Trends in renal function after radical nephrectomy: a multicentre analysis

Authors


Abstract

Objective

  • To evaluate serial changes in renal function by investigating various clinical factors after radical nephrectomy (RN).

Patients and Methods

  • The study population consisted of 2068 consecutive patients who were treated at multiple institutions by RN for renal cortical tumour without metastasis between 1999 and 2011.
  • We measured the serial change in estimated glomerular filtration rate (eGFR) and clinical factors during a 60-month follow-up period.
  • The changes in eGFR over time were analysed according to baseline eGFR (eGFR ≥60 and 15–59 mL/min/1.73m2) using a linear mixed model.
  • The independent prognostic value of various clinical factors on the increase in eGFR was ascertained by multivariate mixed regression model.

Results

  • Overall, there was a subsequent restoration of renal function over the 60 months.
  • The slope for the relationship between the eGFR and the time since RN was 0.082 (95% confidence interval [CI] 0.039–0.104; P < 0.001) and 0.053 (95% CI 0.006–0.100; P = 0.038) in each baseline group, indicating that each month after RN was associated with an increase in eGFR of 0.082 and 0.053 mL/min/1.73m2, respectively.
  • When we analysed renal function based on various factors, postoperative eGFR of patients with diabetes mellitus, old age (≥70 years) or a preoperative eGFR of <30 mL/min/1.73 m2, was decreased or maintained at a certain level without any improvement in renal function.
  • Preoperative predictors of an increase in eGFR after RN were young age, no DM, no hypertension, a preoperative eGFR of ≥30 mL/min/1.73m2 and time after surgery (≥36 months).

Conclusions

  • Renal function recovered continuously during the 60-month follow-up period after RN.
  • However, the trends in functional recovery change were different according to various clinical factors and such information should be discussed with patients when being counselled about their treatment for renal cell carcinoma (RCC).

Introduction

Radical nephrectomy (RN) is still considered the standard treatment for large RCCs. It is accepted that patients undergoing RN for RCC are at an increased risk for the development of renal insufficiency [1, 2]. Renal insufficiency contributes to many relevant clinical sequelae, an increased risk of noncancerous-related deaths and higher rates of cardiovascular disease [1, 3, 4]. Therefore, a paradigm shift has been witnessed in understanding as to how RN affects kidney function, and how renal function eventually affects cardiovascular events and overall survival.

Previous studies have shown an increase in the serum creatinine level (SCr) by 20% above baseline after RN [5]. However, it has also been shown that RN in patients results in functional adaptation and compensatory hypertrophy of the remaining kidney. The range of the GFR after RN has been reported to be 75–80% of its baseline level, when patients were followed-up for 20 years after RN [6-9]. Moreover, the effective renal plasma flow increased by ≈30% as early as 1 week after RN and remained at a greater level than before RN, even after 10 years of follow-up [10].

However, most of these results were obtained from donor nephrectomy patients. These patients were healthy young people with a potentially greater ability for compensation. They have different characteristics that are not comparable with patients with renal cancer. To date, only limited information is available on renal function and compensatory mechanism after RN. We cannot draw a proper conclusion on the issue of whether renal function would be continuously aggravated or improved because of the compensatory mechanism produced by contralateral renal hyperfunction. Therefore, we investigated unilateral renal function longitudinally after RN until 5 years after RN to observe the postoperative functional changes. We also evaluated the effect of clinical factors on the serial change in renal function.

Patients and Methods

Between 1999 and 2011, 2068 consecutive patients who were treated with RN for renal cortical tumours from five different institutions were evaluated. Patients were excluded if they had bilateral renal tumours, metastasis at surgery, a preoperative estimated GFR (eGFR) of <15 mL/min/1.73 m2 and those on dialysis. Patients were followed until death or censoring. Clinical variables evaluated included: age, sex, diabetes mellitus (DM), hypertension, body mass index (BMI), surgical method (open and laparoscopic) and preoperative eGFR. We obtained Institutional Review Board approval from every centre that participated.

