Predictive factors of complications after robot-assisted laparoscopic partial nephrectomy: a retrospective multicentre study


Correspondence: Romain Mathieu, Service d'Urologie, Centre Hospitalier Universitaire de Rennes, 2, rue Henri Le Guilloux, 35033 Rennes Cedex, France.




  • To analyse the predictive factors of complications after robot-assisted laparoscopic partial nephrectomy (RALPN).

Materials and Methods

  • Data from six French institutions on 240 patients who underwent RALPN between 2009 and 2011 were retrospectively reviewed.
  • Clinical (age, body mass index, American Society of Anesthesiologists and Charlson comorbidity index scores, anticoagulant treatment), tumoral (size, R.E.N.A.L nephrometry score) and operative (surgeon experience, blood loss, opening of the collecting system, operating time) variables were considered.
  • Univariate and multivariate regression models were used to assess the impact of these variables on the occurrence of global and major postoperative complications, classified according to the Clavien system.


  • The median (range) patient age was 61 (26–83) years. Tumours were of low complexity in 62% of cases. Median (range) operating time, blood loss and warm ischaemia time were 161 (45–425) min, 100 (0–2500) mL and 20 (0–59) min, respectively. Postoperative complications occurred in 79 (33%) patients. Complications were ≥grade III in 25 (10%) patients and were mostly haemorrhagic.
  • In multivariate analysis, surgeon's experience (hazard ratio [HR]: 2.14 [1.07–4.27], P = 0.03) and blood loss (HR: 1.002 [1.001–1.003], P < 0.001) were independent predictors of overall complications. When considering major complications, opening of the collecting system was the only factor that was significant (OR: 2.99 [1.2–7.26], P = 0.02). Nephrometry R.E.N.A.L. score was not associated with postoperative complications.


  • In our experience, RALPN is associated with a 30% risk of postoperative complications; surgeon's experience, blood loss and opening of the collecting system were the three predictors of postoperative complications.


Partial nephrectomy (PN) is the standard treatment for renal tumours < 7 cm. Laparoscopic PN (LPN) can avoid the morbidity associated with flank incision and provides oncological efficacy equivalent to that of open PN [1, 2]; however, owing to technical challenges, in particular tumour excision and renorraphy, LPN has remained limited to tertiary care centres [3]. Recently, robotic assistance has been proposed, with the aim of decreasing the technical difficulty of LPN through wristed instrumentation and three-dimensional vision [4, 5].

Numerous centres have reported their experience of robot-assisted LPN (RALPN) [6, 7]. RALPN has been proven to be safe and reproducible [8], yields similar oncological results to those of LPN and, in most series, is associated with shorter ischaemia time [9-11].

Postoperative complications occurring after PN are a major concern that might explain the underuse of nephron-sparing surgery to treat small renal tumours [12]. Several factors, related to the patient (anticoagulant treatment, obesity), the tumour (tumour size or location, nephrometry score) or the surgeon (experience, type of access, clinical environment) can influence postoperative outcomes [13, 14]. In older study series, LPN was associated with a higher rate of postoperative complications compared with open PN [15]. The enhanced precision and skills provided by robotic assistance may confer a better quality of renal repair and thus facilitate laparoscopic surgery for more complex tumours.

Currently, the complication rates reported for RALPN are similar to those observed for LPN [11, 16]. As robot-assisted surgery is steadily increasing worldwide, RALPN has become common practice in many countries. Our objective was to identify the factors that predict postoperative complications in a large multicentre series of patients undergoing RALPN.

Material and Methods

Data Collection

The study included data from six academic centres in France on 240 patients who underwent RALPN between 2009 and 2012. A transperitoneal approach was used for all procedures. After institutional board approval, patient records were extracted from each centre and entered into a standardized database. The following items were queried: age, gender, Charlson comorbidity index score, American Society of Anesthesiologists (ASA) score [17], body mass index (BMI), indication (elective or imperative), operating time, warm ischaemia time, opening of collecting system, estimated blood loss, length of hospital stay, pre- and postoperative renal function (estimated according to the modification of diet in renal disease [MDRD] equation). Tumour stage was determined according to the 2002 Union Internationale Contre le Cancer revised TNM classification [18]. Histological subtype was classified according to the Heidelberg classification and tumours were graded according to the Fuhrman grading scheme by pathologists at each of the six institutions [19, 20]. Tumour complexity was evaluated using the R.E.N.A.L. nephrometry score [21] and stratified into low risk (score 4–6), intermediate risk (score 8–9), and high risk (score 10–12). We evaluated surgeon's experience according to the numbers of RALPN procedures performed: a surgeon was considered experienced when he had performed > 20 procedures.

