1. Department of Medical Imaging, The Canberra Hospital, Canberra
2. Department of Medical Imaging, School of Medicine, Australian National University, Canberra
Magnetic resonance imaging is usefully employed on its own or with complementary technologies to evaluate stage and other characteristics for a range of specific tumour types. These include tumours of the central nervous system, head and neck, breast, prostate and colo-rectum, as well as gynaecological and musculoskeletal malignancies. For each tumour type, optimal usage of magnetic resonance imaging involves particularities of tumour type and various other characteristics. For most tumour categories, there are scenarios in which magnetic resonance imaging has only limited application.
Magnetic Resonance Imaging (MRI) is a safe and painless imaging investigation (test) that produces cross sectional imaging of the tissues of the body. MRI is a valuable tool that can aid in diagnosis of a wide range of conditions and is often used to diagnose cancer. It is most effective in detecting and staging cancer of the brain, spinal cord, head and neck and musculoskeletal system. MRI relies on a large magnetic field and certain people should avoid the test. This includes patients with implanted pacemakers and implantable cardioverter defibrillators. Pregnant women should generally avoid MRI, unless it is necessary, as the risk to a developing foetus is not known.
MRI is now the investigation of choice for the evaluation of cerebral neoplasms. MRI is superior to computed tomography (CT) for tumour detection because of its inherently high sensitivity to altered brain tissue. Although conventional MRI sequences play a major role in determining prognosis,1 MRI is unable to predict tumour type and grade reliably.2 The accuracy is limited by the inherent nature of the majority of common brain tumours, which are diffusely infiltrative; MRI is frequently unable to identify the tumour margins. Since many gliomas contain areas of varying histological type, the aim of imaging should be to identify the area of highest grade and thereby guide the stereotactic biopsy appropriately.
Since there is increasing evidence that complete resection of tumour prolongs survival, especially in low-grade gliomas,3 intraoperative MRI with its ability to provide up-to-date images that reflect intraoperative anatomical change should enable more complete resection.1,4 However the resection would be limited in neurologically eloquent areas where risk of producing neurological deficit increases.
Functional imaging studies using MRI at 1.5 Tesla or higher are being developed which permit non-invasive determination of centres of task activation in the cortex of the brain.5 This may allow the accurate mapping of the relationship of the normal functioning tissue to the tumour and enable larger resections while preserving normal function.6 MR spectroscopy (MRS) has become more readily available and easier to use and is therefore becoming part of preoperative imaging and tumour follow-up. MRS is able to show residual or recurrent tumour outside areas of enhancement seen in gadolinium-DTPA contrast MR scans.
Echoplanar diffusion-weighted imaging (DWI) is routinely used in many institutions. Its main value is to discriminate between an acute infarct and tumour at presentation. Apparent diffusion coefficient (ADC) maps can discriminate between high-grade glioma and normal brain tissue and may help to target the most malignant areas.7 There is, however, some overlap between Grade II and Grade IV astrocytomas.8
There is no scientific evidence to indicate whether MRI or CT is better in the evaluation of head and neck cancers. Each is complementary with its advantages and disadvantages.
CT is reliable to evaluate bony structures. MRI is valuable to evaluate bone marrow involvement. However there is usually bony destruction in CT when tumour invades the marrow space. MRI is more useful around the skull base because of the higher contrast resolution obtained to delineate complex anatomy around the skull base and upper neck. The main disadvantage of MRI compared to CT is the motion artefacts, especially in the region of the lower neck and oral cavity due to swallowing, coughing etc.
MRI has little role to play in the evaluation of thyroid nodules. It cannot reliably differentiate benign from malignant nodules.9 MRI can be used for tumour staging and assessing mediastinal and oesophageal extension of tumours considered to be aggressive or invasive. Introduction of one and two dimensional proton MRS is promising and allows more specific tissue characterisation, which may help to distinguish benign and malignant nodules.10
Staging of all potentially malignant tumours in bone is most accurately achieved by MRI, which should be performed prior to biopsy. This allows measurement of the maximum dimension of the tumour prior to any treatment. CT has a limited role in evaluating the local staging of the tumour but is the examination of choice for evaluation of the chest for metastatic disease. CT is the preferred test where characterisation of the lesion by radiography is inadequate because of inadequate visualisation of the matrix of a lesion. In these circumstances CT imaging may suffice for local staging.11
MRI has become the imaging method of choice in evaluation of soft tissue tumours. This is due to improved soft tissue contrast and multi-planar image acquisition, which allows more accurate anatomical delineation of the tumour and its relationship to neurovascular structures. However, inability to detect soft tissue calcification renders a mass non-specific on MRI, whereas a plain radiograph or CT can make the diagnosis immediately obvious. Pulmonary metastasis is best characterised by CT as 10-20% patients with primary soft tissue cancers have pulmonary metastasis at diagnosis.12 Knowledge of pulmonary metastasis is critical for optimum management of these patients.
MRI has high sensitivity for breast cancer detection that relies on the tendency of malignant tumours to generate neovascularity. Malignant angiogenesis is seen with leaky capillaries that allow the contrast agent to show high intensity peak with rapid washout that is seen in most, but not all, malignancies. False negative examinations have been reported with well-differentiated ductal carcinomas and lobular carcinoma.13 Although sensitivity is high for invasive carcinoma, ductal carcinoma in situ (DCIS) is more difficult to detect, with a sensitivity as low as 40%.14
Breast MRI is best used as an adjunct to conventional imaging, complementing mammography and ultrasound. Its high sensitivity is helpful in detecting multifocal disease and is being looked at as a possible screening investigation in high-risk populations.15,16
False positive breast MRI is seen with fibroadenomas, atypical ductal hyperplasia, lobular carcinoma in situ, papilloma, fibrocystic changes and other benign conditions with focal enhancement. The relatively low specificity is likely to be the greatest impediment to an increase in clinical utility for MRI in breast cancer work-up.