Renal Function Measurements

Longitudinal SCr was used to estimate eGFR. We assessed the SCr level at each respective evaluation point, that was preoperatively, and at 3, 6, 9, 12, 24, 36, 48, 60 months after RN. The eGFR was calculated using the Modification of Diet in Renal Disease (MDRD) equation:

display math

[11]

Four outcomes were measured: (i) presence of new-onset chronic kidney disease (CKD) stages III (eGFR 30–59 mL/min/1.73 m2) or CKD stages IV (15–29 mL/min/1.73 m2); (ii) the overall serial change in eGFR during the 60-month follow-up period according to baseline eGFR (15–59 and ≥60 mL/min/1.73 m2); (ii) the effect of various clinical factors on the serial change in eGFR in all patients and (iv) determination of the clinical factors that predicted for the increase in postoperative renal function. To evaluate the effects of clinical factors on the eGFR after RN, we subdivided the patients based on age at RN (<60, 60–69, and ≥70 years), sex (male and female), BMI (<30 and ≥30 kg/m2), preoperative eGFR (≥80, 60–79, 30–59, and 15–29 mL/min/1.73 m2) and surgical approach (open and laparoscopic). We investigated whether DM or hypertension, as underlying diseases influenced renal function after RN. Furthermore, in multivariate mixed regression model, we performed a subanalysis of our cohort based on baseline eGFR (15–59 and ≥60 mL/min/1.73 m2).

The measurement results are presented as median values. The changes in eGFR and the slope of the increase in renal function over time after RN were analysed using a linear mixed model. This analysis was used because it took into account the correlation of the repeated measurement of the outcome within patients and accounted for any missing data. The slope of the increase in renal function over time was calculated by linear mixed regression analysis of serial eGFR for each patient. The slope was expressed as the regression coefficient (mL/min/month/1.73 m2) with the 95% CI. In our implementation of the mixed model, the intercept and the regression coefficient for the follow-up time were treated as random effects, such that each subject had a unique intercept and regression coefficient. In addition, we used generalised linear mixed model analysis separately for each variable of interest. We created a separate regression line with time slopes and calculated an average line of best fit from the pooled data to compare the differences in the slopes of eGFR increase according to various clinical factors. Finally, the independent prognostic value of various clinical factors on the increase in eGFR was ascertained by multivariate mixed regression model. We improved precision of effect estimates by using all observed data, in contrast to the approach of deleting all observations on participants with at least one missing observation. All the statistical analysis was performed using SAS software (version 9.1.3; SAS Institute Inc., Cary, NC), and statistical significance was assumed at P < 0.05.

Results

Presence of New-onset of CKD Stages III or IV

The median (interquartile, IQR) age of the patients was 56.0 (48–66) years. The median (IQR) preoperative SCr level was 1.0 (0.9–1.2) mg/dL and the median (IQR) preoperative eGFR was 76.5 (64.2–87.9) mL/min/1.73m2. The median (IQR) clinical follow-up was 33.0 (11.0–66.0) months (Table 1). The prevalence of a pre-existing eGFR of <60 mL/min/1.73 m2 before RN was 313 (15.1%). The incidence of a new onset of CKD stages III and IV after RN was 633 (36.1%) and 60 (3.4%), respectively. Our data showed that the frequency of a 50% decline in eGFR was 133 (6.4%).

Table 1. The patients' clinicopathological characteristics
VariableValue
No. of patients2068
Median (IQR) age, years56 (48–66)
Sex, n (%) 
Male1422 (68.8)
Female646 (31.2)
Median (IQR) BMI, kg/m224.0 (21.9–26.1)
N (%): 
DM351 (17.0)
Hypertension828 (40.0)
Operation procedure: 
Laparoscopic or robot988 (47.8)
Open1080 (52.2)
Pathological stage: 
pT11286 (62.2)
pT2333 (16.1)
pT3401 (19.4)
pT448 (2.3)
Median (IQR): 
Preoperative SCr, mg/dL1.0 (0.9–1.2)
Postoperative SCr, mg/dL1.3 (1.0–1.5)
Preoperative eGFR, mL/min/1.73m276.5 (64.2–87.9)
Postoperative ultimate eGFR, mL/min/1.73m258.9 (49.1–69.5)
Follow-up, months33.0 (11.0–66.0)

The Overall Change in Pattern in Renal Function Over Time after RN According to Baseline eGFR (15–59 and ≥60 mL/min/1.73 m2)

At each month after RN the eGFR was significantly lower than the preoperative eGFR. There was a significant drop in the eGFR in the first 3 months after RN. However, renal function generally began to stabilise and gradually increase after a nadir eGFR value was reached in both groups (Fig. 1). A longer duration since RN was associated with a higher eGFR. Overall, there was a subsequent improvement in renal function over the following 60 months, even in patients with preoperative eGFR of 15–59 mL/min/1.73 m2 (Fig. 2). The slope for the relationship between the eGFR and the time since RN was 0.082 (95% C, 0.039–0.104; P < 0.001), indicating that each month after RN was associated with an increase in eGFR values of 0.082 mL/min/1.73 m2 in the patients with a preoperative eGFR of ≥60 mL/min/1.73 m2 (Fig. 2A). There was also an association between the time after RN and an increase in eGFR in patients with preoperative eGFRs of 15–59 mL/min/1.73 m2; the slope value was 0.053 (95% CI 0.006–0.100; P = 0.038) (Fig. 2B).

