Introduction

Nephron sparing strategies (NSS), such as partial nephrectomy (PN), have progressively replaced historical radical nephrectomy (RN) for treatment of kidney tumor, becoming the standard procedure for small renal masses, and whenever technically feasible for larger ones [1–3]. Indeed, retrospective studies have shown that PN, compared to RN, improved overall and cancer-related mortality, and was associated with fewer cardio-vascular events [4–8]. However, the only randomized controlled trial (EORTC trial) comparing PN and RN reported that PN resulted in better renal function [9] but did not improve oncological outcomes or overall survival [9, 10]. It is therefore considered that PN allows to preserve nephron capital, with at least equivalent oncological outcomes and survival. In parallel, mini-invasive approaches, including laparoscopic and more recently robot-assisted laparoscopic procedures, have also been increasingly used. Retrospective studies have shown that mini-invasive NSS resulted in non-inferior oncological outcomes compared to open NSS, with fewer post-operative complications [11], with some studies suggesting a benefit of robot-assisted over laparoscopic-only procedures [12–14]. NSS and mini-invasive techniques have consequently been used in patients with chronic kidney disease (CKD) or with solitary kidney (SK) in order to preserve renal parenchyma. However, the need of temporary vascular clamping during PN has raised concern about the risk of renal ischemia-reperfusion injuries, potentially leading to acute kidney injury (AKI) and worsening of CKD. Retrospective studies have reported an incidence of post-operative AKI following PN with renal clamping that ranged from 20 to 40%, associated with a worse renal function 1 year after surgery [15–17]. Robot-assisted PN seemed interesting since it was associated in some studies with a shorter ischemia time than open or laparoscopic PN [18, 19] but studies failed to demonstrate its benefit on renal function preservation [11, 12, 18, 20]. In a recent prospective randomized study, super selective clamping of tumor-targeted arteries also failed to provide better renal function preservation compared to conventional robot-assisted PN [21], in addition, achieving zero ischemia during NSS through “off clamp” procedures has also been proposed to avoid renal ischemia reperfusion injuries, but it remains technically demanding [22].

Little is known about the effects on renal function of these recent surgical procedures in the specific population of patients with CKD. Retrospective studies have reported a beneficial effect of PN on CKD progression in stage 3A CKD patients compared to RN [23], or no effect [6], and on the requirement of permanent dialysis in severe CKD patients [24]. PN was also associated with lower renal function deterioration in CKD patients compared to those with normal kidney function [25]. These studies were limited by short [25] or incomplete [6] follow-up, lack of power [24], or the exclusion of severe CKD [6, 23].

We therefore conducted a retrospective study to assess the renal and clinical outcomes after a RN or PN in patients with pre-existing moderate-to-severe CKD or with SK. The second aim was to identify the major determinants of adverse outcomes in this population.

Methods

Results

Patient characteristics

Among 1259 surgical procedures conducted for a renal mass, 173 (171 patients) were included: there were a total of 67 RN in CKD patients (RN-CKD) and 106 PN; among the latter 79 were in CKD patients (PN-CKD) and 27 were in SK (PN-SK; Fig 1). Patients undergoing RN-CKD were older (median age 74 [68–79] vs 68 [64–76] years, p = 0.004) and had lower rate of hypertension (60 vs 80%, p = 0.02) compared to PN-CKD (Table 1).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Study flow-chart.

eGFR estimated glomerular filtration rate, CKD chronic kidney disease.

https://doi.org/10.1371/journal.pone.0300367.g001

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Baseline patient characteristics.

https://doi.org/10.1371/journal.pone.0300367.t001

Preoperative eGFR was assessed a median 12 [5–23] days before surgery, using the CKD-EPI formula in 82% of cases. The median preoperative eGFR in the RN-CKD and PN-CKD groups were 50 [41–55] and 49 [42–55] mL/min/1.73m² respectively, showing no significant difference, and CKD stages were also not significantly different between the two groups. PN-SK patients had more frequently a known preoperative referral to a nephrologist (74%) than patients in the RN-CKD (25%) or PN-CKD (27%) groups (p < 0.001 for both; Table 1).

