What's known on the subject? and What does the study add?
Retrospective data have suggested an increased survival benefit for patients undergoing partial nephrectomy compared to radical nephrectomy, possibly as a result of the avoidance of long-term renalin sufficiency and subsequent sequelae. However, recent level-one evidence has questioned this benefit. Both retrospective studies and randomized controlled trials are not without limitations.
There are few population-based data available with respect to the outcomes of partial nephrectomy vs radical nephrectomy. Additionally, there are no population-based studies analyzing the surgical approach (minimally-invasive vs open), as well as other modalities, such as ablation and surveillance. Finally, there is very little information available on the potential differences in cost for each approach. The present study comprises the first comprehensive population-based analysis of the trends, outcomes and costs of all treatment modalities for T1a renal masses from 2005 to 2007.
To perform a comprehensive analysis of the outcomes and costs for treatments for small renal masses (SRM) using a population-based approach. Partial nephrectomy may be associated with improved survival, although level-one evidence has questioned this survival advantage.
Patients and Methods
Using Surveillance, Epidemiology and End Results–Medicare data, we identified 1682 subjects who were diagnosed with SRM from 2005 to 2007.
Treatment included open radical nephrectomy (ORN; n = 404), minimally-invasive radical nephrectomy (MIRN; n = 535), open partial nephrectomy (OPN; n = 330), minimally-invasive partial nephrectomy (MIPN; n = 160), ablation (n = 211) and surveillance (n = 42).
Postoperative complications, renal insufficiency diagnosis, overall mortality, cancer-specific mortality and postoperative costs were compared.
Covariates were balanced before outcomes analysis using propensity score methods.
Although the use of nephron-sparing surgery (NSS) increased over the study period, radical nephrectomy remained the predominant approach for SRM in 2007.
Minimally-invasive approaches had shorter lengths of stay (P < 0.001), whereas open approaches had more overall complications, respiratory complications and intensive care unit admissions (all P < 0.003).
MIRN and ORN were associated with more peri-operative medical complications, acute renal failure, haemodialysis use and long-term chronic renal insufficiency diagnosis vs NSS (all P < 0.001).
Ablation, MIRN and ORN were associated with the highest overall mortality rates (P < 0.001), whereas MIRN and ORN were associated with the highest cancer-specific mortality rates (P < 0.001).
Treatment costs were lowest for surveillance ($2911) followed by ablation ($10 730), MIRN ($15 373), MIPN ($15 695), OPN ($16 986) and ORN ($17 803).
Although not the predominant treatment approach for SRM over the study period, the use of NSS increased and was associated with improved survival, fewer complications and less renal insufficiency.
The surgical management of RCC has evolved dramatically subsequent to the description by Robson et al.  of radical nephrectomy (RN) with ipsilateral adrenalectomy as the gold standard of care. Although partial nephrectomy (PN) was historically reserved for imperative indications, elective PN became accepted after reports of comparable clinical outcomes and survival in patients with T1 disease [2, 3]. Laparoscopic radical nephrectomy (LRN) was introduced in 1991  and was followed by laparoscopic partial nephrectomy (LPN) in 1992 . Technical demands hindered the widespread application of LPN; however, the addition of robot-assisted partial nephrectomy (RAPN) in the 2000s shortened the learning curve and has led to the increased use of PN . Additionally, ablative techniques and surveillance have been advocated in select patients with small renal masses (SRM) (i.e. <4 cm).
Retrospective data have suggested that nephron-preservation after PN for SRM leads to improved overall survival as a result of decreased renal insufficiency, as well as the subsequent avoidance of renal, cardiovascular and skeletal morbidity [7, 8]. Subsequently, AUA guidelines recommend PN as a standard treatment in patients presenting with T1 RCC when feasible . However, PN still remains underused, potentially because of the ease of radical nephrectomy (RN) and the increased popularity of LRN. Population-based estimates show that only 20% of renal tumours sized 2–4 cm received PN from 2000 to 2001, suggesting a potential quality of care concern . Although recent data from large-volume academic centres have indicated that 90% of patients with T1a disease now undergo PN , this has not been verified within a community setting. Additionally, level-one evidence comparing PN vs RN outcomes has questioned the overall survival benefit conferred by PN , whereas cancer-specific survival outcomes remain unknown.
