The first 2 authors contributed equally to this work.
Survival rates after resection for localized kidney cancer: 1989 to 2004
Article first published online: 9 MAY 2008
Copyright © 2008 American Cancer Society
Volume 113, Issue 1, pages 84–96, 1 July 2008
How to Cite
Russo, P., Jang, T. L., Pettus, J. A., Huang, W. C., Eggener, S. E., O'Brien, M. F., Karellas, M. E., Karanikolas, N. T. and Kagiwada, M. A. (2008), Survival rates after resection for localized kidney cancer: 1989 to 2004. Cancer, 113: 84–96. doi: 10.1002/cncr.23520
- Issue published online: 20 JUN 2008
- Article first published online: 9 MAY 2008
- Manuscript Accepted: 19 FEB 2008
- Manuscript Revised: 13 JAN 2008
- Manuscript Received: 18 OCT 2007
- National Institutes of Health Ruth Kirchstein National Research Service Award. Grant Number: T32 CA 82088-07
- lymph node dissection;
- overall survival;
- pathologic tumor classification;
- progression-free survival;
- tumor size
Mortality rates from kidney cancer have continued to rise despite increases in the detection of smaller renal tumors and rates of renal surgery. To explore the factors associated with this treatment-outcome discrepancy, the authors evaluated how changes in tumor size have affected disease progression in patients after nephrectomy for localized kidney cancer, and they sought to identify the factors associated with disease progression and overall patient survival after resection for localized kidney cancer.
In total, 1618 patients with localized kidney cancer were identified who underwent nephrectomy at Memorial Sloan-Kettering Cancer Center from 1989 to 2004. Patients were categorized by year of surgery: from 1989 to 1992, from 1993 to 1996, from 1997 to 2000, and from 2001 to 2004. Tumor size was classified according to the following strata: <2 cm, from 2 cm to 4 cm, from 4 cm to 7 cm, and >7 cm. Disease progression was defined as the development of local recurrence or distant metastases. Five-year progression-free survival (PFS) was calculated for patients in each tumor size strata according to the year of operation using the Kaplan-Meier method. The patient-, tumor-, and surgery-related characteristics associated with PFS and overall survival (OS) were explored using univariate analysis, and all significant variables were retained in a multivariate Cox regression analysis.
Overall, the number of nephrectomies increased for all tumor size categories from 1989 to 2004. A tumor size migration was evident during this period, because the proportion of patients with tumors <2 cm and with tumors from 2 cm to 4 cm increased, whereas the proportion of patients with tumors >7 cm decreased. One hundred seventy-nine patients (11%) developed disease progression after nephrectomy. Sixteen patients (1%) developed local recurrences, and 163 patients (10%) developed distant metastases. When 5-year PFS was calculated for each tumor size strata according to 4-year cohorts, trends in PFS did not improve or differ significantly over time. Compared with historic cohorts, patients in more contemporary cohorts were more likely to undergo partial nephrectomy rather than radical nephrectomy and were less likely to undergo concomitant lymph node dissection and adrenalectomy. Multivariate analysis demonstrated that pathologic stage and tumor grade were associated with disease progression, whereas patient age and tumor stage were associated with overall patient survival.
Despite an increasing number of nephrectomies and a size migration toward smaller tumors, trends in 5-year PFS and OS did not improve or differ significantly over time. These findings require further research to identify causative mechanisms, and they argue for the consideration of active surveillance for patients who have select renal tumors and a re-evaluation of the current treatment paradigm of surgically removing solid renal masses on initial detection. Cancer 2008. © 2008 American Cancer Society.
Kidney cancer is the third most common genitourinary tumor, with 51,190 new cases and 12,890 deaths estimated for 2007.1 Incidence rates have increased steadily since the 1970s, in part because of the widespread use of noninvasive imaging modalities, such as ultrasonography (US), computed tomography (CT), and magnetic resonance imaging (MRI).2, 3
Single-center, multi-institutional, and national incidence trends have noted a higher proportion of kidney cancers diagnosed at smaller tumor sizes and at earlier, presymptomatic stages.2, 4-8 Furthermore, the rising incidence of kidney cancer largely has been attributed to an increase in small kidney tumors.9 Despite these findings and the concurrent increases in rates of renal operations for small renal masses that presumably are curable, kidney cancer mortality rates, paradoxically, have continued to rise.9 To explore the factors associated with this treatment-outcome discrepancy, we sought to describe the changes in tumor size of localized kidney cancers in patients who presented at our institution from 1989 to 2004, to evaluate the impact of size migration on progression-free survival (PFS) trends after nephrectomy for localized disease, to describe trends in surgery-related and tumor-related characteristics, and to identify patient demographic and clinical characteristics associated with disease progression and overall survival (OS).
