Enhancement Characteristics of Papillary Renal Neoplasms Revealed on Triphasic Helical CT of the Kidneys

Transcription

1 Brian R. Herts 1 Deirdre M. Coll 1,2 Andrew C. Novick 3 Nancy Obuchowski 1,4 Grant Linnell 1 Susan L. Wirth 1 Mark E. Baker 1 Received June 29, 2001; accepted after revision August 23, Department of Radiology-H66, The Cleveland Clinic Foundation, 9500 Euclid Ave., Cleveland, OH Address correspondence to B. R. Herts. 2 Present address: Department of Radiology, Weill Hospital, Cornell University Medical Center, 525 E. 68th St., New York, NY Urological Institute, The Cleveland Clinic Foundation, Cleveland, OH Department of Biostatistics, The Cleveland Clinic Foundation, Cleveland, OH AJR 2002;178: X/02/ American Roentgen Ray Society Enhancement Characteristics of Papillary Renal Neoplasms Revealed on Triphasic Helical CT of the Kidneys OBJECTIVE. The purpose of this study was to determine whether renal tumor enhancement or heterogeneity on triphasic helical CT scans is predictive of the papillary cell subtype or nuclear grade of renal cell carcinoma. MATERIALS AND METHODS. We reviewed the CT scans of 90 consecutive patients with renal masses who had undergone triphasic renal helical CT before a complete or partial nephrectomy (12 with papillary renal cell carcinomas, 66 with nonpapillary renal cell carcinomas, and 12 with benign lesions). Three radiologists who were unaware of the patients diagnoses retrospectively and independently measured the attenuation of each patient s tumor, abdominal aorta, and normal renal parenchyma on the scans obtained during all three phases. Ratios of tumor-to-aorta enhancement and tumor-to normal renal parenchyma enhancement were calculated for both of the phases performed after contrast material had been administered. Tumor heterogeneity was calculated as the difference between the highest and lowest attenuation values divided by the value of the enhancement of the aorta. Values were correlated with cell type and nuclear grade found at surgical pathology. RESULTS. Low tumor-to-aorta enhancement and low tumor-to normal renal parenchyma enhancement ratios on the vascular phase scans significantly correlated (p < 0.001) with papillary renal cell type carcinoma. Homogeneity and tumor-to-parenchyma enhancement ratios on the parenchymal phase scans also significantly correlated (p < 0.001) with papillary renal cell type carcinoma. Heterogeneity and tumor enhancement ratios did not correlate with the nuclear grade of the carcinoma. CONCLUSION. Papillary renal cell carcinomas are typically hypovascular and homogeneous. A high tumor-to-parenchyma enhancement ratio ( 25%) essentially excludes the possibility of a tumor being papillary renal cell carcinoma. A low tumor-to-aorta enhancement ratio or tumor-to normal renal parenchyma enhancement ratio is more likely to indicate papillary renal cell carcinoma. T he most important prognostic factors for patients with renal cell carcinoma are the stage of the disease at diagnosis and nuclear grade [1, 2]. Papillary renal cell carcinomas are associated with a better prognosis than other cell types at a similar stage, but papillary renal cancers are more frequently multiple and bilateral and can be hereditary [3 7]. A papillary renal cell carcinoma may be preferentially treated with nephronsparing surgery because of the propensity of the disease to be bilateral and to be diagnosed at an earlier stage. In addition, because of the possibility of synchronous or metachronous lesions, patients with papillary renal tumors should be followed up more closely after undergoing partial nephrectomy. Therefore, it is clinically important for both surgical planning and prognosis to determine whether a renal neoplasm is potentially a papillary renal cell carcinoma or to effectively exclude that possibility. CT remains the standard modality to use in diagnosing and staging renal neoplasms, and the widespread use of CT for abdominal imaging has led to earlier discovery of many small renal neoplasms [8 12]. In 1984, Press et al. [13] reported that papillary renal cell carcinoma should be suspected in patients with renal lesions that have soft-tissue attenuation, central calcifications, no or poor contrast enhancement, and low clinical stage at the time of disease presentation. Papillary renal cell carcinomas also tend to be hypovascular on CT and angiography [4, 14 16], but other studies have failed to establish a reliable association between morphology and appearance of renal neoplasms on CT and renal carcinoma cell subtype [9, 16 18]. Size is the most reliable factor that correlates with nuclear AJR:178, February

