Role of computed tomography-calculated intraparenchymal tumor volume in assessment of patients undergoing partial nephrectomy

Transcription

1 International Journal of Urology (2018) 25, doi: /iju Original Article: Clinical Investigation Role of computed tomography-calculated intraparenchymal tumor volume in assessment of patients undergoing partial nephrectomy Raj Vikesh Tiwari, 1 Chia Ming Ho, 2 Hong Hong Huang, 1 Henry Sun Sien Ho 1 and Allen Soon Phang Sim 1 1 Department of Urology, and 2 Department of Diagnostic Radiology, Singapore General Hospital, Singapore Abbreviations & Acronyms CT = computed tomography IPV = intraparenchymal tumor volume PN = partial nephrectomy ROC = receiver operating curve Correspondence: Raj Vikesh Tiwari M.B.B.S., M.R.C.S. (Ed), Department of Urology, Singapore General Hospital, 20 College Road, Academia, Level 5, Singapore rajvikeshtiwari.pkt@ mohh.com.sg Received 28 August 2017; accepted 12 January Online publication 8 February 2018 Objectives: To determine the complexity of renal masses by using an objective novel imaging parameter (intraparenchymal tumor volume) based on computed tomography scans, to correlate this parameter to perioperative outcomes and to the RENAL nephrometry score. Methods: After institutional review board approval, 87 patients who underwent partial nephrectomy between 2012 and 2016 at Singapore General Hospital, Singapore, were retrospectively analyzed. Preoperative computed tomography intravenous pyelogram scans were reviewed by a single senior radiologist who calculated the intraparenchymal tumor volume. Once the intraparenchymal tumor volume scores were obtained, they were compared with perioperative renal and surgical outcomes, and nephrometry scores. Furthermore, intraparenchymal tumor volume was subdivided into two categories, low and high intraparenchymal tumor volume, both using the 89th percentile. Results: The mean patient age was 60 years, and the mean tumor size was 2.9 cm. The mean nephrometry score was 7.8, and the mean intraparenchymal tumor volume score was 12.7 cm³. The cut-off for high intraparenchymal tumor volume was >27.26 cm³. Asa continuous variable, intraparenchymal tumor volume showed a significant relationship with the percentage of creatinine change (P = 0.009) and nephrometry scores (P < 0.001). As a categorical variable, high intraparenchymal tumor volume showed significance when compared with absolute creatinine change (P = 0.018), percentage of creatinine change (P = 0.004) and nephrometry score (P < 0.001). Conclusions: Intraparenchymal tumor volume represents a novel objective tool based on computed tomography imaging to determine the complexity of a renal mass. This tool correlates with renal functional outcomes of partial nephrectomy, and it also shows good correlation with RENAL nephrometry score. Key words: intraparenchymal volume, nephrometry, partial nephrectomy, renal outcomes, surgical outcomes. Introduction Nephron-sparing surgery is now the standard of care for most small renal masses, with oncological equivalency comparable with radical nephrectomy. 1 Various nephrometry scoring systems, such as RENAL, PADUA and contact surface area, are in use to objectively quantify tumor anatomy and complexity. 2 4 In addition to facilitating standardized academic reporting, nephrometry scores can also predict perioperative PN outcomes, such as ischemic time, blood loss and complications. 5 Several previous studies have alluded to the impact of parenchymal volume loss as the strongest determinant of renal functional decline after PN. 6 Some volumetric CT studies, in which parenchymal volume loss was directly measured after PN, also support this perspective. 7 In the present study, we developed a novel concept of IPV to capture a radiologically measurable parameter for objective assessment of tumor complexity before PN. IPV combines the relevant complexities of two anatomical parameters the size and degree of intraparenchymal extension of a tumor into a single radiologically measurable parameter. This we believe will provide a more objective and easily derived preoperative analysis of tumor complexity and thus perioperative outcomes when compared with current nephrometry scores The Japanese Urological Association

