| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Commentary |
1 Hammersmith Hospital, London, 2 Mount Vernon Cancer Centre, Northwood, 3 University of Kent, Christ Church, Canterbury, 4 St Thomas' Hospital, London, 5 The Plymouth Oncology Centre, Plymouth, 6 Bristol Oncology Centre, Bristol and 7 Patterson Institute, Christie Hospital, Manchester, UK
Correspondence: Dr Bleddyn Jones, Oncology Centre, Hammersmith Hospital, London W12 0HS, UK
The first United Kingdom Radiation Oncology (UKRO) conference was held in York on 23–25 April 2001 under the combined auspices of the Institute of Physics and Engineering in Medicine, the British Institute of Radiology, the College of Radiographers and the Royal College of Radiologists. There were 247 delegates from the various disciplines involved in the process of radiotherapy (Table 1
). The conference theme was accuracy and uncertainty in radiotherapy, a vast multidisciplinary topic. It is not possible to cover every presentation in this report, but the invited reviews and some proffered papers are summarized. The abstracts of oral presentations given at the meeting have already been published in Clinical Oncology [1] and can be referred to by the interested reader for further details.
|
Treatment planning overviews
The first plenary session included a brief general overview by Bleddyn Jones (Hammersmith), who emphasized the need to consider accuracy and uncertainty throughout the process of radiotherapy by a fusion of disciplines including physics, pathology, radiobiology and statistics. A modelling illustration of the above approach was used to show that variations in dose across a planning target volume (PTV) have a smaller effect on tumour control probability than the same variation in dose precision, thus emphasizing the need for good quality control within radiation physics.
The contribution of radiotherapy physics was described by Ben Mijneer (Amsterdam). Head and neck cancer dose–response curves are characterized by slopes (gamma factors) of between 1–3% additional tumour control for a 1 Gy increment in dose; a wider variation of 1–12% per Gy has been found for a variety of normal tissues [2]. Linear accelerator output calibration can be inaccurate by as much as 7%. Errors arising within the various parts of the treatment planning process may compound to produce significant inaccuracy [3] and standard deviations may be a better way to express dose variation than the imposition of ranges (such as ±5% in International Commission on Radiation Units and Measurements (ICRU) guidelines). The greatest inaccuracy occurs in the definition of gross target volume (GTV), which varies with imaging technique as well as between clinicians. In the dose calculation procedure, normalization of dose to 100% at 10 cm depth on the central beam axis is preferable to using 100% as the maximum value. Variations between different commercial treatment planning software can produce up to 5% changes in the predicted dose, and greater differences can be found in more complex plans, for example variations in dose under the thick ends of wedges can become significant when many wedged fields are used. Lung density correction is a further issue owing to significant field edge dose variations.
For errors related to patient movement, systematic errors appear to be more significant than random errors. In vivo dosimetry has revealed a variation of between 1.075 and 0.975 for conformal prostate radiotherapy [4]. Correction protocols can reduce geometric inaccuracies, for example when there is greater than 6 mm and 4 mm deviation in three dimensions on the first and second fractions, respectively, as assessed by portal imaging. The "state of the art" for achievable geometric error is a standard deviation of 2 mm (head and neck cancers), 2.5 mm (prostate), 3 mm (other pelvic targets) and 3.5 mm (lung). For intensity modulated radiotherapy (IMRT), large potential errors are possible owing to a fall in output with field size. Consequently there is considerable scope for further refinements in radiotherapy physics and the translation of these concepts into routine use.
Christopher Nutting (London) listed the clinical causes of planning errors as well as the ICRU 50 and 62 requirements, and he emphasized that several clinical target volumes (CTVs) should be considered for some cancers according to their tendency to local and regional spread. Random errors cause increased "blurring" of dose at target margins, whereas systematic errors (that is repeated similar errors) cause a significant CTV underdose, resulting in a geographical miss. Recently, a progressive improvement in the accuracy of radiotherapy within sequential clinical research studies has been confirmed at The Royal Marsden Hosptital. This is very reassuring, particularly since treatment-related toxicities have been reduced, although the patterns of recurrence will need to be studied carefully to determine the appropriateness or otherwise of the CTVs, especially as imaging techniques are also providing better resolution with time. New techniques to improve accuracy include devices such as lung gating using radiomarkers [5]. In prostate cancer, a radio-opaque urinary catheter, a rectal balloon and interactive ultrasound checks may improve accuracy [6, 7]. Target definition remains a major clinical problem: for prostate cancer, the apex, base and seminal vesicles are the most difficult regions to identify, the standard deviations being around 10% of the GTV dimensions [8]. The use of three-dimensional (3D) imaging for head and neck cancer, with lymph node outlining, is a further important contribution [9, 10] allowing specific knowledge of dose and this will permit selective dose escalation by IMRT (see below).