2 Herts et al. grade [19]. Because our institution and many others now routinely perform triphasic helical CT of the kidneys for diagnosis and staging, we sought to determine whether the degree of tumor enhancement or heterogeneity on triphasic helical CT scans of the kidneys could be used to predict papillary cell subtype or nuclear grade. Materials and Methods Patient Population Ninety consecutive patients with renal masses who were referred for CT before undergoing complete or partial nephrectomy were retrospectively evaluated, 89 of whom were included in the final study group. One patient with a mixed cell type carcinoma was excluded because we could not determine from the pathology report whether this carcinoma was predominantly clear cell or papillary cell subtype. All patients had preoperative triphasic helical CT of the kidneys performed at our institution. Enhancement and homogeneity characteristics (that we will later define) were correlated with tumor cell type and nuclear grade found at final surgical pathology. The group studied was composed of 52 men and 37 women whose ages ranged from 29 to 80 years. The mean and median ages of the patients were 60.7 years and 62.3 years, respectively. At final pathology, findings for the group were oncocytoma (n = 9), benign complex renal cyst (n = 2), angiomyolipoma without fat attenuation on CT (n = 1), and renal cell carcinoma (n = 77). The subtypes of renal cell carcinoma subtypes were clear cell carcinoma (n = 52), papillary carcinoma (n = 12), and undifferentiated or unspecified cell type (n = 13). CT Protocol Triphasic renal helical CT was performed on a Somatom Plus 4 scanner (Siemens Medical Systems, Forchheim, Germany) at 120 kvp and revolution time of 0.75 sec. No oral contrast material was administered. First, unenhanced CT of the kidneys was performed using a 5-mm collimation and a pitch of 1 (table speed of 5 mm per revolution) at ma, depending on body habitus of the patient and the discretion of the technologist. The unenhanced CT scans localized the kidneys, allowing us to obtain baseline tumor attenuation measurements and to detect calcifications. Next, a test bolus of 20 ml injected at 4 ml/sec was used to calculate the delay time for the vascular (corticomedullary) phase. Nonionic contrast media (iopromide, Ultravist 300; Berlex, Wayne, NJ)) was used for all scans. A single CT slice was performed in the patient s upper abdominal aorta, and images were obtained every second beginning at 10 sec for period of 40 sec. A region of interest was placed over the aorta, and the time to peak enhancement was calculated. The delay time for the vascular A B C phase was the time to peak aortic enhancement plus an additional 5 sec to allow venous opacification. We waited 2 min after the test injection before beginning the contrast medium injection for the vascular phase to permit filling of the collecting system. The vascular phase scan was then obtained using the delay time (calculated as previously described) after the initiation of the contrast medium injection. One hundred twenty ml of contrast medium was injected at 4 ml/ sec. This scan was obtained using 3-mm collimation and a pitch of (table speed of 3 5 mm per revolution) at 240 ma. Thinner collimation was used for the vascular phase than for the parenchymal and unenhanced phases to improve visualization of small renal vessels. Finally, a nephrographic (or parenchymal) phase scan was obtained using a 140-sec delay time from the initiation of the bolus