Despite the uncertainty regarding the best treatment for SRM, data are scarce with respect to contemporary practice patterns, outcomes and costs for the treatment of these patients. In the present study, using a population-based approach, we analyzed the outcomes (including overall and cancer-specific mortality) and costs of all treatment options for SRM ≤4 cm from 2005 to 2007.
Patients and Methods
The present study was approved by the Brigham and Women's Institutional Review Board; patient data were de-identified and requirement for consent was waived. We analyzed Surveillance, Epidemiology and End Results (SEER)–Medicare data, comprising a linkage of population-based cancer registries from 20 SEER areas covering ≈28% of the US population with Medicare administrative data . Medicare provides healthcare benefits to most Americans aged ≥65 years. SEER–Medicare captures ≈97% of incident cancer cases and collects data such as patient demographics, tumour characteristics and the initial course of treatment .
We identified 31 145 subjects aged ≥66 years with the first and only cancer diagnosis of RCC from 2005 to 2007 with follow-up until 2009. After excluding subjects with tumours >4 cm, 18 377 subjects remained for analysis. Subjects were excluded if they were diagnosed at autopsy or death, had Medicare entitlement on the basis of end-stage renal disease, or were not enrolled continuously in both Medicare A and B throughout the study period because outcomes may not be accurately captured, reducing the sample size to 10 675. After excluding patients with multiple renal surgeries or undefined renal procedures within the SEER–Medicare database, 5841 remained. Finally, after excluding patients without a final histological diagnosis, the final study cohort included 1682 subjects.
Using Current Procedural Terminology (4th edition) (CPT-4) codes, we identified 404 open radical nephrectomy (ORN; 50220, 50230), 535 minimally-invasive radical nephrectomy (MIRN; 50546), 330 open partial nephrectomy (OPN; 50240), 160 minimally-invasive partial nephrectomy (MIPN; 50543) and 211 radiofrequency ablation and/or cryoablation (50250, 50542) approaches. Because there is no identifier for robot-assistance during the study period, MIRN and MIPN comprised both laparoscopic and robotic surgical approaches. There were 42 subjects diagnosed with RCC via renal biopsy (CPT 50200, 50205) without subsequent definitive therapy who were classified as undergoing surveillance.
Age (66–69, 70–75, >75 years) was obtained from the Medicare denominator file, whereas race, US Census region, education level, household income, population density (urban vs rural) and marital status were obtained from SEER. Co-morbidity was assessed using the Klabunde modification of the Charlson index based on inpatient, outpatient and physician services the year before RCC diagnosis . Baseline renal insufficiency was defined by International Classification of Diseases (9th Revision) and CPT-4 codes identifying acute renal failure, haemodialysis, diabetic nephropathy, hypertensive nephropathy, chronic renal insufficiency and miscellaneous other renal diseases, as described previously .
Tumour grade (well-, moderately- and poorly-differentiated), clinical stage and histology (clear cell, papillary, chromophobe, other) were obtained from SEER.
We captured complications (cardiac, respiratory or vascular events; genitourinary complications; bleeding; miscellaneous surgical/medical complications; wound infection; and death) and intensive care unit admissions within 30 days of treatment, consistent with previous studies [17, 18]. Hospital length of stay was defined as the interval between hospital admission to discharge. We identified those subjects requiring postoperative ureteric stent placement (CPT-4 52332) and/or nephrostomy tube placement (CPT-4 50393) within 30 days as a surrogate for postoperative urine leak.
Costs associated with each treatment modality were determined by summing all Medicare healthcare expenditures from inpatient, outpatient and physician services within 3 months following treatment and subtracting expenditures in the 3 months before treatment. For patients undergoing surveillance, costs were determined using the date of biopsy as the treatment date. All costs were adjusted to US dollars as of 2007.
Renal insufficiency and overall and disease-specific survival
Survival data were obtained based on date and cause of death provided by Medicare and defined as the interval between surgery/biopsy and death. Postoperative renal insufficiency outcomes were defined by the same codes outlined above. Because overall mortality, cancer-specific mortality and the renal insufficiency diagnosis do not have an upper time limit and the length of follow-up varied, we compared the rate of events occurring per 100 person-years.
