¹²³I-Hippuran Renal Scintigraphy with Evaluation of Single-Kidney Clearance for Predicting Renal Scarring After Acute Urinary Tract Infection: Comparison with ^99mTc-DMSA Scanning

Patient Population

Among children admitted to our Pediatric Unit during a 2-y period, 58 (21 boys, 37 girls; age range, 0.5–54 mo; mean age, 7.2 ± 12.4 mo) with signs and symptoms suggestive of APN (3) were selected for this study. The inclusion criteria were the following: age less than 10 y, no history of previous UTI, no hypodysplasia pattern on either prenatal or routine sonography, no urinary tract obstruction causing secondary reflux, and no recurrent infection between the initial (during the APN phase) and subsequent ^99mTc-DMSA examinations.

With a renal unit defined as 1 kidney and its ureter, 116 renal units were examined.

The diagnosis of APN was based on both clinical examination (unexplained fever, irritability, vomiting, diarrhea, and failure to thrive) and laboratory examinations (quantitative C-reactive protein, sedimentation rate, white blood cell count, and catheterized urine culture) (3). The urine culture was considered positive if the number of colonies of a single organism (CFU) was ≥10⁵/mL.

For confirmation of the clinical and laboratory diagnosis of APN, all children underwent renal and pelvic sonography, ^99mTc-DMSA scanning, and RSS within a week of the acute episode. On the basis of laboratory findings, clinical symptoms, and imaging studies, all children received appropriate medical treatment (3). For detection and graduation of vesicoureteral reflux, all patients underwent x-ray voiding cystourethrography 1 mo after resolution of the acute episode. All patients entered the follow-up protocol, including both periodic clinical examinations and urine cultures (3).

Because an APN lesion takes about 5 mo to stabilize (9), ^99mTc-DMSA scanning and RSS were repeated at least 5 mo after UTI was diagnosed (follow-up range, 5–19 mo; mean follow-up, 10 ± 3.5 mo) and after having excluded any other UTI by multiple urine cultures. Renal scarring was diagnosed whenever a cortical defect, detected at the first ^99mTc-DMSA examination, persisted on the follow-up one. Therefore, according to the results of follow-up ^99mTc-DMSA scanning, the examined renal units were retrospectively classified into 2 groups: scarred kidney (SK) and nonscarred kidney (NSK).

^99mTc-DMSA Scanning and RSS Technical Features

^99mTc-DMSA scintigraphy was performed according to our usual clinical protocol. The administered dose was 0.74 MBq/kg of body weight (at least 10 MBq). One posterior projection and 2 posterior oblique projections were recorded beginning 3 h after dose administration, with the patient lying supine and the γ-camera head directly beneath the examination bed. A large-field-of-view γ-camera (ZLC; Siemens) interfaced with an ICON workstation (Siemens) was equipped with a low-energy (140 keV), ultra-high-resolution collimator for a 128 × 128 matrix acquisition. The total count of each image was 500,000. Images were stored on computer to allow digital image enhancement for better evaluation. An experienced observer, unaware of the other patient data, evaluated the images. Renal parenchymal involvement seen on ^99mTc-DMSA scans was graded as 0 (normal), 1 (1 lesion), 2 (2 lesions), or 3 (diffuse damage with renal parenchymal subversion).

RSS was performed using the same modalities and γ-camera as were used for ^99mTc-DMSA scintigraphy. Twenty to 75 MBq of ¹²³I-OIH were prepared in a 5-mL syringe, and the contained volume was increased to 2 mL with saline. Before the patient examination began, the syringe was positioned on a specially prepared 5-cm-thick acrylic support and placed in contact with the collimator surface on the examination table. Activity was measured by the γ-camera; one 50-s static image was acquired in a 64 × 64 matrix by a 20% window centered over the ¹²³I 159-keV energy peak. After this measurement, ¹²³I-OIH was injected as a bolus and data were acquired by a 2-phase dynamic acquisition (80 frames × 5 s per frame and 65 frames × 20 s per frame). After the examination, the depth of the kidney was measured on 1 lateral view by a ⁵⁷Co-marker.