Helen McNair (Royal Marsden) reviewed geometrical errors in radiography. Regular quality assurance of CT planning scanners (apertures, couch, lasers) is essential, since systematic errors of up to 4.5 mm can occur. Digitally reconstructed radiographs (DRRs) can now be used instead of simulator films, but they require good resolution with small slice thickness as well as additional time on the first treatment fraction. Heavy patients may cause distortion of the couch scaling, which is calibrated using patients of normal weight. The problem of using skin markings on curved skin surfaces can be largely overcome by marking on planar stereotactic frames. Several ergonomic immobilization techniques were shown and the potential importance of the Beam Cath "lung active breathing control" technique for reduction of cardiac irradiation in breast patients was highlighted. In a later talk, Helen McNair also described IMRT by the "step and shoot" and dynamic collimator techniques, and stressed the importance of positioning and reproducibility [11–14].
John Wong (USA) summarized the possible errors that may occur with IMRT. The main concern was shrinking margins around a cancer. The process of adaptive radiotherapy [15, 16], where interfraction organ motion is assessed, with redefinition of the PTV during the first week of radiotherapy, is likely to alter radiotherapy practice considerably. By this means, the PTV is progressively optimized to a maximum permitted 2% error by five sequential CT scans during the first week of treatment to yield a final treatment volume that is used to set the total dose. This complex repetitive process cannot be achieved in approximately 20% of patients owing to non-compliance. Normal tissue sparing can allow more hypofractionated radiotherapy in a shorter overall treatment time, which may be an advantage in lung cancer. There is little evidence that shorter overall times are required in prostate cancer where conventional fractionation to 94 Gy is now possible, although long-term outcomes are not available. The use of dose–volume histograms (DVHs) can be confusing, since different IMRT distributions can provide the same equivalent uniform dose (EUD). The use of large animal models to test IMRT was suggested, although the audience appeared critical of this suggestion. Dr Wong did not suggest that IMRT was the treatment of choice for all patients or at all tumour sites. Rather, there were concerns about the capacity to deliver high-dose IMRT accurately and with safety. Adequate accessibility to frequent CT planning scans in the UK was thought to be a potential problem. The prospect of an undetected error, with resultant overdosage by IMRT, was of real concern.
Geometrical uncertainties
Alan McKenzie (Bristol) outlined the aims and structure of the British Institute of Radiology Working Party on Geometric Uncertainties in Radiotherapy. A future publication will contain guidelines enabling centres to determine the size of the uncertainties for the sites they treat and the techniques they use. Six site-specific centres are contributing to the report, which will show how to combine the uncertainties into a margin that converts the CTV into a PTV. Systematic (treatment preparation) uncertainties should account for about 80% of the margin distance, with the remainder coming from daily (treatment execution) uncertainties.
Jane Dobbs (London) reviewed the application of the ICRU 62 volume concepts to breast cancer. Limitations of imaging modalities, such as CT in distinguishing between glandular or adipose tissue, contribute to CTV uncertainty, as does organ motion. Random variations in set-up position necessitate a PTV margin. Planning organ-at-risk volumes (PRVs), including the lung, heart and contralateral breast, were described, with techniques such as elevation of the arms above the head and deep inspiratory breath-holding to reduce cardiac radiation doses. The START Cardiac study will measure cardiac volume and dose prospectively using CT. CT simulation with accurate immobilization permits 3D target volume delineation of the CTV, PTV and PRVs as a preliminary to breast IMRT.
Elizabeth Winfield (START Trial QA Team, Mount Vernon Hospital) assessed the dose in the junction region between breast tangential fields and an anterior lymph node field. Junctional region dose (up to 7 cm depth) was estimated by film dosimetry in a phantom. Radiographers were asked to treat the phantom according to local departmental techniques, resulting in 18 different variations. The developed film profiles showed the characteristics both of treatment technique and machine type and showed that a good match was achieved with a variety of techniques that attempted compensation for beam divergence from all three fields.
