- Email: [email protected]
Prognostic assessment of nonmetastatic renal cell carcinoma: a clinically based model
RAPID COMMUNICATION
PROGNOSTIC ASSESSMENT OF NONMETASTATIC RENAL CELL CARCINOMA: A CLINICALLY BASED MODEL OZGUR YAYCIOGLU, WILLIAM W. ROBERTS, THERESA CHAN, JONATHAN I. EPSTEIN, FRAY F. MARSHALL, AND LOUIS R. KAVOUSSI
ABSTRACT Objectives. Determining the recurrence risk in patients treated for renal cell carcinoma (RCC) is important for providing prognostic information and planning potential surveillance strategies. The pathologic stage has been the most widely used single prognostic variable. However, with minimally invasive treatment modalities, the pathologic stage may not be readily available. We developed a biostatistical prognostic model for postoperative RCC that is independent of the pathologic stage. Methods. The records of 296 patients who underwent open nephrectomy for RCC at Johns Hopkins Hospital between 1990 and 1999 were reviewed. Cox proportional hazards regression analysis was used to generate a prognostic model. Results. The recurrence risk (Rrec) was determined from this model: Rrec ⫽ 1.55 ⫻ presentation (0-1) ⫹ 0.19 ⫻ clinical size (in centimeters). Using this equation, 79% of patients were identified as low risk compared with 45% of patients considered low risk by pathologic stage (pT1). Moreover, the separation between the high and low-risk survival curves increased. Conclusions. This model is the first to our knowledge that uses purely clinical variables to assess the postoperative prognosis in patients with RCC. These results, although not validated, provide substantial evidence that preoperative clinical variables may be used instead of the pathologic stage to determine the risk of recurrence. Uncoupling the reliance on pathologic stage for prognostic information removes a potential barrier to novel minimally invasive treatments for renal malignancy and provides a standard to which observation protocols can be compared. In the future, this model may facilitate selection of appropriate patients for less toxic adjuvant or neoadjuvant therapies. UROLOGY 58: 141–145, 2001. © 2001, Elsevier Science Inc.
C
areful assessment of the prognosis in patients after surgery for renal cell carcinoma (RCC) is helpful in selecting the appropriate surveillance protocols. The early detection of recurrence and subsequent aggressive treatment may provide the greatest chance for extended survival, as effective adjuvant therapies are not currently available for patients with locally advanced disease.1 Although many attempts have been made to identify better pathologic and biomolecular prognostic markers, the pathologic stage remains the most widely used From the James Buchanan Brady Urological Institute and Department of Pathology, Johns Hopkins Medical Institutions, Baltimore, Maryland; and Department of Urology, Emory University School of Medicine, Atlanta, Georgia Reprint requests: William W. Roberts, M.D., 600 North Wolfe Street, Suite 161, Jefferson Street Building, Baltimore, MD 21287-8915 Submitted: April 9, 2001, accepted (with revisions): April 23, 2001 © 2001, ELSEVIER SCIENCE INC. ALL RIGHTS RESERVED
prognostic factor for patients with RCC after nephrectomy.2,3 The 1997 TNM staging system has been shown to be a predictor of cause-specific survival and has been the basis of several proposed postoperative surveillance algorithms.4 – 6 Minimally invasive approaches have been increasingly used for the surgical treatment of RCC.7–11 Unfortunately, the very nature of minimally invasive surgery limits the pathologic assessment of the specimen. For example, during laparoscopic nephrectomy, specimen retrieval can be accomplished by en bloc removal, necessitating an expanded incision, or by morcellation through a port site. Morcellation, although preserving the minimally invasive nature of the operation, may potentially interfere with the determination of the exact pathologic stage of the tumor. We sought to develop a biostatistical model based on readily available preoperative clinical variables and independent of the pathologic stage 0090-4295/01/$20.00 PII S0090-4295(01)01207-9 141