Cl was then computed by a method and proprietary program previously validated at our institution (10, 11). This method is based on time–activity curves measured on the heart and kidney areas by the region-of-interest technique. Reference Cl values for an age-matched population were previously established to be 333 ± 49 mL/min/1.73 m² of body surface area (BSA) (12).

X-ray voiding cystourethrography was performed without sedation 1 mo after APN, and the results were reported by an expert pediatric radiologist. All children were previously prepared by preventive treatment with broad-spectrum antibiotics. The bladder was catheterized by a Foley catheter and gently filled with radiographic contrast material. The children then voided during fluoroscopy. According to the recommendations of the International Reflux Study in Children (13), a grade of 1–5 was assigned to indicate the severity of vesicoureteral reflux.

Statistical Analysis

All results are expressed as mean ± SD. The relationship between ^99mTc-DMSA scores and the Cl variable was assessed by the Spearman nonparametric regression coefficient. ANOVA and Tukey post hoc testing were used for between-group comparisons. Proportions were compared through the Fisher exact test. Univariate discriminant analysis was performed on the variable Cl (relative to the inflammatory acute phase) to linearly best discriminate (predictive value) between healthy and scarred renal units on the control ^99mTc-DMSA scans. This procedure computes a linear function yielding the optimal possible classification of observations, taking the result of control ^99mTc-DMSA scanning (NSK or SK) as the grouping factor. The achieved classification, after cross-validation, was submitted to a jackknife validation procedure to limit bias in the number of patients correctly classified. Paired analysis of receiver operating characteristic (ROC) curves was used to compare tests (^99mTc-DMSA and RSS).

The STATISTICA (STATSOFT) and the MiniTab Statistical Software (MINITAB) packages were used for statistical data analysis. The software used for paired analysis of ROC curves was Computational Methods for Diagnostic Tests (Freie Universität Berlin). P < 0.05 was considered statistically significant.

Staging Evaluation

All children presented with high fever (≥38.5°C), with 9 showing a failure to thrive and 5 having hip pain. All children had a positive urine culture (10⁵–10⁶ CFU/mL) and an abnormal level of C-reactive protein (0.3–20 mg/L) accompanied by high values for both sedimentation rate (>30) and white blood cell count (1.4–2.0 × 10⁴/mm³). The patients showed no sign of acute renal failure or dehydration on either ^99mTc-DMSA scanning or RSS.

Sonography showed signs suggestive of APN in 40 kidneys (bilateral in 16 children; unilateral in 8 children) and had negative findings in the remaining 76 kidneys.

^99mTc-DMSA scanning showed renal damage in 76 of 116 kidneys (bilateral in 20 children; unilateral in 36 children): 40 renal units were graded 0, 49 were graded 1, and 21 and 6 were graded 2 and 3, respectively (Table 1). The remaining 40 kidneys did not show parenchymal damage (2 patients had bilaterally negative findings). None of the ^99mTc-DMSA findings were equivocal. Table 1 shows the Cl values measured for the different classes of renal damage according to the ^99mTc-DMSA results. Cl values differed significantly from each other (P < 0.001), with the exception of the values for grade 1 versus grade 2 (Fig. 1).

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 1.

Staging ¹²³I-OIH Cl in all 116 examined kidneys, classified into 4 groups on basis of uptake pattern of ^99mTc-DMSA scan obtained during APN. * = outliers.

Regression analysis showed a significant relationship between morphologic (^99mTc-DMSA grading) and functional (Cl values) renal damage (R = 0.69; P < 0.0001).

According to the voiding cystourethrography, 40 of 116 examined kidneys had grade I–IV primary reflux. ^99mTc-DMSA scanning had positive findings in 34 kidneys affected by vesicoureteral reflux (85%). Of 76 kidneys without signs of reflux, ^99mTc-DMSA scanning had positive findings in 42 (55%) (Table 2). The incidence of parenchymal ^99mTc-DMSA deficits was significantly different according to the presence or absence of reflux (P < 0.02).