2 Role of intraparenchymal tumor volume Our primary aim was to devise an objective novel imaging parameter (IPV) based on CT-calculated volume, and correlate it with perioperative renal and surgical outcomes. Our secondary aim was to assess how it compared with RENAL nephrometry scoring systems. The method of calculating IPV is using image rendering software as described. Methods Patient population and statistical methods On approval of the institutional review board, we evaluated the scans of 87 patients who underwent PN between 2012 and 2016 at Singapore General Hospital, Singapore. All had preoperative CT intravenous pyelogram scans carried out, which were used for calculation of IPV by a single senior radiologist. Perioperative and follow-up data were obtained from a prospectively maintained renal cell carcinoma database and analyzed retrospectively. The techniques used at our institution for PN have been described previously. Approaches might include transperitoneal or retroperitoneal from a wide array of methods including robot-assisted laparoscopic, pure laparoscopic or open surgery. Cold, warm or zero ischemia was applied according to surgeon preference. Demographic and perioperative data were obtained, including age, sex, tumor characteristics, surgery date, procedure type (robotic, laparoscopic, open), operative time, margin positivity, ischemic time (warm, cold or zero), postoperative length of stay, complications (measured using Clavien Dindo grading) and RENAL nephrometry scores. Renal functional scoring was also obtained by recording preoperative serum creatinine levels within 2 months of surgery, and postoperative serum creatinine levels within 1 month of surgery. CT scans used for volumetric analysis were carried out within 2 months before the surgery. Univariate analysis of IPV was carried out both as a continuous and categorical variable. As a continuous variable, IPV was studied against the various outcome measures (absolute creatinine change, percentage creatinine change, ischemic time, length of stay, complications, margin positivity and RENAL nephrometry scores) using simple linear regression. As a categorical variable, IPV was dichotomized into high IPV (>27.26 cm 3 ) and low IPV (<27.26 cm 3 ). This value was chosen after various IPV percentiles were tested against outcomes, and this value was found to have the highest ROC value and sensitivity/specificity to predict changes in outcome measures. High IPV scores were studied for significance with outcome measures using Mann Whitney U-tests. Multivariate analysis was carried out using logistic regression of significant outcomes on univariate analysis. All statistics were carried out with SPSS software (IBM SPSS Statistics for Windows, Version 22.0, Armonk, NY, USA). All statistics were considered significant at level of P < Volumetric analysis A total of 87 renal tumors in the selection criteria were shown on CT with pre-contrast, nephrographic and pyelographic phases, using an automated power injector. These CT scans were carried out on Siemens Somatom Definition Flash 128, Siemens Somatom Definition 64 slice, Siemens Somatom Sensation 16 slices (Siemens AG, Munich, Germany) or Philips Brilliance ict 256 CT scanners (Royal Philips, Amsterdam, The Netherlands). Fine slices were loaded into Toshiba Vital Vitrea Core software version (Toshiba Medical Systems, Tochigi, Japan) for manual contouring and volumetric measurements. These were reviewed with a three-dimensional multiplanar reformatting layout on the workstation. Window width and level was first adjusted on loading to delineate the border between normal parenchyma and the tumor. These were correlated on the different phases, with the tumor contour best seen and determined in the nephrographic and pyelographic phases (Figs 1,2). The contour was then traced manually using a sculpting tool for border tagging and interpolation, thereby deriving the total volume of the tumor enclosed by the lines of the region of interest. A curved resection line was drawn along the boundary of the kidney to separate and obtain intraparenchymal and extraparenchymal volume, respectively (Figs 3 5). These data were then correlated to the surgical outcome. Results Demographic, perioperative renal, surgical and volumetric outcome data The baseline clinical features among all 87 patients are shown in Table 1. The present study cohort is commensurate with conventional PN populations. The mean age was 60 years (range years). As expected, the male : female ratio was approximately 2:1. The mean total tumor size was 2.9 cm (range cm). The complexity of the cases was intermediate in most cases, with a mean RENAL nephrometry score of 7.8. Overall, 43% of PN cases were open, with the remainder being minimally invasive surgery. A total of 21 patients had zero ischemia PN, one had