Renal Cell Carcinoma: Attenuation Values on Unenhanced CT

Transcription

1 Genitourinary Imaging Original Research Pooler et al. Attenuation Values of Unenhanced CT of Renal Cell Carcinoma Genitourinary Imaging Original Research B. Dustin Pooler 1 Perry J. Pickhardt 1 Stacy D. O Connor 1 Richard J. Bruce 1 Sutchin R. Patel 2 Stephen Y. Nakada 2 Pooler BD, Pickhardt PJ, O Connor SD, Bruce RJ, Patel SR, Nakada SY Keywords: attenuation, CT, renal cell carcinoma, renal mass, unenhanced CT DOI: /AJR Received July 14, 2011; accepted after revision October 4, Department of Radiology, University of Wisconsin School of Medicine and Public Health, E3/311 Clinical Science Center, 600 Highland Ave, Madison, WI Address correspondence to P. J. Pickhardt (ppickhardt2@uwhealth.org). 2 Department of Urology, University of Wisconsin School of Medicine and Public Health, Madison, WI. AJR 2012; 198: X/12/ American Roentgen Ray Society Renal Cell Carcinoma: Attenuation Values on Unenhanced CT OBJECTIVE. The purpose of this study is to analyze the attenuation values of pathologically proven renal cell carcinomas (RCCs) on unenhanced CT and to determine the range of values wherein malignancy should be considered. MATERIALS AND Methods. A retrospective review was performed of 189 consecutive patients with 193 pathologically proven RCCs 1 cm or larger on unenhanced CT. For each RCC, attenuation values were assessed throughout the tumor by continuous sampling with a mm 2 region of interest (ROI), avoiding foci of calcification and peritumoral volume averaging. The lowest and highest ROI attenuation values per lesion were recorded. Each tumor was categorized as either homogeneous or heterogeneous on the basis of visual inspection with soft-tissue window settings. RESULTS. The 193 malignant tumors ranged in size from 1.1 to 20.1 cm (mean [± SD], 5.1 ± 3.4 cm). Eighteen RCCs (9.3%) were homogeneous in appearance on unenhanced CT. The minimum and maximum ROI attenuation values obtained by sampling throughout each tumor were 27.5 ± 10.4 HU (range, 4 67 HU) and 39.7 ± 10.6 HU (range, HU), respectively. Regional areas of minimum attenuation less than 20 HU and maximum attenuation greater than 70 HU were seen in 24.9% (48/193) and 2.1% (4/193) of RCCs, respectively. However, all 193 RCCs (100%) were predominantly composed of noncalcific regions within HU; 72.5% (140/193) fell entirely within this HU danger zone, including all 18 homogeneous lesions. CONCLUSION. All proven RCCs in this series contained substantial noncalcified regions that measured HU in ROI attenuation on unenhanced CT. Indeterminate renal lesions on unenhanced CT measuring within this HU danger zone warrant further workup, whereas lesions that fall entirely outside this range may be considered benign. T he incidence of renal cell carcinoma (RCC) has increased in the United States over the last few decades [1]. Renal lesions are common incidental findings on CT [2], and, consequently, radiologists are frequently faced with the question of how to manage these incidentalomas [3]. This can be even more vexing for lesions found on unenhanced CT studies, because the internal structure of the lesion can be more difficult to assess without the benefit of IV contrast agent [4]. This issue is particularly relevant because unenhanced CT is being increasingly used for a variety of indications, including diagnosis and monitoring of kidney stones [5, 6] and CT colonography [7]. Previous studies have suggested that an average unenhanced attenuation of greater than 70 HU or less than 20 HU in a homogeneous lesion generally indicates benignity and no further workup is required [8 10]. However, the corollary of how to manage heterogeneous renal lesions or those that extend into the HU range is not as well established. Characteristics suggestive of a benign renal lesion on unenhanced CT have been well studied [8, 11]. However, most studies have focused primarily on benign cystic lesions, with relatively few centered on pathologically proven RCC. Furthermore, the sample sizes of proven RCCs have been relatively small [9, 12]. In studies where larger series of pathologically proven RCCs were examined, the focus has largely been on contrast enhancement and imaging features [13]. The purpose of this study is to specifically analyze the unenhanced CT appearance of a large series of pathologically proven RCCs to determine the range of attenuation values AJR:198, May