to stratify patients by recurrence risk. This model was compared with pathologic staging as a prognostic indicator in patients after surgical treatment for nonmetastatic RCC. MATERIAL AND METHODS The records of all the patients in the Johns Hopkins Cancer Registry who were surgically treated for RCC from January 1990 to December 1999 were retrospectively reviewed. Patients with bilateral synchronous or metastatic disease at the time of surgery or von Hippel-Lindau disease and those who underwent surgery elsewhere were excluded. Patients with radiographic evidence of lymphadenopathy, tumor extension beyond Gerota’s fascia, or vena caval involvement above the diaphragm were also excluded. The patient presentation was categorized as symptomatic or incidental. Tumors that caused pain, hematuria, abdominal mass, or weight loss were identified as symptomatic tumors. All patients had undergone preoperative computed tomography or magnetic resonance imaging, the findings of which were used to assign the clinical stage in accordance with the 1997 TNM staging system.4 The largest diameter of the tumor measured on the computed tomography scan was recorded as the clinical size. Postoperatively, tumors were pathologically staged according to the 1997 TNM system and assigned a Fuhrman nuclear grade. Patients were followed up at 6 to 12-month intervals with a physical examination, routine laboratory evaluation, chest x-ray, and computed tomography scan. Information on patients followed up elsewhere was obtained through the Cancer Registry. The time to recurrence was defined as the interval from surgery to the first evidence of disease recurrence. Kaplan-Meier statistical analyses were performed using Stata 6.0 (Stata, College Station, Tex) to estimate the recurrence-free survival after surgical treatment of RCC. Patients last known alive with recurrence and those dead without recurrence had their follow-up time censored. The Cox proportional hazards regression analysis was used to model the time to recurrence after surgery and to develop an equation for recurrence risk (Rrec), which was used to stratify patients into high and low-risk groups.
RESULTS From 1990 through 1999, 296 patients met the entry criteria and underwent surgery for the treatment of clinically nonmetastatic RCC. The clinical and pathologic information is presented in Table I. The median follow-up time after surgery was 48 months (range 5 to 129). Thirty-eight patients (12.8%) developed recurrences a median of 17 months after surgery (range 1 to 77). The diseasefree survival curves for patients with pathologic Stage T1 tumors versus those with pathologic Stage T2-4 tumors are shown in Figure 1. Presentation, clinical stage, grade, pathologic stage, clinical size, age, and sex were evaluated individually (Table II). They were then considered together using backward stepwise logistic regression analyses to remove the variables that failed to explain clinically important differences in the risk of recurrence (Rrec). The remaining variables (presentation and clinical size) were incorporated into 142
TABLE I. Clinical and pathologic patient characteristics (n ⴝ 296) Sex Male Female Mean age (yr) Nephrectomy type Radical Partial Presentation Symptomatic Incidental Mean clinical size (cm) Clinical T stage (n) T1 T2 T3a T3b Pathologic T stage (n) T1 T2 T3a T3b Tumor grade (n) I II III IV Undetermined*
185 (63) 111 (37) 60.1 ⫾ 12.5 236 (80) 60 (20) 157 (53) 139 (47) 6.0 ⫾ 3.5 186 39 47 24
(62.8) (13.2) (15.9) (8.1)
132 25 106 33
(44.6) (8.4) (35.8) (11.1)
13 169 94 19 1
(4.4) (57.1) (31.8) (6.4) (0.3)
Numbers in parentheses are percentages. * No grade assigned because of extensive tumor necrosis.
the final risk equation: Rrec ⫽ 1.55 ⫻ presentation (0-1) ⫹ 0.19 ⫻ clinical size (in centimeters). Patient presentation, scored as asymptomatic (0) or symptomatic (1), and the clinical tumor size (in centimeters) were inserted into the equation. Each patient received an Rrec score that was used for stratification. Patients with an Rrec score of 3.0 or less were categorized as low risk; those with an Rrec score greater than 3.0 were categorized as high risk. This cutoff was selected to generate a low-risk group with equivalent survival rates to patients with pathologic T1 tumors (specifically 91% 5-year survival and 89% 10-year survival rates). Similar survival analyses have suggested that patients with pT1 disease are at lower risk of recurrence and may be candidates for less rigorous surveillance regimens.5,12 Using this prognostic model, Rrec, 79% of the patients were identified as low risk compared with 45% classified as low risk by the pathologic criteria (pathologic Stage T1). Furthermore, the low and high-risk survival curves were statistically distinct (P ⬍0.001, Wilcoxon test), with survivor functions for the low and high-risk groups of 99% versus 77% at 1 year, 98% versus 67% at 2 years, and 92% versus 57% at 5 years of follow-up. This results in increased separation between the curves UROLOGY 58 (2), 2001
FIGURE 1. Kaplan-Meier disease-free survival curves (based on stratification by pathologic stage) for patients with Stage pT1 and pT2-4 are statistically distinct (P ⬍0.008, Wilcoxon test). The disease-free survivor percentages for patients with Stage pT1 and pT2-4 were 99% versus 91% at 1 year, 98% versus 86% at 2 years, and 91% versus 80% at 5 years of follow-up, respectively.
