Partial Nephrectomy Is Associated with Improved Overall Survival
Compared to Radical Nephrectomy in Patients with Unanticipated
Benign Renal Tumours
Christopher J. Weight, Gregory Lieser, Benjamin T. Larson, Tianming Gao, Brian R. Lane,
Steven C. Campbell, Inderbir S. Gill, Andrew C. Novick, Amr F. Fergany*
Glickman Urologic and Kidney Institute, Cleveland Clinic, Cleveland, Ohio, USA
EUROPEAN UROL OG Y 58 (2010) 293–298
available at www.sciencedirect.com
journal homepage: www.europeanurology.com
Accepted April 29, 2010
Published online ahead of
print on May 18, 2010
europeanurology to read and
answer questions on-line.
The EU-ACME credits will
then be attributed
Background: Partial nephrectomy (PN) has been associated with improved overall
survival (OS) in select cohorts with localised renal masses when compared to radical
nephrectomy (RN). The driving forces behind these differences have been difficult to
elucidate given the heterogeneity of previously compared cohorts.
Objective: Compare OS in a subset of patients with unanticipated benign renal
masses to minimise the confounding effect of cancer.
Design, setting, and participants: We retrospectively evaluated 2608 consecutive
clinical T1 enhancing renal masses that were treated with extirpative surgery at our
institution between 1999 and 2006. Of these, 499 tumours (19%) were found to be
benign on final pathology. Preoperative data and renal functional data were used to
generate a propensity model that was then plugged into a multivariate model of
survival. Median follow-up fortheentire cohort was 50 mo (interquartile range [IQR]:
Intervention: All patients underwent PN or RN.
Measurements: We measured OS and cardiac-specific survival.
Results and limitations: Five-year OS estimates for the PN (n = 388) and RN (n = 111)
cohorts were 95% (95% confidence interval [CI], 93–98) versus 83% (95% CI, 74–90),
respectively (P < 0.0001). On multivariate analysis, controlling for both comorbidity
and age, RN was associated with a 2.5-fold increased risk of death compared to PN
(hazard ratio [HR]: 2.5; 95% CI, 1.3–5.1). Postoperativeestimated glomerular filtration
rate(eGFR) was also anindependentpredictor ofOSandcardiac-specificsurvival(HR:
nature of this analysis limits the strength of the conclusions.
Conclusions: PN was associated with better OS when compared to RN in patients
with unanticipated benign tumours. This observed survival advantage appears partly
to be the result of better preservation of eGFR, but other kidney functions or
unmeasured factors may also play a role. These data indicate that PN should be
aggressively pursued in any patient where PN is technically feasible.
# 2010 European Association of Urology. Published by Elsevier B.V. All rights reserved.
Cleveland, OH 44195, USA. Tel. +216 444 0414.
E-mail address: Fergana@ccf.org (A.F. Fergany).
0302-2838/$ – see back matter # 2010 European Association of Urology. Published by Elsevier B.V. All rights reserved.doi:10.1016/j.eururo.2010.04.033
A good deal of research has attempted to compare survival
in patients undergoing extirpative surgery for localised
renal masses treated by either partial nephrectomy (PN) or
radical nephrectomy (RN) [1–8]. These comparisons have
been challenging to interpret because of small numbers,
found PN to be associated with better overall survival (OS)
and have hypothesised that this result is because of better
preservation of renal function, the strength of these claims
has been limited by the lack of renal function outcomes
in the models and the preponderance of larger, more
aggressive tumours found on final pathology in the RN
cohort [1,3,6,9]. We therefore set out to study the effect of
surgerytype on cardiac-specific survival and OS in a cohort
of patients who underwent surgery for enhancing renal
masses suspected of being malignant but on final
pathology were benign. By focusing on the unanticipated
benign renal tumours, we have removed the confounding
effect of cancer-specific death from our analysis and can
better understand whether PN—and its attendant preser-
vation of renal function—contributes to a survival benefit
compared to RN.
