BJR review of the year – 2009

This has been a most successful year for the journal, with more than 1000 submissions received for the first time, leading to a substantial increase in the rejection rate. The diversity of interesting papers has been maintained in all areas covered by the journal, and this review attempts to highlight what have been considered by the editors and deputy editors to be the most noteworthy papers.

The development of conformal radiotherapy techniques leads inevitably to continuing interest in the geometric aspects of treatment delivery. Intuitively, the value of conformal radiotherapy must be compromised if the volume irradiated is uncertain, or if the geometric errors (both systematic and random) involved in field placements are imprecisely known. Several papers published in 2009 have addressed these issues. Osei et al [1] studied the magnitude of interfraction set-up errors and prostate displacement in 118 patients using three gold seeds implanted within the prostate. Comparing electronic portal images and digitally reconstructed radiographs, they found that random errors were generally greater than systematic errors, and that antero-posterior displacements were generally greater than left–right or superior–inferior displacements. In the same field, McGarry et al [2] studied two methods of determining set-up errors in prostate radiotherapy using portal imaging devices before and after a move to a new cancer centre. They suggest that correcting a patient's set-up by applying a 5 mm rather than a 3 mm action level would not be detrimental.

In evaluating geometric accuracy, two imaging techniques may be better than one. Webster et al [3] fused MR and CT images of the head and neck for radiotherapy target delineation. They found that image quality was superior when surface coils were used rather than body coils, with distortions <1 mm out at approximately 90 mm radius and image registration accuracy ∼2 mm.

The need for geometric accuracy, which can be improved by a multiplicity of imaging techniques, is accompanied by the requirement to justify and minimise doses to organs at risk. Two papers address this dosimetric issue: Sawyer et al [4] estimated skin and effective doses from kilovoltage cone beam CT, whereas Roxby et al [5], using a similar system, were able to demonstrate that dose was reduced significantly by using a copper filter.

Controversies surrounding the assurance of dosimetric accuracy and the prevention of serious errors continue to attract comment. In particular, the debate about the cost-effectiveness of in vivo dosimetry continues, following the recommendations of the Chief Medical Officer [6] and previous suggestions in this journal that in vivo dosimetry is a cost-effective technique for the prevention of otherwise undetected dosimetric errors [7]. By contrast, Mackay and Williams [8] argue that both under- and over-doses must be considered, and that the impact of an in vivo dosimetry programme on a 0.002% rate of serious dose maladministration would not lead to demonstrable improvements in outcome. Practical issues in this same subject were addressed by Edwards and Mountford [9] who drew attention to the effects of the electron contamination and X-ray energy response of diodes and lithium fluoride thermoluminescence dosimeters, showing that there are both scientific and political issues still to be addressed.

Radiological examinations in which the optimisation of patient dose and image quality continue to attract attention are those that produce high values of patient dose (e.g. cardiac interventional examinations), include tissues that are especially sensitive to radiation, or in which new technologies have been adopted recently, such as computed radiography (CR) and multislice computed tomography (MSCT). For instance, Moore et al [10] recommend that, where practical, the outcomes of optimisation studies that are based on theoretical simulations or on phantom or animal measurements should be clinically validated before routine implementation.

Diagnostic reference levels (DRLs) give an indication of the scope for patient dose optimisation, and Hart et al [11] provided a comprehensive summary of the national reference doses that form the basis of the current DRLs in the UK. They concluded that more dose data were required for paediatric radiographs, and that more information was needed on the effect of patient height and weight or body thickness on the X-ray beam. More detailed information is also required to describe the complexity and anatomical location of common adult interventional and angiographic procedures to allow reference doses to be derived for more precisely specified techniques. Using the dose–area product (DAP) values recorded for 1804 adult patients, D'Helft et al [12] derived preliminary DRLs for use in four common types of cardiac interventional procedures. Bogaert et al [13] recorded the skin-dose distribution from a grid of 70 thermoluminescent dosimeters (TLD) secured to 318 patients undergoing cardiac interventional procedures, together with the corresponding DAP values and other relevant factors. The mean skin dose exceeded the 2 Gy threshold dose for deterministic effects in 3% of the patients. A patient follow-up strategy was proposed, based on two DAP action levels, and a DRL of 71.3 Gy cm² was derived for diagnostic procedures and 106.0 Gy cm² for all types of therapeutic procedures (single, multiple, and combined with a diagnostic procedure).