Figure 1.

Sequential changes in eGFR after RN. (A) Patients with a preoperative eGFR of ≥ 60 mL/min/1.73 m2. (B) Patients with preoperative eGFR of 15–59 mL/min/1.73m2. Data are presented as means with sds. *P < 0.05 for the mean eGFR vs preoperative eGFR.

Figure 2.

The eGFR according to time since RN. Slopes are based on regression coefficients from the linear mixed models. The solid line indicates the regression line, and the dotted line, the 95% CI. (A) Patients with a preoperative eGFR of ≥60 mL/min/1.73 m2; (B) Patients with a preoperative eGFR of 15–59 mL/min/1.73 m2.

Comparison of Renal Function over time According to Various Clinical Factors

When we analysed in detail the changes in the increasing rate of renal function by various factors, the trends in functional recovery change were different. Mixed model regression analysis showed a statistically significant interaction between various clinical factors and time on renal function. Patients aged <70 years had significant increases in their eGFRs during the follow-up period (Fig. 3A). Patients without DM also showed significant increases in renal function (Fig. 3C). There was a continuous recovery in the renal function of patients aged <70 years, with preoperative eGFRs of >30 mL/min/1.73 m2 and with no DM, whereas the postoperative eGFR in patients with DM, old age (≥70 years) and preoperative eGFR of <30 mL/min/1.73 m2 was seen to decrease or maintain at a certain level without any improvement in renal function. For patients aged ≥ 70 years, with preoperative eGFRs of <30 mL/min/1.73 m2 and DM, there were unclear associations between postoperative eGFR and the time since RN. The eGFR increased by 0.007 mL/min/1.73 m2 (95% CI –0.04 to 0.06, P = 0.781), –0.002 mL/min/1.73 m2 (95% CI –0.06 to 0.05, P = 0.931) and 0.191 mL/min/1.73 m2 (95% CI –0.10 to 0.48, P = 0.192) per month, in patients aged ≥ 70 years, DM, and a preoperative eGFR of <30 mL/min/1.73 m2, respectively (Fig. 3A,C,E). The tendency for an increase in renal function over time was similar for other clinical variables, e.g. sex (male and female), hypertension, and surgical approach (open and laparoscopic) (Fig. 3B,D,F).

Figure 3.

Changes in the eGFR over time after RN in all patients according to age (A), sex (B), DM (C), hypertension (D), preoperative eGFR (E) and surgical method (F). Slopes are based on regression coefficients from the linear mixed models. [*Correction added on 20 January 2014, after first online publication: ‘HTN’ was corrected to ‘DM’]

Independent Prognostic Value of Various Clinical Factors on the Increase in eGFR after RN

Table 2 shows the results of the multivariate linear mixed model for postoperative increase in eGFR in patients with a preoperative eGFR of ≥ 60 mL/min/1.73 m2. Certain variables were significantly related to the increase in postoperative eGFR, including younger age (<70 years), time after surgery (≥24 months), no DM, no hypertension, and a preoperative eGFR of ≥80 mL/min/1.73 m2. The impact of time after surgery, especially at ≥ 24 months, on the increase in renal function remained significant even after adjustment for age, sex, DM, hypertension, preoperative eGFR and surgical methods (β=2.276, standard error 0.466, 95% CI 1.361–3.192, P < 0.001). This result means that the increased rates of the eGFRs at 24–60 months after RN were statistically higher than those during the first 3 months, indicating improvement in postoperative renal function. There was no significant effect of BMI on the increase in renal function in the multivariate mixed model analysis. There were similar results in patients with preoperative eGFRs of 15–59mL/min/1.73 m2. Young age (<60 years), time after surgery (≥36 months), no DM, no hypertension, and a preoperative eGFR of ≥30 mL/min/1.73 m2 were significantly related to an increase in postoperative eGFR (Table 3).