Karen Venables (START QA Team, Mount Vernon Hospital) presented a study that checked breast dose accuracy. Three purpose-built phantoms were used to measure the dose in the central, superior and inferior planes. The average reference point dose close to the centre of the breast was 2% less than that prescribed, with 3D planning systems performing better than 2D systems. In vivo thermoluminescent dosemeter measurements showed that the average dose for trial patients was 0.99 of that prescribed. A large range of values was found, depending on the planning system used, the breast density and the positioning technique. These studies associated with START show the important benefits that can accrue with national trials and their quality assurance components.
Peter Canney (Glasgow) presented a study of two different patient positioning methods: (a) conventional arm-rest position abducted at 90°; and (b) breast board with arms elevated to grip a T-bar over the head. The standard deviations of the positions of each technique were very similar, showing no significant difference in set-up accuracy. The greatest variation in position was on Day 1, suggesting that portal imaging should be delayed until Day 2. The heart DVH was better using the T-grip CT-compatible positioning device than with the standard position.
Jeanette Dickson (Mount Vernon Hospital) discussed the results of a questionnaire sent to all radiotherapists in the British National Lymphoma Investigation. A compliance of 73% revealed a median of five mantle treatments and one inverted Y treatment per year for respondents. Field borders and prescription parameters were relatively uniform, but shielding practice varied from 98% (cervical cord) to 71% (thoracic cord) and 78% (humeral head). All respondents used individualized lung blocks but only 8 out of 42 included lung density corrections, introducing uncertainty in the outcome analysis because of variations in the delivered dose. This report again demonstrated the importance of a quality assurance team in radiotherapy trials.
P Tai (Regina, Canada) reported that special training given to radiation oncologists can improve the consistency with which they outlined the length of target volumes but not the transverse sections in oesophageal cancer.
George Gerrard (Leeds) reported on the results of a survey of UK clinical neuro-oncologists. 76% of the questionnaires were returned and these revealed significant differences in approach among the respondents. 66% use CT planning for patients with high grade lymphomas. 50% treat volumes based on pre-operative tumour target volume; the remainder base their treatment on the post-operative volume. 39% do not shield the optic chiasm when treating to 60 Gy in 30 fractions. As a result of these and other differences, the use of CT–MRI image fusion is recommended.
Michael Brada (Royal Marsden Hospital) delivered a plenary lecture on volume in central nervous system (CNS) irradiation. Routine immobilization devices reduced the set-up error to 0.5–3 mm for the CNS, the main uncertainty being in the definition of GTV and CTV. For malignant gliomas, GTV is defined using MRI with contrast enhancement. For benign intracranial tumours, e.g. pituitary adenomas, MRI is superior to CT in defining the GTV, but image interpretation is highly operator dependent and training is critical for oncologists for the success of radiotherapy. High relocation accuracy and clear tumour delineation has led to a CTV equal to the GTV and a small CTV–PTV margin of 2–5 mm. These narrow margins are dependent on accurate immobilization and treatment delivery as well as specialized image interpretation. Prospective studies of this type require careful long-term follow-up, as well as studies of the pattern of failures and their correlation with the details of technique and the biological equivalent doses achieved.
Physics teaching lectures
An explanation of the various forms of DVHs, with numerous practical examples from the Royal Marsden Hospital, was given by Margaret Bidmead [17, 18].
David Thwaites (Edinburgh) covered in vivo dosimetry [19], a controversial topic among physicists. Entrance diode dose estimations effectively check monitor units, beam modifiers, patient position and machine performance, whereas exit dosimetry depends upon patient factors such as inhomogeneities and treatment planning algorithms. Diodes should be calibrated using a water phantom. Doses to target volumes estimated from diode measurements are normally within 2.5% (1 standard deviation) of the planned target dose. Occasional unexpectedly high or low measured doses may occur. These discrepancies can reveal faults in the radiotherapy process, for example improper output correction with blocked fields. For routine diode use, one entrance dose measurement per patient may be sufficient and cost effective if the radiotherapy processes have already been audited. Entrance and exit doses should be measured when new treatment techniques, machines or major changes are being introduced. In vivo dosimetry is a power fulmethod for verifiying individual treatments and ensuring consistency in different centres.