MATERIALS AND METHODS
Patients and Variables
From January 1989 to December 2004, we identified 1618 patients who underwent radical nephrectomy (n = 1050) or partial nephrectomy (n = 568) for clinically localized kidney cancer at Memorial Sloan-Kettering Cancer Center. Pathologic tumor size was classified according to the following strata: <2 cm, from 2 cm to 4 cm, from 4 cm to 7 cm, and >7 cm. Patients were categorized by year of operation according to the following 4-year cohorts: from 1989 to 1992, from 1993 to 1996, from 1997 to 2000, and from 2001 to 2004. Tumor stage was determined according to the 2002 American Joint Committee on Cancer staging system.10 All pathologic subtypes of renal cell carcinoma were included. Fuhrman tumor grade was defined as the worst grade within a tumor. Data regarding patient age, sex, race, tumor laterality, histology, year of surgery, type of surgery, surgery approach, whether patients underwent concomitant adrenalectomy, and disease progression status were available for all 1618 patients. Data concerning surgical margins and whether patients underwent concomitant lymph node dissection at the time of nephrectomy were unavailable in 24 patients and 14 patients, respectively, from the 1989 to 1992 cohort. Progression was defined as the development of local recurrence or distant metastases. Patients who had bilateral masses at diagnosis were excluded. Metachronous disease in the contralateral kidney was not considered a recurrence.
Summary statistics were described using frequencies and proportions for categorical variables. We used chi-square tests as appropriate to examine whether select demographic or clinical patient characteristics were associated with tumor size. PFS was defined as the time from nephrectomy to disease progression. OS was defined as the time from nephrectomy to the date of death. Patients were censored at the date of the last follow-up if they were free of disease (when determining PFS) or if they were alive (when determining OS). The median follow-up after surgery was 50.4 months (interquartile range, 61.3 months). Five-year PFS rates were calculated for patients in each tumor size strata, according to the year of surgery, using the Kaplan-Meier method, and trends in PFS were compared using the log-rank test. Patient-, tumor-, and surgery-related characteristics that were associated with PFS and OS were explored first using univariate analysis, and all significant variables were retained subsequently in a multivariate Cox regression analysis. P values were 2-sided and were considered statistically significant at <.05. All statistical analyses were performed using Stata 8.2 software (Stata Corp., College Station, Tex).
Patient-, Tumor-, and Surgery-related Characteristics at the Time of Nephrectomy
Demographic and clinical characteristics of cohort members are shown in Table 1 according to tumor size. There were no significant differences among tumor size groups with respect to patient age or tumor laterality. Across all tumor size strata, >85% of patients were white (P = .019), and the majority of patients (66%–73%) had clear cell tumor histology (P = .002). Larger tumor size was associated with higher Fuhrman grade.
|Characteristic||Tumor size, cm|
|<2 (n = 167)||2-4 (n = 632)||>4-7 (n = 453)||>7 (n = 366)|
|% of patients||% of patients||% of patients||% of patients||P|
|Mean age ± SD, y||60.4 ± 12.2||61.3 ± 11.8||62.1 ± 12.2||61.3 ± 12.5|
|Type of surgery|
|Surgical margin status||.004|
|Lymph node dissection||<.0001|
Patients who had smaller tumors were more likely to undergo partial nephrectomy than patients who had larger tumors. For instance, among patients with tumors that measured <2 cm and tumors that measured from 2 cm to 4 cm, 82% and 55%, respectively, underwent partial nephrectomy (P < .0001). Patients who had larger tumors were more likely to undergo concomitant adrenalectomy and/or lymph node dissection. For example, among patients who had tumors that measured <2 cm, 11% and 8% underwent adrenalectomy and lymph node dissection, respectively, compared with 75% and 60%, respectively, of patients who had tumors that measured >7 cm (adrenalectomy, P < .0001; lymph node dissection, P < .0001).
