1991, The British Journal of Radiology, 64, 683-689
Magnetic resonance imaging in the staging of renal cell carcinoma By J. E/Kabala, MRCP, FRCR, *D. A. Gillatt, FRCS, *R. A. Persad, FRCS, J. B. Penry, FRCR, \J. C. Gingell, FRCS and *D. Chadwick, FRCS Departments of Radiology and 'Urology, Southmead Hospital, Bristol, UK {Received October 1990 and in revised form February 1991) Keywords: Magnetic resonance imaging (MRI), Carcinoma kidney, Staging
Abstract. A prospective study has been carried out to examine the role of magnetic resonance imaging (MRI) in the investigation of renal cell carcinoma in 24 patients. In all cases the inferior vena cava (IVC) was well demonstrated with MRI. In 14 out of 15 patients where surgical correlation was available, the MRI and operative staging were in agreement. Magnetic resonance imaging and computed tomographic (CT) staging were in agreement in 16 out of the 17 patients where both were performed. In one case, CT suggested hepatic invasion but this was found not to be present on MRI and at operation. Magnetic resonance imaging also provided substantial additional information in three patients, including two cases where MRI demonstrated a patent IVC that appeared occluded on CT (one of which also had vertebral metastases seen on MRI but missed on CT) and one case where CT failed to demonstrate minimal involvement of the IVC. Magnetic resonance imaging is an accurate means of staging renal cell carcinoma with clear advantages over CT. In no case in this series was inferior vena cavography found to be necessary.
In the vast majority of patients with suspected renal cell carcinoma, intravenous urography (IVU) and ultrasound will produce the diagnosis (Demas et al, 1988), occasionally with the addition of percutaneous biopsy. Further investigations will generally be directed towards staging the tumour. Involvement of the renal vein and inferior vena cava (IVC) will affect the surgical technique used (McDonald, 1982; Woodhouse et al, 1985; Pritchett et al, 1986) and the presence of invasion of regional lymph nodes or viscera and distant metastases has considerable prognostic significance (Skinner et al, 1972; Boxer et al, 1979; Lieber et al, 1981). Both ultrasound and computed tomography (CT) (especially dynamic CT) can be used to evaluate the IVC, regional lymph nodes and local invasion of the tumour. Although an accuracy of up to 9 1 % for CT has been reported (Johnson et al, 1987), an accuracy of 72-78% may be more realistic (Frohmuller et al, 1987; Didier et al, 1987; London et al, 1989; Miles et al, 1990). Problems with CT include inadequate or mistimed injection of intravenous contrast medium and limitations of the transverse plane of the section may result in difficulty in evaluating the interface between the tumour and adjacent viscera (Demas et al, 1988). The purpose of this study was to determine the accuracy of magnetic resonance imaging (MRI) staging of renal cell carcinomas, in particular with reference to IVC involvement and to see if inferior vena cavography can be regarded as obsolete in this situation. Address correspondence to: J. E. Kabala, Department of Radiodiagnosis, Bristol Royal Infirmary, Marlborough Street, Bristol BS2 8HW, UK. Vol. 64, No. 764
Methods
Magnetic resonance imaging was performed on 24 patients with renal cell carcinoma. A T{ -weighted spin-echo sequence (TR 600 ms, TE 26 ms) in transverse and coronal planes and a STIR (short tau inversion-recovery) sequence (TR 1900 ms, TE 30 ms, TI 100 ms) in the coronal plane were performed using a Picker Vista 2055 HP 0.5 T scanner. Further sequences in additional planes were performed depending on the appearances on the standard views. The tumours were staged using the conventional classification described by Robson et al (1969) (Table I). The" results were compared with operative findings in 15 cases. In 17 cases comparison with CT was available but only 11 of these were operated on. The CT scans were performed on a Siemens Somatom DRH or a GE 9000 in one of two referring hospitals using 10 mm consecutive cuts through the kidney immediately following an intravenous injection of 50 ml of Iopamidol 370 mg/ml. In three patients who did not undergo surgery, only MRI and ultrasound were performed. Table I. Staging of renal cell carcinoma (Robson et al 1969) Stage I II Ilia Hlb IIIc IV
Limited by the renal capsule Perirenal involvement including the adrenal but limited by Gerota's fascia Involvement of renal vein or IVC Involvement of regional lymph nodes Vascular and lymphatic involvement Invasion of adjacent viscera or distant metastases
683
J. E. Kabala, D. A. Gillatt, R. A. Persad, J. B. Penry, J. C. Gingell and D. Chadwick
Figure 2. Transverse STIR sequence (TR 1900 ms, TE 30 ms, TI 100 ms) showing high signal intensity from the normal kidney and the large renal carcinoma, seen protruding from the posteromedial aspect of the right kidney. Figure 1. Coronal Tx -weighted spin-echo sequence (TR 600 ms, TE 26 ms) showing normal renal architecture on the left; medium-signal-intensity cortex, low-signal-intensity pyramids. This is replaced in the mid and upper part of the right kidney by medium-signal-intensity tumour (vertical arrow). Note similar signal intensity lymph node immediately adjacent to the right lateral border of the first lumbar vertebra (horizontal arrow).
Results
of the carcinoma (Demas et al, 1988; Choyke & Pollack, 1988) (Fig. 3). The MRI staging of the tumours is shown in Table II. They included seven patients with renal vein involvement, five of whom showed extension into the IVC. Three patients with locally operable disease (all Stage I on MRI) did not undergo surgery: in two cases because of ultimately fatal medical conditions (cerebrovascular accident in one patient, bronchopneumonia on a background of chronic obstructive airways disease in the
In all 24 cases the IVC was well demonstrated on MRI. The renal cell carcinomas produced a signal intensity similar to renal tissue on all sequences and therefore were seen as masses of medium signal intensity (slightly higher than the signal from skeletal muscle) on r,-weighted sequences and high signal on STIR sequences (Figs 1 and 2). 2",-weighted sequences demonstrated normal renal architecture well and this was lost in areas occupied by tumour. A common finding was a leash of low-signal-intensity vessels lying just outside the margins of the tumour and presumably representing collateral blood supply or parasitic neovascularization Table II. MRI staging of 24 renal cell carcinomas Stage I II Ilia IIIc IV Total
Number
2 2 1
11* 24
""Includes four with renal vein involvement, two of which showed tumour extension into the IVC.