contrast injection. The parenchymal phase scan was obtained at 5-mm collimation and a pitch of 1.0 (table speed of 5 mm per revolution) using the same techniques as used for the unenhanced scan to allow accurate characterization of all masses. All scans were reconstructed with a 50% image overlap and were reviewed on a picture archiving and communication system workstation (MagicView 1000; Siemens Medical Systems). Determination of Tumor Enhancement and Heterogeneity Three radiologists who were unaware of the patients diagnoses independently and retrospectively Fig year-old-man with papillary renal cell carcinoma (arrow). A C, Unenhanced (A), vascular phase (B), and nephrographic phase (C) CT scans show minimal enhancement. Enhancement was 15 H on both B and C. Tumor-to-aorta enhancement ratio and tumor-to-kidney enhancement ratio are 0.08 and 0.09, respectively, giving positive predictive value for papillary renal cell carcinoma of approximately 50% and specificity of 95%. A B C Fig year-old woman with papillary renal cell carcinoma (arrow). A C, Unenhanced (A), vascular phase (B), and nephrographic phase (C) CT scans. Region of interest circles were drawn on observer-estimated areas of highest and lowest densities on contrast-enhanced scans (B and C) to determine tumor heterogeneity. During vascular phase, tumor enhancement is 22 H compared with aortic enhancement of 232 H, giving a 0.1 heterogeneity ratio. 368 AJR:178, February 2002

3 Helical CT of Papillary Renal Neoplasms measured a region of interest of the highest and lowest attenuation in each renal mass (as perceived by the individual interpreter) on scans obtained during all three phases unenhanced (NC), vascular phase (VP), and parenchymal phase (PP). The interpreters also measured the attenuation of the abdominal aorta and normal renal parenchyma on scans obtained during all phases (Figs. 1 3). The observers were instructed to exclude calcifications, maximize the regions of interest, and keep the sizes and locations of the regions of interest the same for each patient s images from the three scanning phases. The reviewing radiologists received no training session and no other instructions. Definition of Tumor Enhancement Ratios Four different measures of tumor enhancement were calculated, two for each post contrast injection phase: the ratio of the tumor enhancement to the aorta enhancement and the ratio of the tumor enhancement to normal renal parenchyma enhancement. The highest observer-perceived attenuation value in a region of interest was used for the tumor in all three scanning phases. The equations used were as follows: the tumor-to-aorta enhancement ratio during the vascular phase (equation 1) was calculated as [tumor (VP) tumor (NC)] / [aorta (VP) aorta (NC)]; the tumor-toaorta enhancement ratio during the parenchymal phase (equation 2) was calculated as [tumor (PP) tumor (NC)] / [aorta (PP) aorta (NC)]; the tumor-to-kidney enhancement ratio during the vascular phase (equation 3) was calculated as [tumor (VP) tumor (NC)] / [kidney (VP) kidney (NC)]; and the tumor-to-kidney enhancement ratio during parenchymal phase (equation 4) was calculated as [tumor (PP) tumor (NC)] / [kidney (PP) kidney (NC)]. Definition of Tumor Heterogeneity Ratios A B C Fig year-old man with clear cell carcinoma (arrow) of the kidney. A C, Unenhanced (A), vascular phase (B), and nephrographic phase (C) CT scans. Measurements of Hounsfield density show avid enhancement on both vascular and nephrographic phase scans. Enhancement is 131 H on vascular phase and 123 H on nephrographic phase. Tumor-to-aorta enhancement ratio and tumor-to-kidney enhancement ratio are 0.66 and 0.54, respectively. These values have negative predictive value for papillary renal cell carcinoma of 98%, essentially excluding possibility of papillary cell type carcinoma. Tumor homogeneity was calculated as the difference between the highest and lowest attenuation values for the tumor divided by the enhancement of the aorta for each post contrast injection phase. The equations used were as follows: The