Trends in Tumor- and Surgery-related Characteristics
Tumor size and tumor stage distributions
Figure 1a,b demonstrates the changes in tumor size and tumor stage distributions, respectively, of localized kidney cancers from 1989 to 2004. A tumor size migration was evident during this period (Fig. 1a), because the proportion of patients with tumors that measured <2 cm and tumors that measured from 2 cm to 4 cm increased 9% and 11%, respectively, whereas the proportion of patients with tumors that measured from 4 cm to 7 cm and tumors that measured >7 cm decreased 5% and 15%, respectively, between the periods 1989 to 1992 and 2001 to 2004 (P < .0001). Similarly, a stage migration toward T1a lesions was observed (Fig. 1b). For example, the proportion of patients with T1a tumors increased 23%, whereas the proportion of patients with T1b and T2 tumors decreased 3% and 8%, respectively, between the periods 1989 to 1992 and 2001 to 2004 (P < .0001). However, the correlation between tumor size and pathologic tumor stage was imperfect. Among patients with tumors that measured <2 cm and tumors that measured from 2 cm and 4 cm, 93% and 90%, respectively, were classified as pathologic T1a tumors (data not shown).
Rates of renal surgery
Figure 2 demonstrates the changes in the number of radical and partial nephrectomies performed from 1989 to 2004. The number of renal operations (radical and partial nephrectomy) increased with time. For patients with localized tumors who presented for nephrectomy from 1989 to 1992, radical nephrectomy comprised 88% of renal operations, whereas partial nephrectomy comprised 12%. In the later years, partial nephrectomy became the predominant renal operation performed at our institution for localized kidney tumors. For instance, partial nephrectomy comprised 53% of all renal operations from 2001 to 2004 compared with 47% for radical nephrectomy.
Renal operations were performed laparoscopically first in 2002. Laparoscopic radical nephrectomies comprised 1.9%, 14.8%, and 6.7% of renal operations performed in 2002, 2003, and 2004, respectively, whereas laparoscopic partial nephrectomies comprised 0.6%, 2.1%, and 3.3%, respectively (data not shown).
Other tumor- and surgery-related characteristics
Tumor histology did not vary significantly with time (Fig. 3a). Clear cell tumors predominated in each time era, approximating 70% of all localized tumors, irrespective of the year of surgery. Tumors with papillary histology comprised 13% to 19% of all tumors, depending on the year of surgery, followed by chromophobe tumors, which comprised 6% to 13% of all tumors (P = .099).
In general, patients who underwent nephrectomy in more contemporary eras were less likely to undergo a concomitant lymph node dissection (Fig. 3b) and adrenalectomy (Fig. 3c) compared with cohorts from earlier eras. When excluding historic data with incomplete ascertainment (ie, the 1989-1992 cohort), 37% and 73% of patients from the 1993 to 1996 cohort underwent lymphadenectomy and adrenalectomy, respectively, compared with 27% and 23% of patients, respectively, from the 2001 to 2004 cohort.
Overall, 179 patients (11%) developed disease progression after nephrectomy, including 16 patients (1%) who developed an isolated local recurrence and 163 patients (10%) who developed distant metastasis. Among the patients who underwent radical nephrectomy, 161 of 1050 patients (15%) had disease progression compared with 18 of 568 patients (3%) who had disease progression after partial nephrectomy (P < .0001)
The PFS for the entire cohort of patients who presented with localized renal cortical tumors, stratified by 4-year cohorts, is shown in Figure 4a. The actuarial 5-year PFS rates after nephrectomy (Table 2) were 88%, 84%, 89% and 90% for patients treated from 1989 to 1992, from 1993 to 1996, from 1997 to 2000, and from 2001 to 2004, respectively (P = .309).
|Characteristic||No. of patients||5-year PFS||5-year OS|
|No. of events||%||95% CI||P||No. of events||%||95% CI||P|
|Age at surgery, y||.8315||<.0001|
|Type of surgery||<.0001||.0004|
|Surgical margin status||.1176||.1403|
|Lymph node dissection||<.0001||.0093|
Figure 4b-e shows the 5-year PFS for patients with renal tumors that measured <2 cm, from 2 cm to 4 cm, from 4 cm to 7 cm, and >7 cm. There were no significant differences in 5-year PFS with respect to tumor size over time.