684
Figure 3. Coronal T,-weighted spin-echo sequence (TR 600 ms, TE 26 ms) showing a leash of new vessels (small arrow) related to the inferior aspect of the left renal cell carcinoma (large arrow). The British Journal oj Radiology, August 1991
Magnetic resonance imaging and renal cell carcinoma Table III. Operative and MRI staging Operative Renal vein involved IVC involved Overall stage I II Ilia Ilk IV Total
MRI
6 5
5 5
6 1 2 1 5 15
5 2 2 1 5 15
other) and in one case because the tumour was in a solitary kidney. Surgery was not performed in six patients with Stage IV disease, in three cases because of distant metastases (two skeletal and one hepatic) and in three cases because of very extensive local tumour. Operative correlation therefore was available in 15 patients. Comparison with CT was available in 17 patients. In three patients only ultrasound was available to correlate the MRI demonstration of an uninvolved IVC. In the 15 patients where operativefindingswere available, the MRI stage agreed with the pathological stage in 14 cases. In one case a Stage I tumour was classified as Stage II on MRI and CT. In all these cases MRI defined accurately IVC involvement and the presence or absence of invasion of regional lymph nodes and adjacent viscera. In one case (overall Stage IV) the MRI was not of sufficiently good quality to be confident of the state of the renal vein, and at operation it was shown to be involved proximally by tumour. The staging of these patients is shown in Table III. The overall staging accuracy of MRI compared with operation was 93%, its specificity and sensitivity for IVC involvement being 100%. In the 17 patients where CT correlation was available, there was agreement with the overall MRI stage in 16 cases. In one case CT suggested hepatic invasion by the tumour, shown not to be present on MRI and at operation. Both MRI and CT suggested a Stage I tumour was Stage II. In two further cases, CT suggested extensive IVC invasion by tumour but MRI demonstrated a compressed but patent IVC (Fig. 4). In one of these patients MRI elegantly demonstrated vertebral metastases not seen on CT (Fig. 5) and in another case MRI defined the upper limit of IVC involvement, which ultrasound and CT had been unable to do (Fig. 6). In one further case CT failed to demonstrate a nodule of tumour in the IVC, shown on MRI and at operation (Fig. 7). This, however, did not change the stage of these tumours. In one patient both MRI and CT demonstrated an irregular tumour continuous with the peritoneum in the presence of ascites. The assumption was made that this was a Stage IV tumour rather unusually directly invading the peritoneum. Unfortunately the patient died Vol. 64, No. 764
Figure 4. Sagittal r,-weighted spin-echo sequence (TR 600 ms, TE 26 ms) showing extension of tumour (vertical arrow) behind a patent IVC (horizontal arrow), producing forward deviation of the vessel.
Figure 5. Coronal STIR sequence (TR 1900 ms, TE 30 ms, TI 100 ms) showing a large left renal cell carcinoma of mixed but predominantly high signal (curved arrows) and high-signal vertebral metastases (horizontal straight arrows).
685
J. E. Kabala, D. A. Gillatt, R. A. Persad, J. B. Penry, J. C. Gingell and D. Chadwick
(a)
Figure 6. Coronal r r weighted spin-echo sequence (TR 600 ms, TE 26 ms) showing right renal cell carcinoma with extension along the right renal vein and into the IVC up to the level of the diaphragm.
without operation and no post-mortem examination was performed. Renal adenocarcinoma cells were, however, demonstrated in the ascites. There was agreement between CT and operative staging in nine out of 11 patients where both were performed, giving a CT accuracy of 82% (Table IV). The two cases of disagreement are described above. In three patients who did not undergo surgery (two Stage IV, one Stage I) only ultrasound and MRI were performed; both demonstrated an absence of IVC involvement. Discussion
A variety of methods is available for staging renal cell carcinoma. Excretion urography and angiography are at best 59% and 57% accurate, respectively (Frohmuller et al, 1987). Ultrasound is almost invariably performed as part of the diagnostic process for renal cell carcinomas. Table IV. Operative and CT staging
Stage I II Ilia IIIc IV Total
686
Operative
CT
4 1 2 1 3 11
3 2 2 0 4 11
Figure 7. (a) Transverse and (b) coronal Tx -weighted spin-echo sequences (TR 600 ms, TE 26 ms) showing tumour extending along an expanded left renal vein [white arrows, Fig. 7(a)]. Tumour is seen to just protrude into the patent IVC [black horizontal arrows, (a) and (b)]. In the coronal plane the nodule of tumour entering the IVC from the left renal vein is well shown [curved arrows, Fig. 7(b)].
Its accuracy for staging is up to about 78% (Frohmuller et al, 1987), with some studies suggesting a lower figure of around 50% (London et al, 1989). Excess bowel gas and adverse patient build may prevent an adequate examination. Clearly the performance of an ultrasound examination depends on the operator's skill but all means of staging are to some extent subjective. Although Johnson et al (1987) reported an accuracy of 91% for CT staging (using a dynamic contrast technique) a number of comparative studies have suggested much lower values. Frohmuller et al (1987) reported a The British Journal of Radiology, August 1991
Magnetic resonance imaging and renal cell carcinoma
staging accuracy of 78% for ultrasound compared with study was 93%. The single error of staging was 72% for CT (although the CT technique used is not mistaking a Stage I tumour for Stage II and was also described). More recently, London et a( (1989) reported made on CT. Others have reported this difficulty with a 50% accuracy for conventionally enhanced CT and both techniques (Fein et al, 1987; Johnson et al, 1987) 72% accuracy for dynamic CT. The lower figure for and it is likely that at the current stage of development conventionally enhanced CT may be partly due to their neither CT nor MRI can reliably distinguish between technique of performing the post-contrast scan 5 Stage I and Stage II. It is, however, not a significant minutes after the intravenous injection. The same group problem since the prognosis and treatment of both of workers suggest a single-location dynamic scan in categories of tumour are virtually identical unless a addition to an incremental scan during infusion of partial nephrectomy is being planned (Robson et al intravenous contrast medium may increase CT accuracy 1969; Boxer et al, 1979; Guiliani et al, 1990). above 72%, especially for the vascular stage (Miles et al, In all patients studied, the IVC was well seen. In five 1990). patients undergoing surgery, IVC involvement was In addition to the general problems of CT compared present and accurately defined. In one further case with with ultrasound or MRI (the use of intravenous iodine- proximal renal vein involvement the MRI was not of containing contrast media and ionizing radiation), diagsufficiently good quality to stage the renal vein. Because nostic errors in CT staging may arise because of the of the signal void in flowing blood, patent vessels are limitations of scanning only in the transverse plane easily seen in good-quality scans and vascular invasion especially when dealing with polar tumours (Demas et by tumour is elegantly demonstrated by MRI al, 1988). Using strict CT criteria—change in organ size (Karstaedt et al, 1986; Fein et al, 1987; Pritchett et al, or density—adjacent organ invasion is only identified in 1987). The MRI sensitivity