was distributed to 43 pathology laboratories. A pathology workshop was organized in December 1995 with the attendance of Dr Quirke. We also produced a sheet with a step-by-step protocol to be used at the dissecting table. The pathology co-ordinator has established a Pathology Review Committee, to discuss problems and prepare standards for the review of the specimen.

How to learn a new surgical technique? Firstly, we produced a videotape on radicality and autonomic nerve preservation, with operations performed by professor Moriya. Mr R.J. Heald from Basingstoke, UK, was installed as a visiting professor, funded by the Dutch Cancer Society. He has performed many operations throughout The Netherlands, attended our six workshops so far and produced two videotapes which were distributed to all participating hospitals.

The Netherlands is divided into nine comprehensive cancer center regions. From each region, a few experienced TME surgeons were selected. In total, 20 instructor surgeons were chosen. Their task was to introduce, teach and control the TME operations in their region. In the first year of the trial, 66% of operations were attended by an instructor. This instructor system has also successfully been used in the Dutch D1-D2 gastric cancer trial, where during the first four months all D2 dissections had been performed by Professor Sasako and thereafter all D2 operations were attended by an instructor surgeon and all D1 operations by the surgical co-ordinator (23). To facilitate the selection of eligible patients, we made pocket size cards with all criteria, the randomization scheme and the phone number for randomizations at our Datacenter.

START OF THE TRIAL

The rectal cancer trial was launched in January 1996. In our country each hospital has its own Medical Ethical Committee and each committee had to approve the protocol. In The Netherlands, we have the availability of an excellent datamanagement network, financed by the Dutch Cancer Society. In practice, this means that every datacenter linked to one of the comprehensive cancer centers receives a fee per randomized patient in their region. All Dutch hospitals are covered by this system. Datamanagers working in the field of cancer clinical trials are all members of the Dutch Working group for Datamanagers (DWD), similar to a network in the UK (24). The DWD group organizes biannual meetings and a yearly course. At those meetings, new protocols and several aspects of cancer treatments are discussed and uniformity on case report form completion is accomplished.

Randomization is done centrally at our Datacenter using predefined lists with stratification per hospital and expected resection type. Every person in our team is familiar with the randomization procedure. Therefore, we can easily guarantee a fast randomization and can always be contacted for questions. This procedure is not as computerized as at major datacenters (25), but much faster. All in- and exclusion criteria are checked by phone and the allocated treatment and trialnumber are given immediately thereafter.

Case report forms are mailed after randomization for every patient separately, including patient identification and trialnumber, to the several doctors involved in the treatment of the patient. Our system of data collection differs from most systems, which have one responsible investigator or datamanager for all forms and the forms are delivered in bulk. We have arranged that the surgeon is responsible for the completion of the surgery forms, the pathologist for the pathology form and the radiotherapist for the radiotherapy form. The on-site datamanager receives a copy of the randomization form and is responsible for the on-study and follow-up forms and can assist the local investigator. The main advantages of this extensive system are that the investigator knows exactly which forms have to be filled in and all involved disciplines are prospectively aware that their patient is participating in a trial.

Before randomization, the surgeon should have handed over a pre-treatment quality of life questionnaire to the patient since quality of life is a secondary endpoint in the trial. In reality, it is more practical and easier for the surgeon that the questionnaire is mailed via the datacenter with a self-addressed return envelop to the patient's home. Follow-up questionnaires are also sent directly to the patient at 3, 6, 12, 18 and 24 months after surgery to unburden the surgeon as much as possible. So far, the compliance has been 85%. Furthermore, a sub-set of patients is visited by the quality of life co-ordinator for face-to-face interviews on treatment preferences and costs.

Returned forms are checked manually for correctness and completeness. On top of the case report forms, surgeons are asked to send a copy of the surgery report and discharge letter. Pathologists are asked to send a copy of the pathology report, a hematoxylin-eosin slide of the tumour for later revision by the Pathology Review Committee, photographs of the specimen and two paraffin blocks: one with tumour and one which contains normal tissue. We also collect fresh frozen tumour and normal tissue, which is collected batchwise by visiting the different pathology laboratories. Paraffin blocks and fresh frozen material are being used for additional research.

Data are entered in our database program (Medical Research Data Management) and automatic range- and cross-checks are carried out. Statistical analyses are performed with the Statistical Product and Service Solutions, SPSS 7.5 for Windows, 1986, SPSS Inc., Chicago, IL. Every six months we send two overviews to all involved disciplines. First, we present an `administrative' overview with patient identifications, number of randomized patients and received forms. In the other overview, some important variables per patient are listed. For the surgeon, we show, for example, TNM stage, last date of follow-up, date of locoregional recurrence, date of detection of distant metastases, date of death and overall survival. For the radiotherapist and the pathologist, similar overviews containing specific data are presented. Approximately one month thereafter we send a reminder to those who did not respond to our missing forms requests. Instructor surgeons are involved in the quality control and education of surgeons in their region. The instructors also stimulate the centers to enter their patients in the trial. Therefore, we also send overviews to the instructors to provide them with a tool to follow the status of the trial in their region.

CONCLUSIONS

A close contact between a datacenter, including co-ordinators for separate disciplines, and all participating physicians is essential for the efficient running of a quality controlled multicenter, multidisciplinary trial. Continuous enthusiasm is kept by the organization of regular workshops and distribution of newsletters. The accrual is very satisfying with 1413 randomized patients in three years time: 1212 patients from 83 Dutch hospitals, 161 from 11 Swedish hospitals and 40 from other countries. Recruitment will continue until 1400 Dutch patients, and this will probably be reached in May/June 1999. After this local treatment study, we are now planning a successor trial on systemic treatment using the same standardized local treatment.