How can we predict lymphorrhoea and clinically significant lymphocoeles after radical prostatectomy and pelvic lymphadenectomy? Clinical implications

Authors

  • Umberto Capitanio,

    1. Department of Urology, University Vita–Salute, San Raffaele Hospital, Milan, Italy, *Cancer Prognostics and Health Outcomes Unit, University of Montreal, Montreal, Quebec, Canada
    Search for more papers by this author
  • Federico Pellucchi,

    1. Department of Urology, University Vita–Salute, San Raffaele Hospital, Milan, Italy, *Cancer Prognostics and Health Outcomes Unit, University of Montreal, Montreal, Quebec, Canada
    Search for more papers by this author
  • Andrea Gallina,

    1. Department of Urology, University Vita–Salute, San Raffaele Hospital, Milan, Italy, *Cancer Prognostics and Health Outcomes Unit, University of Montreal, Montreal, Quebec, Canada
    Search for more papers by this author
  • Alberto Briganti,

    1. Department of Urology, University Vita–Salute, San Raffaele Hospital, Milan, Italy, *Cancer Prognostics and Health Outcomes Unit, University of Montreal, Montreal, Quebec, Canada
    Search for more papers by this author
  • Nazareno Suardi,

    1. Department of Urology, University Vita–Salute, San Raffaele Hospital, Milan, Italy, *Cancer Prognostics and Health Outcomes Unit, University of Montreal, Montreal, Quebec, Canada
    Search for more papers by this author
  • Andrea Salonia,

    1. Department of Urology, University Vita–Salute, San Raffaele Hospital, Milan, Italy, *Cancer Prognostics and Health Outcomes Unit, University of Montreal, Montreal, Quebec, Canada
    Search for more papers by this author
  • Firas Abdollah,

    1. Department of Urology, University Vita–Salute, San Raffaele Hospital, Milan, Italy, *Cancer Prognostics and Health Outcomes Unit, University of Montreal, Montreal, Quebec, Canada
    Search for more papers by this author
  • Ettore Di Trapani,

    1. Department of Urology, University Vita–Salute, San Raffaele Hospital, Milan, Italy, *Cancer Prognostics and Health Outcomes Unit, University of Montreal, Montreal, Quebec, Canada
    Search for more papers by this author
  • Claudio Jeldres,

    1. Department of Urology, University Vita–Salute, San Raffaele Hospital, Milan, Italy, *Cancer Prognostics and Health Outcomes Unit, University of Montreal, Montreal, Quebec, Canada
    Search for more papers by this author
  • Andrea Cestari,

    1. Department of Urology, University Vita–Salute, San Raffaele Hospital, Milan, Italy, *Cancer Prognostics and Health Outcomes Unit, University of Montreal, Montreal, Quebec, Canada
    Search for more papers by this author
  • Pierre I. Karakiewicz,

    1. Department of Urology, University Vita–Salute, San Raffaele Hospital, Milan, Italy, *Cancer Prognostics and Health Outcomes Unit, University of Montreal, Montreal, Quebec, Canada
    Search for more papers by this author
  • Francesco Montorsi

    1. Department of Urology, University Vita–Salute, San Raffaele Hospital, Milan, Italy, *Cancer Prognostics and Health Outcomes Unit, University of Montreal, Montreal, Quebec, Canada
    Search for more papers by this author

Umberto Capitanio, University Vita-Salute San Raffaele, Via Olgettina 60, 20132, Milan, Italy.
e-mail: capitanio.umberto@hsr.it

Abstract

Study Type – Therapy (case series)
Level of Evidence 4

OBJECTIVE

  • • To identify clinical and pathological variables that may help clinicians in predicting, preventing and managing lymphorrhoea and clinically significant lymphocoeles (CSL), which are reported complications after pelvic lymphadenectomy (PLND) and retropubic radical prostatectomy (RRP).

PATIENTS AND METHODS

  • • We prospectively analysed 552 consecutive men with prostate cancer who underwent RRP and PLND (2006–2008).
  • • All patients had detailed clinical and pathological data prospectively recorded in an electronic database. Drains were removed when the amount of lymph was <20 mL in the previous 24 h. A CSL was defined as the presence of a symptomatic lymphocoele requiring treatment. Lymphorrhoea was defined as the total amount of lymph drained by the drains until their removal.
  • • Univariable and multivariable logistic regression models were used to test the association between all the predictors (age, body mass index, American Society of Anesthesiologists score, prostate volume, clinical stage, number of LNs removed, surgeon, pathological T and N stage) and the presence of CSL.
  • • Univariable and multivariable linear regression models were also used to test the association between the available predictors and lymphorrhoea.