for distal renal vein and IVC 60% of cases; diagnosing invasion on the basis of a loss involvement is around 100%, sensitivity and specificity of fat line alone results in a rate of false positives of for proximal renal vein involvement being 86 and 96%, around 15% (Johnson et al, 1987; Demas et al, 1988). respectively (Hricak et al, 1988). In this group of Although inferior vena cavography can detect IVC patients the operative approach depends on adequate and renal vein involvement with an accuracy of 93% demonstration of the intravenous extension. and 88%, respectively (Horan et al 1989), it offers no The prognosis for renal vein extension alone is probinformation about local invasion and only detects ably similar to that for Stage I disease (Skinner et al, lymph nodes sufficiently enlarged to impinge on the 1972); the prognosis for patients with IVC involvement IVC. Horan et al (1989) compared MRI accuracy for is still debated. Some studies suggest it approaches that vascular staging with inferior vena cavography and of Stage I disease (Skinner et al, 1972; Boxer et al, 1979; found it was not significantly different (MRI accuracy Cherrie et al, 1982; Pritchett et al, 1986). Other studies for IVC and renal vein involvement being 93% and are more pessimistic (Robson et al, 1969; Kearney et al, 86%, respectively). 1981). The most recent and one of the largest studies Our study examined the role of MRI in the investi- (Guiliani et al, 1990) shows a poor prognosis with IVC gation of renal cell carcinomas. The tumours were involvement, similar to that for patients with distant found to have a medium signal on T,-weighted metastases. sequences and high signal intensity on the STIR Neither ultrasound nor CT can distinguish tumour or sequence. This was very close to the signal from normal thrombus extension into the IVC from invasion of the renal tissue. The similarity between signal from normal vessel wall (Didier et al, 1987) and there is no evidence kidney and tumour on all sequences including proton to suggest MRI can either. density and T^-weighted sequences has been noted Although the operative approach may be unaffected before (Quint et al, 1988). Tumours, therefore, are not by the presence of involved lymph nodes, this has a generally visible on MRI until they are large enough to significant effect on prognosis (Skinner et al, 1972; distort renal architecture (well seen on the Tx -weighted Guiliani et al, 1990). Two patients in this study had sequence in high-quality scans; Fig. 1) or the renal extensive para-aortic lymphadenopathy and did not outline (Quint et al, 1988; Choyke, 1988). undergo operation. In both, the CT scan suggested IVC A number of studies have described the use of MRI in occlusion but MRI demonstrated a patent, albeit studying renal disease. It was quickly realized that MRI compressed and deviated, IVC. could identify focal renal masses and distinguish solid Of the 15 patients undergoing surgery in this series from cystic lesions (Choyke et al, 1984). Although only one had regional lymph node involvement by initially it was thought that histologically specific signals tumour (Fig 8.) and this was correctly predicted by both might be obtainable from renal tumours (Hricak et al, MRI and CT. Only limited comments can be made on 1983), this promise has not yet been fulfilled and differ- the basis of this experience but it is probable that MRI entiation between benign and malignant renal tumours will be at least as accurate as CT in detecting involved cannot be made on the basis of signal intensity (Hricak lymph nodes. Hricak et al (1988) reported an accuracy et al, 1988). of 99% in predicting lymph node involvement preoperatively with MRI. They considered any lymph node The accuracy of MRI staging of renal cell carcinoma has been reported at 82-96% (Hricak et al, 1985, 1988). in the perivascular or perihilar areas with a diameter The overall accuracy in the 15 patients in this current above 10 mm (plane not specified) to be involved, Vol. 64, No. 764
687
J. E. Kabala, D. A. Gillatt, R. A. Persad, J. B. Penry, J. C. Gingell and D. Chadwick
Figure 8. Transverse T,-weighted spin-echo sequence (TR 600 ms, TE 26 ms) showing an enlarged lymph node (curved arrow) lying medial to the right kidney (containing a large renal cell carcinoma anterolaterally, vertical arrow) and posterior to an expanded IVC occluded with tumour thrombus (horizontal arrow).
commenting that previously they had encountered lymph nodes between 10 and 15 mm diameter that were involved with tumour. Collateral vessels with their signal void on MRI are unlikely to be confused with lymph nodes (Fein et al, 1987). However, MRI and CT are both currently unable to distinguish reactive lymph
node enlargement from those involved with tumour (Choyke & Pollack, 1988; Hricak et al, 1988). In two patients who did not undergo surgery, bone metastases were identified on MRI scans. Although skeletal secondaries can occur outside the area of the scan, the routine inclusion of much of the marrowcontaining skeleton in MRI scans of the kidneys and IVC increase the chance of such metastases being identified coincidentally. Magnetic resonance imaging correctly staged five patients undergoing surgery as Stage IV (Fig. 9). The ability of MRI to produce images in multiple planes as well as the use of different sequences gives it an advantage over CT in identifying local visceral invasion (Hricak et al, 1988; Demas et al, 1988). This is almost certainly present if there is a change of signal intensity or morphology in the adjacent organ and invasion is likely if there is loss of the normal sharp change in signal intensity at the interface between tumour and adjacent viscera (Hricak et al, 1985). Magnetic resonance imaging has been shown to have an accuracy of up to 97-100% for demonstrating direct invasion of a wide range of structures including liver, spleen, psoas and back muscles, mesentery and bowel, although in a study with a relatively small number of Stage IV tumours (Hricak et al, 1988). False positives and false negatives were also encountered in that study, particularly with respect to missing invasion of mesentery and bowel. Conclusion
Magnetic resonance imaging is an accurate means of staging renal cell carcinoma, particularly with respect to IVC involvement. Ideally, pre-operatively, a patient with renal cell carcinoma requires an intravenous urogram, renal ultrasound, chest radiography, an MRI scan and no other routine imaging. Where MRI is available neither CT nor inferior venacavography are required. Acknowledgments The authors gratefully acknowledge the co-operation and encouragement of the Trustees, Dr A. L. T. Beddoes, Dr P. Goddard, Dr C. Johnson and Dr J. L. G. Thomson and the radiographers, Miss A. Case (Superintendent), Miss Joanne Waring (Deputy Superintendent) and Mrs M. Riley of the Bristol MRI Centre. References BOXER R. J., WAISMAN, J., LIEBER, M. M., MAMPASO, F. M. &
SKINNER, D. G., 1979. Renal carcinoma: computer analysis of 96 patients treated by nephrectomy. Journal of Urology, 122, 598-560. CHERRIE, R. J., GOLDMAN, D. G., LINDNER, A. & DEKIRNION,
J. B., 1982. Prognostic implications of vena caval extension of renal cell carcinoma. Journal of Urology, 128, 910-912. CHOYKE, P. L., 1988. MR imaging in renal cell carcinoma. Radiology, 169, 572-573. CHOYKE, P. L., KRESSEL, H. L., POLLACK, H. M., ARGER, P. M.,
Figure 9. Transverse T,-weighted spin-echo sequence (TR 600 ms, TE 26 ms) showing a large left renal cell carcinoma invading the left diaphragmatic crus.