2 Pooler et al. TABLE 1: Patient and Lesion Characteristics Characteristic Value Patients (n = 189) Sex Male 123 (65.1) Female 66 (34.9) Age (y), mean ± SD (range) 59.6 ± 12.9 (14 83) Tumors (n = 193) Location Right side 100 (51.8) Left side 92 (47.7) Transplant 1 (0.5) Histologic subtype Clear cell 155 (80.3) Papillary 24 (12.4) Chromophobe 10 (5.2) Sarcomatoid 3 (1.6) Medullary 1 (0.5) Pathologic grade Grade I 44 (22.8) Grade II 76 (39.4) Grade III 26 (13.5) Grade IV 8 (4.1) Not reported 39 (20.2) Note Except where noted otherwise, data are no. (%). wherein malignancy should be reasonably suspected and further workup performed. Materials and Methods This HIPAA-compliant study was performed under an institutional review board approved protocol. The requirement for informed consent was waived because of the retrospective nature of the study. Review of the electronic medical record at our institution from January 1, 2000, through December 31, 2009, identified a total of 556 consecutive patients with a pathologically proven RCC 1 cm or larger. Previous research has shown that incidental renal lesions 1 cm or smaller are clinically unimportant and are generally too small to adequately characterize [14]. Of these 556 cases, 308 were excluded because of the lack of a unenhanced CT series on file, and an additional 59 patients were excluded because of unenhanced studies that did not completely image the tumor or were substantially obscured by artifact. The study cohort was formed by the remaining 189 patients (mean [± SD] age, 59.6 ± 12.9 years; range, years; 123 male and 66 female patients with 193 total proven RCCs) who had a preintervention unenhanced CT scan that completely imaged the tumor without significant artifacts on file within the PACS. The histologic subtype and grade of each tumor was noted from the pathology report in the electronic medical record. Table 1 details the demographic and tumor specifics of this cohort. Clear cell histopathology was the predominant tumor subtype, representing 80.3% of cases. Nearly 40% of the reported histologic tumor grading was classified as grade II. All patients were imaged using a 120-kVp technique at CT allowing meaningful interpatient comparison of attenuation numbers, which are measured in Hounsfield units (HU). All CT scanners were calibrated for attenuation values on a daily basis to ensure measurement consistency and accuracy. The tube current level for the unenhanced series was dependent on the specific protocol and patient body habitus, which affects the SD (noise) of the attenuation measurement but not the mean HU within a measured region of interest (ROI). The unenhanced series were typically reconstructed with 5-mm slice thickness at 3-mm intervals. All CT images were reviewed on standard PACS. After correlation with pathologic and surgical reports to ensure that the correct lesion had been identified, the measurable range of mean attenuation values (in HU) within an ROI (referred to hereafter as ROI attenuation ) was obtained for each lesion by continuously sampling throughout the tumor parenchyma in multiple axial slices with a circular ROI mm 2 in cross-sectional area, appropriately scaled to the size of the lesion. Care was taken to exclude areas of focal calcification and to avoid the extreme periphery of the tumor to minimize volume-averaging effects from surrounding normal tissues. The lowest and highest ROI attenuation values obtained for each lesion were recorded. The ROI attenuation range was calculated by determining the difference between the highest and lowest recorded ROI attenuation values. The maximum linear tumor diameter was measured with electronic calipers in the axial plane. Each tumor was also categorized in a binary fashion as either homogeneous or heterogeneous according to subjective visual inspection; soft-tissue window settings (400 HU window width, centered at 50 HU) were used in all cases to maintain consistency, as well as to best simulate the clinical setting in which these lesions would be encountered. Criteria for defining lesions as heterogeneous included visual identification of any complex internal structures (i.e., septations, mural nodules, calcifications, and visible attenuation differences), whereas lesions identified as homogeneous appeared uniform in composition throughout on visual assessment with soft-tissue windowing. Statistical analysis was performed using publicly available online calculators. p values for continuously measured variables were determined using the