Diuresis renography is established as the standard approach for assessing suspected renal outflow obstruction in adults, having clear advantages over IVU as it distinguishes true obstruction from chronic pelvic dilatation due to other causes in most cases. It also provides a measure of split renal function. However, in 10–15% of patients the renographic response to frusemide is equivocal, often through poor renal function or gross renal pelvic dilatation . The traditional next step with these patients has been to take a further renogram, administering frusemide 15 min beforehand (F-15) . This gives an increased diuresis during the study period, thus allowing resolution of equivocal outcomes in most cases. However, it entails the inconvenience, cost and increased radiation dose associated with a second separate investigation.
One proposed method for overcoming these problems is to calculate renal output efficiency (OE) , which is defined for an individual kidney as the total output up to time t, expressed as a percentage of the total input up to that time. A recent simulation study confirmed the validity of this measurement . The present study describes the methods for calculating OE and some data related to our initial clinical experience.
Renography was carried out according to guidelines produced by the Consensus Committee on Diuresis Renography, recently updated . Patients were pre-hydrated with 500 mL of water and allowed to void just before starting the test. Patients were generally seated, with their backs against the face of the gamma camera. A special imaging chair was used, with careful attention given to the comfort of the patient. If there was genuine doubt about the patient's ability to sit still for 30 min, they were placed supine instead (with the camera under a transparent imaging couch). In either case it was ensured, by reference to anatomical landmarks, that both the heart and kidneys would be within the field of view. With modern large field-of-view cameras it is usually possible to include the whole of the bladder as well, but this is regarded as desirable rather than essential, so technical staff should be trained to concentrate primarily on the heart and kidneys, particularly when using a relatively small field-of-view camera.
After a rapid bolus injection with 80 MBq 99mTc-MAG3, 90 20-s digital frames were acquired on the computer, giving a total dynamic study duration of 30 min. Frusemide (at 0.5 mg/kg) was given intravenously 15 min after injection, thus allowing the response of the kidneys to be assessed. We use the notation (F+15, t= 30) to indicate both the timing of the diuretic (relative to the radiopharmaceutical injection) and the duration of the dynamic study. A summary (pseudo-analogue) film was produced in which the dynamic study was condensed into 16 frames, allowing a quick overview for reporting. Static views before and after voiding (showing the kidneys and bladder) were obtained at the end of the dynamic study (with the patient erect, either sitting or standing) and the mean urinary flow rate was also measured from a timed urine collection. The dynamic images were processed using standard software. Time-activity curves were derived from computer-generated regions of interest (ROIs) and relative renal function measured using the Patlak-Rutland method . An experienced nuclear-medicine physicist and radiologist interpreted the renogram curves according to standard criteria. A curve that was clearly descending before frusemide injection was generally classified as not obstructive. Curves that continued rising before intervention were classified as obstructive, not obstructive or equivocal (Fig. 1). Extensive use was also made of the renogram images and any previous IVU films when reporting. Genuine hydronephrosis and a simple prominent extra-renal pelvis could thus usually be distinguished. Hydronephrosis could then be classified as obstructive or not obstructive.
The method of calculating OE was described previously . Essentially, the presumed cumulative (or ‘integrated’) input to the kidney was derived by integrating the blood clearance curve (derived from a ROI placed over the left ventricle) and fitting (i.e. fixing) the first part to the second phase of the background-corrected renogram using an iterative least-squares technique (Fig. 2). Theoretically, the resultant continuously rising curve represents the shape of renogram that would be produced if there were no output. The cumulative output from the kidney was calculated by subtracting the actual background corrected kidney curve from the cumulative input. Finally, the OE was calculated as the ratio (expressed as a percentage) of the cumulative output to cumulative input at t = 30 min. Objective criteria used for interpreting the OE were: > 78%, no obstruction; 70–78% equivocal, < 70% obstructed. No equivocal range was mentioned in the original paper , so this was set as above, after subsequent informal discussion with the authors. The OE was calculated using commercial software and without knowing the standard renogram report.
The OE was calculated for 320 renal units in 162 patients undergoing F+15 renography. Of these, 28 (9%) renal units showed an equivocal frusemide response; a representative result is given in Fig. 3. Use of the OE reclassified 19 (66%) of these (10 not obstructed, nine obstructed). To validate the use of the OE a clinical trial is planned where it will be compared with F-15 renography.
COMPARISON WITH OTHER METHODS
Other methods have been used to reduce the need for repeat renography in equivocal cases. As explained above, the underlying problem is thought to be poor isotope excretion in those with impaired renal function or grossly dilated upper tracts. Views after voiding might be of value in resolving problems of stasis and are relatively easy to obtain. However, they will only resolve those cases in which the retention of activity in the renal pelvis is a result of simple postural or lower urinary tract effects. Giving frusemide at the start of the study (F+0) might be expected to reduce the equivocal rate by increasing diuresis during the whole study, but this means giving frusemide to all patients (in F+15 renography this decision can be delayed until the need is clear) and it does not allow the effect of the diuretic on the renogram curves to be determined. Furthermore, a recent study suggested that in adult patients F+0 renography did not provide much additional information over conventional studies .
The use of the OE has been assessed in children, where it was clearly shown to be the best predictor of histologically confirmed obstruction . In that study of 74 children, 20 of 92 (22%) renograms in hydronephrotic kidneys were initially equivocal. The calculation of OE allowed correct classification in 12 of them, a result in keeping with the findings of the present study. The OE was particularly useful in those kidneys with reduced differential function. Other groups have described similar methods to the OE that also purport to be minimally affected by poor renal function, e.g. the pelvic excretion efficiency is a simpler variable than OE that has shown encouraging results in children with hydronephrosis . The normalized residual activity is that at the end of the renogram, expressed as a percentage of the renal activity 2 min after tracer injection , and is reported to be comparable in accuracy to the OE. Although both of these other methods are computationally simpler than OE, the calculation of OE has a much sounder theoretical basis. Furthermore, the software required is not complex and is now commercially available, based on several nuclear medicine computer systems.
ADVANTAGES AND DISADVANTAGES
It has been suggested that estimates of OE might not be comparable among centres , i.e. the OE values obtained might be affected by differences in background correction or analysis of the heart curve. However, the errors described were most significant at low excretion rates, and in the equivocal range of most interest, and only affected OE by 1–2%. Perhaps of more concern in comparing studies is the difference in the timing of frusemide administration that has been reported. There is currently a paucity of published information on this aspect, so further studies will be required before a consensus can be reached. Interestingly, giving frusemide at 15 min in this study (F+15, t = 30) worked just as well (in relation to the discriminatory ability of OE) as in the original study , where frusemide was given at 18 min. The OE clearly depends on the time at which it is measured (OE20, OE30, etc.), so this must be standardized at 30 min to achieve reproducible results.
In conclusion, this initial evaluation of the use of the OE suggests that it is a useful quantitative variable for resolving equivocal F+15 (or F+20) renograms.