Baseline demographic and clinical characteristics were compared using chi-squared tests and were adjusted for using propensity score methods  that enable control for observed confounding factors influencing both group assignment and outcome via a single composite measure and attempt to balance patient characteristics between groups. We used a logistic regression model to calculate the propensity of undergoing the various treatment modalities based on all covariates described above and then weighted each subject's data based on the inverse propensity of being in one of the two treatment groups . Covariate balance was checked after adjustment to ensure that there were no statistically significant differences. All statistical analyses were performed using SAS, version 9.2 (SAS Institute, Cary, NC, USA). P < 0.05 was considered statistically significant.
Patterns of Care for SRM
Figure 1 presents temporal trends in the management of SRM. Although the use of MIRN (34.1% to 30.7%) and ORN (28.7% to 21.2%) decreased over the study period (P < 0.001), RN remained the predominant treatment for SRM in 2007. The use of ablation (6.1% to 15.4%), MIPN (8.3% to 9.9%) and surveillance (2.0% to 3.8%) increased over the study period (P < 0.001), whereas rates of OPN remained relatively stable (20.7% to 20.2%).
Unadjusted demographic and pathological data are presented in Table 1. Subjects undergoing MIPN and OPN were younger (P < 0.001), whereas those undergoing ablation and surveillance were older and had more co-morbidities (both P < 0.001). Additionally, subjects undergoing OPN and surveillance had a higher probability of baseline renal insufficiency (P < 0.001). Subjects undergoing OPN, ORN and MIRN had higher stage tumours (P < 0.001). Ablation, MIRN, ORN and surveillance were more commonly employed in the west, whereas MIPN and OPN were most common in the north-east (P < 0.001).
Table 1. Unadjusted demographic and pathological data
Ablation (n = 211
MIPN (n = 160)
OPN (n = 330)
MIRN (n = 535)
ORN (n = 404)
Surveillance (n = 42)
AJCC, American Joint Committee on Cancer; DS, data suppressed as a result of insufficient numbers; MIPN, minimally-invasive partial nephrectomy; MIRN, minimally-invasive radical nephrectomy; OPN, open partial nephrectomy; ORN, open radical nephrectomy.
Year of diagnosis, n (%)
Age (years), n (%)
Sex, n (%)
Race/ethnicity, n (%)
Charlson co-morbidity score, n (%)
Baseline renal insufficiency (180 days), n (%)
AJCC stage, n (%)
Tumour grade, n (%)
Histology subtype, n (%)
Region, n (%)
Population density, n (%)
Marital status, n (%)
Median household income ($), n (%)
35 000–44 499
45 000–59 999
Propensity-adjusted postoperative complications, renal insufficiency and mortality outcomes are presented in Table 2. Shortest lengths of stay were observed with ablation, MIPN and MIRN (P < 0.001). Overall and respiratory complications, as well as intensive care unit admissions, were most frequent for OPN and ORN (all P < 0.003). MIRN and ORN were associated with more miscellaneous medical complications, peri-operative acute renal failure and haemodialysis (all P < 0.001), whereas MIPN and OPN were associated with more urine leaks (P < 0.001).
DS, data suppressed as a result of insufficient numbers; ICU, intensive care unit; MIPN, minimally-invasive partial nephrectomy; MIRN, minimally-invasive radical nephrectomy; OPN, open partial nephrectomy; ORN, open radical nephrectomy.
Length of stay (days), mean (sd)
30-day postoperative complications, n (%)
Postoperative stent/nephrostomy placement
Acute renal failure
Renal insufficiency 31–365 days post procedure, n (%)
Acute renal failure
Chronic renal failure
Events per 100-person years, n
Renal insufficiency diagnosis
Beyond the peri-operative setting, MIRN (35.6%) and ORN (35.9%) were associated with most frequent chronic renal insufficiency diagnosis (P < 0.001), whereas there were no differences in acute renal failure or haemodialysis. MIRN and ORN were also associated with greater rates of new renal insufficiency diagnosis (38.3 and 35.9 events per 100 person-years, respectively; P < 0.001).
Ablation, MIRN, and ORN were associated with higher overall mortality (P < 0.001), whereas MIRN and ORN were associated with higher cancer-specific mortality (3.4 and 4.7 events per 100 person-years, respectively; P < 0.001). The proportion of overall mortality between RN and PN cohorts over the study period was 27.5% vs 13.7% (P = 0.002).