Student t test ( ttest1.cfm), p values for categoric variables were determined using the Fisher exact test for contingency tables ( html), and where multiple variables were compared, p values were determined using analysis of variance ( Calculation of correlation coefficients and respective p values were performed using Microsoft Excel. Results The 193 pathology-proven RCCs ranged in maximum transverse cross-sectional diameter from 1.1 to 20.1 cm (mean, 5.1 ± 3.4 cm). On unenhanced CT, 90.7% (175/193) were categorized as heterogeneous, and 9.4% (18/193) were categorized as homogeneous on visual inspection. Internal foci of calcification were noted in 15.5% (30/193) of tumors, all of which were categorized as heterogeneous. The average maximum unenhanced ROI attenuation of all lesions was 39.7 ± 10.6 HU (range, HU), and the average minimum ROI attenuation was 27.5 ± 10.4 HU (range, 4 67 HU). The average attenuation span per lesion (i.e., maximum ROI minus minimum ROI) was 13.0 ± 7.3 HU for heterogeneous tumors and 4.6 ± 2.3 HU for homogeneous tumors (p < 0.001). Minimum ROI attenuation was less than 20 HU in 24.9% (48/193) of tumors, and maximum ROI attenuation was greater than 70 HU in 2.1% (4/193) of tumors. However, all 193 RCCs had ROI attenuation values that crossed into the attenuation zone of HU. Of all tumors, 72.5% (140/193) fell entirely within this HU zone, including all 18 homogeneous lesions. These findings are summarized in Figure 1. Examples of homogeneous and heterogeneous tumors are shown in Figure 2 and Figure 3, respectively. When comparing heterogeneous (n = 175) and homogeneous (n = 18) tumors using the Student t test, there were no statistically significant differences between the two groups in terms of patient age at time of CT (59.6 vs 59.4 years; p = ), average maximum ROI attenuation (40.1 vs 35.7 HU; p = ), or average minimum ROI attenuation (27.1 vs 31.1 HU; p = ). Statistically significant differences were found in average cross-sectional diameter (5.4 vs 2.3 cm; p < ) and average ROI attenuation span (13.0 vs 4.6 HU; p < ). The comparison between heterogeneous and homogeneous tumors is summarized in Table AJR:198, May 2012

3 Attenuation Values of Unenhanced CT of Renal Cell Carcinoma CT Attenuation (HU) Lesions (n = 193) Heterogeneous Homogeneous Fig. 1 Range of unenhanced CT region of interest (ROI) attenuation values for renal cell carcinoma (RCC). Plot showing ROI attenuation span for each of 193 pathologyproven RCCs shows that all tumors have regions that cross into HU danger zone. Most lesions fall entirely within this zone, including all 18 homogeneousappearing tumors (blue diamonds). A C Fig year-old man with renal cell carcinoma (RCC) with homogeneous appearance on unenhanced CT. A C, Unenhanced (A), nephrographic (B), and excretory (C) phase CT images show 2.7-cm homogeneous left renal lesion that measures HU on unenhanced region of interest attenuation measurement. This mildly borderdeforming lesion is most conspicuous on excretory phase image (C). Note also benign cyst extending off lower pole of right kidney, which is less than 20 HU and does not enhance after administration of contrast agent. Indeterminate left renal lesion proved to be chromophobe RCC. B AJR:198, May

4 Pooler et al. Fig year-old man with renal cell carcinoma with heterogeneous appearance on unenhanced CT. A and B, Unenhanced CT images show 8-cm heterogeneous right renal mass that ranges from 18 to 41 HU in region of interest (ROI) attenuation. C and D, Contrast-enhanced CT images at same levels show increased heterogeneity, with large area of low attenuation that corresponds to ROI attenuation measurement under 20 HU. Regardless, majority of tumor crosses into HU danger zone on unenhanced CT. Comparison of histologic subtypes (excluding the single case of medullary RCC) revealed no statistically significant difference in maximum linear cross-sectional diameter (p = 0.37) or average ROI attenuation span (p = 0.68) among clear cell, papillary, chromophobe, and sarcomatoid tumors. Comparison of the histologic subtypes using the two-tailed Fisher exact test revealed no statistically significant differences in the proportion of heterogeneous versus homogeneous lesions among clear cell, papillary, chromophobe, and sarcomatoid tumors (p = 0.64). Additionally, among all lesions there was no statistically significant correlation noted between maximum linear diameter and maximum ROI attenuation (R 2 = 0.062; p = 0.39) or