Postoperative complications were defined as any abnormal medical or surgical event <30 days after surgery. Complications were classified according to the Clavien system [22].

Statistical Analysis

Qualitative and quantitative variables were compared by using a chi-squared (Fischer's exact test) and Mann–Whitney U-test, respectively. Relationships between complications and patient, tumour or treatment variables were first analysed using univariate regression logistic analysis. Multivariate analyses included covariates with a P value <0.2 in univariate analysis. All P values were two-sided and a P < 0.05 was considered to indicate statistical significance. All data analyses were processed using the SPSS 18.0 statistical software (Chicago, IL, USA).


Patients and Tumour Characteristics

A total of 240 RALPN procedures were performed. Patient and tumour characteristics are shown in Table 1. The median patient age was 61 years. Most of the patients had few comorbidities (12% of patients had an ASA score > 2). The median tumour size was 30 mm and 62% of renal masses were of low complexity (R.E.N.A.L. score 4–6). A total of 80% of the tumours were staged pT1a and 16% pT1b. The median blood loss was 100 mL and the median warm ischaemia time was 20 min. The median (range) operating time was 161 (45–425) min. Opening of the collecting system occurred in 32 cases (13.2%). The median hospital stay was 4 days.

Table 1. Patient and tumour characteristics
Median (sd) age, years61 (11.6)
Sex, male n (%)154 (63.6)
Median (sd) BMI26.1 (4.2)
ASA score ≥ 3 (n = 236), %12.3
Performance status ≥ 2 (n = 236), %3.8
Median (sd) Charlson comorbidity index score2 (2.8)
Median (sd) preoperative MDRD, mL/min84 (23.8)
Anticoagulant treatment (n = 238), %11.3
Median (sd) tumour size, mm30 (13.5)
Nephrometry score (n = 227), % 
Elective procedures (n = 228), %86.8
Histological subtype (n = 226), % 
Benign tumours15.5
T stage (n = 235), % 
T1 a79.6
Fuhrman grade in malignant tumours (n = 187), % 
Low grade79.1
High grade20.9
Surgeon's experience, % 
≤20 cases57.5
>20 cases42.5
Median (range) blood loss, mL100 (0–2500)
Median (range) warm ischaemia time, min20 (0–59)
Median (range) operating time, min161 (45–425)
Median (range) hospital stay, days4 (2–45)

Description of Complications

Postoperative complications occurred in 79 patients (32.6%). Complications were minor (Clavien I or II) in 54 patients (grade I in 40 (16.5%), grade II in 14 (5.8%) patients) and major in 25 patients (grade III in 23 (9.5%) and grade IV in two (0,8%) patients). Major complications (grade ≥ III) included haemorrhagic events (haematoma or pseudo-aneurysm) in 15 patients, urinary fistula in three patients, urinary sepsis in one patient, and myocardial infarction in one patient (Table 5).

Comparison of Patients with and without Complications

A higher proportion of patients were receiving anticoagulant treatment in the group of patients with complications than in the group without, although the difference was not significant (16 vs 8%, respectively, P = 0.08). According to R.E.N.A.L. score, there were more intermediate and highly complex tumours in the group with complications (43 vs 35%, P = 0.4). The mean operating time was longer (187 vs 161 min, P = 0.001), mean blood loss was higher (427 vs 200 mL, P < 0.001) and opening of the collecting system was more frequent (43 vs 26% of the cases, P = 0.01) in the group of patients who had a complication. Mean hospital stay was significantly longer in patients who had complications (8.6 vs 4.6 days, P < 0.001). There was no difference in terms of age, comorbidities, warm ischaemia time and postoperative renal function (Table 2).