Table 3 compares median costs within 3 months of treatment. Treatment costs were lowest for surveillance ($2911) followed by ablation ($10 720). MIPN ($15 695) and MIRN ($15 373) were less costly than OPN ($16 986) and ORN ($17 803).
Table 3. Postoperative costs by procedure within 3 months of treatment
MIPN, minimally-invasive partial nephrectomy; MIRN, minimally-invasive radical nephrectomy; OPN, open partial nephrectomy; ORN, open radical nephrectomy.
Total cost ($), mean (sd)
10 720 (14 997)
15 695 (11 564)
16 986 (16 258)
15 373 (13 117)
17 803 (15 217)
2 911 (25 899)
Recent studies illustrate the potential long-term consequences of renal insufficiency after RN for RCC, namely an increased overall mortality compared to PN [8, 21]. Accordingly, the AUA has recommended that PN be considered as a first-line treatment in patients presenting with masses ≤4 cm when feasible . However, recent level-one evidence questioned the overall mortality benefit of PN vs RN, although that trial was underpowered and included tumours >4 cm . Additionally, ablative techniques have become popular as a result of the short-term benefits of less anaesthesia, lower blood loss and less pain. However, concerns regarding long-term oncological efficacy, local recurrence and incomplete tumour ablation  challenge its acceptance as a first-line therapy for patients who are adequate surgical candidates. In the present study, we report a comprehensive population-based approach comparing outcomes and costs of the various treatment modalities available for SRM in a contemporary cohort.
The present study has several important findings. First, it shows a cancer-specific mortality benefit associated with PN and ablative techniques vs RN for SRM, regardless of approach. This is the first study to find report a cancer-specific survival benefit among nephron-sparing approaches. The results of the present study support previously published findings showing that PN confers an overall mortality benefit compared to RN, perhaps as a result of the better preservation of overall renal function and avoidance of subsequent sequelae [8, 21]. Using a similar propensity-matched population from 1998 to 2005, Sun et al.  found a decrease in other-cause mortality for PN vs RN. More recently, Tan et al.  reported a benefit in overall survival for patients undergoing PN vs RN. However, these studies did not examine the effect of surgical approach (minimally-invasive vs open) on these outcomes, and ablative techniques were not included, whereas the present study represents a more contemporary cohort and provides temporal trends in management. Moreover, previous studies have not found a significant cancer-specific mortality benefit comparing PN vs RN. These results are somewhat unexpected, and may be a result of differences in the baseline and pathological characteristics between PN and RN cohorts. However, the present study did control for such differences using propensity-score methods.
The findings of the present study differ from those of Van Poppel et al. , who reported a greater overall mortality in subjects undergoing PN vs RN. However, when the analyses were adjusted to include only those subjects with RCC, this mortality difference became insignificant. The trial of Van Poppel et al.  has been criticized for being underpowered (n = 541) and for a lack of homogeneity, with a mean of one patient per institution per year. Finally, although ablation was associated with a greater overall mortality, this may be the result of more co-morbidities in this treatment group, and these findings did not extend to cancer specific mortality. The mortality benefit shown in the present study contributes to the growing evidence that nephron-sparing surgery (NSS) is preferable whenever technically possible.
Second, NSS vs RN, regardless of surgical approach, was associated with decreased rates of peri-operative acute renal failure and haemodialysis, as well as lower rates of long-term postoperative renal insufficiency. This is consistent with the findings reported by Huang et al. , who reported increased renal insufficiency with RN vs PN, as measured by postoperative glomerular filtration rate . Miller et al.  studied SEER–Medicare and also found fewer adverse renal outcomes (less dialysis and renal transplant) associated with PN vs RN . Finally, Sun et al.  also noted a higher incidence of renal insufficiency and anaemia after RN vs PN. The avoidance of postoperative renal insufficiency is crucial because chronic kidney disease is significantly associated with increased cardiovascular and all-cause death in a dose-dependent manner . This further highlights the benefit of NSS.