TABLE II. Results of Cox proportional hazards univariate and multivariate analysis for prediction of recurrence 95% Hazard Standard P Confidence Ratio Error Value Interval Univariate analysis Presentation Clinical stage Grade Pathologic stage Clinical size Age Sex Multivariate analysis Presentation Clinical size
6.17 2.27 2.05 1.93 1.22 1.07 0.63
2.96 0.31 0.45 0.32 0.04 0.17 0.23
0.001 0.001 0.001 0.001 0.001 0.67 0.20
2.41–15.80 1.73–2.98 1.33–3.16 1.39–2.67 1.15–1.30 0.78–1.46 0.31–1.28
4.73 1.21
2.28 0.04
0.001 1.84–12.15 0.001 1.13–1.30
compared with pathologic stratification (Figs. 1 and 2). COMMENT Laparoscopic radical nephrectomy has longterm efficacy comparable to open nephrectomy and provides a shorter and more comfortable convalescence.10,13 New ablative modalities such as raUROLOGY 58 (2), 2001
diofrequency ablation, high-intensity focused ultrasound, cryoablation, and interstitial laser therapy have the potential to cause tumor destruction noninvasively or with only a single needle puncture.11,14 These ablative technologies, however, do not produce a specimen for pathologic analysis, which may be an impediment to their widespread application. For similar reasons, the method of specimen retrieval for laparoscopic radical nephrectomy varies. Intact retrieval preserves the kidney for pathologic analysis but necessitates an extended abdominal incision. With morcellation, the specimen is removed piecemeal through a port site, preserving the minimally invasive nature of the procedure but increasing the difficulty of accurate tumor staging.4 Landman and coworkers15 showed that limited morcellation of radical nephrectomy specimens in vitro did not alter the determination of the histologic features, grade, or local invasiveness of the tumor. Unfortunately, our routine clinical experience shows that this is not always practical with in vivo morcellation. Limited data suggest that morcellation results in decreased postoperative pain and shorter convalescence. Dunn and coworkers8 noted decreased analgesic use and a shorter hospital stay for patients 143
FIGURE 2. Kaplan-Meier disease-free survival curves (based on prognostic model stratification) for low and high-risk clinical groups are statistically distinct (P ⬍0.001, Wilcoxon test). The disease-free survival percentages for the low and high-risk groups were 99% versus 77% at 1 year, 98% versus 67% at 2 years, and 92% versus 57% at 5 years of follow-up, respectively.