heterogeneity ratio during vascular phase (equation 5) was calculated as [tumor s highest density (VP) tumor s lowest density (VP)] / [aorta (VP) aorta (NC)] and the heterogeneity ratio during the parenchymal phase (equation 6) was calculated as [tumor s highest density (PP) tumor s lowest density (PP)] / [aorta (PP) aorta (NC)]. Statistical Methods Logistic regression models using generalized estimating equations (SAS; SAS Institute, Cary, NC) were fit so that the relationships between the enhancement and heterogeneity variables and the probability of a papillary lesion could be assessed. Methods for clustered binary data [20] were used to estimate and construct pointwise 95% confidence intervals for sensitivity, specificity, positive predictive value, and negative predictive value for various values for the variables. TABLE 1 Results Correlation of Tumor Enhancement with Papillary Renal Cell Carcinoma a Statistically significant at p < b Ratios used for further statistical evaluation and reported in results shown in Tables 2 4. In general, low enhancement ratios correlated with papillary renal cell carcinoma, indicating that these are hypovascular tumors. Three of the four enhancement ratios had statistically significantly correlations with papillary renal cell carcinoma (Table 1). On the vascular phase scans, a low tumor-to-aorta enhancement ratio and a low tumor-to renal parenchyma enhancement ratio both significantly correlated with papillary renal cell type (p < 0.001). On the nephrographic (parenchymal) phase scans, only a low tumor-to renal parenchyma enhancement ratio significantly correlated with papillary renal cell type (p < 0.001). The tumor-to-aorta ratio on the nephrographic scans did not significantly correlate with the papillary cell subtype of renal cell carcinoma (Table 1). Statistical Evaluation of the Calculated Predictors of Papillary Tumors Calculated Enhancement Ratios Phase Scanning p a Comments 1. Tumor-to-aorta enhancement b Vascular < As ratio increases, likelihood of papillary 2. Tumor-to-aorta enhancement Parenchymal Not significant 3. Tumor-to-kidney enhancement Vascular < As ratio increases, likelihood of papillary 4. Tumor-to-kidney enhancement b Parenchymal < As ratio increases, likelihood of papillary 5. Tumor heterogeneity b Vascular < As ratio increases, likelihood of papillary 6. Tumor heterogeneity Parenchymal Not significant AJR:178, February

4 Herts et al. Sensitivity varied depending upon the cutoff ratio used, but specificity and negative predictive values were high (Tables 2 and 3). As an example, for a tumor-to-aorta ratio (vascular phase) or a tumor-to renal parenchyma ratio (nephrographic phase) of less than 0.25, the sensitivity of triphasic helical CT was 86% and 72%, respectively. At these same ratios, the specificity was 85% and 87%, respectively. TABLE 2 Tumor-to-Aorta Enhancement Ratio TABLE 3 Tumor-to-Kidney Enhancement Ratio TABLE 4 Tumor Heterogeneity Enhancement Ratio The negative predictive value of a tumor-toaorta ratio of less than 0.25 was 98%, and the negative predictive value of a tumor-to renal parenchyma ratio of less than 0.25 was 95%. Sensitivity and Specificity of Calculated Tumor-to-Aorta Enhancement Ratio in the Vascular Phase Sensitivity Specificity Note. Tumor-to-aorta enhancement ratio in vascular phase as defined as equation 1 in the text and in Table 1. Values in boldface represent the suggested cutoff for sensitivity, specificity, and negative predictive value. CI = confidence interval. Sensitivity and Specificity of the Calculated Tumor-to-Kidney Enhancement Ratio in the Parenchymal Phase Sensitivity Specificity Note. Tumor-to-kidney enhancement ratio in the parenchymal phase as defined as equation 4 in the text and in Table 1. Values in boldface represent the suggested cutoff for sensitivity, specificity, and negative predictive value. CI = confidence interval. Sensitivity and False-Positive Rate of the Calculated Tumor Heterogeneity Enhancement Ratio in the Vascular Phase Sensitivity Specificity Positive Positive Positive Predictive Value