AXEL, L. & MAMOURIAN, A. C , 1984. Focal renal masses: magnetic resonance imaging. Radiology, 152, All-All. CHOYKE, P. L. & POLLACK, H. M.,
1988. The
role of MRI
in
The British Journal of Radiology, August 1991
Magnetic resonance imaging and renal cell carcinoma diseases of the kidney. Radiologic Clinics of North America, 26, 622-623.
DYER, R. B., 1986. Magnetic resonance imaging of the renal mass. Journal of Urology, 136, 566-570.
DEMAS, B. E., STAFFORD, B. A. & HRICAK, H., 1988. Kidneys.
KEARNEY, G. P., BEDFORD WATERS, W., KLEIN, L. A., RICHIE,
In Magnetic Resonance Imaging, ed by D. D. Stark & W. G. Bradley (Mosby, St Louis), pp. 1198-1212.
J. P. & GITTES, R. F., 1981. Results of inferior vena cava resection for renal cell carcinoma. Journal of Urology, 125, 769-772.
DIDIER, D., RACLE, A., ETIEVENT, J. P. & WEILL, F., 1987.
Tumour thrombus of the inferior vena cava secondary to maligHant abdominal neoplasms: US and CT evaluation. Radiology, 162, 83-89. FEIN, A. B., LEE, J. K. T., BALFE, D. M., HEIKEN, J. P., LING, D., GLAZER, H. S. & MCCLENNAN, B. L., 1987. Diagnosis
and staging of renal cell carcinoma: a comparison of MR imaging and CT. American Journal of Roentgenology, 148, 749-753. FROHMULLER, H. G. W., GRUPS, J. W. & HELLER, V., 1987.
Comparative value of ultrasonography, computerised tomography, angiography and excretion urography in the staging of renal cell carcinoma. Journal of Urology, 138, 482^84. GUILIANI, L., GIBERTI, C , MARTORANO, G. & ROVIDA, S., 1990.
Radical extensive surgery for renal cell carcinoma: long-term results and prognostic factors. Journal of Urology, 143, 468-474. HORAN, J. J., ROBERTSON, C. N., CHOYKE, P. L., FRANK, J. A., MILLER, D. L., PASS, H. I. & LINEHAN, W. M., 1989. The
detection of renal carcinoma extension into the renal vein and inferior vena cava: a prospective comparison of venocavography and MRI. Journal of Urology, 142, 943-949. HRICAK, H., DEMAS, B. E., WILLIAMS, R. D., MCNAMARA, M. T., HEDGCOCK, M. W., AMPARO, E. G. & TANAGHO,
E. A., 1985. Magnetic resonance imaging in the diagnosis and staging of renal and perirenal neoplasms. Radiology, 154, 709-715.
LIEBER, M. M., TOMERA, F. M., TAYLOR, W. F. & FARROW,
G. M., 1981. Renal adenocarcinoma in young adults: survival and variables affecting prognosis. Journal of Urology, 125, 164-168. LONDON, N. J. M., MESSIOS, N., KINDER, R. B., SMART, J. G., OSBORN, D. E., WATKIN, E. M. & FLYNN, J. T., 1989. A
prospective study of the value of conventional CT, dynamic CT, ultrasonography and arteriography for staging renal cell carcinoma. British Journal of Urology, 64, 209-217. MCDONALD, M. W., 1982. Current therapy for renal cell carcinoma. Journal of Urology, 127, 211-217. MILES, K. A., LONDON, N. J. M., MESSIOS, N. & SMART, J. G.,
1990. Staging of renal cell carcinoma by dynamic computed tomography: a prospective comparison of two techniques. Clinical Radiology, 42, 122-127. PRITCHETT, T. R., LIESKOVSKY, G. & SKINNER, D. G., 1986.
Extension of renal cell carcinoma into the vena cava: clinical relevance and surgical approach. Journal of Urology, 135, 460-464. PRITCHETT, T. R., RAVAL, J. K., BENSON, R. C , LIESKOVSKY, G., COLLETTI, P. M., BOSWELL, W. D. & SKINNER, D. G.,
1987. Preoperative magnetic resonance imaging of vena caval tumour thrombi: experience with 5 cases. Journal of Urology, 138, 1220-1222. QUINT, L. E., GLAZER, G. M., CHENEVERT, T. L., FECHNER, K. P., GIKAS, P. W., SHIREMAN, P. K., GROSSMAN, H. B. &
KING, C. L., 1988. In vivo and in vitro MR imaging of renal tumours: histopathologic correlation and pulse sequence optimisation. Radiology, 169, 359-362.
HRICAK, H., THOENI, R. F., CARROLL, P. R., DEMAS, B. E., MAROTTI, M. & TANAGHO, E. A., 1988. Detection and
ROBSON, J. R., CHURCHILL, B. M. & ANDERSON, W., 1969. The
staging of renal neoplasms: a reassessment of MR imaging. Radiology, 166, 643-649.
results of radical nephrectomy for renal cell carcinoma. Journal of Urology, 101, 297-301.
HRICAK, H., WILLIAMS, R. D., MOON, K. K., MOSS, A. A., ALPERS, C , CROOKS, L. E. & KAUFMAN, L., 1983. Nuclear
SKINNER, D. G., PFISTER, R. F. & COLVIN, R.,
1972. The
magnetic resonance imaging of the kidney: renal masses. Radiology, 147, 765-772.
extension of renal cell carcinoma into the vena cava: the rationale for aggressive surgical management. Journal of Urology, 107, 711-716.
JOHNSON, C. D., DUNNICK, N. R., COHAN, R. H. & ILLESCAS,
WOODHOUSE, C. R. J., HENDRY, W. F. & BLOOM, H. J. G.,
F. F., 1987. Renal adenocarcinoma: CT staging of 100 tumours. American Journal of Roentgenology, 148, 59-63.
1985. Renal carcinoma. In A Textbook of Genitourinary Surgery, ed. by H. N. Whitfield & W. F. Hendry (Churchill Livingstone, Edinburgh), pp. 952-953.