Organ motion during radiotherapy
This topic is of increasing importance in radiotherapy. Although movements may be catered for by extending field margins, there are means by which these additional volumes can be minimized. John Graham (Bristol) stressed that for successful conformal radiotherapy, methods for minimizing both organ motion and motion during delivery need to be developed. There is a need not only for sophisticated devices, such as active breathing control (ABC), but also for the use of simple immobilization such as specially designed leg stocks, bladder filling and rectal emptying.
Marcel van Herk (Amsterdam) used movie loops to illustrate target movement and its impact on tumour control probability (TCP), which demonstrated the need for adequate margins. An incorrect conformally designed margin, or too tight a margin, could lead to a reduced TCP. His margin formula was based on individual centre values for a technique with a measured systematic and random error rate. It was concluded that the emphasis in the entire planning process should be to reduce the systematic errors that can lead to a geographical miss.
Christine McKenzie (Ipswich) and Cathy Hall (Bristol) studied bladder cancer patients treated conformally. Many patients were unable to retain the original volume of fluid and some required re-scanning because of changes in rectal distension. Further work is required in this important area and decisions need to be taken on the frequency of check CT scans during radiotherapy.
Peter Bownes (Mount Vernon) described the checks that have been made on movement of high dose rate brachytherapy catheters in eight prostate cancer patients. CT showed that the catheters had moved an average 14±2.5 mm inferiorly, mainly owing to tissue oedema between the prostate apex and the perineum rather than movement of the prostate itself. Individual adjustment of the catheters is now routinely performed. This study again illustrated that useful small-scale practical studies can provide information that could potentially improve the prospects of cure.
Quality control
Edwin Aird (Mount Vernon) and David Thwaites (Edinburgh) discussed the various parameters that may affect the accuracy and precision of radiotherapy by a multidisciplinary approach for commissioning, quality assurance, verification and delivery [20]. Dosimetry quality control comparisons from a Scottish Audit Group (nine centres) has shown an overall accuracy of 95% [21, 22], but dose escalation in techniques such as IMRT needs to be carefully monitored [23].
Janet Johnson (Sheffield) compared the use of a CT simulator with a conventional simulator to demonstrate which techniques can be effectively "virtually" planned and then verified on a linear accelerator with portal imaging [24]. The accuracy of a digitally reconstructed radiograph (DRR) to act as the standard against which to compare portal images requires further study. Removal of a separate physical simulation step may reduce transferral errors [25, 26].
Vicky Walker (Royal Marsden) discussed the advantages of electronic portal imaging devices (EPIDs): "real-time" acquisition, image manipulation, recording of actual treated field and the possibility of treatment intervention. The main disadvantages appeared to be limited field sizes and a pre-set position. The use of DRRs [27] and the role of EPIDs in verifying conformal and IMRT [28, 29] was considered using written image assessment protocols and radiographer-led reviews.
Susan Hilton (Bristol) assessed whether tattooing lateral marks at verification rather than at initial simulation/CT, which has the potential for transfer errors [30], improved treatment set-up accuracy. It was noted that 63% did require a change in the reference site to be tattooed at verification and that this reduced the number of changes required at the first treatment. Susan Griffiths (Leeds) considered the effects of staffing levels, technology and workload on errors. The results of a local audit on treatment errors, even on machines with computerized verification, highlighted the importance of radiographer checks (details of which have been published elsewhere [31]).
Two studies examined the use of CT planning for head and neck tumours and non-small cell lung cancer (NSCLC). Christopher Scrase described the implementation of 3D CT-based planning using a phased approach where the PTV is progressively defined during therapy to enable dose escalation. Mark McJury (Sheffield) compared conventional with virtual simulation in palliative treatment for NSCLC and demonstrated that virtual simulation permits more accurate definition of the target volume and a reduction in geographical misses and treatment-related toxicity.
It is clear that radiographic techniques continue to evolve, with greater responsibility for decision-making about the accuracy of field positions.