Negative < ( ) 0.95 ( ) 0.48 ( ) 0.89 ( ) < ( ) 0.92 ( ) 0.54 ( ) 0.94 ( ) < ( ) 0.88 ( ) 0.51 ( ) 0.97 ( ) < ( ) 0.85 ( ) 0.48 ( ) 0.98 ( ) < ( ) 0.83 ( ) 0.45 ( ) 0.98 ( ) < ( ) 0.81 ( ) 0.43 ( ) 0.98 ( ) < ( ) 0.73 ( ) 0.35 ( ) 0.98 ( ) Negative < ( ) 0.95 ( ) 0.25 ( ) 0.89 ( ) < ( ) 0.92 ( ) 0.32 ( ) 0.94 ( ) < ( ) 0.88 ( ) 0.47 ( ) 0.97 ( ) < ( ) 0.85 ( ) 0.46 ( ) 0.97 ( ) < ( ) 0.83 ( ) 0.40 ( ) 0.99 ( ) < ( ) 0.81 ( ) 0.35 ( ) 0.99 ( ) < ( ) 0.73 ( ) 0.34 ( ) 0.99 ( ) Negative Predictive Value < ( ) 0.87 ( ) 0.44 ( ) 0.94 ( ) < ( ) 0.85 ( ) 0.45 ( ) 0.97 ( ) < ( ) 0.78 ( ) 0.38 ( ) 0.97 ( ) < ( ) 0.75 ( ) 0.36 ( ) 0.98 ( ) < ( ) 0.71 ( ) 0.33 ( ) 0.98 ( ) < ( ) 0.65 ( ) 0.29 ( ) 0.98 ( ) < ( ) 0.59 ( ) 0.26 ( ) 0.98 ( ) Note. Tumor heterogeneity enhancement ratio in the vascular phase as defined equation 5 in the text and in Table 1. Values in boldface represent the suggested cutoff for sensitivity, specificity, and negative predictive value. CI = confidence interval. Correlation of Heterogeneity with Papillary Renal Cell Carcinoma In general, low heterogeneity ratios correlated with papillary renal cell carcinoma, indicating that these are homogeneous tumors (Table 1). A low heterogeneity score on the vascular phase scan also significantly correlated with papillary cell type (p < 0.001). Sensitivity and specificity of triphasic helical CT for papillary renal cell carcinoma were 72% and 87%, respectively if the tumor heterogeneity ratio in the vascular phase was less than 0.25 (Table 4). The heterogeneity score did not significantly correlate with papillary renal cell carcinoma on the nephrographic phase (p = 0.185). Nuclear Grade The tumor-to-aorta enhancement ratio in the vascular phase significantly correlated with nuclear grade (Pearson correlation coefficient = 0.195, p = 0.003), but the range of values was small (from 0.53 for grade 1 to 0.42 for grade 4) and not thought to be clinically useful. Tumor heterogeneity was not predictive of nuclear grade (p = 0.27). Discussion Nephron-sparing surgery for renal neoplasms is performed using open or laparoscopic partial nephrectomy, cryoablation, or radiofrequency ablation and is the main treatment option for patients with bilateral tumors, a solitary kidney, or any condition causing renal insufficiency. This surgery has also been shown to be an effective therapeutic option for patients with small, low-stage, low-grade tumors and is the procedure of choice at some institutions [21]. The increased use of CT for diagnostic imaging has led to the identification of many small incidental renal neoplasms [10, 11]. Most of these tumors are small, low stage, and low grade [10 12], and therefore, patients with such tumors are ideal candidates for nephron-sparing surgery. Papillary renal cell carcinomas are typically low stage and low grade at presentation, and may be multifocal or bilateral [3, 4, 6]. In most studies, papillary renal cell carcinoma has been reported to be associated with a better prognosis than nonpapillary renal cell carcinomas, especially when compared with the prognosis for other renal tumors at the same stage [3 5, 7, 22]. Although the stage at presentation and nuclear grade remain the most important factors for the prognosis of renal cell carcinoma [1], it nevertheless is important to be able to identify or effectively exclude a papillary cell subtype preoperatively. Papillary renal cell carcinomas are frequently found at a low stage, making nephron-sparing surgery an appropriate treatment. In addition, the possibility 370 AJR:178, February 2002