KARSTAEDT, N., MCCULLOUGH,
Vol. 64, No. 764
D.L., WOLFMAN, N. T. &
689
Magnetic resonance imaging in the staging of renal cell carcinoma By J. E/Kabala, MRCP, FRCR, *D. A. Gillatt, FRCS, *R. A. Persad, FRCS, J. B. Penry, FRCR, \J. C. Gingell, FRCS and *D. Chadwick, FRCS Departments of Radiology and 'Urology, Southmead Hospital, Bristol, UK {Received October 1990 and in revised form February 1991) Keywords: Magnetic resonance imaging (MRI), Carcinoma kidney, Staging
Abstract. A prospective study has been carried out to examine the role of magnetic resonance imaging (MRI) in the investigation of renal cell carcinoma in 24 patients. In all cases the inferior vena cava (IVC) was well demonstrated with MRI. In 14 out of 15 patients where surgical correlation was available, the MRI and operative staging were in agreement. Magnetic resonance imaging and computed tomographic (CT) staging were in agreement in 16 out of the 17 patients where both were performed. In one case, CT suggested hepatic invasion but this was found not to be present on MRI and at operation. Magnetic resonance imaging also provided substantial additional information in three patients, including two cases where MRI demonstrated a patent IVC that appeared occluded on CT (one of which also had vertebral metastases seen on MRI but missed on CT) and one case where CT failed to demonstrate minimal involvement of the IVC. Magnetic resonance imaging is an accurate means of staging renal cell carcinoma with clear advantages over CT. In no case in this series was inferior vena cavography found to be necessary.
In the vast majority of patients with suspected renal cell carcinoma, intravenous urography (IVU) and ultrasound will produce the diagnosis (Demas et al, 1988), occasionally with the addition of percutaneous biopsy. Further investigations will generally be directed towards staging the tumour. Involvement of the renal vein and inferior vena cava (IVC) will affect the surgical technique used (McDonald, 1982; Woodhouse et al, 1985; Pritchett et al, 1986) and the presence of invasion of regional lymph nodes or viscera and distant metastases has considerable prognostic significance (Skinner et al, 1972; Boxer et al, 1979; Lieber et al, 1981). Both ultrasound and computed tomography (CT) (especially dynamic CT) can be used to evaluate the IVC, regional lymph nodes and local invasion of the tumour. Although an accuracy of up to 9 1 % for CT has been reported (Johnson et al, 1987), an accuracy of 72-78% may be more realistic (Frohmuller et al, 1987; Didier et al, 1987; London et al, 1989; Miles et al, 1990). Problems with CT include inadequate or mistimed injection of intravenous contrast medium and limitations of the transverse plane of the section may result in difficulty in evaluating the interface between the tumour and adjacent viscera (Demas et al, 1988). The purpose of this study was to determine the accuracy of magnetic resonance imaging (MRI) staging of renal cell carcinomas, in particular with reference to IVC involvement and to see if inferior vena cavography can be regarded as obsolete in this situation. Address correspondence to: J. E. Kabala, Department of Radiodiagnosis, Bristol Royal Infirmary, Marlborough Street, Bristol BS2 8HW, UK. Vol. 64, No. 764
Methods
Magnetic resonance imaging was performed on 24 patients with renal cell carcinoma. A T{ -weighted spin-echo sequence (TR 600 ms, TE 26 ms) in transverse and coronal planes and a STIR (short tau inversion-recovery) sequence (TR 1900 ms, TE 30 ms, TI 100 ms) in the coronal plane were performed using a Picker Vista 2055 HP 0.5 T scanner. Further sequences in additional planes were performed depending on the appearances on the standard views. The tumours were staged using the conventional classification described by Robson et al (1969) (Table I). The" results were compared with operative findings in 15 cases. In 17 cases comparison with CT was available but only 11 of these were operated on. The CT scans were performed on a Siemens Somatom DRH or a GE 9000 in one of two referring hospitals using 10 mm consecutive cuts through the kidney immediately following an intravenous injection of 50 ml of Iopamidol 370 mg/ml. In three patients who did not undergo surgery, only MRI and ultrasound were performed. Table I. Staging of renal cell carcinoma (Robson et al 1969) Stage I II Ilia Hlb IIIc IV
Limited by the renal capsule Perirenal involvement including the adrenal but limited by Gerota's fascia Involvement of renal vein or IVC Involvement of regional lymph nodes Vascular and lymphatic involvement Invasion of adjacent viscera or distant metastases
683
J. E. Kabala, D. A. Gillatt, R. A. Persad, J. B. Penry, J. C. Gingell and D. Chadwick
Figure 2. Transverse STIR sequence (TR 1900 ms, TE 30 ms, TI 100 ms) showing high signal intensity from the normal kidney and the large renal carcinoma, seen protruding from the posteromedial aspect of the right kidney. Figure 1. Coronal Tx -weighted spin-echo sequence (TR 600 ms, TE 26 ms) showing normal renal architecture on the left; medium-signal-intensity cortex, low-signal-intensity pyramids. This is replaced in the mid and upper part of the right kidney by medium-signal-intensity tumour (vertical arrow). Note similar signal intensity lymph node immediately adjacent to the right lateral border of the first lumbar vertebra (horizontal arrow).
Results
of the carcinoma (Demas et al, 1988; Choyke & Pollack, 1988) (Fig. 3). The MRI staging of the tumours is shown in Table II. They included seven patients with renal vein involvement, five of whom showed extension into the IVC. Three patients with locally operable disease (all Stage I on MRI) did not undergo surgery: in two cases because of ultimately fatal medical conditions (cerebrovascular accident in one patient, bronchopneumonia on a background of chronic obstructive airways disease in the
In all 24 cases the IVC was well demonstrated on MRI. The renal cell carcinomas produced a signal intensity similar to renal tissue on all sequences and therefore were seen as masses of medium signal intensity (slightly higher than the signal from skeletal muscle) on r,-weighted sequences and high signal on STIR sequences (Figs 1 and 2). 2",-weighted sequences demonstrated normal renal architecture well and this was lost in areas occupied by tumour. A common finding was a leash of low-signal-intensity vessels lying just outside the margins of the tumour and presumably representing collateral blood supply or parasitic neovascularization Table II. MRI staging of 24 renal cell carcinomas Stage I II Ilia IIIc IV Total
Number
2 2 1
11* 24
""Includes four with renal vein involvement, two of which showed tumour extension into the IVC.