The female breast is associated with increased radiation sensitivity and is always included in thoracic CT images without usually being the organ of interest. Vollmar and Kalender [14] concluded from thoracic CT simulations and phantom measurements that a reduction in the tube voltages used would yield a significant decrease in the breast dose without affecting the contrast-to-noise ratio. Ledenius et al [15] pointed out, first, that although modern multidetector computed tomography (MDCT) units have the capacity to modulate the tube current automatically according to patient size and region of interest, image quality requirements will dictate the final radiation dose. Second, they showed that adjustment of tube current is particularly valuable for the investigation of paediatric patients because of their increased sensitivity to radiation. By adding artificial noise (in steps representing dose reductions of 20 mA each) to raw MDCT brain images of paediatric patients aged 1–5 years, they produced an acceptable reproduction of low- and high-contrast structures at CT dose index values by volume (CTDI_vol) of 20 and 10 mGy, respectively. Karambatsakidou et al [16] showed that, in order not to underestimate the dose to very young patients, age-dependent factors are required to convert a DAP value to an effective dose in paediatric interventional cardiology, and that the effective dose is of much greater concern than the skin dose.

Moore et al [10] measured the frequency-dependent normalised noise–power spectra, signal-to-noise ratio and tissue-to-rib ratio in images of a chest phantom obtained with an Agfa CR system. They found that processing was optimal with an exposure (speed) class of 600 over a wide range of effective doses (0.05–0.8 mSv). Using factorial experiments to evaluate several parameters concomitantly, Geijer et al [17] showed that when a flat-panel detector is used for lumbar spine radiography, provided the system speed is also increased, a reduction in the tube potential leads to a lower effective dose and/or increased image quality.

Arthurs et al [18] drew attention to growing concern over the radiation dose from CT imaging, and to the continual pressure for radiology services to provide the latest CT technology in order to allow faster procedures and a greater range of examinations. They found, however, that a 64-slice CT system produced significantly better image-quality scores and lower effective doses than a 16-slice CT system when 15 children and young adults underwent thoracic examinations on both systems. From a study using photodiode measurements in two anthropomorphic phantoms, one representing an adult, the second a 6-year-old, Fujii et al [19] found that a 64-slice CT system produced the same CTDI_vol, dose–length product and effective dose values for chest and abdominopelvic examinations as 4-, 8- and 16-slice CT systems.

MSCT can be used as a non-invasive method for imaging the whole coronary artery tree but it is susceptible to motion artefacts, which can be overcome by β-blockade to reduce the heart rate. Schnapauff et al [20] demonstrated that multisegment reconstruction significantly reduced motion artefacts compared to half-scan reconstruction in 126 consecutive patients with suspected coronary artery disease. Furthermore, multisegment reconstruction produced greater image quality and better diagnostic accuracy than half-scan reconstruction even without β-blocker administration. Dual-energy subtraction radiography, consisting of two separate images (one at low kV_p, the other at high kV_p), can be used for the detection of calcification within primary nodules, of bone and pleural abnormalities, and of hilar and mediastinal masses, and for the localisation of implants. The subtracted images can, however, include motion artefacts, which can be avoided by the use of electrocardiogram gating. Kyriakou et al [21] used correspondence matching of dual-energy images of a thorax-motion phantom and of a rat taken with a C-arm flat detector system to produce motion-free bone and soft-tissue subtracted images without the need for gating. The use of cone beam CT for dental examinations is increasing in the UK. Roberts et al [22] found that effective doses derived from thermoluminescence dosimetry (TLD) measurements in a phantom for a range of imaging protocols were low compared to the effective doses from conventional CT examinations, but they were significantly greater than those from conventional dental radiography.