2.1.Patient data and follow-up
From 1999 to 2006, 2608 patients with an enhancing cT1 renal mass
underwent surgery. Though certain imaging and patient characteristics
heighten our suspicion of a benign renal tumour, we have tended to
regard enhancing renal masses as likely malignant and proceed with
extirpation in most cases. Of these patients, 499 (19%) were found to
have unanticipated benign tumours, including all patients either treated
by RN (n = 111) or PN (n = 388). Perioperative and pathologic data were
obtained from our internal review board–approved institutional kidney
cancer patient registry. Estimated glomerular filtration rate (eGFR) was
calculated using the abbreviated Modification of Diet in Renal Disease
(MDRD) study equation (GFR in ml/min per 1.73 m2= 186 ? serum
creatinine [sCr] ? 1.154 ? age ? 0. 203 ? (0.742 if female) ? (1.210 if
Choice of extirpative surgery was left to surgeon and patient
preference after consideration of tumour size, radiographic appear-
ance, overall patient health, life expectancy, and surgeon comfort.
Comorbidity was evaluated using the Charlson-Romano Comorbidity
Index or the Age-adjusted Charlson Comorbidity Index as indicated.
For each patient, vital status was obtained using the Social Security
death index, and cause-of-death information was determined by
reviewingthe patient’s medical records and information obtained from
the National Death Index. Patients without a Social Security number
were excluded from the analysis. Cardiac-related deaths included
deaths attributable to ischemic heart disease, congestive heart disease,
ischemic stroke, and peripheral vascular disease (International
Classification of Disease, 9th Edition [ICD-9] codes 398, 402, 410-1,
414, 428, 433-4, 436, and 440; ICD, 10th Edition (ICD-10) codes I10-3,
I20-I25, I46, I48, I50-1, I63-4, I69-71, and I73). Renal failure deaths
were those deaths in which renal failure, renal insufficiency, or
nephritic syndrome played a role in the patient’s death as determined
by the physician pronouncing the patient dead and including the codes
Because these groups were not randomised, we used a propensity score
model, including only preoperative clinical parameters to control for
selection bias. The propensity to undergo PN was calculated using a
multivariable logistic regression model and the following predictive
variables: age, tumour size, surgery type (laparoscopic vs open), solitary
kidney status, and Charlson Comorbidity Index. Each patient was then
assigned a propensity score based on these preoperative characteristics.
The patients were stratified into similar risk quintiles based on
propensity scores and compared according to treatment group within
each quintile in order to calculate a hazard ratio (HR) of OS.
A Cox multivariate hazard analysis of OS was performed using both
renal function outcomes and propensity scores to attempt to control for
both pre- and postoperative predictors of survival. Cardiac-specific
survival analysis was limited by only 14 events, allowing only univariate
Proportions were analysed by x2or Fisher exact test as indicated.
The Wilcoxon/Kruskal-Wallis test was used to compare nonparametric
continuous data, and the Student t test was used if the data were
parametric. The Kaplan-Meier analysis was used to evaluate OS, along
with the log-rank test. A random coefficient mixed-effect model was
used to fitthe observed decrease ineGFR aftersurgery as demonstrated
previously . All patients had at least one pre- and postoperative sCr
measurement. The drop in eGFR was calculated by first modelling all
preoperative sCr values, generating a line, and projecting it toward the
day of surgery. All postoperative sCr values starting 3 wk after surgery
were also modelled. Thismodeldemonstrated thatthe entire decline in
renal function associated with surgery occurred within the first 3 wk.
After this postoperative steady state was reached, no statistically
significant decline in renal function was noted (data not shown).
Therefore, the drop in eGFR associated with extirpative surgery could
then be calculated by taking the difference between the preoperative
eGFR obtained by extrapolating a line forward to the day of surgery
(based on preoperative measurements) and the postoperative eGFR,
obtained by extrapolating a line backward to the day of surgery.
HRs and their 95% confidence intervals (CI) were reported. A two-
sided p value < 0.05 was considered statistically significant. All analyses
for this study were performed using SAS (SAS Institute, Cary, NC, USA) or
JMP 8.0 (SAS Institute, Cary, NC, USA) statistical software.
As expected in a nonrandomised cohort, there were
significant differences between the groups (Table 1), with
those selected for RN more likely to be older, have higher
comorbidity scores, have larger tumours, and be treated by
laparoscopic surgery. Those with solitary kidneys were
much more likely to be treated by PN. Each of these
variables significantly predicted the propensity score.
Renal preservation was optimised in the PN cohort, with
the majority of patients treated by PN maintaining an eGFR
above 60 (65%), compared with only 30% of the RN cohort
(p < 0.0001). The average drop ineGFR wasmorethantwice
as great in the RN cohort (22.3 vs 10.1 ml/min per 1.73 m2;
p < 0.0001).