minimum ROI attenuation (R 2 = 0.021; p = 0.77). Comparison of pathologic grades (again excluding the single case of medullary RCC) A C revealed statistically significant differences in maximum linear cross-sectional diameter (grade I = 4.2 cm, grade II = 4.9 cm, grade III = 7.3 cm, and grade IV = 9.9 cm; p < 0.001) and average ROI attenuation span (grade I = 11.4 HU, grade II = 12.2 HU, grade III = 14.9 HU, and grade IV = 19.6 HU; p = 0.02). Additional comparison of pathologic grades revealed no statistically significant difference in the proportion of heterogeneous versus homogeneous lesions (p = 0.57). Discussion Irrespective of heterogeneous or homogeneous visual appearance on unenhanced CT, all proven RCCs in this large series contained noncalcified areas within the tumor that clearly measured HU in ROI attenuation on unenhanced CT, with nearly three-quarters falling entirely within this zone. Therefore, we think that a renal mass with ROI attenuation values that cross into this HU danger zone on unenhanced CT should raise suspicion for possible malignancy and further workup should be considered. To be clear, although 25% of tumors had focal regions less than 20 HU and 2% of tumors had areas greater than 70 HU, all 193 cancers had substantial noncalcified regions that measured HU. As a corollary, none of the pathologically proven RCCs in this large series were noted to have attenuation values that fell completely outside of the HU danger zone, implying that incidental lesions found on unenhanced CT with internal ROI attenuation values entirely outside this zone may confidently be called benign and the patient spared the cost and anxiety of further workup. Specifically, homogeneous lesions consistently measur- B D 1118 AJR:198, May 2012

5 Attenuation Values of Unenhanced CT of Renal Cell Carcinoma TABLE 2: Comparison of Heterogeneous and Homogeneous Tumors Characteristic Heterogeneous (n = 175) Homogeneous (n = 18) p Patient characteristics Sex Male, no. (%) 115 (65.7) 12 (66.6) Female, no. (%) 60 (34.3) 6 (33.3) Age (y) 59.6 ± ± ROI attenuation (HU) Maximum 40.1 ± ± Minimum 27.1 ± ± ROI attenuation span per lesion 13.0 ± ± 2.3 < Lesion diameter (cm) 5.4 ± ± 0.9 < Note Except where noted otherwise, data are mean ± SD. ROI = region of interest. ing less than 20 HU should represent benign cysts (or angiomyolipomas if > 0 HU because of macroscopic fat) and homogeneous lesions measuring greater than 70 HU should represent high-attenuation benign cysts. A recent complementary study evaluating more than 400 patients with incidental renal lesions found on unenhanced CT for screening colonography showed that all lesions outside the HU range were benign, whereas all four proven RCCs were within this danger zone [15]. Not surprisingly, a number of benign lesions also entered the danger zone, with a corresponding positive predictive value for RCC of 10% in this study of asymptomatic adults. Although the HU range is clearly not specific for RCC, it appears that ROI attenuation values entirely outside this range are specific for benignity. Our data add to this previous study by confirming that the HU range is a very sensitive indicator for RCC detection. Together, these two articles provide a useful framework for clinical management. There is surprisingly little information in the existing literature regarding the unenhanced attenuation of pathology-proven RCC. Until recently, this issue had not been of major clinical importance, because most patients undergoing CT evaluation of the abdomen and pelvis generally received IV contrast agents for a wide variety of indications. With the increasing use of unenhanced CT for indications, including urolithiasis [5, 6] and colorectal evaluation [7], as well as in the setting of renal insufficiency or concern for contrast agent reactions, the issue of which renal lesions warrant further workup has become highly relevant in clinical practice. Our results build on the prior evidence, although most data regarding unenhanced attenuation of RCC are buried in studies that focused on contrast enhancement. One small series reported a mean attenuation of 20 HU or higher in 26 of 27 (96.3%) pathology-proven RCCs on unenhanced CT, with a mean attenuation of 18 HU in the remaining lesion [12]. Another small series reported a mean attenuation on unenhanced CT of 31.9 ± 7.0 HU for clear cell RCC (n = 29), 33.5 ± 5.6 HU for papillary RCC (n = 5), and 40.3 ± 3.0 HU for chromophobe RCC (n = 2) [16]. A