cold ischemia PN (25 min) and 65 had warm ischemia PN. The ischemic method used was clamping only the renal artery. The mean warm ischemic time was 23 min (range 6 53 min). A total of 13.6% of patients had margin positivity on histological analysis. The mean postoperative length of stay was 5 days (range 2 30 days). Clavien Dindo complications of grade 2 were seen in 29.8% of patients. A total of 7% of patients had either conversion from minimally invasive surgery to open surgery or repeat operation required within 30 days of initial PN. In terms of renal outcomes, the mean preoperative creatinine was 80.1 lmol/l and mean postoperative creatinine was 93.8 lmol/l. Percentage creatinine change is measured by: postoperative creatinine preoperative creatinine / preoperative creatinine. The mean creatinine change was 17.1%. As described previously, intraparenchymal tumor volume was measured and the mean total tumor volume was measured at 31.6 cm 3. The mean IPV was measured at 12.7 cm 3. Univariate and multivariate analysis for IPV as a continuous variable Table 2 details the results of univariate and multivariate analysis of IPV and its relationship to outcome measures 2018 The Japanese Urological Association 437

3 RV TIWARI ET AL. Fig. 1 The renal tumor is identified with several dots marked around the circumference of the lesion on contiguous slices. Fig. 4 The contour of the renal tumor and its intersection line is also examined on the coronal plane to ensure accurate measurement. Adjustment can be made on selected slices to correct any malaligned contour and resultant step deformities. Region Volume (ml) Percent (%) Total % Region Total Extra Intra Volume (ml) Percent (%) % % % Fig. 2 The total volume of the renal tumor is derived on the software. Fig. 5 Volumetry measurements of the intrarenal and extrarenal components of the renal mass can be expressed as a percentage of the total renal tumor volume. percentage change in creatinine (P = 0.009) and RENAL nephrometry score (P < 0.001). For every 1-cm 3 increase in IPV, there was a increase in RENAL nephrometry score. IPV did not reach statistical significance for absolute change in creatinine (P = 0.114), ischemic time (P = 0.528), length of stay (P = 0.724), margin positivity (P = 0.447), surgical approach (P = 0.283), Clavien Dindo grade 2 (P = 0.861) and conversion/reoperation (P = 0.078). On multivariate analysis using generalized linear models, IPV once again showed significance when compared with nephrometry scores (P < 0.001), but no significance (P = 0.075) when compared with the percentage of creatinine change. Fig. 3 A curved intersection line is drawn along the expected boundary of the affected kidney, thereby dividing the intrarenal and extrarenal component of the renal mass. previously mentioned as a continuous variable. On univariate analysis of IPV as a continuous variable using simple linear regression, IPV was shown to be significantly related to Finding the cut-off for high IPV For analysis of IPV as a categorical variable, analysis was first carried out looking for the ideal value for a high IPV cut-off. This was achieved using ROC curves, and plotting sensitivity and specificity data of IPV percentiles in relation to certain outcomes. Various percentiles were analyzed looking for the one with the highest ROC values and thus predictive ability. We found the 89th percentile at IPV value of The Japanese Urological Association

4 Role of intraparenchymal tumor volume Table 1 Demographic, perioperative and volumetric data Patient characteristics (n = 87) Mean (range) Age (years) 60 (28 81) Preop creatinine (lmol/l) 80.1 (42 320) Postop creatinine (lmol/l) 93.8 (38 309) Creatinine change (%) 17.1 ( 18 67) Ischemic time (min) Zero ischemia (n = 21) 0 Cold ischemia (n = 1) 25 Warm ischemia (n = 65) 23 (6 53) Postoperative length of stay (days) 5 (2 30) Nephrometry score 7.8 (4 12) Mean tumor size (cm) 2.9 ( ) Mean total tumor volume (cm 3 ) 31.6 (0 639) Mean IPV (cm 3 ) 12.7 ( ) n (%) Male 59 (67.8%) Positive surgical margins 12 (13.6%) Mode of PN Open 38 (43.7) Laparoscopic 26 (29.9) Robot-assisted laparoscopic 23 (26.4) Clavien Dindo complications Grade 1 61 (70.1) Grade 2 26 (29.9) Conversion/reoperation 7 (8.0) Table 2 Analysis of IPV as a continuous variable (linear regression) P-values OR (95% CI) Univariate analysis Absolute change in creatinine ( ) Percentage change in creatinine ( ) Ischemic time ( ) Length of stay ( ) Margin positivity ( ) Surgical approach ( ) Clavien Dindo grade ( ) Conversion/reoperation ( ) RENAL