Third, MIPN and MIRN were associated with shorter lengths of stay, fewer postoperative complications and fewer intensive care unit admissions compared to their open counterparts. Therefore, although the use of robotic-assistance may be cost-prohibitive and criticized for other procedures, this approach leads to fewer complications than open procedures after renal surgery. LRN has long been associated with improved outcomes compared to ORN. Sun et al.  similarly noted fewer inpatient complications after LPN vs OPN using population-based data, although tumour characteristics were not accounted for. Simhan et al.  found that patients undergoing RAPN vs OPN for complex renal masses had comparable outcomes but shorter hospital stays. Finally, Benway et al.  reported less blood loss, shorter operating times and an improved warm ischaemia time for RAPN vs LPN Similar to the study of Sun et al. , however, the present study complements these findings in a national rather than institutional cohort.
Fourth, we observed lower costs for minimally-invasive vs open procedures. ORN and OPN were both more costly than their minimally-invasive counterparts, whereas ablation and surveillance remain the least costly. In the present study, fewer postoperative complications may have contributed to the observed lower costs compared to open approaches. Park et al.  noted only slightly higher costs for LPN vs OPN ($8450 vs $8019), noting that lower complication rates offset the costs of laparoscopic equipment. Comparing RAPN, LPN and OPN costs, Mir et al.  noted that LPN was cheaper ($10 311) than OPN ($11 427) and RAPN ($11 962). Most recently, Anderson  found no differences in costs between RAPN, OPN and ORN, concluding that fewer complications offset the cost of newer technology. Although the overall costs in the present study are higher than those of previous studies, we assessed costs up to 3 months after surgery and therefore account for readmissions, as well as costs incurred as a result of subsequent medical problems, such as renal insufficiency. The findings reported in the present study provide further evidence that minimally-invasive surgery may lead to lower overall costs as a result of fewer complications, as well as shorter lengths of stay. However, it should be noted that these are costs within the US healthcare system and may not necessarily translate to medical costs outside of the USA.
Fifth, although MIRN and ORN remained the predominant treatment modality for SRM over the study period, their use decreased over time, whereas the use of surveillance and NSS increased. Overall, RN use decreased from 62.8% to 51.9%, whereas NSS increased from 35.1% to 45.5% from 2005 to 2007. This is probably a reflection of more urologists accepting the benefits of nephron-preservation in light of the aforementioned studies indicating an overall benefit, as well as more urologists adhering to suggested guidelines as they become more comfortable with performing NSS. However, the use of OPN remained relatively stable, possibly reflecting increasing comfort with minimally-invasive approaches. The findings of the present study also show that the use of PN has increased after 2000–2001 , at which time PN was performed for only 42% of renal tumours <2 cm and 20% of tumours in the size range 2–4 cm. However, the population-based results of the present study also show that PN is used less frequently relative to estimates of 90% PN use for SRM in high-volume tertiary care centres . This disparity in rates of NSS between tertiary care centre data and population-based data warrants further study.
The findings of the present study must be considered in the context of the study design. First, an observational study design was used and therefore we were unable to control for unmeasured confounders. Second, the study was limited to Medicare beneficiaries aged >65 years and may not be generalizable to younger patients. However, as noted previously, the greatest increase in the incidence of SRM has been noted in patients aged ≥70 years, making our findings even more pertinent . Third, administrative data designed for billing purposes may lack detailed clinical information; however, Medicare has a high sensitivity with respect to capturing both surgical complications, as well as renal function outcomes. Fourth, with our cohort being treated between 2005 and 2007, with follow-up to 2009, our oncological follow-up in this cohort is relatively short. However, even within this short period of time, differences in cancer-specific and overall mortality were found, and these results may very well be more pronounced with a longer follow-up. Finally, the cost analysis in the present study covers up to 3 months after surgery. With a longer follow-up, RN medical costs may differ significantly given higher rates of renal insufficiency.
In conclusion, NSS is becoming more popular for the treatment of SRM and is associated with an improved overall and cancer-specific mortality, as well as fewer postoperative complications, including a lower risk of renal insufficiency. The present study complements other studies suggesting the overall survival benefits of NSS, although it is the first to identify a cancer-specific benefit. Furthermore, minimally-invasive approaches for SRM confer lower costs, resulting in significant healthcare cost savings.
Jim C. Hu receives salary support from Department of Defense Physician Training Award W81XWH-08-1-0283.