having morcellation versus intact removal. Walther and coworkers7 found that the postoperative narcotic requirement was similar for patients who underwent open nephrectomy and laparoscopic nephrectomy with intact removal, but was significantly less for those who had morcellation. In this study, symptomatic presentation and clinical size were the two most important prognostic factors for predicting cancer recurrence and thus were included in the final risk equation. The results of this model, Rrec, were compared with stratification by pathologic stage alone. Interestingly, the pathologic grade when combined with the pathologic stage in a multivariate fashion failed to appreciably improve on the stratification based on the pathologic stage alone. This is probably because 88.9% of the tumors were pathologic grade II/III and the prognostic distinction between grade II and grade III tumors was not statistically significant. Moreover, this model (Rrec) has not been validated, as the sample was not of sufficient size to allow rigorous internal validation. Therefore, external validation will be important in determining the reproducibility of this model. This model classifies a greater number of patients as low risk than does pathologic stratification while achieving the same survival rates. It therefore may allow a greater number of patients to fol144
low less rigorous surveillance protocols after surgery. In the future, this model may also help in selecting candidates to receive new, potentially less toxic, adjuvant and neoadjuvant therapies. CONCLUSIONS Using the Cox proportional hazard regression analysis, we constructed a biostatistical model that stratifies patients according to their risk of recurrence. This model is the first to our knowledge that uses purely clinical variables without reliance on the pathologic stage to assess the postoperative prognosis in patients with RCC. Seventy-nine percent of the patients were classified as low risk, potentially needing less rigorous surveillance postoperatively and unlikely to require future adjuvant treatment. Although not yet validated, these results provide substantial evidence that preoperative clinical variables may be used instead of the pathologic stage in determining the risk of recurrence after surgery for RCC. Uncoupling the reliance on pathologic stage for prognostic information will allow a greater role for minimally invasive treatment of renal malignancies and in the future may help to select appropriate candidates for less toxic adjuvant immunotherapies. UROLOGY 58 (2), 2001
ACKNOWLEDGMENT. To Anne L. Kammer of Johns Hopkins Cancer Registry for providing assistance with data collection. REFERENCES 1. Tanguay S, Pisters LL, Lawrence DD, et al: Therapy of locally recurrent renal cell carcinoma after nephrectomy. J Urol 155: 26 –29, 1996. 2. Thrasher JB, and Paulson DF: Prognostic factors in renal cancer. Urol Clin North Am 20: 247–262, 1993. 3. Gelb AB: Renal cell carcinoma current prognostic factors. Cancer 80: 981–986, 1997. 4. Guinan P, Sobin LH, Algaba F, et al: TNM staging of renal cell carcinoma: Workgroup No. 3. Union International Contre le Cancer (UICC) and the American Joint Committee on Cancer (AJCC). Cancer 80: 992–993, 1997. 5. Levy DA, Slaton JW, Swanson DA, et al: Stage specific guidelines for surveillance after radical nephrectomy for local renal cell carcinoma. J Urol 159: 1163–1167, 1998. 6. Gettman MT, Blute ML, Spotts B, et al: Pathologic staging of renal cell carcinoma: significance of tumor classification with the 1997 TNM staging system. Cancer 91: 354 –361, 2001. 7. Walther MM, Lyne JC, Libutti SK, et al: Laparoscopic cytoreductive nephrectomy as preparation for administration of systemic interleukin-2 in the treatment of metastatic renal cell carcinoma: a pilot study. Urology 53: 496 –501, 1999.
UROLOGY 58 (2), 2001
8. Dunn MD, Portis AJ, Shalhav AL, et al: Laparoscopic versus open radical nephrectomy: a 9-year experience. J Urol 164: 1153–1159, 2000. 9. Janetschek G, Jeschke K, Peschel R, et al: Laparoscopic surgery for stage T1 renal cell carcinoma: radical nephrectomy and wedge resection. Eur Urol 38: 131–138, 2000. 10. Cadeddu JA, Ono Y, Clayman RV, et al: Laparoscopic nephrectomy for renal cell cancer: evaluation of efficacy and safety—a multicenter experience. Urology 52: 773–777, 1998. 11. Gill IS, Novick AC, Meraney AM, et al: Laparoscopic renal cryoablation in 32 patients. Urology 56: 748 –753, 2000. 12. Sandock DS, Seftel AD, and Resnick MI: A new protocol for the followup of renal cell carcinoma based on pathological stage. J Urol 154: 28 –31, 1995. 13. Portis AJ, Yan Y, Chen C, et al: Long-term follow up after laparoscopic radical nephrectomy. J Endourol 14(suppl 1): A31, 2000. 14. Zlotta AR, Wildschutz T, Raviv G, et al: Radiofrequency interstitial tumor ablation (RITA) is a possible new modality for treatment of renal cancer: ex vivo and in vivo experience. J Endourol 11: 251–258, 1997. 15. Landman J, Lento P, Hassen W, et al: Feasibility of pathological evaluation of morcellated kidneys after radical nephrectomy. J Urol 164: 2086 –2089, 2000.