684
Figure 3. Coronal T,-weighted spin-echo sequence (TR 600 ms, TE 26 ms) showing a leash of new vessels (small arrow) related to the inferior aspect of the left renal cell carcinoma (large arrow). The British Journal oj Radiology, August 1991
Magnetic resonance imaging and renal cell carcinoma Table III. Operative and MRI staging Operative Renal vein involved IVC involved Overall stage I II Ilia Ilk IV Total
MRI
6 5
5 5
6 1 2 1 5 15
5 2 2 1 5 15
other) and in one case because the tumour was in a solitary kidney. Surgery was not performed in six patients with Stage IV disease, in three cases because of distant metastases (two skeletal and one hepatic) and in three cases because of very extensive local tumour. Operative correlation therefore was available in 15 patients. Comparison with CT was available in 17 patients. In three patients only ultrasound was available to correlate the MRI demonstration of an uninvolved IVC. In the 15 patients where operativefindingswere available, the MRI stage agreed with the pathological stage in 14 cases. In one case a Stage I tumour was classified as Stage II on MRI and CT. In all these cases MRI defined accurately IVC involvement and the presence or absence of invasion of regional lymph nodes and adjacent viscera. In one case (overall Stage IV) the MRI was not of sufficiently good quality to be confident of the state of the renal vein, and at operation it was shown to be involved proximally by tumour. The staging of these patients is shown in Table III. The overall staging accuracy of MRI compared with operation was 93%, its specificity and sensitivity for IVC involvement being 100%. In the 17 patients where CT correlation was available, there was agreement with the overall MRI stage in 16 cases. In one case CT suggested hepatic invasion by the tumour, shown not to be present on MRI and at operation. Both MRI and CT suggested a Stage I tumour was Stage II. In two further cases, CT suggested extensive IVC invasion by tumour but MRI demonstrated a compressed but patent IVC (Fig. 4). In one of these patients MRI elegantly demonstrated vertebral metastases not seen on CT (Fig. 5) and in another case MRI defined the upper limit of IVC involvement, which ultrasound and CT had been unable to do (Fig. 6). In one further case CT failed to demonstrate a nodule of tumour in the IVC, shown on MRI and at operation (Fig. 7). This, however, did not change the stage of these tumours. In one patient both MRI and CT demonstrated an irregular tumour continuous with the peritoneum in the presence of ascites. The assumption was made that this was a Stage IV tumour rather unusually directly invading the peritoneum. Unfortunately the patient died Vol. 64, No. 764
Figure 4. Sagittal r,-weighted spin-echo sequence (TR 600 ms, TE 26 ms) showing extension of tumour (vertical arrow) behind a patent IVC (horizontal arrow), producing forward deviation of the vessel.
Figure 5. Coronal STIR sequence (TR 1900 ms, TE 30 ms, TI 100 ms) showing a large left renal cell carcinoma of mixed but predominantly high signal (curved arrows) and high-signal vertebral metastases (horizontal straight arrows).
685
J. E. Kabala, D. A. Gillatt, R. A. Persad, J. B. Penry, J. C. Gingell and D. Chadwick
(a)
Figure 6. Coronal r r weighted spin-echo sequence (TR 600 ms, TE 26 ms) showing right renal cell carcinoma with extension along the right renal vein and into the IVC up to the level of the diaphragm.
without operation and no post-mortem examination was performed. Renal adenocarcinoma cells were, however, demonstrated in the ascites. There was agreement between CT and operative staging in nine out of 11 patients where both were performed, giving a CT accuracy of 82% (Table IV). The two cases of disagreement are described above. In three patients who did not undergo surgery (two Stage IV, one Stage I) only ultrasound and MRI were performed; both demonstrated an absence of IVC involvement. Discussion
A variety of methods is available for staging renal cell carcinoma. Excretion urography and angiography are at best 59% and 57% accurate, respectively (Frohmuller et al, 1987). Ultrasound is almost invariably performed as part of the diagnostic process for renal cell carcinomas. Table IV. Operative and CT staging
Stage I II Ilia IIIc IV Total
686
Operative
CT
4 1 2 1 3 11
3 2 2 0 4 11
Figure 7. (a) Transverse and (b) coronal Tx -weighted spin-echo sequences (TR 600 ms, TE 26 ms) showing tumour extending along an expanded left renal vein [white arrows, Fig. 7(a)]. Tumour is seen to just protrude into the patent IVC [black horizontal arrows, (a) and (b)]. In the coronal plane the nodule of tumour entering the IVC from the left renal vein is well shown [curved arrows, Fig. 7(b)].
Its accuracy for staging is up to about 78% (Frohmuller et al, 1987), with some studies suggesting a lower figure of around 50% (London et al, 1989). Excess bowel gas and adverse patient build may prevent an adequate examination. Clearly the performance of an ultrasound examination depends on the operator's skill but all means of staging are to some extent subjective. Although Johnson et al (1987) reported an accuracy of 91% for CT staging (using a dynamic contrast technique) a number of comparative studies have suggested much lower values. Frohmuller et al (1987) reported a The British Journal of Radiology, August 1991
Magnetic resonance imaging and renal cell carcinoma
staging accuracy of 78% for ultrasound compared with study was 93%. The single error of staging was 72% for CT (although the CT technique used is not mistaking a Stage I tumour for Stage II and was also described). More recently, London et a( (1989) reported made on CT. Others have reported this difficulty with a 50% accuracy for conventionally enhanced CT and both techniques (Fein et al, 1987; Johnson et al, 1987) 72% accuracy for dynamic CT. The lower figure for and it is likely that at the current stage of development conventionally enhanced CT may be partly due to their neither CT nor MRI can reliably distinguish between technique of performing the post-contrast scan 5 Stage I and Stage II. It is, however, not a significant minutes after the intravenous injection. The same group problem since the prognosis and treatment of both of workers suggest a single-location dynamic scan in categories of tumour are virtually identical unless a addition to an incremental scan during infusion of partial nephrectomy is being planned (Robson et al intravenous contrast medium may increase CT accuracy 1969; Boxer et al, 1979; Guiliani et al, 1990). above 72%, especially for the vascular stage (Miles et al, In all patients studied, the IVC was well seen. In five 1990). patients undergoing surgery, IVC involvement was In addition to the general problems of CT compared present and accurately defined. In one further case with with ultrasound or MRI (the use of intravenous iodine- proximal renal vein involvement the MRI was not of containing contrast media and ionizing radiation), diagsufficiently good quality to stage the renal vein. Because nostic errors in CT staging may arise because of the of the signal void in flowing blood, patent vessels are limitations of scanning only in the transverse plane easily seen in good-quality scans and vascular invasion especially when dealing with polar tumours (Demas et by tumour is elegantly demonstrated by MRI al, 1988). Using strict CT criteria—change in organ size (Karstaedt et al, 1986; Fein et al, 1987; Pritchett et al, or density—adjacent organ invasion is only identified in 1987). The MRI sensitivity