Imaging
The interaction of the consultant (diagnostic) radiologist with the clinical oncologist was discussed by Brendan Carey (Leeds), including the roles of various imaging techniques in determining correct TNM staging, accurate GTV definition and follow-up evaluation. GTV is not a simple line, but the defined edges will reflect the uncertainties within each imaging technique and also the observer's opinion. Anatomical imaging, such as CT, MRI and ultrasound, needs to be complemented with biological/functional imaging techniques in the future (see below). CT is used extensively for treatment planning purposes [32] to determine target volume and patient outline as well as for diagnosis. MRI provides good soft tissue definition and is particularly useful for head and neck, prostate and bladder tumours as well as being useful for follow-up information [33]. Position emission tomography (PET) is useful for lung and breast cancer staging and MRI spectroscopy is proving to be valuable for IMRT. Good communication between radiologist and oncologist is required to choose the best and most appropriate imaging technique(s) before, during and after treatment. There is a tendency for each department to devise local solutions for problems that are, in essence, generic. There needs to be a mechanism through which information and expertise are shared nationally.
Functional/biological imaging
Pat Price (Manchester) summarized the potential of PET scanning in tumour staging (e.g. whole body PET with labelled thymidine is the equivalent of a soft tissue "bone scan") and for quantification of tumour characteristics (e.g. hypoxia or vascular endothelial growth factor (VEGF) status). The pharmacokinetics of labelled drugs such as temozolamide or 5-fluorouracil (5-FU) can be studied within patients, and labelled thymidine can also be used to measure response. MRI can also provide vascular kinetic measurements as well as biochemical spectroscopy data. In radiotherapy planning, there are studies comparing the use of CT/MRI and PET, in which lung appears to be the most promising site [34, 35]. Fusion of images may eventually provide the opportunity to give additional radiotherapy to a specific biological target volume, e.g. a severely hypoxic focus. Such additional partial tumour boosts would need to be delivered by advanced focused forms of radiotherapy such as IMRT or protons.
Radiochemotherapy
Two excellent literature reviews were given. Tom Keane (Vancouver) summarized the difficulties encountered in trying to define the optimal treatment of oesophageal cancer [36, 37]. Radiochemotherapy is now regarded as the treatment of choice, although there is little randomized control trial (RCT) evidence. There is also no convincing evidence that the addition of further cytotoxic drugs to 5-FU results in further benefit. Attempts at the addition of brachytherapy have further increased morbidity, although total doses were high and there remains the potential for selective lower dose brachytherapy.
Trevor Roberts (Newcastle) reviewed the role of radiochemotherapy in bladder cancer [38], with examples of many types of multimodality treatments available that have not been tested in RCTs but have been published to demonstrate high tumour control rates. Neoadjuvant therapy is not effective, but concomitant radiochemotherapy appears to be the best available option in patients that are sufficiently fit to tolerate the treatment.
Biological/molecular uncertainties
Allan Price (Edinburgh) explained the idiosyncratic, genetically determined forms of radiation DNA repair disorders that confer extreme radiosensitivity (there are now five established syndromes). These concern patients who suffer normal tissue damage that cannot be explained by the dose given and the volume treated [39]. Retrospective studies on tissues from 829 patients (overall 11% serious morbidity; 3% at <75 Gy and 3% at >85 Gy) showed low expression of the catalytic unit of the enzyme DNApk, particularly in breast cancer patients. There is a need for prospective studies and the development of protocols for the detailed investigation of such patients in future or established clinical trials.
Paul Workman (Sutton) summarized progress in molecular biology and the search for novel drugs that inhibit multiple biochemical processes known to be important in cancer cells. A detailed description was given of the HSP90 molecular chaperone that simultaneously controls folding of proteins that have functions relating to the cell cycle, e.g. rRaf, CDK4, ser 473 and p53. The HSP90 inhibitor 17 AAG has already been used in clinical trials [40, 41]. The new genomic and proteomic techniques are producing very large data sets that will allow the study of multiple gene expression as well as a clearer understanding of the biochemical dynamics that govern outcomes and will also allow a more rational approach to pharmacological interventions.
The importance of novel forms of drugs that modify tumour proliferation, or enhance tumour cell kill independently of radiation, or preferentially radiosensitize tumours will need careful testing in patients and may eventually have important roles in radical radiotherapy schedules. It is important that key members of all the disciplines within radiotherapy be aware of these developments and how they may impact on future practice.
