Abstract
Background and Purpose:
The centrality-index (C–Index), preoperative aspects and dimensions used for anatomic (P.A.D.U.A.) classification, and radius.exophyic/endophytic.nearness.anterior/posterior.location (R.E.N.A.L.) nephrometry schemes were developed as standardized scoring systems (SS) to quantify anatomic characteristics of kidney tumors. The objective of this study was to establish reliability and assess relationships between these three SS and perioperative and postoperative variables.
Patients and Methods:
A retrospective chart review was performed in 101 patients who underwent laparoscopic partial nephrectomy. The nephrometry schemes were correlated with intraoperative and postoperative parameters using Spearman correlations. In addition, interobserver reliability was assessed on 50 of the patients by interclass correlations comparing the scores assigned by two residents and one fellow who reviewed preoperative CT studies of these patients.
Results:
The interobserver correlation was 0.84 for the C-Index, 0.81 for the P.A.D.U.A., and 0.92 for the R.E.N.A.L. scoring systems, demonstrating excellent interobserver reliability. All three SS were significantly associated with warm ischemia time (WIT) (C-Index, P=−0.44; P.A.D.U.A., P=0.25; R.E.N.A.L., P=0.32) and percent change in creatinine level (C-Index, P=− 0.33; P.A.D.U.A., P=0.37; R.E.N.A.L., P=0.37). There were no significant associations between any of the three SS assessed and the occurrence of complications, operative time, or estimated blood loss. No significant correlation was found between the P.A.D.U.A. and R.E.N.A.L. SS and length of stay; however, C-Index did show a significant relationship for patients with lower scores having longer hospital stays (P=−0.21).
Conclusions:
All three scoring systems demonstrated reliability among observers and represent novel methods of quantitatively describing renal tumors. They were all associated with WIT, percent change in creatinine level, and tumor size. They did not, however, correlate with any other perioperative parameters investigated. At this time, these SS provide a common language for describing renal tumors.
Introduction
Tumor complexity remains the primary modality by which urologists determine surgical approach and treatment strategy. Recently, three different scoring systems have been reported to standardize and quantitate tumor complexity. 2 –4 These were developed to serve as a common vocabulary when discussing anatomic geometry and complexity of renal masses. Moreover, these scores have the potential to influence treatment selection if the scores predicted operative outcomes.
Preoperative aspects and dimensions used for anatomic (P.A.D.U.A.) classification and radius.exophyic/endophytic.nearness.anterior/posterior.location (R.E.N.A.L.) scoring systems (SS) involve similar components and methodology, enabling a comprehensive description of the tumor size, polarity, anterior/posterior location, closeness to the collecting system; each component is assigned by score, providing a continuous variables. The centrality index (C-Index), however, is a completely different system that involves a relatively complex mathematical concept yet represents simple practical measurement and characterizes tumor centrality based on the ratio of the distance between the tumor and kidney center and tumor radius.
This study was undertaken to validate these renal tumor SS with respect to interobserver reliability and correlation with perioperative outcomes in an external patient cohort.
Patients and Methods
This study reviewed 461 consecutive patients who underwent LPN between January 2006 and March 2010, of whom 101 patients had preoperative images available for retrospective review and SS assignment. Our technique of LPN has been described previously. 5 After Institutional Review Board approval, CT or MRI scans were reviewed for all patients. C-Index, P.A.D.U.A., and R.E.N.A.L. SS were calculated for each scan by a urology fellow. Two urology residents also independently reviewed 50 of these scans. Each observer was blinded to the results of the other observers' assessments.
Statistical analyses were performed using SAS 9.2 (SAS Institute, Inc, Cary, NC). Interobserver reliability (IR) was established on the 50 patients with interclass correlations (ICC) between the three observers for the three scoring systems. The remainder of analyses was conducted on the 101 patients using only scores computed by the urology fellow. For each SS, Spearman correlations were conducted with the following variables: Warm ischemia time (WIT), percent change in creatinine level, estimated blood loss (EBL), operative time (OT), postoperative complications, and length of hospital stay (LOS).