On univariate analysis (Table 2), factors that were associated with disease progression included patient race (P < .0001), tumor stage (P < .0001), Fuhrman tumor grade (P < .0001), tumor histology (P < .0001), type of operation (P < .0001), having undergone adrenalectomy (P < .0001), and having undergone lymphadenectomy (P < .0001). When significant variables were retained in a multivariate Cox regression analysis (Table 3), only tumor stage and Fuhrman tumor grade were associated with disease progression. Compared with patients who had T1 tumors, the odds of having disease progression were 4.5 for patients who had T2 tumors (95% confidence interval [95% CI], 2.6-7.8; P < .0001), 2.3 for patients who had T3 tumors (95% CI, 1.4-3.9; P = .001), and 11.4 for patients who had T4 tumors (95% CI, 3.5-37.2; P < .0001). Compared with patients who had Fuhrman grade 1 tumors, the odds of having disease progression were 2.7 for patients who had grade 3 tumors (95% CI, 1.0-7.1; P = .043) and 4.0 for patients who had grade 4 tumors (95% CI, 1.3-12; P = .015). There were no significant differences between the likelihood of disease progression for patients with grade 1 and grade 2 tumors. Figure 5a,b illustrates 5-year PFS according to tumor stage and Fuhrman grade, respectively.
|Patient characteristics||No. of events||OR (95% CI)||P|
|Clear cell||143||1.0 (Referent)|
|Type of surgery|
|Radical nephrectomy||161||1.0 (Referent)|
|Partial nephrectomy||18||0.69 (0.32–1.47)||.337|
|Lymph node dissection|
In total, 376 patients (23%) died from any cause after nephrectomy at a median follow-up of 50.4 months (interquartile range, 61.3 months). The actuarial 5-year OS rates after nephrectomy (Table 2) were 83%, 80%, 84%, and 79% for patients treated from 1989 to 1992, from 1993 to 1996, from 1997 to 2000, and from 2001 to 2004, respectively (P = .0811).
On univariate analysis (Table 2), factors that were associated with OS included patient age (P < .0001), tumor stage (P < .0001), Fuhrman tumor grade (P < .0001), tumor histology (P < .0001), type of operation (P = .0004), having undergone adrenalectomy (P = .0034), and having undergone lymphadenectomy (P = .0093). When significant variables were retained in a multivariate Cox regression analysis (Table 4), only patient age and tumor stage were associated with OS. Patients aged ≥70 years were at greater odds of dying from any cause after renal surgery than patients aged <70 years. Compared with patients who had T1 tumors, the odds of dying from any cause were 1.7 for patients who had T2 tumors (95% CI, 1.1-2.6; P = .028), 1.4 for patients who had T3 tumors (95% CI, 1.0-1.9; P = .037), and 5.3 for patients who had T4 tumors (95% CI, 2.0-14.2; P = .001). Figure 6a,b illustrates 5-year OS according to patient age and tumor stage, respectively.
|Patient characteristic||No. of events||OR (95% CI)||P|
|Patient age, y|
|Clear cell||283||1.0 (Referent)|
|Collecting duct||2||7.10 (0.95–53.2)||.057|
|Type of surgery|
|Radical nephrectomy||312||1.0 (Referent)|
|Partial nephrectomy||62||0.86 (0.55–1.33)||.492|
|Lymph node dissection|
The treatment paradigm of surgically removing solid renal masses on detection recently has been questioned. Investigators have demonstrated that the rising incidence of kidney cancer in the past 20 years largely is attributable to the increased detection of small renal masses; and, despite parallel increases in surgical treatment, mortality rates for kidney cancer have continued to rise.9 Furthermore, detection of advanced tumors, including those with regional involvement and distant metastasis, also has increased in incidence, calling into question whether incidental detection has had a positive impact on mortality and whether the increased detection of asymptomatic tumors by imaging procedures can explain fully the rising incidence trends of kidney cancer.2, 11
A prior study demonstrated that the greatest absolute increase in kidney cancer mortality is in patients with lesions >7 cm. Although cancer-specific mortality rates in that study also rose in patients who had tumors <2 cm, from 2 cm to 4 cm, and from >4 cm to 7 cm, the increases were not as pronounced as among patients who had lesions >7 cm.9
However, other investigators have argued that survival rates, compared with mortality rates, should reflect more accurately the clinical course of patients after a cancer diagnosis. By using data from the Surveillance, Epidemiology, and End Results (SEER) Program, those investigators observed improved 5-year relative survival rates over time for patients with kidney cancer, particularly for patients with small tumors.12
Similar to reported national trends,9 we demonstrate a size migration toward smaller tumors (Fig. 1a) and an increasing number of nephrectomies performed from 1989 to 2004 (Fig. 2). Presumably, mortality rates from kidney cancer would be even higher if earlier and incidental detection were not so common.2, 11 However, survival rates did not improve or differ significantly over time at our institution (Fig. 4a). To account for these observed trends in PFS, we evaluated changes in tumor- and surgery-related factors with respect to time. Whereas Fuhrman tumor grade (Fig. 3d) and histology (Fig. 3a) did not vary significantly over time, we observed a stage migration toward T1a lesions (Fig. 1b). Coincident with the favorable stage migration, the proportion of patients who underwent concomitant lymphadenectomy and adrenalectomy at the time of nephrectomy decreased significantly over time (Fig. 3b,c, respectively) when excluding historic data with incomplete ascertainment (ie, the 1989-1992 cohort). The observable trends in tumor- and surgery-related characteristics with respect to time could not account for why PFS did not improve or differ significantly over time despite a favorable size migration and an increase in the number of operations performed.