5 Helical CT of Papillary Renal Neoplasms of bilateral tumors makes nephron-sparing surgery preferable to nephrectomy, despite the possibility of synchronous or metachronous tumors in the same kidney [13, 23, 24]. Many CT features of renal cell carcinomas have previously been evaluated in an attempt to predict stage, cell type, and nuclear grade. Enhancement, lack of tumor margins, and necrosis all have been found to correlate with increasing tumor size [18]. Increasing size and lack of sharp tumor margins have both been found to correlate with higher nuclear grade [19, 25]. High attenuation in tumors on CT has also been shown to correlate with clear cell carcinomas [26]. Papillary renal cell carcinomas are hypovascular on CT and angiography [4, 14 16]. Although calcifications and hypovascularity are frequently seen CT features of papillary renal tumors, most CT features have not been shown to be reliable predictors of the renal cell carcinoma subtypes [17, 18]. Furthermore, percutaneous needle aspiration is unreliable for determining nuclear grade and is generally inaccurate for distinguishing the papillary subtype of renal cell carcinoma [27, 28] from other subtypes. In this study, we sought to determine whether we could reliably predict papillary renal cell carcinomas on the basis of enhancement characteristics. We sought to identify a numeric value that would be simple to use and have a high positive or negative predictive value. This value should be normalized relative to the contrast material bolus to ensure that it is independent of the variability frequently seen when evaluating the enhancement of small renal lesions [29]. Therefore, we used the enhancement values of the aorta and normal renal parenchyma as divisors to determine a relative enhancement ratio that might be more sensitive for and predictive of papillary renal cell carcinomas. We found that three of the four enhancement ratios significantly correlated with papillary subtype and chose two ratios as the most clinically relevant the enhancement ratio of the tumor to the aorta during the vascular phase and the enhancement ratio of the tumor to the normal renal parenchyma during the parenchymal phase. Tumor enhancement correlated most to the vascular concentration of contrast material on the bolus phase, and the enhancement of the tumor is compared with normal renal parenchyma on the scans obtained during the nephrographic phase when the normal kidney is concentrating the filtered contrast material. Triphasic renal helical CT studies have shown that the scans obtained during nephrographic phase remain the most sensitive for lesion identification and characterization [24, 30, 31]. Our results show that a low ratio of tumor-to-aorta enhancement during the vascular phase and enhancement of the tumor-to normal renal parenchyma during the nephrographic phase have a high sensitivity, specificity, and negative predictive value for papillary renal cell carcinomas. The best use for the results of our study is as a method of excluding the possibility that a lesion is a papillary renal cell carcinoma; the negative predictive value and specificity are high even in instances in which the positive predictive value is less than 50%. This result occurs because the prevalence of disease in this patient population (those with papillary renal cell carcinoma) is small (13%), and the findings of these hypovascular masses overlap with those of complex cysts or poorly enhancing nonpapillary tumors. We found that if the enhancement ratio of the tumor-to aorta or the tumor-to renal parenchyma is more than 0.25, the likelihood that a lesion is a nonpapillary tumor is greater than 95% (negative predictive value). If the enhancement ratio of the tumor-to aorta or the tumor-to renal parenchyma is less than 0.25, the likelihood of the lesion being a papillary tumor is almost 50% (positive predictive value), a percentage more than three times higher than that would be expected from its incidence compared with other types of renal cell carcinomas in this study. We also found that papillary tumors are more often homogeneous than other renal cell carcinomas. This finding is compatible with the small size, low stage, and hypovascular nature of papillary renal cell carcinomas; it is consistent with previous reports [9, 13, 16, 18, 19] and is most likely a byproduct of the lack of vascularity. Our study has several limitations. As Macari and Bosniak [32] correctly note, tumor enhancement is dependent upon the contrast bolus, among other factors, for delivery of the contrast medium to the tumor and