Median follow-up for the entire cohort was 50 mo
from any cause. On univariate analysis, using both the log-
rank analysis (Fig. 1) and the Cox proportional hazards
model (Table 2), RN was significantly associated with
an increased risk of death from any cause (HR: 3.6; 95% CI,
E U RO P E A N U R O L O GY 58 ( 2 0 1 0 ) 2 9 3 – 2 9 8
1.9–6.7). When stratified according to propensity score class
and combined into a multivariate Cox proportional hazards
model, patients with unanticipated benign renal tumours
treated by RN were still 2.5 times more likelyto die from any
cause than those treated by PN. Postoperative eGFR was also
an independent predictor of OS (Fig. 2) and accounted for
(HR: 0.97 per unit loss of eGFR; 95% CI, 0.95–0.99).
Table 1 – Perioperative characteristics of the 499 patients with unanticipated benign tumors stratified according to treatment type
(n = 111)(n = 388)
Patient characteristics: mean (IQR) or no. (percentage)
Age, yr (range)
Male gender, no. (%)
Preoperative tumour size, cm (range)
Preoperative eGFR, ml/min per 1.73 m2(range)
Average decrease in eGFR at 95% CI, no., (range)
Charlson Group, no. (%)
Preoperative coronary artery disease, no. (%)
Solitary kidney, no. (%)
Laparoscopic, no. (%)
Pathology, no. (%):
Benign cyst/multilocular cysts
Other benign lesions (eg, adenoma, inflammatory mass, hematoma)
Median follow-up, mo (IQR)
IQR = interquartile range; eGFR = estimated glomerular filtration rate; CI = confidence interval.
Table 2 – Unstratified univariate and stratified multivariate Cox proportional hazards analysis of overall survival, stratified according to
propensity score class
StratifierFactorHR 95% CI
No propensity Score adjustment RN vs PN
Propensity score classRN vs PN
HR = hazard ratio; CI = confidence interval; RN = radical nephrectomy; PN = partial nephrectomy; GFR = glomerular filtration rate; eGFR = estimated glomerular
*For each unit of eGFR loss, there was a 3% decrease in survival.
Fig. 2 – Kaplan-Meier analysis of overall survival stratified according to
postoperative estimated glomerular filtration rate.
OS = overall survival; eGFR = estimated glomerular filtration rate.
Fig. 1 – Kaplan-Meier analysis of overall survival in 499 patients with
unanticipated benignrenal tumours stratified accordingtotreatment type.
OS = overall survival; PN = partial nephrectomy; RN = radical nephrectomy.
EUROPEAN UR OLOGY 58 (2010) 293– 298
with decreasing postoperative renal function (HR: 0.96 per
unit loss of eGFR; 95% CI, 0.93–0.99; Fig. 3).
Previous analyses of the role that postoperative renal
insufficiency may play in the OS of patients with renal mass
have been somewhat clouded by the nonrandomisation of
patientsandtheheterogeneous pathologicstage oftumours
[1–9,11]. Nearly every report to date that attempts to
compare RN to PN has been hindered by the fact that the RN
cohort has had the larger and more aggressive tumours and
often less healthy patients. Although multivariate analyses
attempt to control for this, there remains some uncertainty
that statistical models can adequately eliminate these
obvious confounding factors. Therefore, we set out to study
a subset of patients who had clinical T1 enhancing renal
masses on preoperative imaging but benign tumours on
final pathology, allowing for a cleaner evaluation of renal
function preservation on OS and cardiac-specific survival.
Although there were significant differences between the
cohorts with regard to meaningful clinical parameters
(Table 1) such as age, comorbidity status, and contralateral
kidney status, when controlling for these factors using
propensity score class and multivariate analyses, RN was
still associated with a 2.5-fold increased risk of death from
any cause. This result falls in line with the findings of Go
et al, who demonstrated that in a dose-dependent manner,
decreasing kidney function was associated with increased
risk of cardiovascular death and death from any cause in the
general medical patient . However, it had been unclear
whether nephrectomy-induced chronic kidney disease
harbours similar increased risks to those observed in a
patient with medical renal disease . Our data argue that
regardless of the pathway taken to poor renal function,
these patients are at increased risk cardiac-specific death
and death from any cause. It is instructive to note that the
cardiac-specific survival and OS curves stratified according
to postoperative kidney function nearly overlap for the first
3–4 yr after surgery. After this point, they separate
markedly accordingto postoperative
(Fig. 2 and 3), suggesting that the deleterious effects of
decreased renal function may take some time to translate
into a cardiovascular milieu that leads to earlier death.