somewhat larger series focusing on enhancement characteristics of RCC reported mean unenhanced attenuation values of 35 HU (range, HU) for clear cell RCCs (n = 108), 39.5 HU (range, HU) for papillary RCCs (n = 30), and 39 HU (range, HU) for chromophobe RCCs (n = 24) [13]. Another study focusing on enhancement characteristics indicated that the mean unenhanced attenuation of 56 pathologically proven RCCs was HU, depending on the Bosniak class [17]. Finally, a recent study focusing on differentiation of hyperdense renal cysts from RCCs on unenhanced CT reported an average attenuation of HU (range, HU) for 54 RCCs scored by two different readers [9]. In addition to featuring a larger series, our study expands on these previous studies by focusing particularly on the attenuation span of individual lesions and suggests a management strategy for lesions of any density discovered incidentally on unenhanced CT. One potential limitation of these previous studies in characterizing renal lesions on unenhanced CT lies in the use of a single ROI measurement to assess attenuation [9, 12, 13, 16, 17]. Because most of the pathology-proven RCCs in this study were heterogeneous in composition, we think that our method of continuously sampling throughout the tumor parenchyma to determine an attenuation range offers a better characterization of these lesions. Additionally, in these previous studies, the region of tumor sampled on unenhanced CT was usually analogous to the area that appeared to most vigorously enhance on contrast-enhanced CT. Because a large number of our studies had no corresponding contrastenhanced image (e.g., CT colonography and nephrolithiasis protocol), this sampling strategy was not possible. As the volume of these unenhanced-only studies continues to increase in clinical practice, an evidence-based system for triaging incidentally discovered lesions without the benefit of IV contrast agent is becoming increasingly crucial. Malignancy will undoubtedly be a concern for many incidental renal lesions that are found to be heterogeneous in appearance on unenhanced CT. However, incidental lesions that appear homogeneous can present a greater diagnostic challenge, because most incidentalomas found on CT will be benign cysts, leaving the question of when to suspect malignancy in a homogeneous lesion. Prior studies have suggested that homogeneous lesions with attenuation values less than 20 HU and greater than 70 HU can be considered benign and no further workup is required [8 10, 15]. The data from this series are supportive of these conclusions because all 18 of the homogeneous pathologically proven RCCs presented here had attenuation values that fell entirely within the HU danger zone. In the present study, homogeneous RCCs occurred over the same age and sex distributions as their heterogeneous counterparts. They also possessed attenuation values over virtually the same spectrum as their heterogeneous counterparts, with similar values for average maximum and minimum ROI attenuation (Table 2). As expected, homogeneous lesions were found to have a narrower range of attenuation values, which correlates well with their subjective uniform appearance on CT. They were also found to be significantly smaller in diameter than the heterogeneous lesions, which would be logically expected, because complex internal structures would be more difficult to identify in smaller lesions, and larger tumor size likely facilitates heterogeneity. There was no significant difference in ROI attenuation range among the various histologic subtypes, illustrating the difficulty of attempting to predict histologic subtype on the basis of unenhanced CT findings. With AJR:198, May

6 Pooler et al. the addition of IV contrast agent, the degree of tumor enhancement can predict histologic subtype to some degree, with papillary tumors having a lesser degree of enhancement [18]. In contrast, there were significant differences seen among pathologic grades in both average maximum linear diameter and average midrange ROI attenuation. Overall, higher grade tumors were found to be larger and, on average, denser than their lower-grade counterparts, while also having a larger average ROI attenuation span. These results support earlier research showing that higher grade RCCs, on average, grow faster than lower grade RCCs and, consequently, are generally larger at presentation [19]. The limitations of this study are mostly related to its single-center retrospective nature, raising