145
PROGNOSTIC ASSESSMENT OF NONMETASTATIC RENAL CELL CARCINOMA: A CLINICALLY BASED MODEL OZGUR YAYCIOGLU, WILLIAM W. ROBERTS, THERESA CHAN, JONATHAN I. EPSTEIN, FRAY F. MARSHALL, AND LOUIS R. KAVOUSSI
ABSTRACT Objectives. Determining the recurrence risk in patients treated for renal cell carcinoma (RCC) is important for providing prognostic information and planning potential surveillance strategies. The pathologic stage has been the most widely used single prognostic variable. However, with minimally invasive treatment modalities, the pathologic stage may not be readily available. We developed a biostatistical prognostic model for postoperative RCC that is independent of the pathologic stage. Methods. The records of 296 patients who underwent open nephrectomy for RCC at Johns Hopkins Hospital between 1990 and 1999 were reviewed. Cox proportional hazards regression analysis was used to generate a prognostic model. Results. The recurrence risk (Rrec) was determined from this model: Rrec ⫽ 1.55 ⫻ presentation (0-1) ⫹ 0.19 ⫻ clinical size (in centimeters). Using this equation, 79% of patients were identified as low risk compared with 45% of patients considered low risk by pathologic stage (pT1). Moreover, the separation between the high and low-risk survival curves increased. Conclusions. This model is the first to our knowledge that uses purely clinical variables to assess the postoperative prognosis in patients with RCC. These results, although not validated, provide substantial evidence that preoperative clinical variables may be used instead of the pathologic stage to determine the risk of recurrence. Uncoupling the reliance on pathologic stage for prognostic information removes a potential barrier to novel minimally invasive treatments for renal malignancy and provides a standard to which observation protocols can be compared. In the future, this model may facilitate selection of appropriate patients for less toxic adjuvant or neoadjuvant therapies. UROLOGY 58: 141–145, 2001. © 2001, Elsevier Science Inc.
C
areful assessment of the prognosis in patients after surgery for renal cell carcinoma (RCC) is helpful in selecting the appropriate surveillance protocols. The early detection of recurrence and subsequent aggressive treatment may provide the greatest chance for extended survival, as effective adjuvant therapies are not currently available for patients with locally advanced disease.1 Although many attempts have been made to identify better pathologic and biomolecular prognostic markers, the pathologic stage remains the most widely used From the James Buchanan Brady Urological Institute and Department of Pathology, Johns Hopkins Medical Institutions, Baltimore, Maryland; and Department of Urology, Emory University School of Medicine, Atlanta, Georgia Reprint requests: William W. Roberts, M.D., 600 North Wolfe Street, Suite 161, Jefferson Street Building, Baltimore, MD 21287-8915 Submitted: April 9, 2001, accepted (with revisions): April 23, 2001 © 2001, ELSEVIER SCIENCE INC. ALL RIGHTS RESERVED
prognostic factor for patients with RCC after nephrectomy.2,3 The 1997 TNM staging system has been shown to be a predictor of cause-specific survival and has been the basis of several proposed postoperative surveillance algorithms.4 – 6 Minimally invasive approaches have been increasingly used for the surgical treatment of RCC.7–11 Unfortunately, the very nature of minimally invasive surgery limits the pathologic assessment of the specimen. For example, during laparoscopic nephrectomy, specimen retrieval can be accomplished by en bloc removal, necessitating an expanded incision, or by morcellation through a port site. Morcellation, although preserving the minimally invasive nature of the operation, may potentially interfere with the determination of the exact pathologic stage of the tumor. We sought to develop a biostatistical model based on readily available preoperative clinical variables and independent of the pathologic stage 0090-4295/01/$20.00 PII S0090-4295(01)01207-9 141