Uncertainty in normal tissue responses and radiation morbidity
Stanley Dische (Mount Vernon) described the greater awareness of morbidity associated with radiotherapy patients who have been cured and who have an expectation of complication-free survival. Although clinical trials are now likely to incorporate morbidity scoring, there are many different systems available and none have been validated as an international tool. Robin Hunter (Manchester) described some of the many factors contributing to radiation morbidity that add to uncertainties in scoring. These included differences between patients in normal tissue anatomy, radiation dose, irradiated volume and the radiosensitivity of normal tissues. There is a clear need to identify the most important factors in order to separate patients who are curable by standard techniques from those capable of tolerating more aggressive multimodality or adjuvant treatment. A national audit of late morbidity in patients with carcinoma of the cervix treated in 1993 [42] highlighted the additional importance of tumour stage and the type of radiotherapy given. Jane Maher (Mount Vernon) stressed the need for an integrated care pathway for treating complications arising from treatment, as well as the commitment of the clinical oncology community to monitor and reduce morbidity. An international scoring system has been developed that has resulted in publication of the LENT SOMA scales [43].
Although groups are starting to adopt the new system, there has been no widespread validation exercise. Susan Davidson (Manchester) showed, for carcinoma of the cervix, that LENT objective scores correlated well with results obtained using an established scoring system (the French–Italian glossary), while subjective data correlated well with established quality of life scores. Neil Burnet (Cambridge) stressed the need to address the analytical component of the LENT SOMA scales, which is becoming increasingly important with the advent of advanced radiotherapy treatment planning such as IMRT.
Radiobiological modelling
An overview of the concept of biologically effective dose (BED) by Bleddyn Jones (Hammersmith) stressed the need for clinical oncologists to be aware of the potential benefits of developing a quantitative radiobiological approach [44]. The many practical uses include multiple BED evaluations to reflect dose variation adequately throughout a large PTV, the effect of chemoradiotherapy in BED units and the use of different 
/ß ratios for different histological types of cancers. Trials of different radiotherapy schedules can now be simulated using randomly generated radiobiological parameters. Future clinical trials should investigate whether predictive assay determined treatment and biological optimization studies can produce substantial improvements in tumour control, in contrast to the small gains found in conventional randomized trials [45].
Stephen Roberts (Manchester) described modelling studies that have been carried out using clinical data sets. The analysis of clinical series has produced greater radiobiological understanding as well as parameter values that can be used in practical radiobiological models. A retrospective dose-fractionation study in breast cancer was presented by Jackie Worville (Nottingham). The elevated BED values of the hypofractionated arm correctly predicted an increase in normal tissue toxicity.
In many radiotherapy departments there is still uncertainty regarding how best to apply a reasonable correction for extended overall treatment time caused by interruptions. Methods of compensation, using the BED concept, were described by Roger Dale, and new guidelines and examples of these approaches should be published in 2002 [46]. Li Tee Tan and Neil Burnett (Cambridge) warned that the use of treatment gap compensation schemes and increasing treatment complexity (as well as other factors such as linear accelerator numbers and radiographer staffing levels) were leading to an increase in waiting times for radiotherapy and that this delay would also cause a significant reduction in the probability of tumour control.
Clinical radiobiology of tumours
Over the past decade there has been considerable interest in radiosensitivity, proliferation and hypoxia. In an overview, Catharine West (Manchester) provided experimental and clinical evidence to show the importance of all three parameters as determinants of the outcome of radiotherapy [47]. So far no test has been introduced into routine clinical practice owing to the inaccuracy of available assays. Although there is no method currently available for the routine measurement of tumour radiosensitivity, Peter Hoskin (Mount Vernon) described an interesting method for measuring tumour proliferation that involved examining the pattern of BudR staining in tumours. Work in head and neck cancers showed an association between BudR staining pattern and response to accelerated fractionation, with a marginal pattern predicting a favourable outcome to CHART compared with standard fractionation. There is interest in the assesment of tumour oxygenation using oxygen electrodes and the hypoxia-specific probe pimonidazole. From a clinical perspective, tumour hypoxia can be asessed by imaging. Juliette Loncaster (Manchester) suggested that dynamic contrast enhanced MRI data correlated with the level of tumour oxygenation [48] and prognosis in patients with cervix cancer. The dynamics of re-oxygenation can be followed during therapy, allowing identification of tumours that do not efficiently re-oxygenate and where increasing dose or use of adjuvant treatments (e.g. bioreductive drugs) may be required. Michelle Saunders (Mount Vernon) provided an excellent account of fractionation including the Royal College of Radiologists fractionation survey [49]. She described developments in accelerated radiotherapy and the evolution of the CHARTWELL (weekend-less) schedule that is predicted to provide superior outcomes to the CHART schedule [50]. Future fractionation of IMRT is a problem that may be clarified by the use of normal tissue complication probability–tumour control probability (NTCP–TCP) plots for different fractionation schedules.