Table 2. Multivariate mixed regression analysis based on the group of patients with a preoperative eGFR of >60 mL/min/1.73 m2 for postoperative increase in eGFR
VariableβStandard error95% CIP
Intercept48.5371.37645.831–51.240 
Time after RN (months):    
3 (referent)0   
6–0.5750.450–1.458 to 0.3060.201
90.2620.496–0.710 to 1.2350.597
120.6500.449–0.230 to 1.5310.147
242.2760.4661.361 to 3.192<0.001
364.0300.4933.061 to 4.998<0.001
485.7580.5404.698 to 6.818<0.001
605.6040.6074.412 to 6.795<0.001
Age (years):    
<607.0501.0075.074 to 9.026<0.001
60–692.4621.0810.342 to 4.5820.022
≥70 (referent)0   
Sex:    
Male–1.9240.766–3.427 to 0.4780.185
Female (referent)0   
DM:    
No2.0121.0540.054 to 4.0800.036
Yes (referent)0   
Hypertension:    
No1.7080.7530.230 to 3.1870.023
Yes (referent)0   
Preoperative eGFR (mL/min/1.73m2):    
≥806.8572.0974.012 to 9.8040.005
60–79 (referent)0   
BMI (kg/m2):    
<30–1.2621.777–4.749 to 2.223.0.477
≥30 (referent)0   
Surgical method:    
Open0.2730.709–1.116 to 1.6640.699
Laparoscopic (referent)0   
Table 3. Multivariate mixed regression analysis based on the group of patients with a preoperative eGFR of 15 to <60 mL/min/1.73 m2 for postoperative increase in eGFR
VariableβStandard error95% CIP
Intercept19.8734.86710.290 to 29.456 
Time after RN (months):    
3 (referent)0   
6–1.0390.624–2.265 to 0.1870.103
9–0.4620.659–1.756 to 0.8310.483
12–0.8190.623–2.042 to 0.4030.201
240.6450.654–0.640 to 1.9300.324
361.4570.6980.120 to 2.8250.037
481.2200.7010.119 to 2.5910.039
601.3290.6960.203 to 2.9530.042
Age (years):    
<608.7262.0204.760 to 12.692<0.001
60–692.9061.722–0.473 to 6.2860.092
≥70 (referent)0   
Sex:    
Male–2.4651.546–5.500 to 0.5690.111
Female (referent)0   
DM:    
No5.1081.6841.801 to 8.4140.003
Yes (referent)0   
Hypertension:    
No3.8091.5990.669 to 6.9480.017
Yes (referent)0   
Preoperative eGFR (mL/min/1.73m2):    
30–5916.1394.4937.320 to 24.958<0.001
15–29 (referent)0   
BMI (kg/m2):    
<301.0290.885–0.352 to 2.102.0.122
≥30 (referent)0   
Surgical method:    
Open2.6571.484–0.255 to 5.5690.083
Laparoscopic (referent)0   

Discussion

In the present study, we sequentially evaluated eGFR to determine postoperative changes in renal function and investigated clinical factors affecting the natural course of renal function. In previous donor nephrectomy studies, the range of GFR after RN has been reported to be 75–80% of its baseline level when patients were followed up for 20 years [6-9]. Similarly, Clark et al. [12] reported a 32% decrease in renal function in the RN patient group. Shirasaki et al. [13] reported a 33% increase in SCr concentration. They also insisted that an adaptive hyperfunction occurs soon after RN that continues for at least 1 year. In the present study, there was a significant deterioration in renal function at 3 months after RN. We documented a statistically significant decrease at 3 months after RN when compared with the mean preoperative eGFR. However, renal function then stabilised quickly and gradually increased after a nadir eGFR was reached. Overall, there was a subsequent improvement in renal function over the following 60 months.

In the present study, the prevalence of pre-existing renal insufficiency (eGFR< 60 mL/min/1.73 m2) was 15.1%. These findings are consistent with those of previous studies, which state that 10–30% of patients with RCC had renal insufficiency with a GFR of <60 mL/min/1.73 m2 preoperatively [14-17]. However, the incidence of a new onset of CKD III and IV after RN was 633 (36.1%) and 60 (3.4%), respectively, and this finding is not consistent with those reported by other studies. Previous studies showed a 60–69% incidence of de novo renal insufficiency (eGFR of <60 mL/min/1.73 m2) [1, 14]. Malcolm et al. [18] also reported that 44.7% of patients developed new onset renal insufficiency (eGFR of <60 mL/min/1.73 m2). Unlike these studies, the present cohort of patients were younger than those in most of the other RCC cohorts (median age, 57–62 years) and baseline renal function in our patients was higher than in other patients cohorts whose preoperative eGFR values were 73.3–74.8 mL/min/1.73 m2. In the present study, there was a continuous recovery in renal function irrespective of BMI values. However, in the multivariate mixed model analysis, we found that the effect of BMI on the increase in renal function was not statistically significant. The narrow range of BMI values (IQR 21.9–26.1 kg/m2) may affect the negative result. Therefore, it should be noted that further study is necessary to evaluate how renal function can be improved after surgery in cohorts with different racial and geographical compositions.

