Applications of magnetic resonance spectroscopy in radiotherapy treatment planning
G S Payne, DPhil and M O Leach, PhD, FInstP, FMedSci
Cancer Research UK Clinical Magnetic Resonance Research Group, Institute of Cancer Research and Royal Marsden NHS Trust, Downs Road, Sutton, Surrey SM2 5PT, UK
Following advances in conformal radiotherapy, a key problem now facing radiation oncologists is target definition. While MRI and CT provide images of excellent spatial resolution, they do not always provide sufficient contrast to identify tumour extent or to identify regions of high cellular activity that might be targeted with boost doses. Magnetic resonance spectroscopy (MRS) is an alternative approach that holds great promise for aiding target definition for radiotherapy treatment planning, and for evaluation of response and recurrence. MRS is able to detect signals from low molecular weight metabolites such as choline and creatine that are present at concentrations of a few mM in tissue. Spectra may be acquired from single voxels, or from a 2D or 3D array of voxels using spectroscopic imaging. The current state of the art achieves a spatial resolution of 6-10 mm in a scan time of about 10-15 min. Co-registered MR images are acquired in the same examination. The method is currently under evaluation, in particular in brain (where MRS has been shown to differentiate between many tumour types and grades) and in prostate (where cancer may be distinguished from normal tissue and benign prostatic hypertrophy). The contrast achieved with MRS, based on tissue biochemistry, therefore provides a promising alternative for identifying tumour extent and regions of high metabolic activity. It is anticipated that MRS will become an essential tool for treatment planning where other modalities lack the necessary contrast.
Encouraged by positive responses to our reviews of 2004 and 2005, the Editors again offer a selection of papers which they feel have made a mark during 2006. These have been chosen for the immediate impact they have had on individual Editors, rather than by any formal assessment or comparison, and we hope that they will remind readers of some key radiological issues of 2006, or at least alert them to what they might otherwise have missed.
Since the introduction of online submission, the British Journal of Radiology continues to receive an increased flow of papers, with 2006 exceeding all previous records. Submissions were received from 53 countries, with just over one quarter from the UK. Encouraged by this, the Editorial Board has explored several ways in which the overall quality of published papers may be enhanced and we look forward to implementing these during 2007. Although we may view the Journal Impact Factor with some scepticism, we are conscious of its standing, particularly in the UK, and most of our proposals for raising standards will, we hope, be reflected in an increased Impact Factor in future years.
A mention of these initiatives gives us the opportunity to express our appreciation and thanks to Dr Philip Dendy, Honorary Scientific Editor and a driving force for improved quality, who retired from his position in September 2006, after many years of invaluable service to the British Journal of Radiology, including the last eight as Deputy, and then Honorary Scientific Editor.
One of the many highlights of the year, the annual British Institute of Radiology President's conference, reflects the interests of the incumbent President, and the year opened with an overview of the 2005 Conference "Technology in Imaging and Radiotherapy - towards improved workflow and productivity". The wide ranging nature of the subject matter, from the status of CT to IMRT (intensity-modulated radiation therapy), and with tributes to Sir Godfrey Hounsfield, is typical of the British Institute of Radiology's multidisciplinary ethos and was summarized by Dendy [1]. The CT papers, in particular, offer an insight into the development of CT and illustrate the unpredictability (at least to those not intimately involved) of the development of this technology. After the initial surge in development in the 1970s, immediately following the announcement of the invention, the pace slackened in the 1980s, only to gather renewed momentum in the following decade with the emergence of spiral CT and multidetector arrays.
One can speculate about the future development cycles of other less well established technologies, one such being particle radiotherapy, also covered in the President's Conference. Jones [2] pointed out the recent disappointments of rejected bids for UK particle therapy centres in contrast with similar facilities abroad, and suggests that 5000-12 000 patients might need therapy abroad if no UK centres are forthcoming.
Diagnostic radiology
The broad range of topics that are met on a day-to-day basis by clinical and academic radiologists has again been reflected in last year's published diagnostic papers. The majority of papers have focused on the ever increasing armamentarium of diagnostic and therapeutic techniques available and have, as one would expect, demonstrated that modern technology is of benefit, but this may be at a cost.
The paper by Taylor et al [3] on CT colonography reflects the trend away from conventional barium examinations in the investigation of bowel disease, and confirmation that this trend is justified is increasingly seen in the published literature and in clinical practice. Their paper suggests that CT colonography is both more accurate than barium enema in the detection of polyps greater than 6 mm in size, and importantly may also be reported with a high degree of confidence by experienced observers. Interestingly, two of five colonic cancers in the study group were only detected by the CT examination.