to the normal kidney. Although we used ratios rather than absolute enhancement to attempt to correct for differences in each patient s body habitus and cardiac output in the contrast bolus, these results may not necessarily be applicable to lower flow rate injection techniques (i.e., 1 2 ml/sec). Possibly, our results are better than those reported in some previous studies because the combination of injecting the contrast medium at the high-flow rate of 4 ml/sec and more rapid scanning was not available until the introduction of helical CT. Higher levels of enhancement in the aorta and normal renal parenchyma might widen the gap between the papillary tumor and both the normal kidney and vascular tumors, increasing sensitivity, specificity, and negative predictive values as we found. Another limitation of this study is that negative and positive predictive values are dependent on disease prevalence in the population. As a tertiary referral center for renal cell carcinoma and nephron-sparing surgery, we might see a larger share of patients with small, lowgrade, or multiple tumors that could increase the prevalence of papillary tumors in our patient population. Other reports have estimated the prevalence of papillary renal cell carcinomas to be 5 15%. With a prevalence of 14% (excluding the renal cysts and angiomyolipoma), ours is at the high end of what is generally estimated [13]. A final limitation is that we only tested a limited number of enhancement ratios. We could potentially have had different results with different definitions of tumor enhancement that we did not test, such as more complex combinations of factors or different ratios. For example, in our study, the best way to discriminate papillary from nonpapillary lesions was to use a combination of all three variables reported in Tables 2 4. Using the same cutoff values of 0.25 for enhancement and 0.20 for heterogeneity, a papillary lesion was essentially excluded as a possible finding for patients who tested negative on all three ratios (a negative predictive value of 1). However, we sought to keep the calculations as simple as possible so that they could be used simply and easily. We suggest using a tumor-to-aorta enhancement ratio in the vascular phase and a tumor-to-kidney enhancement ratio in the parenchymal phase of 0.25 as a good cutoff value for sensitivity, specificity, and negative predictive value. In general, a tumor-to-aorta ratio or a tumor-to-kidney ratio of less than 0.25 suggests a higher degree of likelihood of papillary renal cell carcinomas, and conversely, a ratio of more than 0.25 generally excludes the possibility of a papillary renal cell carcinoma. We also suggest using a heterogeneity ratio in the vascular phase of 0.20 as a cutoff value with similar implications. In conclusion, papillary renal cell carcinoma is more often hypovascular and homogeneous on CT than other subtypes of renal tumors. The ratios we described can be used to predict an increased likelihood of the presence of papillary renal cell carcinoma or to exclude papillary renal cell carcinoma as a possible di- AJR:178, February

6 Herts et al. agnosis. Such findings should be communicated to the referring urologist, especially if nephron-sparing surgery is an option. Acknowledgments We thank Jennifer Klauer for her assistance with manuscript preparation. References 1. Thrasher JB, Paulson DF. Prognostic factors in renal cancer. Urol Clin North Am 1993;20: Fuhrman SA, Lasky LC, Limas C. Prognostic significance of morphologic parameters in renal cell carcinoma. Am J Surg Pathol 1982;6: Kletscher BA, Qian J, Bostwick DG, Andrews PE, Zincke H. Prospective analysis of multifocality in renal cell carcinoma: influence of histological pattern, grade, number, size, volume and deoxyribonucleic acid ploidy. J Urol 1995;153: Mancilla-Jimenez R, Stanley RJ, Blath RA. Papillary renal cell carcinoma: a clinical, radiologic, and pathologic study of 34 cases. Cancer 1976;38: Mydlo JH, Bard RH. Analysis of papillary renal adenocarcinoma. Urology 1987;30: Zbar B, Glenn G, Lubensky I, et