for distal renal vein and IVC 60% of cases; diagnosing invasion on the basis of a loss involvement is around 100%, sensitivity and specificity of fat line alone results in a rate of false positives of for proximal renal vein involvement being 86 and 96%, around 15% (Johnson et al, 1987; Demas et al, 1988). respectively (Hricak et al, 1988). In this group of Although inferior vena cavography can detect IVC patients the operative approach depends on adequate and renal vein involvement with an accuracy of 93% demonstration of the intravenous extension. and 88%, respectively (Horan et al 1989), it offers no The prognosis for renal vein extension alone is probinformation about local invasion and only detects ably similar to that for Stage I disease (Skinner et al, lymph nodes sufficiently enlarged to impinge on the 1972); the prognosis for patients with IVC involvement IVC. Horan et al (1989) compared MRI accuracy for is still debated. Some studies suggest it approaches that vascular staging with inferior vena cavography and of Stage I disease (Skinner et al, 1972; Boxer et al, 1979; found it was not significantly different (MRI accuracy Cherrie et al, 1982; Pritchett et al, 1986). Other studies for IVC and renal vein involvement being 93% and are more pessimistic (Robson et al, 1969; Kearney et al, 86%, respectively). 1981). The most recent and one of the largest studies Our study examined the role of MRI in the investi- (Guiliani et al, 1990) shows a poor prognosis with IVC gation of renal cell carcinomas. The tumours were involvement, similar to that for patients with distant found to have a medium signal on T,-weighted metastases. sequences and high signal intensity on the STIR Neither ultrasound nor CT can distinguish tumour or sequence. This was very close to the signal from normal thrombus extension into the IVC from invasion of the renal tissue. The similarity between signal from normal vessel wall (Didier et al, 1987) and there is no evidence kidney and tumour on all sequences including proton to suggest MRI can either. density and T^-weighted sequences has been noted Although the operative approach may be unaffected before (Quint et al, 1988). Tumours, therefore, are not by the presence of involved lymph nodes, this has a generally visible on MRI until they are large enough to significant effect on prognosis (Skinner et al, 1972; distort renal architecture (well seen on the Tx -weighted Guiliani et al, 1990). Two patients in this study had sequence in high-quality scans; Fig. 1) or the renal extensive para-aortic lymphadenopathy and did not outline (Quint et al, 1988; Choyke, 1988). undergo operation. In both, the CT scan suggested IVC A number of studies have described the use of MRI in occlusion but MRI demonstrated a patent, albeit studying renal disease. It was quickly realized that MRI compressed and deviated, IVC. could identify focal renal masses and distinguish solid Of the 15 patients undergoing surgery in this series from cystic lesions (Choyke et al, 1984). Although only one had regional lymph node involvement by initially it was thought that histologically specific signals tumour (Fig 8.) and this was correctly predicted by both might be obtainable from renal tumours (Hricak et al, MRI and CT. Only limited comments can be made on 1983), this promise has not yet been fulfilled and differ- the basis of this experience but it is probable that MRI entiation between benign and malignant renal tumours will be at least as accurate as CT in detecting involved cannot be made on the basis of signal intensity (Hricak lymph nodes. Hricak et al (1988) reported an accuracy et al, 1988). of 99% in predicting lymph node involvement preoperatively with MRI. They considered any lymph node The accuracy of MRI staging of renal cell carcinoma has been reported at 82-96% (Hricak et al, 1985, 1988). in the perivascular or perihilar areas with a diameter The overall accuracy in the 15 patients in this current above 10 mm (plane not specified) to be involved, Vol. 64, No. 764
687
J. E. Kabala, D. A. Gillatt, R. A. Persad, J. B. Penry, J. C. Gingell and D. Chadwick
Figure 8. Transverse T,-weighted spin-echo sequence (TR 600 ms, TE 26 ms) showing an enlarged lymph node (curved arrow) lying medial to the right kidney (containing a large renal cell carcinoma anterolaterally, vertical arrow) and posterior to an expanded IVC occluded with tumour thrombus (horizontal arrow).
commenting that previously they had encountered lymph nodes between 10 and 15 mm diameter that were involved with tumour. Collateral vessels with their signal void on MRI are unlikely to be confused with lymph nodes (Fein et al, 1987). However, MRI and CT are both currently unable to distinguish reactive lymph
node enlargement from those involved with tumour (Choyke & Pollack, 1988; Hricak et al, 1988). In two patients who did not undergo surgery, bone metastases were identified on MRI scans. Although skeletal secondaries can occur outside the area of the scan, the routine inclusion of much of the marrowcontaining skeleton in MRI scans of the kidneys and IVC increase the chance of such metastases being identified coincidentally. Magnetic resonance imaging correctly staged five patients undergoing surgery as Stage IV (Fig. 9). The ability of MRI to produce images in multiple planes as well as the use of different sequences gives it an advantage over CT in identifying local visceral invasion (Hricak et al, 1988; Demas et al, 1988). This is almost certainly present if there is a change of signal intensity or morphology in the adjacent organ and invasion is likely if there is loss of the normal sharp change in signal intensity at the interface between tumour and adjacent viscera (Hricak et al, 1985). Magnetic resonance imaging has been shown to have an accuracy of up to 97-100% for demonstrating direct invasion of a wide range of structures including liver, spleen, psoas and back muscles, mesentery and bowel, although in a study with a relatively small number of Stage IV tumours (Hricak et al, 1988). False positives and false negatives were also encountered in that study, particularly with respect to missing invasion of mesentery and bowel. Conclusion
Magnetic resonance imaging is an accurate means of staging renal cell carcinoma, particularly with respect to IVC involvement. Ideally, pre-operatively, a patient with renal cell carcinoma requires an intravenous urogram, renal ultrasound, chest radiography, an MRI scan and no other routine imaging. Where MRI is available neither CT nor inferior venacavography are required. Acknowledgments The authors gratefully acknowledge the co-operation and encouragement of the Trustees, Dr A. L. T. Beddoes, Dr P. Goddard, Dr C. Johnson and Dr J. L. G. Thomson and the radiographers, Miss A. Case (Superintendent), Miss Joanne Waring (Deputy Superintendent) and Mrs M. Riley of the Bristol MRI Centre. References BOXER R. J., WAISMAN, J., LIEBER, M. M., MAMPASO, F. M. &
SKINNER, D. G., 1979. Renal carcinoma: computer analysis of 96 patients treated by nephrectomy. Journal of Urology, 122, 598-560. CHERRIE, R. J., GOLDMAN, D. G., LINDNER, A. & DEKIRNION,
J. B., 1982. Prognostic implications of vena caval extension of renal cell carcinoma. Journal of Urology, 128, 910-912. CHOYKE, P. L., 1988. MR imaging in renal cell carcinoma. Radiology, 169, 572-573. CHOYKE, P. L., KRESSEL, H. L., POLLACK, H. M., ARGER, P. M.,
Figure 9. Transverse T,-weighted spin-echo sequence (TR 600 ms, TE 26 ms) showing a large left renal cell carcinoma invading the left diaphragmatic crus.