The additional value or otherwise of CT colonography, namely the ability to detect extracolonic abnormalities, has been less well studied. The paper by Xiong et al [4] reviews the additional financial cost of detecting disease warranting investigation and suggests that the additional findings may result in a doubling of the examination cost, and also result in potentially significant morbidity and possibly even mortality. This is unsurprising given the high prevalence of additional unrelated abnormalities, 116 of 225 patients investigated in their report, and is not dissimilar to reports on additional abnormalities discovered in patients in lung cancer screening programmes [5]. Perhaps such examinations may only be regarded or reported as normal in the future, if an additional and clinically irrelevant abnormality is detected!
Riddell et al [6] reported on the utilization of magnetic resonance imaging (MRI) in the investigation of oesophageal disease in a similar manner to its use in the investigation of rectal cancer. They are to be congratulated on their meticulous application and assessment of different scanning parameters, including cardiac gating, and this paper is a lesson to us all in the careful investigation and application of new technologies in disease investigation.
PET/CT is becoming more widely available. Its value in colorectal cancer is well recognized, but two papers showing novel use of the technique in investigating gastrointestinal disease are worthy of mention. Goshen et al [7] described the use of PET/CT to detect abdominal wall and port site metastases. They concluded that the technique seems to be a sensitive tool for the diagnosis of abdominal wall recurrence with accurate localization enabling functional aspects of PET supplemented by the anatomical features of CT to help such recurrences to be resected with curative results. The same technique was also described by Zissin et al [8] who utilized it for diagnosing mesenteric panniculitis. The technique allows differentiation between mesenteric panniculitis and mesenteric tumoural involvement, an almost impossible task using CT alone.
It is always interesting to read of a technique which has the potential to replace a more invasive one. Hollingsworth et al [9] described the technique of rapid non-invasive MR assessment of hepatic lipid content. They used it in assessing the liver of those on low carbohydrate diets, using 10 volunteers. They showed a strong correlation between the initial fat content and the percentage of fat content reduction in the first 3 days of the diet. Weight loss was not correlated however with hepatic fat loss after 3 days or 10 days of dieting. The technique is very promising for hepatic fat quantification in longitudinal studies and may reduce the need for liver biopsies in some cases.
With the increasing use of interventional radiology, it was instructive to read the paper by Vaño et al [10] on occupational radiation doses in interventional cardiology spanning a 15 year period. It was perhaps gratifying to note that a 14% reduction has been achieved in the doses under and over the lead apron from 1989 to 1992. The most effective action in reducing operator dose has been patient dose reduction and the systematic use of radiation protection measures, particularly the use of ceiling suspended protective screens.
Hiorns et al [11] reminded us that it may well be possible to keep radiation doses below a national standard. They reviewed current local dose-area product (DAP) levels in a paediatric setting and reported marked reductions, by a factor of between 5 and 25 below the current national reference doses, over a 21 month period. They make the point that with appropriate education, national recommendations can sometimes be bettered.
All radiologists believe that they are good at detecting pulmonary nodules. Manning et al [12] studied this with volunteers consisting of radiologists, experienced radiographers and novices. Observers' eye movements were tracked and correlated. True negative decisions from all observers were associated with shorter fixation times than falsenegative decisions. No correct negative decisions were made after fixations exceeding 3 s. True negative decisions were made more quickly than false ones. Perhaps rapid reporting has some advantages!
An increasing number of papers reporting in vitro or animal studies are submitted to the British Journal of Radiology, and these often provide an insight to future studies in man. The report by Lee et al [13] on the investigation of different radiofrequency electrodes for ablating disease in lungs will inform future investigators when, as is increasingly the case, this technique is used to ablate both primary and metastatic disease in the lungs. Meticulous attention to detail is necessary to inform us of best practice, and this report provides an example of just such methodology. An interesting report of the characteristics of gold nanoparticles as a new X-ray contrast agent in mice [14] noted a threefold improvement in contrast relative to iodine at 100 keV, coupled with, amongst other characteristics, a greater tumour retention and lower muscle accumulation. This suggests the possibility of enhanced tumour detection, and the extension from mouse to human studies is awaited.
Diffusion-weighted MRI
One of the aims of new and improved imaging techniques is to differentiate between malignant and non-malignant tissues at all stages in a patient's care. A cluster of papers in the August 2006 issue looked at this aspect of some recent advances in the functional imaging technique of diffusion-weighted MRI (DWI).
A commentary by Koh and Padhani [15] gives a good overview. At a fundamental level, DWI provides information on the random motion of water molecules in tissues and this will manifest itself as a standard deviation about some mean position over a period of time characterized by a diffusion coefficient. In tissues, diffusion is primarily in the extracellular space, but is modified by interactions with hydrophobic cellular membranes and macromolecules. The resulting effect is sometimes referred to as "apparent diffusion" and differences in apparent diffusion can be a means of tissue differentiation.