MRI is promising for diagnosis and staging of prostate cancers. MRI should not be performed for three to four weeks after prostate biopsy to minimise error from signal alteration related to post biopsy haemorrhages.17 Prostate cancer is usually seen as low signal focus in T2 weighted sequences. Cancer may not be detected if it does not show low signal in T2 weighted images, if it is located in the central gland or if the peripheral zone is compressed by advanced benign prostatic hyperplasia (BPH). MRI can detect extracapsular or seminal vesicle invasion. Overall the results of MRI for prostate cancer staging have varied over the last decade. A meta-analysis showed the maximum sensitivity for extracapsular extension was 64% and for seminal vesicle invasion was 82%.18
MR spectroscopy has been reported to be valuable in diagnosis of prostate cancer. The combination of MRS and anatomical information from phased array and endo-rectal coils can improve the localisation of cancer within the prostate and may improve prediction of extracapsular extension. Combined MRS and MR have a reported positive predictive value of 88-92% and a negative predictive value of 80-86% for depiction of foci of prostate cancer within the gland. They have resulted in increased accuracy of tumour detection from 53% to 75%.19 High specificity of MRS helps in the distinction of post-biopsy haemorrhage and other benign abnormalities from tumour. MRS may be able to assess tumour aggressiveness. Significant correlation has been shown between Gleason score and MRS choline levels.20
Tumour staging of colorectal cancer can be achieved with endo-rectal coil with accuracies of 80% or greater. T2 weighted images provide better contrast between the tumour and rectal wall than T1 weighted images. Higher resolution obtained with endo-rectal coil demonstrates the different anatomical layers of bowel wall. Using a high resolution technique, thin slice MRI can be used to measure the depth of extramural spread accurately and show good correlation with resected pathological specimens.21 High spatial and contrast resolution of this technique provide detailed anatomic imaging, which permits assessment of the relationship of the tumour to the mesorectal fascia.21 This provides a preoperative ‘roadmap’ for the surgeons.22 A recent prospective study has shown MRI predicts the histological status of colorectal malignancy with a positive predictive value of 92%.23 Ability of MR to identify extramural venous invasion, peritoneal infiltration and depth of extramural spread23 allows a more specific preoperative treatment strategy.
MRI has shown considerable promise in identifying hepatic metastasis which are < 1 cm and are difficult to characterise by CT or ultrasound. The availability of liver specific contrast agents, such as magnafodipir trisodium (Mn-DPDP), has resulted in further improvements in detection of metastatic disease. In a study comparing the performance of Mn-DPDP MR with CT and intraoperative ultrasound, MR influenced the operative decision in 74% of cases.24 Compared to the histopathology, sensitivities for CT, MR and intraoperative ultrasound were 61%, 83% and 93% respectively.24
MR is the preferred technique for imaging carcinoma of the cervix. The ability to image in oblique planes and superior soft tissue contrast resolution gives MR a major advantage over other imaging techniques. MR has an overall staging accuracy of 79%.25,26 The accuracy for detecting parametrial invasion averages 88% (range 79 – 100%) and of assessing vaginal extension 90% (range 83 – 100%).26 However, the major contribution of MR to planning treatment is the very high negative predictive value for determining parametrial invasion. Compared to CT, MR offers significantly better evaluation of tumour size, stromal involvement and local and regional extent of disease in pre-treatment imaging.27
Contrast enhanced MR is superior to ultrasound in characterising adnexial masses.28 Both techniques are highly sensitive but MR is more specific than ultrasound at identifying malignant masses. Spread of ovarian cancer into uterus, bladder or rectum may be better appreciated on MR than CT. MR has a limited role in the evaluation of intra-abdominal tumour spread; peritoneal deposits greater than 1cm in diameter can probably be identified with a similar sensitivity for both MR and CT.29 Disease within the mesentery or implants on the wall of small and large bowel are better detected by CT. MR remains insensitive for detecting peritoneal, mesenteric or omental tumours in ovarian malignancy. The appropriate role for MR is to characterise the ovarian masses rather than abdo-pelvic staging of proven ovarian cancer.
MR is not an appropriate investigation for diagnosing endometrial mass and should only be carried out for staging purposes when biopsy has given a specific histological diagnosis of endometrial carcinoma. The sensitivity and accuracy of MR in detecting deep myometrial invasion ranges from 82 – 94%.30,31 Contrast enhanced MR performed significantly better than non-contrast MR or ultrasound for myometrial invasion.32 The accuracy of MR is not widely documented in patients with advanced stage III and IV disease.
15. Kuhl CK, Schmutzler RK, Leutner C et al. Breast MR imaging screening in 192 women proved or suspected to be carriers breast cancer susceptibility gene: preliminary results. Radiology 2000; 215: 267 – 279.
16. Brown, Coulthard A, Dixon AK et al. Protocol for a national multi centre study of magnetic resonance imaging (MRI) screening in women of genetic risk of breast cancer. UK MRI breast screening study advisory group. The Breast 2000; 9: 78 – 82.
24. Mann GN, Marx HF, Lai LL et al. Clinical and cost effectiveness of new hepatocellular MR contrast agent, magnafodipir trisodium, in the preoperative assessment of resectability. Ann Surg Oncol 2001; 8: 573 – 579.