Percent change in creatinine level was calculated based on the difference between baseline creatinine and postoperative day 1 serum creatinine levels. In addition to using absolute scores, P.A.D.U.A. scores were categorized into low (6–7), moderate (8–9), and high (≥10) complexity tumors as previously described in Ficarra and associates. 4 The R.E.N.A.L. scores were categorized into low (4–6), moderate (7–9), and high (10–12) based upon Kutikov and colleagues. 3
Kruskal-Wallis test analyses were used to assess the relationships between categorized scores (low, moderate, and high complexity) and the aforementioned perioperative variables. On finding a significant difference, Bonferroni-adjusted pairwise comparisons were conducted using Mann-Whitney tests. Therefore, for these pairwise tests, a preset significance level of 0.0167 was used. The frequency distribution of presence of postoperative complications was compared between the three groups using the chi-square test.
Results
Table 1 summarizes patient demographic, preoperative, and pathologic data. The cohort consisted of 101 patients with a mean age of 57.3, including 66 men and 34 women. Average tumor size was 3.2 cm. There was one intraoperative complication in which the procedure was converted to laparoscopic radical nephrectomy because of excessive bleeding. Postoperatively, complications developed in 18 (18%) patients, including eight Clavien grade I, five grade II, and six grade III complications. Final pathologic examination revealed two patients with positive margins.
ASA=American Society of Anesthesiologists; CR=creatinine; EBL=estimated blood loss; OT=operative time; WIT=warm ischemia time; RCC=renal-cell carcinoma.
All three SS demonstrated excellent IR. The ICC between the three observers on the sample of 50 patients were 0.84 for the C-Index, 0.81 for the P.A.D.U.A., and 0.92 for the R.E.N.A.L. SS. Subsequent analyses including all patients revealed significant correlations between all three tumor SS: C-Index and P.A.D.U.A., P=−0.55, P<0.0001; C-Index and R.E.N.A.L., P=−0.52, P<0.0001; P.A.D.U.A. and R.E.N.A.L., P=0.60, P<0.0001.
Table 2 presents Spearman correlation results between each scoring system and perioperative variables. All three SS showed significant association with WIT and percent change in creatinine level. In addition, C-Index scores correlated with hospital LOS. Table 3 lists perioperative results by each categorized tumor complexity for the R.E.N.A.L. and P.A.D.U.A. SS. For both SS, patients with moderate and high complexity tumors had greater percent change in creatinine level compared with low complexity tumors. Low complexity tumors as classified by the R.E.N.A.L. SS had significantly shorter WIT than moderate and high complexity tumors. This difference was not observed with P.A.D.U.A. low complexity tumors. Interestingly, there were no differences between moderate and high complexity tumors.
C-Index=centrality index; P.A.D.U.A.=preoperative aspects and dimensions used for anatomic; R.E.N.A.L.=radius.exophyic/endophytic.nearness.anterior/posterior.location; EBL=estimated blood loss; WIT=warm ischemia time.
χ2=15.6, P<0.0004, low<moderate (P<0.0002), low<high (P<0.006).
χ2=16.2, P<0.0003, low<moderate (P<0.004), low<high (P<0.008).
χ2=14.5, P<0.0007, low<moderate (P<0.006), low<high (P<0.0003).
R.E.N.A.L.=radius.exophyic/endophytic.nearness.anterior/posterior.location; P.A.D.U.A.=preoperative aspects and dimensions used for anatomic; WIT=warm ischemia time; EBL=estimated blood loss; OT=operative time; LOS=length of stay.
Discussion
In patients with renal cortical neoplasms who are undergoing surgery, preoperative planning and subsequent prognosis are greatly dependent on anatomic tumor characteristics, such as size and location. 6 –8 The advantage of using standardized SS is that they incorporate a variety of important variables to provide a more accurate description of tumor characteristics. In the evaluation of any SS, several factors must be considered. First, the system should not be overly complex in its application and should result in consistent scores between observers. Second, any assessment of tumor complexity based on imaging is valid only if it correlates with objective measurements of technical complexity in the operating room.
The present study aimed to evaluate all three SS based on the above criteria. IR as measured by interclass correlation coefficient has been validated in previou studies as an appropriate assessment of qualitative SS specifically for systems based on quantitative measurements that are made on units that are organized into groups. The present study demonstrates excellent reliability for all three SS in a single cohort of patients. Findings with C-Index were similar to those reported by Simmons and colleagues 2 on 22 patients. In addition, they concluded from further analysis that the learning curve was achieved after approximately 14 patients. Although we did not perform a learning curve analysis, our subjective observations were that the learning curve for all three SS was generally a few patients. It should be noted that C-Index differs significantly from both the R.E.N.A.L. and P.A.D.U.A. SS in that it involves a complex measurement to characterize tumor centrality based on the ratio of the distance between the tumor and kidney center and the tumor radius. Despite this relatively complex measurement, C-Index demonstrated excellent IR.