The scope for optimising staff doses should also be considered when new equipment and procedures are introduced. If such developments involve the administration of radiopharmaceuticals, they are likely to have greater implications for occupational doses because of radiation exposure of clinical staff carrying out processes involved with these procedures (e.g. the preparation and administration of the radiopharmaceutical, contact with the patient after administration, handling of radioactive waste, etc). ⁹⁰Y-ibritumomab tiuxetan (Zevalin) is a radiolabelled therapeutic monoclonal antibody that has become available in recent years for the treatment of non-Hodgkin's lymphoma. Law et al [23] compared TLD measurements on the fingers, forehead and chest of staff involved in measuring radioactivity, in preparing and administering the injection of Zevalin, and in the post-injection radiation survey of the injection room. Zevalin was administered using a locally constructed Perspex injection box or a commercial automatic syringe driver. The injection box significantly reduced the dose to the index finger and thumb of the right hand and to the index finger of the left hand. The annual number of Zevalin administration sessions was limited by the effective dose (i.e. the whole-body exposure), and for the centre described by Law et al [23], which had a maximum annual workload of 20 patients, the annual effective dose would not exceed the limit recommended by the International Commission on Radiological Protection. They did conclude, however, that a pregnant member of their radiation team could exceed the annual dose limit for the surface of her abdomen during pregnancy.

Positron emission tomography (PET)-CT scanners allow functional information from the PET images to be registered with the anatomical detail on the CT images. They avoid the registration uncertainties associated with the acquisition of the two types of image on separate systems. A recent development in PET-CT scanners is the registration of these images for the planning of radiotherapy treatment. In this strategy, treatment-planning radiographers receive a radiation dose additional to that from their other duties. Carson et al [24] implemented a two-stage patient set-up procedure on the scanner, whereby the most time-consuming part of the process was carried out in the first “cold set-up session” (i.e. before administration of the ¹⁸F-FDG). Exposure was minimised by time and distance during the subsequent pre-scan “hot set-up session” (i.e. after ¹⁸F-FDG administration) and when permanent skin marks were made after the scans. It was found that this two-stage process reduced the dose to the radiographers (as recorded by electronic personal dosimeters worn at waist level) by a factor of about three, and that the average dose to a radiographer per patient was 5.1 μSv, which was comparable to that received by nuclear medicine PET scanning staff.

A key highlight of the radiobiological content this year was Professor Wardman's article based on his Sylvanus Thompson Memorial Lecture entitled: “The importance of radiation chemistry and free radical biology” [25]. He described the immediate and definitive role of the free radical chemistry that follows exposure of tissues to ionising radiation, and takes us through the cascade of events which, in turn, can modify the biological consequences of irradiation at the tissue level. These include interactions of water radicals with DNA and with other radicals such as nitric oxide, which he points out may have even more potent radiation-modifying effects than oxygen. Wardman also describes the actions of thiol radical scavengers and radiosensitisers. He then put these observations in a systems biology context and reminded us that “All cell biology reflects chemical kinetics: it's just rather complicated kinetics…”. Finally, Wardman made a strong plea for a renewal of the multidisciplinary approach that used to prevail in radiation biology research; a concept that has always reflected the ethos of the British Institute of Radiology.

Low dose rate hypersensitivity and its associated phenomenon, the inverse dose-rate effect, are thought to be exploited when radiation is given as low doses per fraction or as low dose rate brachytherapy. Leonard and Lucas [26] published an analysis of the influence of dose rate on the relative damage to tumours in organs such as prostate and cervix, and compared that to concomitant injury to adjacent normal tissues such as bladder, rectum, urinary tract and small bowel. Their calculations reveal that as brachytherapy sources decay, the progressively lower dose rates experienced by the normal tissues may give rise to an inverse dose rate effect, whereby these tissues would experience a greater degree of damage that would be predicted from higher dose rates. They conclude that high dose rate brachytherapy, in which seeds are not left in situ, may therefore be preferable to low dose rate therapy with the permanent retention of the seed.