These data may appear to be at odds with a recently
published paper on kidney donation, where the researchers
found no decrease in OS compared to population controls
. We believe that patients with kidney tumours are
quite different from patients undergoing kidney donation
with regard to age, comorbidity, and potential response to
GFR reduction. Notably, only about one-half of our patients
(n = 261) had an eGFR >60 3 wk after surgery, whereas
85.5% of the observed donor population maintained an GFR
>60 an average of 11 yr after kidney donation.
As found in multiple previous studies, PN leads to less
renal functionloss thanRN(Table 1)[7,8,15–17].Theexcess
renal loss associated with RN was on average 12.2 ml/min
per 1.73 m2, which was associated with a 44% increased risk
of death from any cause independent of whether PN was
performed (HR: 1.4; 95% CI, 1.1–1.9). Although this effect
was expected and has been observed both in our previously
published series [7,8] and by others , we were surprised
to find that the better preservation of eGFR did not account
for all of the survival advantage noted with PN. This finding
suggests that there may have been other selection factors
not captured in our multivariate propensity analysis that
favoured survival for those in the PN cohort or that some
function of the kidney beyond filtration may be contribut-
ing to a survival advantage in these patients. Although this
study was not designed to look at this issue in particular, we
note that erythropoesis and vitamin D regulation are also
dependent in large part on the kidney, and it merits
these systems may have deleterious effects.
We were also surprised to find that this cohort of
patients commonly died from some additional malignant
tumour. It is known that sometimes renal cell cancer is
associated with benign renal tumours such as oncocytoma
, but 38% of these patients died from tumours from
various organ systems (Table 3), and it is unclear whether
this was merely a spurious observation or whether benign
renal tumours may be part of a larger syndrome.
In this cohort of patients with benign tumours, cardiovas-
cular death was the most common cause of death. When
out to be malignant, it is important to remember that most
patients will die of some other cause , many of which
appear to be worsened by renal insufficiency. This finding
appears to be especially true of those patients with benign
renal masses. The increased risk of death associated with RN
in this cohort was much higher than that observed in our
previously published malignant cohorts (HR: 2.5 vs 1.25) .
Obviously, the ideal scenario would be not to operate at
of benign histology is invariably an uncertain finding until
final pathology. Even today, the absence of cancer on biopsy
cannot always guarantee that malignant pathology in the
Fig. 3 – Kaplan-Meier analysis of cardiac-specific survival stratified
according to postoperative estimated glomerular filtration rate.
eGFR = estimated glomerular filtration rate.
E U RO P E A N U R O L O GY 58 ( 2 0 1 0 ) 2 9 3 – 2 9 8
same lesion was merely not missed on biopsy. It is
recognised that in certain cohorts with renal lesions, the
chance of the tumour being benign is significant. These
patients include those with fat density on computed
tomography (CT) scan (?20 Hounsfield units), those with
Bosniak 2F and 3 lesions , and younger women with
solid tumours . In these cases, every effort should be
made to perform a PN . It seems prudent, given the
excellent cancer-specific outcomes noted with PN, always
to attempt a PN in patients where it is technically feasible,
because a significant portion (20%) of enhancing renal
masses will harbour no malignancy. We note that most
clinical T1 tumours are ‘‘technically amenable’’ to PN, and
1904 of 2608 (73%) of this cohort were treated by PN.
This study is limited by its retrospective nature and the
fact that the cohorts were not randomised. There were
obvious selection differences between the cohorts, and
although we attempted to control for these differences, it
remains possible that selection bias may explain some of the
survival differences. We note, however, that a randomised
trial of benign renal tumours would be impossible, given the
fact that the diagnosis is not confirmed until after surgery.