RESULTS

  • • The median (range) number of LNs removed was 20 (1–63). Both linear and logistic multivariable regression analysis showed that the number of removed LNs and age were the only two statistically significant predictors of total amount of lymphorrhoea and CSL after RRP and PLND (both P < 0.01).
  • • Specifically, the risk of developing a CSL increased by 5% for every LN removed. Similarly, every year of age increased the risk of having CSL by 5%.
  • • The most informative thresholds for predicting CSL were 65 years of age and 20 LNs removed.
  • • External iliac lymphadenectomy resulted in a higher associated risk of lymphorrhoea and CLS relative to obturator LN removal (P= 0.001 vs P= 0.1, respectively).

CONCLUSIONS

  • • There was a positive association between the number of LNs removed and age at RRP with the amount of lymphorrhoea and the risk of developing a CSL.
  • • The most informative thresholds in predicting CSL were 65 years of age and 20 LNs removed. External iliac lymphadenectomy resulted in a higher risk of lymphorrhoea and CLS relative to obturator LN removal.
Abbreviations
(R)RP

(retropubic) radical prostatectomy

(P)LN(D)

(pelvic) lymph node (dissection)

CSL

clinically significant lymphocoeles

BMI

body mass index

ASA

American Society Anesthesiologists

INTRODUCTION

Currently, radical prostatectomy (RP) represents the most widely used treatment in men with localized prostate cancer [1–3]. In the last decade, surgery has been gaining importance in the treatment of men with high-risk prostate cancer, as well [1,4]. In this context, pelvic lymph node dissection (PLND) is considered the most reliable procedure for the detection of LN metastases in prostate cancer although its therapeutic benefit is currently under debate [5,6]. Recent evidence has shown that if PLND is planned at the time of RRP, this should be extended [7,8]. However, LND may cause lymphorrhoea or the development of lymphocoeles, although controversy exists about the rate of PLND-related complications according to the extent of PLND [5,9].

Lymphorrhoea is the leakage of lymph from the drains after surgery and may affect the perioperative period after retropubic RP (RRP) and extended pelvic lymphadenectomy (PLND). Similarly, lymphocoeles represent a common complication after RRP and PLND, although only in few cases do lymphocoeles become symptomatic [9]. Although several studies have systematically assessed the complications associated with RRP and PLND [5,10–13], to date, no study has specifically focused on lymphorrhoea and clinically significant lymphocoeles (CSL). Moreover, no study has specifically investigated the clinical and pathological variables that may help clinicians in predicting such conditions.

In the present study, we prospectively assessed the amount of lymphorrhoea and the risk of CSL in men treated with RRP and PLND at one academic institution. Moreover, we aimed to identify individual factors that could predict lymphorrhoea, CSL and total days of drainage, to improve treatment decision-making in men with prostate cancer. Indeed, to foresee lymphorrhoea and CSL after RRP and PLND may assist clinicians in selecting the most appropriate surgical technique and management in the postoperative period, when surgery is performed.

PATIENTS AND METHODS

We prospectively analysed 552 consecutive men with prostate cancer who underwent open RRP and PLND between January 2006 and December 2008. All patients received a once-daily dose of 4000 units of low-molecular-weight heparin (s.c.) in the shoulder starting from the night before and until 4 weeks after RRP. RRPs were performed according to the Walsh technique, with the use of a 8-cm mini-laparotomy incision. All patients underwent anatomically defined PLND before RRP. PLNDs consisted of excision of fibro-fatty tissue along the external iliac vein, the distal limit being the deep circumflex vein and femoral canal. Proximally, PLND was performed up to and included the bifurcation of the common iliac artery. Furthermore, all fibro-fatty tissue within the obturator fossa was removed to completely skeletonize the obturator nerve. The lateral limit consisted of the pelvic side-wall; the medial dissection limit was defined by perivesical fat. The LNs along the internal iliac artery were also removed. In all cases lymphadenectomy was performed by using metallic clips and ligation of lymphatic vessels. Sclerotherapy, talc poudrage, fibrin sealant or i.m. injection of octeotride were not used. Two drains were always placed in the obturator fossa bilaterally. Drains were removed when the amount of lymph draining per day (mL/24 h) was <20 mL. For the specific aim of the present study only men in whom the bladder neck was preserved and who showed a perfectly watertight urethrovesical anastomosis intraoperatively were selected. Patients were also excluded when an anastomotic leakage was evidenced in the early postoperative period. A cystogram was routinely performed before removal of the catheter. Contrast extravasation was used to define an anastomotic leak. Those inclusion and exclusion criteria resulted in the exclusion of 51 (9.2%) cases and a final study population of 501 assessable patients.