to stratify patients by recurrence risk. This model was compared with pathologic staging as a prognostic indicator in patients after surgical treatment for nonmetastatic RCC. MATERIAL AND METHODS The records of all the patients in the Johns Hopkins Cancer Registry who were surgically treated for RCC from January 1990 to December 1999 were retrospectively reviewed. Patients with bilateral synchronous or metastatic disease at the time of surgery or von Hippel-Lindau disease and those who underwent surgery elsewhere were excluded. Patients with radiographic evidence of lymphadenopathy, tumor extension beyond Gerota’s fascia, or vena caval involvement above the diaphragm were also excluded. The patient presentation was categorized as symptomatic or incidental. Tumors that caused pain, hematuria, abdominal mass, or weight loss were identified as symptomatic tumors. All patients had undergone preoperative computed tomography or magnetic resonance imaging, the findings of which were used to assign the clinical stage in accordance with the 1997 TNM staging system.4 The largest diameter of the tumor measured on the computed tomography scan was recorded as the clinical size. Postoperatively, tumors were pathologically staged according to the 1997 TNM system and assigned a Fuhrman nuclear grade. Patients were followed up at 6 to 12-month intervals with a physical examination, routine laboratory evaluation, chest x-ray, and computed tomography scan. Information on patients followed up elsewhere was obtained through the Cancer Registry. The time to recurrence was defined as the interval from surgery to the first evidence of disease recurrence. Kaplan-Meier statistical analyses were performed using Stata 6.0 (Stata, College Station, Tex) to estimate the recurrence-free survival after surgical treatment of RCC. Patients last known alive with recurrence and those dead without recurrence had their follow-up time censored. The Cox proportional hazards regression analysis was used to model the time to recurrence after surgery and to develop an equation for recurrence risk (Rrec), which was used to stratify patients into high and low-risk groups.
RESULTS From 1990 through 1999, 296 patients met the entry criteria and underwent surgery for the treatment of clinically nonmetastatic RCC. The clinical and pathologic information is presented in Table I. The median follow-up time after surgery was 48 months (range 5 to 129). Thirty-eight patients (12.8%) developed recurrences a median of 17 months after surgery (range 1 to 77). The diseasefree survival curves for patients with pathologic Stage T1 tumors versus those with pathologic Stage T2-4 tumors are shown in Figure 1. Presentation, clinical stage, grade, pathologic stage, clinical size, age, and sex were evaluated individually (Table II). They were then considered together using backward stepwise logistic regression analyses to remove the variables that failed to explain clinically important differences in the risk of recurrence (Rrec). The remaining variables (presentation and clinical size) were incorporated into 142
TABLE I. Clinical and pathologic patient characteristics (n ⴝ 296) Sex Male Female Mean age (yr) Nephrectomy type Radical Partial Presentation Symptomatic Incidental Mean clinical size (cm) Clinical T stage (n) T1 T2 T3a T3b Pathologic T stage (n) T1 T2 T3a T3b Tumor grade (n) I II III IV Undetermined*
185 (63) 111 (37) 60.1 ⫾ 12.5 236 (80) 60 (20) 157 (53) 139 (47) 6.0 ⫾ 3.5 186 39 47 24
(62.8) (13.2) (15.9) (8.1)
132 25 106 33
(44.6) (8.4) (35.8) (11.1)
13 169 94 19 1
(4.4) (57.1) (31.8) (6.4) (0.3)
Numbers in parentheses are percentages. * No grade assigned because of extensive tumor necrosis.
the final risk equation: Rrec ⫽ 1.55 ⫻ presentation (0-1) ⫹ 0.19 ⫻ clinical size (in centimeters). Patient presentation, scored as asymptomatic (0) or symptomatic (1), and the clinical tumor size (in centimeters) were inserted into the equation. Each patient received an Rrec score that was used for stratification. Patients with an Rrec score of 3.0 or less were categorized as low risk; those with an Rrec score greater than 3.0 were categorized as high risk. This cutoff was selected to generate a low-risk group with equivalent survival rates to patients with pathologic T1 tumors (specifically 91% 5-year survival and 89% 10-year survival rates). Similar survival analyses have suggested that patients with pT1 disease are at lower risk of recurrence and may be candidates for less rigorous surveillance regimens.5,12 Using this prognostic model, Rrec, 79% of the patients were identified as low risk compared with 45% classified as low risk by the pathologic criteria (pathologic Stage T1). Furthermore, the low and high-risk survival curves were statistically distinct (P ⬍0.001, Wilcoxon test), with survivor functions for the low and high-risk groups of 99% versus 77% at 1 year, 98% versus 67% at 2 years, and 92% versus 57% at 5 years of follow-up. This results in increased separation between the curves UROLOGY 58 (2), 2001
FIGURE 1. Kaplan-Meier disease-free survival curves (based on stratification by pathologic stage) for patients with Stage pT1 and pT2-4 are statistically distinct (P ⬍0.008, Wilcoxon test). The disease-free survivor percentages for patients with Stage pT1 and pT2-4 were 99% versus 91% at 1 year, 98% versus 86% at 2 years, and 91% versus 80% at 5 years of follow-up, respectively.