Our study also demonstrated that SS-based tumor complexity correlated WIT and percent change in creatinine level. This result can be expected because it follows that more complex tumors need longer vascular clamp times to achieve safe excision and reconstruction. Interestingly, the SS could only differentiate tumors with low vs moderate/high complexity but not between tumors with moderate vs high complexity, suggesting that a two-tiered complexity classification may be more valid. With regard to renal functional outcomes, the results support previous studies in the literature. In external validation cohort, Waldert and coworkers 9 reported a positive correlation between P.A.D.U.A. scores and WIT for both open and LPN. Using their original cohort of 131 patients, Sampalski and colleagues 10 reported that higher complexity tumors as measured by C-Index were associated with significantly longer WIT and subsequent decrease in estimate glomerular filtration rates. To date, there have been no other studies in the literature correlating R.E.N.A.L. scores with renal functional outcomes.
We also found that tumor complexity based on the SS was not associated with EBL or OT. In our experience, less difficult tumors are excised without vascular clamping to maximally preserve renal function; however, this often results in a higher than expected EBL. With respect to OT, we hypothesize that other patient characteristics, such as body habitus, surgical history, and difficulty of hilar dissection, may be more influential determinants than tumor size and location.
With regard to perioperative complications, our results are not in line with previous reports. In a small cohort of patients, Bruner and associates 11 reported a 35% increased risk of postoperative urine leak after LPN for higher complexity tumors based on the R.E.N.A.L. SS. In their initial publication, Ficcara and coworkers 4 validated the P.A.D.U.A. SS based on the significantly higher complication rate for moderate and high complexity renal tumors. None of the three SS analyzed in our cohort showed significant association with risk of complication, although a lower C-Index score was associated with a longer hospital LOS (P<0.04). The reasons for the discrepancy may be the low rate of complications in our cohort (not enough events) or the experience of the surgeon in performing nephron-sparing surgery for complex tumors. All of the procedures were performed by a single surgeon with experience of more than 800 LPNs.
The limitation of this study is that, first, the size of the cohort was relatively small, and a larger number of patients might yield more power to observe smaller effect sizes. In addition, larger numbers may have shown better performance in discriminating between moderate and high complexity tumors. Second, only patients for whom preoperative imaging was available were included in this cohort. This may have incorporated an unintended bias into the dataset. Finally, it would be interesting to determine if the surgeon's subjective assessment of the tumor would be a more or less accurate measure of tumor complexity.
These SS currently represent the most comprehensive tools for categorizing small renal masses and appear to reliably discriminate between low, moderate, and high complexity tumors. When the clinician is faced with the decision of choosing among the different surgical approaches, the SS may provide with accurate estimation of the tumor complexity. Compared with C-Index, the R.E.N.A.L. and P.A.D.U.A. SS are a more appealing means of calculating accurate predictions without the use of complex formulas and measurement technique. Tumor complexity based on these SS, however, may be presumptive, and surgical outcomes may vary by surgeon experience and skill level for challenging procedures such as LPN.
Conclusions
All three SS represent novel methods of quantitatively describing renal tumors in a standardized manner with reproducible interobserver assessments. While P.A.D.U.A., R.E.N.A.L., and C-Index were all associated with renal functional outcomes, they did not show significant correlations with other measures of operative complexity, such as EBL, OT, and perioperative complication rates. Furthermore, the P.A.D.U.A. and R.E.N.A.L. SS were not able to discriminate between moderate and high complexity tumors.
These SS currently represent the most comprehensive tools for categorizing patients undergoing LPN into low, moderate, and high complexity based on the tumor characteristics. This information may be helpful in preoperative surgical planning, especially for surgeons early in their experience. While the SS do provide a standardized tool for reporting purposes, further modifications may be necessary to more accurately predict operative and patient-related outcomes.
Footnotes
Disclosure Statement
No competing financial interests exist.