Clinical trials
National clinical trials have a major role to play, not just in establishing best practices for the safe and accurate delivery of radiation therapy, but also in facilitating the diffusion of such practices throughout the country. Moreover, meticulous small studies can provide information of general applicability and usefulness.
Max Parmar (London) described the historical evolution of Medical Research Council (MRC) clinical trials, emphasizing that it is the number of events rather than number of patients that is the important determinant of statistically significant changes in outcomes. It has taken 15–30 years to define significant improvements, e.g. for adjuvant 5-FU chemotherapy in rectal cancer. Meta–analyses have helped to clarify results where treatments are compared with no treatments, but these are less effective when different treatments are compared. Trial recruitment is always difficult and, as a general rule, 2000 patients are required to detect a 20% reduction in the relative risk of death in trials of early cancers, whereas for a 30% risk reduction in more advanced cancers approximately 500–600 patients are required.
Future trial design is likely to test for progression-free survival in a few hundred patients only, since this is independent of salvage treatments; and then from the most favourable categories, a survival end-point trial may be designed. Alternatively, large mega-trials may be started with interim analysis to discard arms with insignificant improvements in progression-free survival. In these ways it may be possible to arrive at definite conclusions in some tumour types within 5 years. The MRC are developing new ways to conduct clinical trials, but there is perhaps still a reluctance to consider end-points based on the therapeutic ratio (tumour control coupled with normal tissue effects) in radiotherapy trials. By incorporating morbidity as well as survival into clinical trials of radiotherapy we may, by increasing the number of events available for analysis, increase the efficiency of these trials.
Summary
This conference proved to be an important landmark in the development of multidisciplinary collaboration in UK radiotherapy and there are already plans for a further conference at The University of Bath (6–9 April 2003), the theme of the meeting being "Optimisation in radiotherapy". There were many lessons for UK radiotherapy and some important caveats. Some new developments, such as IMRT, appear to involve substantial resources for safe and effective application: the risk management considerations are considerable, particularly owing to the many assumptions within the processes involved. Enhanced quality assurance levels are clearly indicated in the more sophisticated forms of radiotherapy. Perhaps such advances will only be achieved at large centres that have adequate academic units of physics, clinical oncology and radiography with sufficient charitable and targeted NHS research and development funding. Alternatively, more progress may be made in departments with smaller workloads, where staff can concentrate on developmental issues. There is clearly a danger that many departments will independently attempt to define practical ways to deliver IMRT; national guidelines and trials including quality assurance are urgently required.
The use of mathematical modelling of the various processes in radiotherapy also appears to be of increasing importance in guiding clinical decisions and clinical trial design, and within audit of the overall process of radiotherapy, including workload considerations. Mathematical simulations of the entire process may allow optimal use of resources: techniques such as linear programming and operational research have been successfully used in industry for many decades. Within radiotherapy, such work usually depends on the goodwill and enthusiasm of the contributors and is not funded formally. Adequate resource support is essential.
The definition of target volumes should include the contributions of histopathologists and radiologists. This may be a suitable topic for a future UKRO meetings, involving colleagues from the relevant disciplines.
The professions involved in radiotherapy need to be aware of developments in related fields: the molecular pathology of cancer; advances in diagnostic imaging; and the potential interaction between radiation and novel approaches to drug treatment of cancer.
The conference confirmed that the next decade will be crucial for the development of clinical oncology. There is a wide and increasing opportunity for more research to improve accuracy where necessary and to reduce uncertainty where possible.
Acknowledgments
The authors are grateful for the advice of Dr Victor Barley in the preparation of this article.
Received for publication August 21, 2001. Accepted for publication December 20, 2001.
References
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| BJR | DMFR | IMAGING | ALL BIR JOURNALS |