It is well-known that renal insufficiency is associated with an increased risk of cardiovascular disease and death. Takeshita et al. [4] showed that the risk of cardiovascular events increased proportionally with a decline in eGFR. Huang et al. [1] also reported that RN was involved with cardiovascular events and noncancerous mortality. Their report supported the plausibility of worsened overall and cardiovascular survival after RN. It is presumed that renal insufficiency causes activation of the renin-angiotensin system, oxidative stress, inflammation with increased levels of cytokines, and dyslipidaemia. These factors lead to endothelial dysfunction and atherosclerosis. Consequentially, the risk of heart failure, coronary artery disease and stroke increase [19-21]. Although renal function after RN is a complex entity, we realised that it depended on how good baseline renal function was. In comparison with donor nephrectomy patients, patients with RCC have an increased frequency of risk factors for renal insufficiency, such as DM, hypertension, and increased age. Even though these clinical factors cannot be modified, such information should be discussed with patients when being counselled about their treatment for RCC.

To date, there has been no standardised way of measuring renal function. Although SCr was used as a quantifying tool of renal function in the present study, most investigators suggest that estimation of the SCr level is not a reliable method and that a renal scan is an accurate method for measuring renal function [13, 22]. However, it is very difficult to use renal scans routinely in clinical practice. Therefore, we used eGFR calculated by MDRD formula. It is reasonable to suggest that eGFR could be useful for quantifying postoperative renal function in patients with a solitary kidney. Also, we consider that more specific tests would not substantially change our results.

The present study should be interpreted with consideration of its limitations. First, this is a retrospective analysis of data collected from multiple institutions, where the interval of postoperative measurement of SCr is different. Therefore, during the follow-up periods, a number of cases were missing because SCr data were not available.

Second, we should consider the survival bias. During the follow-up period, healthy patients with a good baseline renal function who are still alive may be included in this study.

The third limitation is that the present median (IQR) follow-up of 33 (11.0–66.00) months, may not be long enough to draw conclusions on long-term postoperative renal function. Therefore, additional longer follow-up investigation is needed to identify preoperatively those patients, who will be at the risk of exhibiting deterioration of renal function after RN.

The present investigation has several distinctive features. To date, very few studies have evaluated the natural course of postoperative renal function and the effect of chronological time on renal function in patients with RCC. The influence of clinical factors on renal function also has not been well studied. Moreover, the timing of changes in renal function after RN has not been exactly assessed. Results from the present study add considerably to current knowledge about renal function after RN. The observation of an apparent association between renal function and time after RN and various clinical factors in RN patients has never been reported before and these novel data are important in renal function research. In the present study, the large data set and the availability of many repeated measurements per person is unique and a strong asset. The extensive amount of repeated data enabled us to estimate very precisely the association between renal function and time, and various clinical factors. By using sophisticated statistical methods to analyse repeated measures, we had the opportunity for the first time to distinguish different changing patterns of renal function with various clinical factors. This study helps us gain insight into factors associated with renal function after RN.

In conclusion, we showed that renal function is aggravated immediately after RN but stabilised soon after and recovered continuously during the 60-months follow-up period. However, the trends in functional recovery change were different according to clinical factors. Preoperative predictor for increase in eGFR after RN were young age, no DM, no hypertension, preoperative eGFR of ≥30 mL/min/1.73 m2) and time after surgery (≥36 months). We should provide such information to patients with RCC and encompass a discussion on lifestyle modifications, medications or referral to nephrology department for the patients at risk.

Conflicts of Interest

None declared.

IRB Numbers

  1. Seoul National University Bundang Hospital: B-1202/145-102.
  2. Seoul National University Hospital: H-1207-005-415
  3. Seoul St. Mary's Hospital: KC12RIMI0778
  4. Korea University Anam Hospital: ED12053
  5. Chungbuk National University Hospital: 2012-2-014
Abbreviations
BMI

body mass index

CKD

chronic kidney disease

DM

diabetes mellitus

eGFR

estimated GFR

IQR

interquartile range

MDRD

modification of diet in renal disease

RN

radical nephrectomy

SCr

serum creatinine

Ancillary