Images are acquired with various b values (derived from the amplitudes, lengths and intervals between diffusion gradients) and the data are processed to obtain "apparent diffusion coefficients" (ADC). It is important to note that tumour shows as a cold area on a functional ADC map (indicating a low diffusion rate) but as a hot area on original images obtained at high b values.
Vandecaveye et al [16] examined the value of DWI for differentiation of persistent or recurrent tumour from post-radiotherapeutic sequelae or complications for four patients with laryngeal squamous cell carcinoma. The two true positive patients demonstrated the appearance of recurrent tumour tissue on the DWI images. The lesions were hyperintense on the b 1000 map (b = 1000 s mm-2) and hypointense on the ADC map, in contrast to the surrounding tissue. In the two true negatives, the diffuse ADC value and the absence of any focal restrictive signal on the b 1000 images were consistent with diffuse laryngeal necrosis and absence of tumour recurrence.
In the same issue Fan et al [17] evaluated the usefulness of DWI and the related technique of perfusion-weighted MRI (PWI) for predicting tumour grading of supratentorial brain gliomas. Some 15-45% of supratentorial gliomas which show no enhancement on conventional contrast-enhanced T1 weighted MRI images were subsequently found to be malignant. 22 patients with such non-enhancing tumours were enlisted as the cohort for this study. Histologically, 14 tumours were low grade gliomas (WHO Grade I and II) and 8 tumours were anaplastic. Low ADC values were found in the solid portions of the anaplastic gliomas but not in low grade tumours. On the other hand, ADC values in the peritumoural regions, although lower than in the contra-lateral normal white matter, were a poor discriminator, showing no significant difference between anaplastic and low grade gliomas.
Unlike conventional contrast-enhanced imaging, perfusion imaging may be used to obtain information on tumour angiogenesis with or without the destruction of the blood-brain barrier. Fan et al [17] expressed perfusion data in terms of relative cerebral blood volume ratios, and these were raised in both solid portions and peritumoural regions of anaplastic gliomas, but not in low grade gliomas.
In the simplified diffusion theory normally applied in DWI, it is assumed, to a first approximation, that diffusion is isotropic. However, in highly structured parts of the body this assumption may not be justified and a more complete mathematical treatment of diffusion is required resulting in MR diffusion tensor imaging (MRDTI), where the diffusion coefficient is resolved along the three principal axes.
The standard methodology is to compute the mean diffusion rate and the fractional or relative anisotropy. In the February 2006 issue, Peña et al [18] proposed an alternative approach in which the diffusion tensor is decomposed into its isotropic (p) and anisotropic (q) components. The method was tested on a healthy volunteer, a sequential study on a patient with a recent stroke, a patient with hydrocephalus and a patient with an intracranial tumour. Distinctive p:q patterns were recorded for the four cases, but further work will be required to determine if a particular pattern is sufficiently unique to be of diagnostic value.
Fan et al [17] conclude that DWI and PWI should be integrated in the diagnostic work-up of non-enhancing gliomas in order to better predict grading and, taken as an ensemble, the papers reviewed here suggest that these forms of functional imaging will make an important contribution to tumour evaluation. However, in respect of DWI, Koh and Padhani [15] draw attention to some outstanding challenges, including (a) overlap in ADC values between malignant and non-malignant tissues, (b) averaged ADC values from selected regions of interest may not adequately characterize tumour heterogeneity. The work of Peña et al [18] shows that in MRDTI there remains a wealth of data to be uncovered, but unfortunately we still do not know which tensor scalar measure or measures best describe brain tissue and its pathological changes.
Radiotherapy and oncology
One clear message that emerges from reviewing the British Journal of Radiology in 2006 is the increasing importance of imaging in the day-to-day practice of radiation oncology. We are a long way from 20 years ago, when a few plain films and a pair of orthogonal films on the simulator were all that it took to get a patient from diagnosis through to the completion of treatment. It is not just the emergence of the new technologies, CT, MRI, PET and online portal imaging that have influenced practice, it is also the way in which these technologies are integrated into departmental protocols and management plans for individual patients. There is usually the assumption that more is better, and the result is that work expands to fill the technology available. There are consequences from all of this and several papers in the British Journal of Radiology in 2006 either deliberately or inadvertently point to some of the difficulties and questions that may arise. A special issue in September 2006 was devoted to issues concerning imaging in radiotherapy treatment planning and delivery. There were important contributions on functional imaging [19, 20] and on how to fuse and manage images acquired using different technologies [21]. But this was not all: there were two substantial publications on the role of imaging in improving the planning of radiotherapy for head and neck cancer [22, 23], and other important contributions on functional imaging [24, 25]. Two papers [26, 27] attempted to provide a clinical context for the use of imaging. Both suggested that, for follow-up at least, excessive imaging of asymptomatic patients is unlikely to be helpful.