Table 2. Comparison of characteristics according to presence of global complications
VariablesGlobal complicationsP
Yes (n = 79)no (n = 161)
Patient characteristics   
Median age, years61.5610.99
BMI > 30 kg/m223100.69
Median preoperative MDRD, mL/min82.585.150.68
ASA score 3 or 4, %17280.20
Charlson 3, %29360.31
Anticoagulant treatment, %1680.08
Treatment characteristics   
Imperative indication, %430.12
Surgeon's experience  0.12
0–20 cases5187 
>20 cases2874 
Median tumour size, mm30300.18
Nephrometry score, %   
Intermediate or high43350.37
Median warm ischaemia time, min19200.71
Median blood loss, mL3001000.0001
Median operating time, min1721530.001
Collecting system opening, %43260.01
Tumour characteristics   
T stage in tumours (n = 235)  0.1
T1 a56131 
Margins (%)790.7
Median 1-month postoperative serum creatinine, mM/L83860.35
Median hospital stay, days540.001

Predictors of Complications

In multivariable analysis, the predictors of global complications were blood loss (hazard ratio [HR]: 1.002) and surgeon's experience (HR: 2.14). When considering only major (i.e. ≥ grade III) complications, the only significant predictor was the opening of the collecting system (HR: 2.99 [Tables 3 and 4])

Table 3. Predictors of overall complications in univariate and multivariate analysis
VariablesGlobal complications
Univariate analysisMultivariate analysis
  1. Inf, inferior; Sup, superior.
Mean age0.991.000.981.02    
Obesity (BMI > 30 kg/m2)0.690.850.381.89    
ASA score (≥3)0.151.770.813.910.570.710.212.37
Charlson comorbidity index score (≥3)0.320.740.411.33    
Anticoagulant treatment0.082.040.914.580.311.810.575.74
Tumour size0.151.010.991.040.440.990.961.02
Nephrometry score (1 vs 2)0.401.290.712.33    
Nephrometry score (1 vs 3)0.222.440.5810.22    
Warm ischaemia time0.340.990.961.02    
Blood loss0.
Operating time0.
Surgeon's experience0.121.550.892.700.
Opening of the collecting system0.
Table 4. Predictors of complications ≥ Clavien grade III in univariate and multivariate analysis
VariablesComplications grade ≥ III
Univariate analysisMultivariate analysis
  1. Inf, inferior; Sup, superior.
Obesity (BMI > 30 kg/m2)0.410.700.301.63    
ASA score (≥3)0.551.420.454.47    
Charlson comorbidity index score (≥3)0.510.740.291.84    
Anticoagulant treatment0.911.070.303.86    
Tumour size0.991.000.971.03    
Nephrometry score (1 vs 2)0.801.130.452.85    
Nephrometry score (1 vs 3)    
Warm ischaemia time0.331.020.981.07    
Blood loss0.391.001.001.00    
Operating time0.
Surgeon's experience0.791.120.482.61    
Opening of the collecting system0.013.431.448.150.022.991.237.26


Laparoscopic partial nephrectomy has been reported to reduce the morbidity associated with flank incision. Expert teams showed there were benefits with regard to blood loss, postoperative pain and length of stay [23, 24], but these benefits came at the cost of higher complication rates and longer ischaemia times [15]. Technical difficulties have probably hampered the development of LPN, which remained mostly performed at tertiary care centres. Over the last 10 years, robotic assistance has emerged and this has overcome the technical complexity of LPN. Several series have been reported to date showing that RALPN was feasible and reproducible [6, 25]. Moreover, RALPN could have a benefit regarding renal function preservation; a recent meta-analysis showed that RALPN yielded similar results to those of LPN regarding postoperative and oncological outcomes but was associated with a reduced ischaemia time [16].

The present study specifically focused on postoperative complications after RALPN. To our knowledge, three multicentre series have reported such information to date (Table 6 [13, 25, 26]). Overall, we had a 30% complication rate and we found that global complications wereprincipally affected by surgeon's experience and blood loss.