Surgery and tumor characteristics, surgical and renal complications

Patients in the RN-CKD group had larger tumors than those in the PN-CKD group (p < 0.001), with fewer T1 lesions and more T3 lesions (p < 0.001 for both), more frequently positive nodal and metastatic status (p < 0.001 for both), logically more collecting duct or urothelial carcinomas (p < 0.001), and less oncocytomas (p = 0.04). The RN-CKD group had more frequently laparoscopic and less frequently robot-assisted procedures than PN-CKD (p = 0.002 for both), and the median operative time was longer (p = 0.03). PN-SK patients had a longer median operative time compared to PN-CKD patients but had shorter median ischemia time (12 [0–17] vs 18 [11–23] min, p = 0.009; Table 2).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Surgery and tumor characteristics, and postoperative complications.

https://doi.org/10.1371/journal.pone.0300367.t002

Patients in the RN-CKD group had more frequently a severe (Clavien-Dindo >2) intra-operative complication than those in the PN-CKD group (p = 0.006), but less frequently a severe postoperative complication (p = 0.03), and a longer median length of hospital stay (p = 0.02). There was no significant difference in terms of severe intra- or post-operative complications between patients in the PN-SK or PN-CKD groups (p > 0.05), but PN-SK patients had a longer median length of hospital stay (p = 0.002; Table 2).

There was no significant difference in the frequency of postoperative AKI between those in the RN-CKD group (61%) and in the PN-CKD group (51%), which were mainly KDIGO stage 1 in both, but AKI was more frequent in the PN-SK group (81%) than in the PN-CKD group (p = 0.0498), and more severe (KDGIGO stages 2 and 3), requiring more frequently postoperative temporary dialysis (15% vs 1% in the PN-CKD, p = 0.004; Table 2). There were 12% of patients who had missing data for the primary outcome or who were lost to follow-up at 1 year.

Primary outcome 1 year after surgery

After a median follow-up of 12.7 [10.3–15.0] months after surgery, patients in the PN-CKD group presented significantly less frequently the primary composite outcome (CKD progression, or MACE, or death) compared to those in the RN-CKD group (43% vs 71%; p = 0.007; Table 3).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. Outcomes one year after surgery.

https://doi.org/10.1371/journal.pone.0300367.t003

Predictors of the primary and renal outcomes at 1 year

In univariate analysis, variables significantly associated with the main outcome were postoperative AKI, and a larger tumor diameter, whereas sex, age, diabetes mellitus, previous history hypertension, and preoperative eGFR (as a continuous variable) were not. In multivariate analysis, postoperative AKI, and a larger tumor diameter remained independent risk factors for the main outcome (Table 4). Type of surgery (RN vs PN) could not be included in the model because it showed a too strong colinear association with tumor diameter.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 4. Logistic regression analysis for the primary outcome.

https://doi.org/10.1371/journal.pone.0300367.t004

Similarly, logistic regression analysis of independent predictors for CKD progression or absolute eGFR loss at 1 year also identified postoperative AKI and larger tumor diameter (S1 Table).

Discussion

In this retrospective cohort study of patients with moderate-to-severe CKD, we report that PN compared to RN led to less frequent adverse outcomes, defined as a composite endpoint including progression of CKD, MACE or death, 1 year after surgery. This was mostly driven by a less frequent progression of CKD, and lower mortality. In the total population including SK patients, we identified postoperative AKI, and larger tumor size as independent risk factors for the primary outcome, CKD progression, and eGFR loss 1 year after surgery, whereas preoperative eGFR was not. However, the influence of type of surgery was not included in the multivariate analysis owing to a too strong collinearity with tumor size.

The greater benefit of PN compared to RN reported herein is consistent with the accumulated data in patients with normal renal function [1–3]. However, the literature is scarce in CKD patients; some studies have reported a greater benefit of PN on CKD progression for patients with stage 3A CKD but not for stage 3B [23] or with stage 4 CKD [24], and no significant difference in mortality [23, 24]. Conversely, a study found no significant difference between RN and PN in the rate of eGFR decrease for patients with moderate CKD, and no significant difference in overall mortality [6], but the major limitation of this study was the frequency of missing data that was up to two-thirds at the end of follow-up.

The strength of the study presented herein is that it included patients treated after the implementation of recent surgical approaches, with few patients with missing data or lost to follow-up. The present study thus reflects recent surgical approaches, which may notably explain the shorter ischemic time that we report compared to previous studies on PN [24]. Moreover, in all of these previously published cohorts the study period was over 20 years (from the 1980s to 2008 [23], 2014 [6] or 2015 [24]) during which surgical procedures were either open only [24] or open and laparoscopic [23], or were unreported [6], and ischemia time during PN was relatively long (median 40 min) [24] or unreported [6, 23]. Furthermore, the cohort herein is one of the largest comparing PN and RN specifically in patients with moderate-to-severe pre-existing CKD, alongside patients with SK. It is noteworthy that we included SK patients regardless of their CKD staging, including a few patients with stage 1 and 2, i.e. even with eGFR > 60 mL/min/1.73m². This was motivated by the fact that SK condition in itself is considered a CKD, even when impairment of eGFR has not occurred yet [30]. Having a SK, even with normal renal function, has indeed been proved as an independent risk factor of renal function deterioration leading to progression of CKD [31]. Likewise, American Urological Association (AUA) and European Association of Urology (EAU) guidelines also associate SK patients regardless of renal function along with patients with renal dysfunction (moderate to severe CKD), as they classify them similarly as having an “imperative” indication of nephron sparing surgery [2, 3]. In the present study, the high incidence of AKI and of the primary outcome in SK patients emphasizes their great susceptibility to renal lesions during and after partial nephrectomy, in the short and long term. Precipitating SK patients from CKD stages with normal to impaired functions is a particularly concerning event, as this will ultimately exposes them to potential complications due to CKD.