So, what are the issues raised by the papers published in 2006 on imaging and radiotherapy? The first issue is that we need to know far more about whether the investment of time and effort in better localization and definition of target volumes is worthwhile. The images are of good quality, registration is perfect throughout treatment, but is the patient any better off? If there are better outcomes, what has the cost been and are the results so much better that extra costs are justified? We also need to address the issue of opportunity cost: imaging capacity is finite, we already have waiting lists for diagnostic imaging procedures. A scanner that is being used for target localization in radiotherapy is a scanner that cannot be used for diagnostic or staging scans. We need to be explicit about the trade-offs we make when we introduce unproven technologies. Nowhere in radiation oncology is this problem so acute as it is in the introduction of new image-based technology into the processes of planning and delivering radiotherapy treatment. Another key issue concerns resources - not just machines, but people and expertise. Who is going to interpret the images and integrate these interpretations into the clinical context: what is the appropriate PTV (planning target volume) for this particular patient with this particular tumour at this particular time? As a journal, the British Journal of Radiology is healthily positioned to facilitate and co-ordinate debates on these issues. By looking back on what we published in 2006 we can easily define the framework for the discussions. Looking forward to 2007, perhaps we will be able to publish more in the way of analysis and less that is simply hopeful expectation, analysis that will inform the important debates that we need to have if we are to ensure that the introduction of these new techniques is to be handled sensibly. It is not just about what we can do, it is also about what we should be doing.
Radiobiology
A strong recurring theme amongst the radiobiology papers this year was the perennial concept of hypoxic radioresistance in radiotherapy. While the importance of this issue has been recognized for decades, new evidence continues to accumulate, reinforcing its importance. We were reminded of the history of the topic, inextricably linked to the work of LH Gray and his colleagues, in a commentary by Dendy and Wardman [28]. It described contributions to an international meeting marking the centenary of the birth of LH Gray and celebrating his contributions to science.
More specifically, the role of hypoxia in determining outcome in head and neck cancer was reviewed in the October 2006 issue [29]. Potential markers for hypoxia in the clinical setting were discussed, together with strategies to reduce levels of hypoxia or to target hypoxic cells with hypoxia-specific cytotoxins. A combination of these approaches could lead to individualized therapy based on hypoxic status of tumours.
The impact of hypoxia on the effectiveness of radiotherapy in prostate cancer was the subject of a modelling study [30]. These authors used measurements of relative hypoxia in prostate tumours to calculate the additional dose that would be needed in the clinical setting to compensate for hypoxia. This "clinical oxygen enhancement ratio" was in the range 1.2-1.5, depending on the exact modelling parameters used, emphasising the importance of hypoxia in this disease.
Radiation protection
The UK Radiological Congress debate in 2004 on low doses and risks [31] seems to have helped to keep alive the controversies surrounding the risks of low dose radiation. Wall et al, in a comprehensive review article on this subject [32], discuss the current consensus of the linear no-threshold (LNT) model and its challenges from those who argue for a non-linear relationship between dose and risk at low doses, due to possible adaptive responses. For practical patient protection in radiology, the emergence of adaptive response hypotheses, interesting though they are, are overshadowed by the need for practical advice. For these purposes, Wall et al argued that the LNT model provides sufficiently reliable risk estimates for patient protection and give four risk bands into which all radiological examinations may be divided.
The implications of the dose-risk relationship at low doses comes into sharp focus when applied to mammography. Heyes et al [33] considered the implications for mammography screening programmes of enhanced relative biological effectiveness (RBE) for mammography X-rays of between 1 and 6, and on this basis advised caution in programmes involving early regular screening in women with a family history of breast cancer. This view was challenged by Law et al [34] and Redpath and Mitchel [35] and defended by Heyes et al [36] in subsequent letters. In the field of low dose risk, controversy is almost guaranteed because of the paucity of hard data and the numerous variables involved (such as in vivo vs in vitro data, non-linear hypotheses, extrapolation from high doses and photon energy dependence). Law and Faulker [37] also discuss radiation benefit and risk at the assessment stage of the UK breast screening programme.
In contrast to today's low dose issues, an historical perspective on patient and staff doses in radiology was given by Kotre and Little [38] who estimated the doses received by radiographic staff between 1899 and 1902 in Newcastle by simulating the X-ray characteristics of a cold cathode X-ray tube and applying the results to radiographic examinations recorded in a log book of the period. The radiographer who carried out these examinations suffered severe radiation injuries and his effective dose was estimated as 940 mSv per year. The median entrance surface dose for posteroanterior views of the chest was 68 mGy, nearly 500 times greater than expected today.