Magnetic resonance imaging of the pelvis

O ver the past decade several clinical applications of magnetic resonance imaging have been developed for benign and malignant conditions of the pelvis. The production of easily recognisable multiplanar images of high soft tissue contrast—without the operator dependence of ultrasonography or the ionising radiation of computed tomography—well suit magnetic resonance imaging for such tasks. Attractive though the technique appears, we must demand more from such an expensive—and still scarce—resource.

The applications of magnetic resonance imaging in the pelvis have received great scrutiny. Several large prospective studies have been performed comparing pelvic magnetic resonance with surgical specimens sectioned and analysed in the planes used for imaging. Its diagnostic performance is thus well validated for cancer of the rectum, cervix, and prostate.^1–3 Its key advantage is prediction of local extension of cancer beyond the primary organ. Here, magnetic resonance imaging competes with tried and tested clinical methods such as digital examination, probing, examination under anaesthetic, or alternative imaging tests such as ultrasonography and computed tomography.

It is now clear that magnetic resonance imaging can identify tumour spread to the circumferential margin of the rectum, the parametrial margin of the cervix, and within periprostatic fat, factors that predict treatment failure with primary surgery alone.^1–3 The impact of this information on management varies with the primary site: for the rectum neoadjuvant therapy may be indicated before surgery, for the cervix or prostate radiotherapy may replace surgery.

It is also emerging that magnetic resonance imaging not only has therapeutic impact in planning therapy for both benign and malignant conditions but may also allow prediction of its success. Data are mainly from observational studies in which magnetic resonance imaging, performed at initiation of therapy, played no part in decision making and outcomes were judged at a predetermined and clinically relevant endpoint. The question is simple: could the findings of magnetic resonance imaging better predict the success or failure of standard therapies than traditional clinical methods and if so, why? In some circumstances inflammation may hamper clinical evaluation or disease may extend beyond the reach of the examining digit or probe. In others treatment selection and its success may be determined by disease volume, a feature which can be accurately estimated by imaging but only hinted at by palpation. With cervical cancer there is an excellent correlation between magnetic resonance volume estimations and those found within surgical specimens.²

Magnetic resonance imaging can predict the success of some medical treatment

The data on predicting therapeutic success and clinical outcomes by magnetic resonance imaging are mixed. For men with prostate cancer treated with radiotherapy rather than surgery, information from magnetic resonance imaging appears to be non-contributory.⁴ By contrast, for women with cervical cancer treated with radiotherapy, magnetic resonance imaging better predicts tumour control than examination under anaesthesia or ultrasound assessment and indicates disease regression rate.^5–7 Magnetic resonance imaging has assumed the pivotal role in clinical decision making for this condition suggested by early work.⁸

Such prediction of outcome is not confined to pelvic malignancy. For complex benign pelvic conditions such as perianal sepsis magnetic resonance imaging enables the extent of disease to be assessed better than any other non-invasive method and predicts outcome better than standard surgical assessment.⁹,¹⁰ In patients with endometriosis magnetic resonance imaging predicts the success of medical treatments.¹¹

The acid test is whether clinicians are so impressed by these findings that they are not prepared to deliver radical therapies without access to magnetic resonance imaging. To truly show an improvement in wellbeing for patients, randomised trials are required in which half the patients—and their clinicians—are denied access to magnetic resonance imaging. Herein lies the problem for evidence based diagnostic radiology.¹² Such trials rarely occur early in the evaluation process of an imaging technique.

Radiologists will have gone through a learning phase in evaluating its diagnostic performance; clinicians will have had a chance to refine their case selection. Even when ethics committees are convinced of the justification for such research, recruitment and randomisation may be so slow that by the time the trial reports, the imaging test is already considered routine in clinical practice. Furthermore, the pace of technological development in magnetic resonance imaging is such that by the time large randomised studies are completed the technology used may have been superseded.

Perhaps it is asking for too much of an evidence base for pelvic magnetic resonance imaging to expect data from randomised clinical trials to underpin its every application. Business cases for expansion in new imaging technologies have traditionally been more concerned with financial than with diagnostic and clinical impact.

Yet the messages are becoming clear for those formulating guidelines for the use of imaging as well as for purchasers of health care. They argue strongly for integration of magnetic resonance imaging in the pathway of care for several pelvic malignancies and complex benign pelvic conditions. To deny patients access to magnetic resonance imaging is likely to result in suboptimal clinical and cost effectiveness.