TABLE II. Results of Cox proportional hazards univariate and multivariate analysis for prediction of recurrence 95% Hazard Standard P Confidence Ratio Error Value Interval Univariate analysis Presentation Clinical stage Grade Pathologic stage Clinical size Age Sex Multivariate analysis Presentation Clinical size
6.17 2.27 2.05 1.93 1.22 1.07 0.63
2.96 0.31 0.45 0.32 0.04 0.17 0.23
0.001 0.001 0.001 0.001 0.001 0.67 0.20
2.41–15.80 1.73–2.98 1.33–3.16 1.39–2.67 1.15–1.30 0.78–1.46 0.31–1.28
4.73 1.21
2.28 0.04
0.001 1.84–12.15 0.001 1.13–1.30
compared with pathologic stratification (Figs. 1 and 2). COMMENT Laparoscopic radical nephrectomy has longterm efficacy comparable to open nephrectomy and provides a shorter and more comfortable convalescence.10,13 New ablative modalities such as raUROLOGY 58 (2), 2001
diofrequency ablation, high-intensity focused ultrasound, cryoablation, and interstitial laser therapy have the potential to cause tumor destruction noninvasively or with only a single needle puncture.11,14 These ablative technologies, however, do not produce a specimen for pathologic analysis, which may be an impediment to their widespread application. For similar reasons, the method of specimen retrieval for laparoscopic radical nephrectomy varies. Intact retrieval preserves the kidney for pathologic analysis but necessitates an extended abdominal incision. With morcellation, the specimen is removed piecemeal through a port site, preserving the minimally invasive nature of the procedure but increasing the difficulty of accurate tumor staging.4 Landman and coworkers15 showed that limited morcellation of radical nephrectomy specimens in vitro did not alter the determination of the histologic features, grade, or local invasiveness of the tumor. Unfortunately, our routine clinical experience shows that this is not always practical with in vivo morcellation. Limited data suggest that morcellation results in decreased postoperative pain and shorter convalescence. Dunn and coworkers8 noted decreased analgesic use and a shorter hospital stay for patients 143
FIGURE 2. Kaplan-Meier disease-free survival curves (based on prognostic model stratification) for low and high-risk clinical groups are statistically distinct (P ⬍0.001, Wilcoxon test). The disease-free survival percentages for the low and high-risk groups were 99% versus 77% at 1 year, 98% versus 67% at 2 years, and 92% versus 57% at 5 years of follow-up, respectively.