Cisplatin is a widely used anticancer agent that has proven efficacy against a wide range of cancers. There is evidence that enhanced residence times could increase efficacy. One approach, described by Ohta et al [27], is the use of gelatin microspheres to conjugate the drug, allowing its release over a period of days. Further efficacy studies appear warranted. Another study, by Masunaga et al [28], involving cisplatin in combination with radiation, examined the effect of two very different agents, tirapazamine (a bioreductive cytotoxin) and hexamethylenetetramine (a formaldehyde-donating antiseptic) as therapeutic enhancers. These authors carried out an in vivo study of mouse tumours, comparing the total tumour cell population and the quiescent (Q) cell subpopulation. They observed that continuous administration of either agent caused sensitisation to combined cisplatin–gamma-ray therapy, and that the effects of continuous administration were greater than those of a single dose. Masunaga et al [28] advocate further studies of hexamethylenetetramine as an enhancer of chemoradiotherapy.

An interesting concept was presented by Anoopkumar-Dukie et al [29]. They hypothesised that oxygen-dependent radiosensitivity may not be dependent solely on DNA damaging events and that non-nuclear mechanisms, probably involving mitochondria, are important. This is consistent with what we already know about cell-death signalling after radiation insult. Anoopkumar-Dukie et al [29] used specific pathway blockers of mitochondrial membrane permeability to show that HeLa, but not breast cancer or melanoma cells, could be radio-protected. Bax/Bcl-2 family proteins were not involved in this phenomenon. These authors therefore suggest that the use of agents to target mitochondrial membrane permeability could be used to enhance radiotherapy for some solid tumours.

There was no particular theme common to any of the oncology papers published in the BJR in 2009. At a time when research in radiation oncology in the UK is (depending upon the assessors' innate optimism) either in the doldrums or about to undergo a major renaissance, it seems appropriate to look at the origins of last year's BJR papers whose authorship included at least one oncologist. There were 29 such papers in total. In three of them, the author was a medical, rather than a radiation, oncologist. The geographical origins of these papers were widely dispersed: only 13 were from the UK, five were from Turkey, four from Japan, three from commonwealth countries (Australia, Canada and India), two from China, and France and the USA contributed one paper each. This geographical pattern reflects the international nature of modern medical journals and demonstrates that there is nothing parochial or exclusively ‘British’ about the journal. Of the papers originating from the UK, six came from the Royal Marsden Hospital, another five from English centres, and Scotland and Northern Ireland contributed one paper each. These figures make salutary reading for clinical oncologists in the UK — we are falling behind the rest of the world and the renaissance, if there is to be one, is long overdue. One major problem is that UK radiation oncologists are, with obvious institutional exceptions, no longer being educated within a culture that encourages critical thinking or publication. Educational targets (based on rote learning, rather than understanding) dominate the five year training programme and, once accredited, other (non-intellectual) targets dominate clinical life. If the BJR can be used as a barometer for change, then let us hope that the next few years see the mercury rising.

On the diagnostic radiology front, two papers published in 2009 highlighted a worrying shortfall in the knowledge of radiology staff in the treatment of serious anaphylactic reactions and in resuscitation skills. Rachapalli et al [30] sent questionnaires to all 222 staff in the department of radiology at the University Hospital of Wales. Of the 66% who responded, 75% had received formal resuscitation training, but this had lapsed in 66% of cases. Only 11% were aware of the location of all of the resuscitation equipment in the department, and although 35% could use a defibrillator, only 6% were aware of changes to guidelines for the use of this equipment introduced in 2006. Only 57% of medical staff and radiographers could manage an anaphylactic reaction. In a similar study, Tapping and Culvewell [31] surveyed 171 radiologists in six NHS trusts in Yorkshire, receiving 105 replies. Only 61% had attended a life-support course in the past four years. Participants were asked eight questions aimed at assessing their training, knowledge and confidence in the management of adult resuscitation, and only 13% answered all of these questions correctly. Both sets of authors conclude that there is a need for more regular life-support training and point out that radiologists have a responsibility to keep their knowledge in this area up-to-date.