nephrometry score < ( ) Multivariate analysis Percentage change in creatinine ( ) RENAL nephrometry score < ( ) cm 3 to have the highest ROC value of (95% CI ) for predicting the percentage of postoperative change in creatinine (sensitivity 100%, specificity 64.1%), and a ROC value of (95% CI ) for predicting the nephrometry score (sensitivity 88.9%, specificity 65.4%). The results are detailed in Tables 3 and 4. Univariate and multivariate analysis of IPV as a categorical variable (high IPV) Table 5 details the univariate and multivariate analysis results for high IPV as a categorical variable. High IPV score (IPV >27.26 cm 3 ) was significantly correlated to absolute change in creatinine (P = 0.018), percentage change in creatinine (P = 0.004) and nephrometry score (P < 0.001) on Mann Whitney U-tests. High IPV score did not reach significance with ischemic time (P = 0.768), length of stay (P = 0.602), margin positivity (P = 0.126), surgical approach (P = 0.586), Clavien Dindo grade 2 (P = 0.862) and conversion/reoperation (P = 0.673). Multivariate analysis using logistic regression showed IPV correlated well with percentage change in creatinine (P = 0.028) and RENAL nephrometry score (P < 0.001). Discussion The anatomical factors of a tumor including its location and size will determine the technical difficulty of PN. Our experience with PN surgery shows that the IPV will be a key factor in determining the technical complexity. Our hypothesis was that the larger the IPV, the greater the kidney parenchyma affected by resection, thus increasing the complexity of surgery, and worse perioperative renal and surgical outcomes. A recent study by Nisen et al. identified renal tumor invasion into parenchyma as a predictor of urological complications during PN. 8 As yet, we are not aware of any standardized objective measure of IPV reported in the literature; as such, we sought to develop the methodology of IPV, and examine if it predicts PN outcomes and complications. IPV is an objective, readily measurable CT-based radiological data-point that numerically combines two important aspects of tumor complexity, being size and percentage of the endophytic component. As such, tumors that are large, but superficial, might have low IPV scores and will not be considered to be complex. Similarly, if two tumors have a similar endophytic property, then the larger one will have a higher IPV, thus accurately reflecting overall tumor complexity. In this initial study, univariate and multivariate analysis has shown IPV to be an independent predictor of postoperative percentage change in creatinine (Table 2). On Table 3 Finding the right cut-off value for high IPV score (predicting percentage creatinine change) IPV percentile IPV value P-value ROC (95% CI) Sensitivity (%) Specificity (%) ( ) ( ) ( ) ( ) ( ) ( ) The Japanese Urological Association 439

5 RV TIWARI ET AL. Table 4 Finding the right cut-off value for high IPV score (predicting nephrometry score) IPV percentile IPV value P-value ROC (95% CI) Sensitivity (%) Specificity (%) ( ) ( ) Table 5 Analysis of high IPV as a categorical variable P-values Univariate analysis Absolute change in creatinine Percentage change in creatinine Ischemic time Length of stay Margin positivity Surgical approach Clavien Dindo grade Conversion/reoperation RENAL nephrometry score <0.001 Multivariate analysis Percentage change in creatinine RENAL nephrometry score <0.001 dichotomizing IPV into a high-ipv category (IPV >27.26 cm 3 ), univariate and multivariate analysis have shown it to be an independent predictor of postoperative absolute change in creatinine and postoperative percentage change in creatinine (Table 5). Not surprisingly, IPV correlated well with the RENAL nephrometry score given that it is also a surrogate for tumor complexity (Tables 2,5). The present study showed a increase in RENAL nephrometry score with every 1-cm 3 increase in IPV. Many investigators have assessed the ability of nephrometry scores to predict preoperative outcomes. In a series of 134 PNs with a 23% complication rate, Hew et al. showed a PADUA score >10 and RENAL score >9 independently predicted perioperative complications. 9 Conversely, in a series of 92 patients undergoing robotic PN, nephrometry scores failed to predict perioperative outcomes. 