The better renal preservation associated with PN was
associated with better cardiac-specific survival and OS
when compared to RN in patients who were found to have
unanticipated benign tumours. In light of the observations
that a substantial proportion of patients with unanticipated
benign renal tumours stand to benefit markedly from PN
and the excellent cancer control PN has demonstrated in
patients with malignant tumours, taken collectively, these
data argue that a patient with a cT1 renal tumour appears to
be best served by PN whenever technically feasible.
Author contributions: Amr F. Fergany had full access to all the data in the
study and takes responsibility for the integrity of the data and the
accuracy of the data analysis.
Study concept and design: Weight, Fergany, Novick, Lane, Gill.
Acquisition of data: Weight, Lieser, Larson, Lane, Gill, Novick.
Analysis and interpretation of data: Weight, Gao, Lane, Campbell, Fergany.
Drafting of the manuscript: Weight, Lane, Campbell, Fergany.
Critical revision of the manuscript for important intellectual content:
Weight, Lieser, Larson, Gao, Lane, Campbell, Gill, Fergany.
Statistical analysis: Gao, Weight.
Obtaining funding: None.
Administrative, technical, or material support: None.
Other (specify): None.
Financial disclosures: I certify that all conflicts of interest, including
affiliation,grantsor funding,consultancies,honoraria, stockownershipor
are the following: None.
Funding/Support and role of the sponsor: None.
Acknowledgment statement: The authors acknowledge Wei Liao, Mary
Federico, and Kay Tucker for their assistance with database management.
 Thompson RH, Siddiqui S, Lohse CM, Leibovich BC, Russo P, Blute
ML. Partial versus radical nephrectomy for 4 to 7 cm renal cortical
tumors. J Urol 2009;182:2601–6.
 Leibovich BC, Blute ML, Cheville JC, Lohse CM, Weaver AL, Zincke H.
Nephron sparing surgery for appropriately selected renal cell car-
cinoma between 4 and 7 cm results in outcome similar to radical
nephrectomy. J Urol 2004;171:1066–70.
 Thompson RH, Boorjian SA, Lohse CM, et al. Radical nephrectomy
for pT1a renal masses may be associated with decreased overall
survival compared with partial nephrectomy. J Urol 2008;179:
468–71, discussion 472–3.
 Patard JJ, Shvarts O, Lam JS, et al. Safety and efficacy of partial
nephrectomy for all T1 tumors based on an international multicen-
ter experience. J Urol 2004;171:2181–5, quiz 435.
 Fergany AF, Hafez KS, Novick AC. Long-term results of nephron
sparing surgery for localized renal cell carcinoma: 10-year follow-
up. J Urol 2000;163:442–5.
 Antonelli A, Cozzoli A, Nicolai M, et al. Nephron-sparing surgery
versus radical nephrectomy in the treatment of intracapsular renal
cell carcinoma up to 7 cm. Eur Urol 2008;53:803–9.
 Weight CJ, Larson BT, Gao T, et al. Elective partial nephrectomy in
patients with clinical T1b renal tumors is associated with improved
overall survival. Urology. In press.
 Weight CJ, Larson BT, Fergany AF, et al. Nephrectomy induced
chronic renal insufficiency is associated with increased risk of
cardiovascular death and death from any cause in patients with
localized cT1b renal masses. J Urol 2010;183:1317–23.
 PahernikS, RoosF, Rohrig B,WiesnerC, ThuroffJW. Electivenephron
sparing surgery for renal cell carcinoma larger than 4 cm. J Urol
2008;179:71–4, discussion 74.
 Lane BR, Babineau DC, Poggio ED, et al. Factors predicting renal
functional outcome after partial nephrectomy. J Urol 2008;180:
2363–8, discussion 2368–9.
 Miller DC, Schonlau M, Litwin MS, Lai J, Saigal CS. Urologic diseases
in America. Renal and cardiovascular morbidity after partial or
radical nephrectomy. Cancer 2008;112:511–20.
 Go AS, Chertow GM, Fan D, McCulloch CE, Hsu CY. Chronic kidney
disease and the risks of death, cardiovascular events, and hospital-
ization. N Engl J Med 2004;351:1296–305.
Table 3 – Causes of death in 40 of 499 patients with benign renal
Cause of deathNo. (%)*
Renal cell (2)
Renal failure/nephrotic syndrome
Other (sepsis, accidents, thyrotoxicosis)
*Patients may have had more than one cause of death listed on the death
EUROPEAN UR OLOGY 58 (2010) 293– 298