Diagnosis of colorectal cancer remains topical, with barium enema, CT colonography and colonoscopy being the diagnostic tests commonly used. In a study assessing the experiences of individual patients, Von Wagner et al [32] compared impressions of the physical sensations, social interactions and information provision relating to these three techniques. Social interactions with staff were perceived as very important, as was the provision of information. Verbal feedback received during or after colonoscopy was invariably found to be very reassuring. On the other hand, patients undergoing CT colonography had little or no verbal feedback and were more likely to be confused regarding the outcome. In terms of physical sensations, patients found all three tests demanding, but all were generally well tolerated as patients considered the examinations necessary for diagnosis of their symptoms. Physical sensations associated with CT colonography were comparable to barium enema but less physically demanding. Overall, CT colonography and colonoscopy, but not barium enema, were found to have advantages. The authors comment that CT colonography could benefit from improved provision of information following examination. They also highlight the importance of communicating the outcomes of diagnostic tests quickly, pointing out that patients reported substantial benefit from instantaneous feedback from the endoscopic team following colonoscopy. They further point out that this preference may be strongly influenced by clinical outcome, as the radiology department may not be the appropriate place to break bad news.

Continuing this theme, radiation dose remains an important consideration, and Hirofugi et al [33] compared patient doses for barium enemas and for ‘routine’ and ‘low-dose’ CT colonography examinations in Japan. For barium enemas, the effective dose for analogue radiography was 10.7 mSv, and this decreased by 12% when digital radiography was used. Routine CT colonography was performed using a relatively high mean effective mAs of 119, looking to detect colorectal cancer and extracolonic lesions (using paired prone and supine scanning of the entire colon region following insufflation of air, which gave an effective dose of 23.4 mSv). This dose is about double that of barium enema. Lowering the tube current to 50 mA reduced the effective dose to 5.7 mSv (i.e. about half that of barium enema). The image quality of the low-dose scans was not, however, evaluated in this study.

Also in the field of CT colonography, Punwani et al [34] published a study comparing colonic movement in patients in prone and supine positions in order to aid development of image-registration techniques. Changes in the coordinate position of colonic segments insufflated by carbon dioxide and following administration of N-butyl-bromide were calculated using the superior mesenteric artery as a fixed point of reference. Punwani et al [34] found minimal variation in colonic length between prone and supine orientations. The transverse colon was the most mobile segment with an average displacement of 4.6 cm. There were, however, significant differences between the sexes for individual colonic segments: the ascending colon, descending colon and rectum being longer in men, whereas the sigmoid colon was longer in women.

In another paper on colorectal radiology, faecal tagging for minimal preparation CT of the colon was evaluated. This technique may be used as an alternative to the more invasive procedures mentioned above in patients, particularly the elderly, who seem unlikely to tolerate full-bowel preparation very well. The technique involves taking a low dose of oral contrast agent to ‘tag’ the colonic contents, so that unenhanced low-attenuation tumours and large polyps can be distinguished on CT from high-attenuation tagged faeces. Slater et al [35] found that extending the length of oral contrast administration before the examination from two days to three provided significantly better tagging of faeces in the rectum and sigmoid colon. This may be important as most tumours that are missed by minimal preparation CT are in the rectosigmoid region.

Any imaging technique that has the ability to predict response to treatment has to be examined closely. Niwa et al [36] looked at the use of apparent diffusion coefficient (ADC) in cases of advanced pancreatic cancer and showed that “the rate of tumour progression was significantly higher in those with a lower b-value ADC than in those with a higher b-value ADC”.

It is sometimes helpful to challenge assumptions that affect clinical practice. Shah et al [37] asked whether there was a contraindication to mixing steroid, iohexol and local anaesthetic for intra-articular injection. They performed chromatography and showed that these products remained stable when mixed.