AXEL, L. & MAMOURIAN, A. C , 1984. Focal renal masses: magnetic resonance imaging. Radiology, 152, All-All. CHOYKE, P. L. & POLLACK, H. M.,
1988. The
role of MRI
in
The British Journal of Radiology, August 1991
Magnetic resonance imaging and renal cell carcinoma diseases of the kidney. Radiologic Clinics of North America, 26, 622-623.
DYER, R. B., 1986. Magnetic resonance imaging of the renal mass. Journal of Urology, 136, 566-570.
DEMAS, B. E., STAFFORD, B. A. & HRICAK, H., 1988. Kidneys.
KEARNEY, G. P., BEDFORD WATERS, W., KLEIN, L. A., RICHIE,
In Magnetic Resonance Imaging, ed by D. D. Stark & W. G. Bradley (Mosby, St Louis), pp. 1198-1212.
J. P. & GITTES, R. F., 1981. Results of inferior vena cava resection for renal cell carcinoma. Journal of Urology, 125, 769-772.
DIDIER, D., RACLE, A., ETIEVENT, J. P. & WEILL, F., 1987.
Tumour thrombus of the inferior vena cava secondary to maligHant abdominal neoplasms: US and CT evaluation. Radiology, 162, 83-89. FEIN, A. B., LEE, J. K. T., BALFE, D. M., HEIKEN, J. P., LING, D., GLAZER, H. S. & MCCLENNAN, B. L., 1987. Diagnosis
and staging of renal cell carcinoma: a comparison of MR imaging and CT. American Journal of Roentgenology, 148, 749-753. FROHMULLER, H. G. W., GRUPS, J. W. & HELLER, V., 1987.
Comparative value of ultrasonography, computerised tomography, angiography and excretion urography in the staging of renal cell carcinoma. Journal of Urology, 138, 482^84. GUILIANI, L., GIBERTI, C , MARTORANO, G. & ROVIDA, S., 1990.
Radical extensive surgery for renal cell carcinoma: long-term results and prognostic factors. Journal of Urology, 143, 468-474. HORAN, J. J., ROBERTSON, C. N., CHOYKE, P. L., FRANK, J. A., MILLER, D. L., PASS, H. I. & LINEHAN, W. M., 1989. The
detection of renal carcinoma extension into the renal vein and inferior vena cava: a prospective comparison of venocavography and MRI. Journal of Urology, 142, 943-949. HRICAK, H., DEMAS, B. E., WILLIAMS, R. D., MCNAMARA, M. T., HEDGCOCK, M. W., AMPARO, E. G. & TANAGHO,
E. A., 1985. Magnetic resonance imaging in the diagnosis and staging of renal and perirenal neoplasms. Radiology, 154, 709-715.
LIEBER, M. M., TOMERA, F. M., TAYLOR, W. F. & FARROW,
G. M., 1981. Renal adenocarcinoma in young adults: survival and variables affecting prognosis. Journal of Urology, 125, 164-168. LONDON, N. J. M., MESSIOS, N., KINDER, R. B., SMART, J. G., OSBORN, D. E., WATKIN, E. M. & FLYNN, J. T., 1989. A
prospective study of the value of conventional CT, dynamic CT, ultrasonography and arteriography for staging renal cell carcinoma. British Journal of Urology, 64, 209-217. MCDONALD, M. W., 1982. Current therapy for renal cell carcinoma. Journal of Urology, 127, 211-217. MILES, K. A., LONDON, N. J. M., MESSIOS, N. & SMART, J. G.,
1990. Staging of renal cell carcinoma by dynamic computed tomography: a prospective comparison of two techniques. Clinical Radiology, 42, 122-127. PRITCHETT, T. R., LIESKOVSKY, G. & SKINNER, D. G., 1986.
Extension of renal cell carcinoma into the vena cava: clinical relevance and surgical approach. Journal of Urology, 135, 460-464. PRITCHETT, T. R., RAVAL, J. K., BENSON, R. C , LIESKOVSKY, G., COLLETTI, P. M., BOSWELL, W. D. & SKINNER, D. G.,
1987. Preoperative magnetic resonance imaging of vena caval tumour thrombi: experience with 5 cases. Journal of Urology, 138, 1220-1222. QUINT, L. E., GLAZER, G. M., CHENEVERT, T. L., FECHNER, K. P., GIKAS, P. W., SHIREMAN, P. K., GROSSMAN, H. B. &
KING, C. L., 1988. In vivo and in vitro MR imaging of renal tumours: histopathologic correlation and pulse sequence optimisation. Radiology, 169, 359-362.
HRICAK, H., THOENI, R. F., CARROLL, P. R., DEMAS, B. E., MAROTTI, M. & TANAGHO, E. A., 1988. Detection and
ROBSON, J. R., CHURCHILL, B. M. & ANDERSON, W., 1969. The
staging of renal neoplasms: a reassessment of MR imaging. Radiology, 166, 643-649.
results of radical nephrectomy for renal cell carcinoma. Journal of Urology, 101, 297-301.
HRICAK, H., WILLIAMS, R. D., MOON, K. K., MOSS, A. A., ALPERS, C , CROOKS, L. E. & KAUFMAN, L., 1983. Nuclear
SKINNER, D. G., PFISTER, R. F. & COLVIN, R.,
1972. The
magnetic resonance imaging of the kidney: renal masses. Radiology, 147, 765-772.
extension of renal cell carcinoma into the vena cava: the rationale for aggressive surgical management. Journal of Urology, 107, 711-716.
JOHNSON, C. D., DUNNICK, N. R., COHAN, R. H. & ILLESCAS,
WOODHOUSE, C. R. J., HENDRY, W. F. & BLOOM, H. J. G.,
F. F., 1987. Renal adenocarcinoma: CT staging of 100 tumours. American Journal of Roentgenology, 148, 59-63.
1985. Renal carcinoma. In A Textbook of Genitourinary Surgery, ed. by H. N. Whitfield & W. F. Hendry (Churchill Livingstone, Edinburgh), pp. 952-953.
KARSTAEDT, N., MCCULLOUGH,
Vol. 64, No. 764
D.L., WOLFMAN, N. T. &
689
