Reproducibility and Fidelity of the R.E.N.A.L. Nephrometry Score

Introduction

D etection and management of renal malignancy has increased in frequency and is a common clinical encounter for practicing urologists. ^1,2 The standard of care is extirpative surgery, which in the past was routinely accomplished by radical nephrectomy. ^{3 –5} With development of new technologies and advanced imaging procedures, open and minimally invasive nephron-sparing surgery ^{6 –8} and thermal ablation methods ^9,10 have emerged as viable treatment options for localized renal-cell carcinoma (RCC).

From an oncologic perspective, nephron-sparing surgery has established itself as an equivalent to radical nephrectomy for T₁ tumors. ^11,12 At advanced centers, even complex lesions near the hilum ¹³ and tumors larger than 4 cm ^{14 –16} are now amendable to minimally invasive resection, maximally preserving normal renal parenchyma, especially important in solitary kidney or bilateral tumors.

Recent evidence further suggests increased risk for chronic kidney disease and associated cardiovascular morbidity ^17,18 for patients undergoing radical nephrectomy, supporting the expanding role for nephron-sparing surgery. ¹⁹ Partial nephrectomy, however, remains a demanding procedure with a steep learning curve when performed laparoscopically. Communication inherent to patient referrals and research efforts for the treatment of small renal masses emphasizes the need for the development of a standard nomenclature to describe renal tumor anatomy.

In an effort to improve communication between clinicians who care for patients with a given renal mass, a new scoring system was developed to describe tumor geometry to allow better communication of anatomic complexity and aid in decision making regarding the best treatment approach. Kutikov and Uzzo ²⁰ developed the R.E.N.A.L. nephrometry score (NS) to systematically describe renal tumor anatomy in a reproducible and quantifiable manner.

The five most reproducible and pertinent features characterizing renal tumor anatomy as related to resectability were selected and combined to form a scoring system designated the R.E.N.A.L. NS: R (radius or maximum tumor diameter), E (location of the mass with relation to the surface of the kidney or percentage being endophytic vs exophytic), N (nearness of the mass to the collecting system), A (location anterior vs posterior) and L (location with reference to polar lines). Except for the component (A) describing anterior vs posterior location, a three point scoring scale is used adding up to the NS sum, ranging from a minimum of 4 to a maximum of 12. In addition, suffixes “x” and “h” are used describing a tumor that is neither anterior nor posterior and to designate hilar location, respectively. The definition for a hilar mass, as proposed by Kutikov and Uzzo, ²⁰ designates tumors that abut the main renal artery or vein.

Initial evaluation of the NS system included 50 cases of resected renal masses and revealed that lesions with lower complexity were more likely to be amendable to partial nephrectomy. To date, no independent studies verifying reproducibility of the R.E.N.A.L. NS have been published.

Our goal in this study was to assess the reproducibility of the nephrometry score by calculating interobserver variability.

Patients and Methods

We identified 149 patients from our prospectively maintained, Institutional Review Board-approved database of patients who were undergoing laparoscopic partial nephrectomy (LPN) with available preoperative imaging, either CT or MRI. Two urology residents at different levels of residency training, one junior and one senior, read the scans, independently assigning values for the individual NS parameters. Each resident was blinded to the scores assigned by the other resident. Before evaluating the 149 scans, both residents received a short tutorial on R.E.N.A.L. NS, and each practiced on 10 scans with evaluation and NS assignment being verified by an experienced observer (LRK), who also assigned NS during this tutorial phase. This attending observer was not involved in the original development and assessment of the nephrometry scoring system.

R.E.N.A.L NS consist of five parameters, four of which are scored on a numerical (1–3) point scale as described previously. ²⁰ Suffixes “a” and “p” are added to describe tumor location in the coronal plane as anterior or posterior, respectively, or x if the coronal location is indeterminate. This comprises the fifth component of the NS. In addition, the suffix h is added to indicate hilar positioning of the mass. The final NS is composed of the added sum of all numerical values as well as the suffixes.

For the purpose of our statistical analyses, we assigned additional numeric scores to the suffixes. All tumors in the hilar location were assigned one additional point, and a numeric point scale was assigned for coronal plane location evaluation with one point for anterior (a), two for intermediate (x), and three for posterior (p) tumors. Assigning scores for coronal location takes into account the more difficult access for posterior tumors when performing LPN. This new total sum was named total nephrometry score (TNS), ranging from a minimum value of 5 to a maximum value of 16. This is different from the NS sum, which includes non-numeric components, and its numeric range is 4 to 12. Complexity of resection was assessed by dividing the NS sum and TNS each into three different cohorts indicating low, intermediate, and high complexity (Table 1).