Cholestasis in pregnancy is an uncommon event. Ultrasound provides much useful information, but Oto et al [38] show that MR cholangiopancreatography can reduce the need for endoscopic investigation in this high-risk group.

MRI in cardiology has a well established role, and Debl et al [39] showed that phase-contrast cine MRI is as accurate as invasive oximetry in quantifying left-to-right shunts in adults.

It has been assumed that the ophthalmopathy associated with Graves' disease is multifactorial. Dodds et al [40] showed a significant reduction in optic nerve diameter in patients with symptoms of optic neuritis compared with those without.

The link between depression and hippocampal choline decrease was suggested by Hong et al [41]. Using an animal model of depression, they showed significant decreases in choline/creatine ratio and choline/N-acetyl aspartate ratio in the left hippocampus when the situation before induced depression was compared with that after.

A comparison of whole-body MRI and bone scintigraphy in renal metastases [42] showed that significantly more bony metastases were picked up by MRI and that a number of soft tissue lesions were also revealed by this technique. Bone scintigraphy did, however, detect more skull, facial bone and bony thoracic abnormalities.

A potentially landmark paper was authored by Au-Yong and colleagues in Nottingham [43]. In this paper, ‘Isolated cerebral cortical tears in children: aetiology, characterisation and differentiation from non-accidental head injury’, the authors described a series of cases in which neonatal MRI showed isolated cortical tears shortly after difficult instrumental delivery. This pattern does not appear to have been reported previously and the authors stress that recognition of this injury pattern is important because of its possible misinterpretation as a marker of non-accidental head injury. The authors also emphasise the need to obtain high-quality cross-sectional imaging in newborn infants presenting with seizures, and that details of the birth history are important in the accurate interpretation of the resultant imaging.

A further paper with the potential to change current imaging practice significantly was authored by Lyle and colleagues from Southampton [44]. ‘MRI of intermittent meniscal dislocation in the knee’ described the use of MR in patients who had recurrent but reproducible locking of the knee to demonstrate meniscal dislocation. Detection of abnormality on MR in such intermittent conditions is notoriously difficult and although the most likely cause of intermittent locking is sporadic meniscal displacement, this is often not apparent on routine MR imaging. Lyle et al [44] demonstrated meniscal dislocation in a series of patients with previously normal MRI scans when repeat imaging was performed in the ‘locked’ position and concluded that imaging in this way has the potential to reveal abnormalities not apparent on conventional scans.

Kumar et al [45] from Portsmouth asked ‘Is there an association between intestinal perfusion and Crohn's disease activity?’ Their paper describes a feasibility study using contrast-enhanced ultrasound in which they collate their initial experience with this modality. They observed a range of abnormalities in flow patterns and flow rates in the superior mesenteric artery, and found a correlation with disease activity. This relation had been suggested by previous Doppler studies, but the use of contrast-enhancement in this context is novel. Kumar et al [45] suggest that the novel use of this technique has the potential to provide an improved marker of disease activity in this patient group.

An important academic radiology paper from Brodoefel et al [46] described a collaborative study from Boston and Tübingen entitled, ‘Accuracy of dual-source CT in the characterisation of non-calcified plaque: use of a colour-coded analysis compared with virtual histology intravascular ultrasound’. In this paper, they reported their experience of analysing plaque volume and composition. Non-invasive assessment of these parameters is important for risk stratification in coronary atherosclerotic disease, and although intravascular ultrasound (IVUS) is the standard of reference for detection of non-stenotic atheroma in particular, multi-slice CT — and perhaps dual-source CT especially — is suggested as a less expensive and non-invasive method of investigation. The authors reported good correlation between CT and IVUS in determining entire plaque and non-calcified plaque volume, although results for calcified plaque were less well correlated. Nevertheless, this paper advances the understanding of the capabilities of dual-source CT, and of its associated image post-processing, in the assessment of this important patient group.