10 It remains unclear whether current established nephrometry scoring systems can predict perioperative outcomes. IPV, if validated, might provide an opportunity to further improve the predictive ability of these scoring systems. One aspect of the study we would like to explain is the use of manual free-hand scripting for volumetric analysis, which was chosen over computerized semiautomatic segmentation techniques (Figs 3 5). This allowed for exclusion of cysts, collecting system and vascular structures, which computer software might not be able to discern. In a comparison of free-hand techniques and computer software-based methods by Mir et al., it was found that free-hand techniques did better with discriminating more complex lesions. 11 The present study had a number of limitations. First, it was retrospective and derived from a single tertiary care center. Second, a wide range of surgical approaches were used for PN with several different methods of ischemia (cold, warm or zero). All these could potentially act as confounders to postoperative renal outcomes. A strength of the present study was the relatively large number of patients analyzed with no exclusions applied. The generalizability of the present results will require further study, in particular, prospective and independent validation. We believe that IPV represents just one component of renal tumor complexity. We are proposing it as a component of a comprehensive nephrometry system that includes other parameters, such as tumor location and proximity to the renal hilum. IPV is a novel, readily measurable, CT-based imaging parameter that helps quantify renal tumor complexity. In our experience, IPV is generalizable and reproducible. It takes <30 min to calculate per scan, with minimal additional cost to patients. All that is required is a volumetric software program. It has shown objectively how parenchymal volume loss will lead to renal functional decline after PN. It also compares very well with existing RENAL nephrometry scoring systems, and will serve to further complement them as a component of a comprehensive nephrometry system. We have shown in particular that high IPV, defined as IPV >27.26 cm 3, is correlated with worse perioperative renal outcomes. Future larger prospective trials will be required to test the validity of IPV in predicting perioperative surgical outcomes. Acknowledgments We acknowledge the contribution of Sharon Kuan (principal radiographer) and Tan Hwee Siah (senior radiographer) from the Department of Diagnostic Radiology, Singapore General Hospital, for their help in processing of tumor volumes. Conflict of interest None declared. References 1 Leibovich BC, Blute ML, Cheville JC, Lohse CM, Weaver AL, Zincke H. Nephron sparing surgery for appropriately selected renal cell carcinoma between 4 and 7 cm results in similar outcome to radical nephrectomy. J. Urol. 2004; 171: Kutikov A, Uzzo RG. The RENAL nephrometry score: a comprehensive standardized system for quantitating renal tumour size, location and depth. J. Urol. 2009; 182: Ficarra V, Novara G, Secco S et al. Preoperative aspects and dimensions used for an anatomical (PADUA) classification of renal tumors in patients The Japanese Urological Association

6 Role of intraparenchymal tumor volume who are candidates for nephron sparing surgery. Eur. Urol. 2009; 56: Leslie S, Gill IS, de Castro Abreu AL et al. Renal tumor contact surface area: a novel parameter for predicting complexity and outcomes of partial nephrectomy. Eur. Urol. 2014; 66: Simhan J, Smaldone MC, Tsai KJ et al. Objective measures of renal mass anatomic complexity predict rates of major complications following partial nephrectomy. Eur. Urol. 2011; 60: Aron M, Gill IS, Campbell SC. A non ischemic approach to partial nephrectomy is optimal. J. Urol. 2012; 187: Sharma N, O Hara J, Novack AC et al. Correlation between loss of renal function and loss of renal volume after partial nephrectomy for tumour in a solitary kidney. J. Urol. 2008; 179: Nisen H, Ruutu M, Glucker E, Visapaa H, Taair K. Renal tumour invasion index as a novel anatomical classification predicting urological complications after partial nephrectomy. Scand. J. Urol. 2014; 48: Hew MN, Baseskiouglu B, Barwari K et al. Critical appraisal of the PADUA classification and assessment of the RENAL nephrometry score in patients undergoing partial nephrectomy. J. Urol. 2011; 186: Muffarij PW, Krane LS, Rajamahanty S, Hemal AK. Does nephrometry scoring of renal tumors predict outcome in patients selected for robot assisted partial nephrectomy? J. Endourol. 2011; 25: Mir MC, Campbell RA, Sharma N et al. Parenchymal volume preservation and ischemia during partial nephrectomy: functional and volumetric analysis. J. Endourol. 2013; 82: The Japanese Urological Association 441