R.E.N.A.L. NS components, NS sum, and TNS as well as complexity cohorts based on NS sum and TNS as assigned by two independent observers were analyzed using simple concordance rates and Kendall tau-b correlations. Interobserver variability was assessed with the Pearson correlation. The two data sets (observer A vs observer B) were compared using the paired t test to determine any existing statistical significance between the groups. A P value of <0.05 was the predetermined threshold for statistical significance.

Results

The demographics as well as perioperative parameters, final pathology findings, and postoperative complications of this LPN cohort are outlined in Table 2. The concordance rates for R.E.N.A.L. NS components, NS sum, and TNS are listed in Table 3. We found that scores were highly consistent as indicated by concordance rates ranging between 86% and 99% for individual tumor characteristics assessed and 74% for the sum of the components or the TNS. The components with highest concordance were size (R), hilar location, and assessment of anterior vs posterior nature of the mass. Assignment of scores for nearness to the collecting system (N) and location with respect to polar lines (L) had the lowest concordance rates.

When lesions were analyzed by complexity cohort, the overall concordance rates increased to 96% for NS sum and 89% for TNS defined cohorts. Interobserver correlation of TNS calculated by the Pearson test was found to be 0.92 (P<0.001). The mean NS sum±standard deviation (SD) were 6.9±2.0 and 6.9±1.9 for observer A and B, respectively, with no statistical significance. Similarly, there was no difference for mean TNS (Table 4). Based on both NS sum and TNS, most tumors evaluated fell within the intermediate complexity cohort.

Discussion

In an effort to standardize the nomenclature for renal tumor anatomy, three different scoring systems have been recently described to aid with assessment of tumor complexity and clinical decision making. ^{20 –22} At our institution, we have adapted the R.E.N.A.L. NS to communicate description of renal tumors among members of the urology team with an effort to further extend use of this NS system to our radiology colleagues. Several factors make the R.E.N.A.L. NS system useful, including its comprehensive nature that addresses important tumor characteristics relevant for clinical decision making, all organized into an easily understood scoring system. What appears complex at first, given its multiple components, has in our experience demonstrated to be easily applicable, reproducible, and generally has an easily achievable learning curve.

From the data presented, we found strong reproducibility of the R.E.N.A.L. NS with high interobserver fidelity (0.92, P<0.001). Individual components with the highest fidelity are tumor location in the coronal plane (anterior vs posterior) and location with respect to the surface of the kidney (exophytic vs endophytic).

The system can be used with either axial CT scan or MRI. Although not verified in this study, we noted that the highest interobserver correlations were achieved when the imaging included axial, coronal, and sagittal images, making it easier for each of the components to be assessed for the nonradiologist. Accurate assessment in only the axial plane is possible but usually takes longer, and interobserver variability may vary more widely, especially for assessing polar location of the tumor as well as nearness to the collecting system.

We used the R.E.N.A.L. NS as initially described by Kutikov and Uzzo ²⁰ and added numeric values to quantify components previously described by suffixes. The TNS calculated consists only of a final numeric value. Assignment of tumor complexity based on either the NS sum or TNS was found to have similar fidelity. No further improvement of the score is associated with computation of a purely numeric value. When used for clinical decision making, interpretation of the entire score with its individual components remains essential, because overall complexity scores can be misleading and may eventually necessitate different numeric score assignments to each component based on clinical outcome parameters as such a nomogram may provide.

We were able to demonstrate reproducibility and interobserver fidelity of a novel scoring system with excellent potential to allow for recording of renal tumor neoplasms in a more standardized manner. Limitations included the retrospective nature of this investigation and the limited number of individual observers. Further verification from academic centers is needed to verify this reproducibility found. It would also be helpful to further test the applicability of the scoring system by comparing scores assigned by board-certified urologists in the community vs those at academic referral centers. Development of spreadsheets or automatic calculation tools would further help to make the NS more applicable for the practicing urologist to use this system in real time while counseling patients. Studies assessing clinical applicability, outcomes, and overall validity of the NS are needed to further its acceptance and widespread use in practice.

Conclusion

The R.E.N.A.L. NS is a comprehensive assessment tool for delineating renal tumor anatomy. The scoring system demonstrates good reproducibility associated with a manageable learning curve and excellent interobserver fidelity. Reporting of the entire NS with all its components in addition to overall score and tumor complexity cohort is important for data collection and comparison of surgical series. In addition, it can aid in clinical decision making.