In contrast to survival data from the SEER Program,12 it is conceivable that more advanced or aggressive tumors are being referred to tertiary care centers for management, that patients who were treated in more contemporary eras at our center had more stringent follow-up and, thus, increased attribution of death to kidney cancer than in the past, or that selection of indolent tumors for observation at tertiary care centers has led to more aggressive tumors presenting for surgery. This may account in part for the differences in 5-year survival rates observed in our data compared with data from the SEER Program. Finally, other unidentified factors may be contributory. For example, it is unclear whether certain environmental, metabolic, or epidemiologic factors, such as obesity, hypertension, smoking, or race, either alone or in a multivariate setting, may have had a negative influence on the biologic behavior of kidney cancers in the past 2 decades and whether patients with more comorbid conditions are more likely to be treated at tertiary care centers. Previously, we reported that, although body mass index was associated with a greater proportion of clear cell histology, comorbidity, and surgical morbidity, it did not have an adverse impact on OS or PFS.13
Factors that were associated with the progression of disease after surgical resection included higher tumor stage and higher Fuhrman grade. Patient age, sex, race, tumor laterality, tumor histology, year of surgery, type of nephrectomy (radical vs partial), surgical margin status, or undergoing concomitant adrenalectomy or lymph node dissection were not associated with disease progression.
Investigators previously noted sex differences in kidney cancer-specific survival after treatment.14, 15 On univariate analysis, we observed that, at 5 years after nephrectomy for localized disease, men were at no greater risk of developing disease progression than women (5-year PFS rate: 86% for men vs 91% for women; P = .0562). However, this trend approached statistical significance; and, with longer follow-up, we may anticipate that men will be more likely to develop disease progression than women. The main factors that contribute to a less favorable prognosis in men are the higher proportion of advanced-stage disease and the higher proportion of higher grade disease in men compared with women. For instance, 64%, 9%, 26%, and 1% of men, compared with 70%, 10.6%, 19%, and 0.4% of women, had pathologic stage T1, T2, T3, and T4 tumors, respectively (P = .009). Fuhrman grade 1, 2, 3, and 4 disease was identified in 7%, 63%, 26%, and 4% of men, respectively, compared with 11%, 70%, 17%, and 2% of women, respectively (P< .0001).
Finally, 5-year OS did not differ for patients who were treated in more contemporary eras compared with historic cohorts (Table 2) and ranged from 79% to 84%. On multivariate analysis, patient age and tumor stage were the only factors associated with OS. Although it was beyond the scope of the current study, competing-cause mortality rises with increasing patient age for patients with renal masses16 and must be taken into consideration during surgical planning and in the selection of appropriate candidates for active surveillance as the initial therapeutic approach for select tumors. Because comorbid conditions such as chronic kidney disease have been associated with increased cardiovascular morbidity and mortality17 and its prevalence in patients with renal cortical tumors has been reported to be higher than was believed previously,18 maximal renal functional preservation is essential. Limitations of our study include selection bias associated with referrals to a tertiary care center, the retrospective nature of the review, and the finding that some patients remain at risk for a recurrence at a median follow-up of 50 months. Perhaps cancer-specific survival would have been a more meaningful clinical endpoint than PFS. However, given the relatively few events (n = 68) available to evaluate cancer-specific survival and the finding that metastatic disease uniformly is deadly in this disease because of the lack of effective systemic therapies, the use of PFS as a proxy for kidney cancer mortality seems reasonable.
The current study findings provide clues for further research into the prognosis for patients with kidney cancer who undergo surgery. Furthermore, the finding that PFS rates did not improve or differ significantly with time despite a favorable tumor size migration and increased numbers of nephrectomies performed requires further research to identify causative mechanisms and argues for the consideration of active surveillance for patients with select renal tumors and a re-evaluation of the current treatment paradigm of surgically removing solid renal masses on initial detection.
- 10GreeneFL,PageDL,FlemingID, et al, eds. AJCC Cancer Staging Manual,6th ed. New York, NY: Springer-Verlag; 2002.