Follow-up Evaluation

Follow-up was completed for all patients 5–19 mo (mean, 10 ± 3.5 mo) after staging. All children underwent ^99mTc-DMSA scanning and RSS. Follow-up ^99mTc-DMSA showed persistent renal damage (scarring) in 40 of 76 kidneys and complete recovery in 36 kidneys, of which 28 and 8 were classified as grades 1 and 2, respectively, at staging (recovery rate, 47.4%). In none of the children with normal ^99mTc-DMSA findings at staging did any parenchymal abnormalities develop. Hence, ^99mTc-DMSA findings were classified as follows: 76 as grade 0, 32 as grade 1, and 6 and 2 as grades 2 and 3, respectively (Table 3). Indeed, among 21 kidneys classified as grade 2 at staging, 5 demonstrated an unchanged uptake pattern and 8 showed an improvement in scar grade (grade 1). Among the 6 kidneys classified asgrade 3 at staging, 2 showed persistent unchanged renal scarring and 4 showed an improvement in scintigraphic pattern to grade 2 (1 kidney) or grade 1 (3 kidneys).

Table 3 shows the values of Cl measured for the different classes of renal damage according to ^99mTc-DMSA scintigraphy. The reported Cl values differed significantly from each other, with the exception of the values for grade 1 versus grade 2 (P < 0.001).

Retrospective Analysis

With the presence or absence of parenchymal scarring at follow-up ^99mTc-DMSA scanning considered the grouping factor, all studied kidneys were retrospectively categorized into 1 of 2 groups: NSK or SK.

The 1-way ANOVA applied to all staging parameters (age, temperature at presentation, duration of fever and symptoms, C-reactive protein, sedimentation rate, white blood cell count, CFU, and Cl) showed a significant difference only for Cl (P < 0.0001).

At staging RSS, mean Cl values were 260 ± 26 and 187 ± 44 mL/min/1.73 m² of BSA for the NSK and SK groups, respectively (P < 0.0002). At the following RSS, mean Cl values were 291 ± 32 and 246 ± 54 mL/min/1.73 m² of BSA for NSK and SK, respectively (P < 0.0002). A significant difference between groups (P < 0.0001) was found for Cl (Table 4; Fig. 2).

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 2.

Comparison of staging and follow-up ¹²³I-OIH Cl in 2 renal subgroups classified on basis of presence or absence of scarring on follow-up ^99mTc-DMSA scan, which was obtained after mean of 10 ± 3.5 mo. Kidneys with normal ^99mTc-DMSA findings at first examination were considered control group. * = outliers.

^99mTc-DMSA grading was best able to detect acute lesions with a favorable evolution when 2 different ^99mTc-DMSA grades were considered as decisional cutoff values. When grade 1 was selected as the cutoff value, kidneys graded 0 or 1 at staging were considered to have a good prognosis and those graded 2 or 3 were considered to have a bad one. When this criterion was applied, 87 of 116 kidneys were correctly classified, with a specificity, sensitivity, positive predictive value, negative predictive value, and overall accuracy of 89%, 47%, 70%, 76%, and 75%, respectively (Table 5).

Grade 2 was selected as the second cutoff value; in this case, grades 0, 1, and 2 were considered indicators of complete parenchymal recovery and grade 3 was considered an indicator of scar development. When this cutoff was used, 82 of 116 kidneys were correctly classified, with a specificity, sensitivity, positive predictive value, negative predictive value, and overall accuracy of 100%, 15%, 100%, 69%, and 71%, respectively (Table 5).

To differentiate the NSK group from the SK group, the Cl cutoff value was determined, for staging Cl values, by univariate discriminant analysis: 232 mL/min/1.73 m² of BSA discriminated best between the SK and NSK groups. Using this Cl cutoff value, the classification matrix was derived. The Cl variable correctly predicted the evolution of acute damage in 110 of 116 kidneys, with a specificity, sensitivity, positive predictive value, negative predictive value, and overall accuracy of 95%, 95%, 90%, 97%, and 95%, respectively (Table 5). RSS failed to correctly classify 6 kidneys, having false-negative findings for 4 and false-positive findings for 2.

Matched-pair analysis of the ROC curves showed a significant difference between those derived from the ^99mTc-DMSA and RSS data (P < 0.0001).