al. Hereditary papillary renal cell carcinoma: clinical studies in 10 families. J Urol 1995;153: Onishi T, Ohishi Y, Goto H, Suzuki M, Miyazawa Y. Papillary renal cell carcinoma: clinicopathological characteristics and evaluation of prognosis in 42 patients. BJU Int 1999;83: Davidson AJ, Hartmen DS, Choyke PL, Wagner BJ. Radiologic assessment of renal masses: implications for patient care. Radiology 1997;202: Silverman SG, Lee BY, Seltzer SE, Bloom DA, Corless CL, Adams DF. Small ( 3 cm) renal masses: correlation of spiral CT features and pathologic findings. AJR 1994;163: Smith SJ, Bosniak MA, Megibow AJ, Hulnick DH, Horii SC, Raghavendra BN. Renal cell carcinoma: earlier discovery and increased detection. Radiology 1989;170: Bosniak MA. The small (< 3.0 cm) renal parenchymal tumor: detection, diagnosis, and controversies. Radiology 1991;179: Curry NS, Bissada NK. Radiologic evaluation of small and indeterminant renal masses. Urol Clin North Am 1997;24: Press GA, McClennan BL, Melson GL, Weyman PJ, Mauro MA, Lee JKT. Papillary renal cell carcinoma: CT and sonographic evaluation. AJR 1984; 143: Blath RA, Mancilla-Jiminez R, Stanley RJ. Clinical comparison between vascular and avascular renal cell carcinoma. J Urol 1976;115: Bard RH, Lord B, Fromowitz F. Papillary adenocarcinoma of kidney: radiographic and biologic characteristics. Urology 1982;16: Yamashita Y, Takahashi M, Watanabe O, et al. Small renal cell carcinoma: pathologic and radiologic correlation. Radiology 1992;184: Wildberger JE, Adam G, Boeckmann W, et al. Computed tomography characterization of renal cell tumors in correlation with histopathology. Invest Radiol 1997;32: Zagoria RJ, Wolfman NT, Karstaedt N, Hinn GC, Dyer RB, Chen YM. CT features of renal cell carcinoma with emphasis on relation to tumor size. Invest Radiol 1990;25: Birnbaum BA, Bosniak MA, Krinsky GA, Cheng D, Waisman J, Ambrosino MM. Renal cell carcinoma: correlation of CT findings with nuclear morphologic grading in 100 tumors. Abdom Imaging 1994;19: Rao JNK, Scott AJ. A simple methods for the analysis of clustered binary data. Biometrics 1992;48: Butler B, Novick AC, Campbell SC, Licht MR. Management of small unilateral renal cell carcinomas: radical versus nephron sparing surgery. Urology 1995;45: Delahunt B, Eble JN. Papillary renal cell carcinoma: a clinicopathologic and immunohistochemical study of 105 tumors. Mod Pathol 1997;10: Whang M, O Toole K, Bixon R, et al. The incidence of multifocal renal cell carcinoma in patients who are candidates for partial nephrectomy. J Urol 1995;154: Novick A. Current surgical approaches, nephron-sparing surgery, and the role of surgery in the integrated immunologic approach to renal-cell carcinoma. Semin Oncol 1995;22: Birnbaum BA, Jacobs JE, Ramchandani P. Multiphasic renal CT: comparison of renal mass enhancement during the corticomedullary and nephrographic phases. Radiology 1996;200: Fujimoto H, Wakao F, Moriyama N, Tobisu K, Sakamoto M, Kakizoe T. Alveolar architecture of clear cell renal cell carcinomas (< 5.0 cm) show high attenuation on dynamic CT scanning. Jpn J Clin Oncol 1999;29: Campbell SC, Novick AC, Herts BR, et al. Prospective evaluation of fine needle aspiration of small, solid renal masses: accuracy and morbidity. Urology 1997;50: Renshaw AA, Lee KR, Madge R. Accuracy of fine needle aspiration in distinguishing subtypes of renal cell carcinoma. Acta Cytol 1997;41: Siegel CL, Fisher AJ, Bennett HF. Interobserver variability in determining enhancement of renal masses on helical CT. AJR 1999;172: Cohan RH, Sherman LS, Korobkin M, Bass JC, Francis IR. Renal masses: assessment of corticomedullary-phase and nephrographic-phase CT scans. Radiology 1995;196: Szolar DH, Kammerhuber F, Altziebler S, et al. Multiphasic helical CT of the kidney: increased conspicuity for detection and characterization of small (< 3 cm) renal masses. Radiology 1997;202: Macari M, Bosniak MA. Delayed CT to evaluate renal masses incidentally discovered at contrastenhanced CT: demonstration of vascularity with deenhancement. Radiology 1999;213: AJR:178, February 2002