Table 5. Surgical and medical complications in the cohort
Clavien gradeMedical complicationsSurgical complications
ThromboembolismMyocardial infarctionOtherHaematomaPseudoaneurysmUrinary fistulaInfectionOther
Table 6. Postoperative complications in the multicentre RALPN series
AuthorsYearNo. of centresNo. of patientsComplication rateMajor complicationsClassification used
Benway et al. [25]200931298.50%NSNS
Spana et al. [26]2011445014.40%3.80%Clavien/Dindo
Ficarra et al. [13]2012434711.80%2.90%Clavien/Dindo

We defined surgeon's experience using a somewhat arbitrary threshold of 20 RALPN procedures, as reported by other publications [27, 28]. Based on that limit, we observed a clear learning curve with a complication rate of 37% among surgeons during their first 20 cases, dropping to 29% after 20 cases. In multivariate analysis, having performed >20 cases reduced the risk of complications by half. Similarly, Ficarra et al. [13] recently showed that surgical experience was an independent predictor of overall complication after RALPN and Mottrie et al. [28] showed that surgeon's experience was correlated to robot console time and duration of warm ischaemia; however, the learning curve for RALPN is expected to be minimal for a surgeon who has a significant experience of LPN [9].

Major complications (i.e. ≥ grade III) occurred in 10% of the cases and were mostly haemorrhagic complications (i.e. perirenal haematoma, false aneurysm or arterio-venous fistula). In multivariate analysis, the main predictor of major complications was the opening of the collecting system (HR: 3.0). Surprisingly, tumour complexity, evaluated using the R.E.N.A.L. score, had no influence, which contrasts with the results of a recent publication [8]. Nevertheless, opening of the collecting system and blood loss can be seen as indirect signs of tumour complexity and technical difficulties during surgery that can influence postoperative outcomes. The impact of nephrometry scores on postoperative complications is not clear. Some authors found that tumour complexity had no impact on on postoperative outcome [29, 30], while others found that it did have an impact [8, 13, 14, 31]. We had a low proportion of highly complex tumours (3.5%) and it is possible that retrospective analysis of imaging in several institutions might have translated into misclassification. Another explanation is that the present series reflects the beginning of RALPN experience in France and therefore there might be a selection by surgeons of easily resectable tumours.

Robotic systems offer enhanced vision and several degrees of freedom that make it easier to repair renal parenchyma after tumour excision. It seems logical that better and faster repair would mean fewer complications postoperatively. It is not certain, however, that these enhanced skills translate into fewer complications compared with LPN. In the meta-analysis by Aboumarzouk et al. [16], complication rates did not differ between LPN and RALPN, but most of the comparative series included in this meta-analysis were of small size or from single centres [6, 9-11]. Benway et al. [25] compared 129 consecutive RALPNs with 118 consecutive LPNs in a multi-institutional study and found similar complication rates between the two procedures (8.6 vs 10.2%, respectively), but those authors did not use the Clavien system. Conversely, in a recent single-centre study, Mullins et al. [7] compared 102 LPN cases vs 105 RALPN cases, performed by the same surgeon and found fewer complications among patients who underwent robot-assisted surgery (4.8 vs 11.8%; P = 0.006).

The present study has some limitations. Most of the patients had tumours of low complexity, which reflects the nature of the population selected to be included in the initial experience of RALPN. Such a selection may have underestimated the impact of nephrometry score as a predictor of complications, as reported in other studies [31]. The study includes a large collection of data from six different centres with the inherent biases associated with large retrospective studies. LPN was commonly performed in all involved institutions before they started using the robot and, therefore, our results can be considered preliminary and additional follow-up and increased numbers of patients are required.

In conclusion, in the present study, RALPN was associated with a significant risk of, mostly minor, postoperative complications. The factors influencing postoperative course were surgeon's experience, blood loss and opening of the collecting system. Longer follow-up and additional prospective studies are needed to confirm these findings. With growing experience and progress in surgical techniques, we may in the future observe a reduction in the frequency of postoperative complications.

Conflict of Interest

None declared.


partial nephrectomy


laparoscopic PN


robot-assisted LPN


hazard ratio


body mass index


American Society of Anesthesiologists


modification of diet in renal disease