having morcellation versus intact removal. Walther and coworkers7 found that the postoperative narcotic requirement was similar for patients who underwent open nephrectomy and laparoscopic nephrectomy with intact removal, but was significantly less for those who had morcellation. In this study, symptomatic presentation and clinical size were the two most important prognostic factors for predicting cancer recurrence and thus were included in the final risk equation. The results of this model, Rrec, were compared with stratification by pathologic stage alone. Interestingly, the pathologic grade when combined with the pathologic stage in a multivariate fashion failed to appreciably improve on the stratification based on the pathologic stage alone. This is probably because 88.9% of the tumors were pathologic grade II/III and the prognostic distinction between grade II and grade III tumors was not statistically significant. Moreover, this model (Rrec) has not been validated, as the sample was not of sufficient size to allow rigorous internal validation. Therefore, external validation will be important in determining the reproducibility of this model. This model classifies a greater number of patients as low risk than does pathologic stratification while achieving the same survival rates. It therefore may allow a greater number of patients to fol144
low less rigorous surveillance protocols after surgery. In the future, this model may also help in selecting candidates to receive new, potentially less toxic, adjuvant and neoadjuvant therapies. CONCLUSIONS Using the Cox proportional hazard regression analysis, we constructed a biostatistical model that stratifies patients according to their risk of recurrence. This model is the first to our knowledge that uses purely clinical variables without reliance on the pathologic stage to assess the postoperative prognosis in patients with RCC. Seventy-nine percent of the patients were classified as low risk, potentially needing less rigorous surveillance postoperatively and unlikely to require future adjuvant treatment. Although not yet validated, these results provide substantial evidence that preoperative clinical variables may be used instead of the pathologic stage in determining the risk of recurrence after surgery for RCC. Uncoupling the reliance on pathologic stage for prognostic information will allow a greater role for minimally invasive treatment of renal malignancies and in the future may help to select appropriate candidates for less toxic adjuvant immunotherapies. UROLOGY 58 (2), 2001
ACKNOWLEDGMENT. To Anne L. Kammer of Johns Hopkins Cancer Registry for providing assistance with data collection. REFERENCES 1. Tanguay S, Pisters LL, Lawrence DD, et al: Therapy of locally recurrent renal cell carcinoma after nephrectomy. J Urol 155: 26 –29, 1996. 2. Thrasher JB, and Paulson DF: Prognostic factors in renal cancer. Urol Clin North Am 20: 247–262, 1993. 3. Gelb AB: Renal cell carcinoma current prognostic factors. Cancer 80: 981–986, 1997. 4. Guinan P, Sobin LH, Algaba F, et al: TNM staging of renal cell carcinoma: Workgroup No. 3. Union International Contre le Cancer (UICC) and the American Joint Committee on Cancer (AJCC). Cancer 80: 992–993, 1997. 5. Levy DA, Slaton JW, Swanson DA, et al: Stage specific guidelines for surveillance after radical nephrectomy for local renal cell carcinoma. J Urol 159: 1163–1167, 1998. 6. Gettman MT, Blute ML, Spotts B, et al: Pathologic staging of renal cell carcinoma: significance of tumor classification with the 1997 TNM staging system. Cancer 91: 354 –361, 2001. 7. Walther MM, Lyne JC, Libutti SK, et al: Laparoscopic cytoreductive nephrectomy as preparation for administration of systemic interleukin-2 in the treatment of metastatic renal cell carcinoma: a pilot study. Urology 53: 496 –501, 1999.
UROLOGY 58 (2), 2001
8. Dunn MD, Portis AJ, Shalhav AL, et al: Laparoscopic versus open radical nephrectomy: a 9-year experience. J Urol 164: 1153–1159, 2000. 9. Janetschek G, Jeschke K, Peschel R, et al: Laparoscopic surgery for stage T1 renal cell carcinoma: radical nephrectomy and wedge resection. Eur Urol 38: 131–138, 2000. 10. Cadeddu JA, Ono Y, Clayman RV, et al: Laparoscopic nephrectomy for renal cell cancer: evaluation of efficacy and safety—a multicenter experience. Urology 52: 773–777, 1998. 11. Gill IS, Novick AC, Meraney AM, et al: Laparoscopic renal cryoablation in 32 patients. Urology 56: 748 –753, 2000. 12. Sandock DS, Seftel AD, and Resnick MI: A new protocol for the followup of renal cell carcinoma based on pathological stage. J Urol 154: 28 –31, 1995. 13. Portis AJ, Yan Y, Chen C, et al: Long-term follow up after laparoscopic radical nephrectomy. J Endourol 14(suppl 1): A31, 2000. 14. Zlotta AR, Wildschutz T, Raviv G, et al: Radiofrequency interstitial tumor ablation (RITA) is a possible new modality for treatment of renal cancer: ex vivo and in vivo experience. J Endourol 11: 251–258, 1997. 15. Landman J, Lento P, Hassen W, et al: Feasibility of pathological evaluation of morcellated kidneys after radical nephrectomy. J Urol 164: 2086 –2089, 2000.
145