possible issues with regard to generalization of results and selection bias. The daily quality assurance testing of our CT scanners ensures more consistent ROI attenuation values. In addition, all scans were performed using a 120-kVp technique, which further improves reproducibility. Scanning at different energy levels will affect CT numbers to some degree, which may be an issue with the dual-energy technique. One caveat involves the direct application of our ROI attenuation data to other CT scanner manufacturers, because vendor-specific differences are known to exist [20, 21]. Ideally, future research would prospectively identify and follow a large number of incidental renal lesions found on unenhanced CT and triaged according to the HU danger zone to better assess patient outcomes and determine the performance statistics of this method. In summary, our results show that all proven RCCs in this series contained substantial noncalcified regions that measured HU in ROI attenuation on unenhanced CT. Given that renal lesions completely outside this range have proven to be benign in previous work, we conclude that indeterminate renal lesions on unenhanced CT that contain areas of ROI attenuation that cross over into this HU danger zone will generally warrant further workup. References 1. Chow WH, Devesa SS, Warren JL, Fraumeni JF. Rising incidence of renal cell cancer in the United States. JAMA 1999; 281: Barrett TW, Schierling M, Zhou C, et al. Prevalence of incidental findings in trauma patients detected by computed tomography imaging. Am J Emerg Med 2009; 27: Berland LL, Silverman SG, Gore RM, et al. Managing incidental findings on abdominal CT: white paper of the ACR incidental findings committee. J Am Coll Radiol 2010; 7: Curry NS, Cochran ST, Bissada NK. Cystic renal masses: accurate Bosniak classification requires adequate renal CT. AJR 2000; 175: Bariol SV, Tolley DA. What is the best imaging for stone management? BJU Int 2005; 95: Catalano O, Nunziata A, Sandomenico F, Siani A. Acute flank pain: comparison of unenhanced helical CT and ultrasonography in detecting causes other than ureterolithiasis. Emerg Radiol 2002; 9: Pickhardt PJ, Hanson ME, Vanness DJ, et al. Unsuspected extracolonic findings at screening CT colonography: clinical and economic impact. Radiology 2008; 249: Israel GM, Bosniak MA. An update of the Bosniak renal cyst classification system. Urology 2005; 66: Jonisch AI, Rubinowitz AN, Mutalik PG, Israel GM. Can high-attenuation renal cysts be differentiated from renal cell carcinoma at unenhanced CT? Radiology 2007; 243: Silverman SG, Israel GM, Herts BR, Richie JP. Management of the incidental renal mass. Radiology 2008; 249: Bosniak MA, Megibow AJ, Hulnick DH, Horii S, Raghavendra BN. CT diagnosis of renal angiomyolipoma: the importance of detecting small amounts of fat. AJR 1988; 151: Silverman SG, Lee BY, Seltzer SE, Bloom DA, Corless CL, Adams DF. Small ( 3 cm) renal masses: correlation of spinal CT features and pathological findings. AJR 1994; 163: Zhang J, Lefkowitz RA, Ishill NM, et al. Solid renal cortical tumors: differentiation with CT. Radiology 2007; 244: Willatt J, Francis IR. Imaging and management of the incidentally discovered renal mass. Cancer Imaging 2009;9(Spec No A):S30 S O Connor SD, Pickhardt PJ, Kim DH, Oliva MR, Silverman SG. Incidental renal masses at unenhanced CT: prevalence and analysis of features for guiding management. AJR 2011; 197: Jinzaki M, Tanimoto A, Mukai M, et al. Doublephase helical CT of small renal parenchymal neoplasms: correlation with pathologic findings and tumor angiogenesis. J Comput Assist Tomogr 2000; 24: Song C, Min GE, Song K, et al. Differential diagnosis of complex cystic renal mass using multiphase computerized tomography. J Urol 2009; 181: Herts BR, Coll DM, Novick AC, et al. Enhancement characteristics of papillary renal neoplasms revealed on triphasic helical CT of the kidneys. AJR 2002; 178: Bosniak MA, Birnbaum BA, Krinsky GA, Waisman J. Small renal parenchymal neoplasms: further observations on growth. Radiology 1995; 197: Sande EPS, Martinsen ACT, Hole EO, Olerud HM. Interphantom and interscanner variations for Hounsfield units: establishment of reference values for HU in a commercial QA phantom. Phys Med Biol 2010; 55: Birnbaum BA, Hindman N, Lee J, Babb JS. Multidetector row CT attenuation measurements: assessment of intra- and interscanner variability with an anthropomorphic body CT phantom. Radiology 2007; 242: AJR:198, May 2012