AKI during hospital stay is a known strong independent predictor of incident CKD or CKD progression [32], and we report it here as a major predictor of progression of CKD, MACE or death (the primary outcome). Previous studies also identified AKI and tumor size as predictors of worse renal function after PN [15, 17, 33, 34] or after RN [35], but in patients with mainly normal renal function. None of the published studies comparing PN and RN in patients with moderate or severe CKD assessed postoperative AKI [6, 23, 24]. We report herein a frequency of AKI higher than that reported in previous studies (from 25 to 40%) in participants with predominantly normal baseline renal function [15–17]. The frequency of AKI was even higher herein in SK patients, and more severe, whereas it was reported to range from 15 to 30% in previous studies with a comparable population [34, 36, 37]. These studies probably underdiagnosed AKI due to restrictive and less sensitive non-standardized definitions, while in our study both baseline eGFR and AKI were appropriately assessed: creatinine assays were mostly based on enzymatic techniques, and eGFR was estimated using the most recent and accurate CKD-EPI formula [38]. Furthermore, we defined AKI according to current KDIGO criteria that better identify subgroups at higher risk of complications. This is also emphasized by the fact that most of the AKI we detected were mild KDIGO stage 1 but still independently predicted worse outcomes, confirming a more sensitive diagnosis that yet remains clinically significant.

We found no association between preoperative eGFR and the primary outcome, CKD progression, or eGFR loss. This was, however, reported in previous studies as a risk factor for renal degradation in patients mainly with normal renal function [33, 35, 39] and in one study with only severe CKD patients [24], but as a categorical variable whereas herein it was considered as a continuous variable which might have influenced the results. More importantly, postoperative AKI was not evaluated and not included in the multivariate analysis, contrary to herein. However, the previously identified risk factors for postoperative AKI after PN (preoperative eGFR [15], tumor size [16, 29], ischemia time [16, 17, 29], operative time [15], comorbidities [age, sex, body mass index, diabetes mellitus, hypertension] [15–17]), most of which were included as covariates in the multivariate logistic and linear regressions. Further studies should therefore investigate predictors of AKI specifically in CKD patients, to identify possible means of nephroprotection for them.

The duration of follow-up is heterogeneous between previous studies making comparison difficult. We chose a 1-year endpoint since it has been proposed that eGFR should be evaluated at least 1 month and 1 year after surgery [40]. Furthermore, it has been previously reported that no significant degradation of renal function occurs from a median of 47 days [15] to 5 months [41] after PN, until a follow-up of 4 years. This suggests that CKD progression occurs predominantly during the first postoperative year, and the importance of early evaluation is supported by the difference in the composite and renal outcomes in the secondary analysis as early as 1 month after PN or RN.

The present study has limitations. Due to its retrospective nature, we cannot rule out selection bias, and bias in data collection. Its monocentric setting limits the generalizability of the results in other different population of CKD patients undergoing nephrectomy. Even though we conducted multivariate analysis, we could not take into account other unknown confounders, or known but unavailable, such as the presence of proteinuria on the risk of CKD progression [42] and use of ACEi/ARBs, or post-operative use of contrast or other nephrotoxic agents. Furthermore, ischemia time could not be analyzed as a potential predictor of the primary or renal outcomes because it was only applicable in PN, and not in RN. In addition, we included every renal tumor requiring surgery, some potentially associated with worse prognosis in the RN group, notably with a higher TNM staging.

In conclusion, we report that in patients with moderate to severe CKD, postoperative AKI was a major independent predictor of the composite outcome including CKD progression, MACE, or death, one year after surgery. Further studies are needed to identify early determinants of AKI in this population. PN led to better outcomes compared to RN, but these results need to be confirmed as the type of surgery could not be integrated in multivariate analysis.

Supporting information

Acknowledgments

We thank Dr Philip Robinson (Direction de la Recherche Clinique et de l’Innovation, Hospices Civils de Lyon) for his help in manuscript preparation and proofreading. We thank Laura Ratenet for her technical support in the process of data collection.