All patients had detailed clinical and pathological data recorded prospectively in an electronic database. The targeted postoperative endpoints were lymphorrhoea, CSL and days of lymph drainage. Lymphorrhoea was defined as the total amount of lymph drained by the drains until their removal. CSL was defined as the presence of a symptomatic lymphocoele requiring any type of invasive treatment (i.e. percutaneous drain, puncture, open or laparoscopic marsupialization). The variable depicting days of drainage was defined as the number of days until the removal of the last drain. Lower moderate-to-severe abdominal pain, moderate-to-severe lymphoedema (2 + or more) and deep venous thrombosis of the lower leg were considered as reliable symptoms to define CSL. No routine ultrasonography was performed. Asymptomatic lymphocoeles incidentally detected by imaging were not considered as CSL. Additional variables recorded for each patient included clinical [age, body mass index (BMI), American Society Anesthesiologists (ASA) score, prostate volume and clinical stage], surgical (surgeon and number of LNs removed) and pathological data (T stage and N status).

Logistic regression analyses were used to test the association between the available predictors and the risk of developing CSL. Linear regression analyses were used to test the association between the available predictors and the variable depicting the days of lymph drainage. Finally, as a widely accepted definition of lymphorrhoea is not currently available, we used the outcome defined as lymphorrhoea a continuously coded endpoint, without categorization. Therefore, in those analyses we relied on univariable and multivariable linear regressions models to test the association between the predictors and total amount of lymphorrhoea.

All statistical tests were two-sided with the significance level set at 0.05.

RESULTS

Table 1 summarizes the clinical, surgical and pathological characteristics of the study population. The median age was 66 years. The median (range) number of LNs removed was 20 (1–63). In all, 56 patients (11.2%) developed a CSL in the perioperative period. The mean (median) amount of lymphorrhoea was 967 (555) mL. Patients with a CSL were significantly older compared with their counterparts without CSL (P= 0.02). The mean (median) number of LNs removed was 23.8 (22) and 20.8 (20) in patients with and without CLS, respectively (P= 0.01). In all, 59% of patients with CLS and 44% of patients without CLS underwent a lymphadenectomy with >20 LNs removed (P= 0.05, Table 1). The median amount of lymphorrhoea was 615 mL and 545 mL in patients with and without CLS, respectively (P= 0.05, Table 1). The mean amount of lymphorrhoea gradually rises as patients’ age and lymphadenectomy extension increase (Fig. 1A,B).

Table 1.  Descriptive characteristics of 501 assessable patients who fulfilled the study inclusion criteria. The results are stratified according the presence of a CSL
VariableOverall populationCSLP
YesNo
No. of patients501  56445 
Clinical characteristics    
 Mean (median; range):    
  Age, years 64.9 (66; 41–83)  67.1 (68; 45–80) 64.6 (66; 41–83)0.02
  BMI, kg/m2 26.3 (26.3; 17.3–39.5)  25.9 (25.1; 20.2–24.4) 26.4 (26.3; 17.3–39.5)0.3
  Prostate volume, mL 50.6 (45; 5–233)  50.3 (44; 14–160) 50.7 (45; 5–233)0.9
 N (%):    
  ASA score:   0.8
   1104 (20.8)  10 (17.9) 94 (21.1) 
   2326 (65.1)  37 (61.1)289 (64.9) 
   3 71 (14.2)   9 (16.1) 62 (13.9) 
  Clinical stage:   0.8
   T1c294 (58.7)  34 (67.7)260 (58.4) 
   T2114 (22.8)   11 (19.6)103 (23.1) 
   T3 93 (18.6)   11 (19.6) 82 (18.4) 
Surgical characteristics    
 Mean (median; range):    
  No. LNs removed 21.2 (20; 1–63)  23.8 (22; 9–49) 20.8 (20; 1–63)0.01
  No. obturator LNs removed  11.0 (10; 1–32)   11.1 (10; 3–27)  11.0 (10; 1–32)0.8
  No. external LNs removed 10.1 (9; 1–45)  12.7 (12; 1–38)  9.9 (9; 1–45)0.02
 N (%):    
  No. of LNs removed:   0.05
   1–10 32 (6.4)   4 (7.1) 28 (6.3) 
   11–20241 (48.1)  19 (33.9)222 (49.9) 
   >20228 (45.5)  33 (58.9)195 (43.8) 
Pathological characteristics    
 N (%):    
  RRP Gleason score:   0.4
   5–6213 (42.5)  20 (35.7)193 (43.4) 
   7233 (46.5)  31 (55.4)202 (45.4) 
   8–10 55 (11.0)   5 (8.9) 50 (11.2) 
  Extraprostatic disease165 (32.9)  20 (35.7)145 (32.6)0.6
  Positive LN status 61 (12.2)   5 (8.9) 56 (12.6)0.4
Follow-up characteristics    
 Mean (median; range):    
  Days of drainage  6.4 (5; 2–25)   6.1 (5; 2–17)  6.4 (6; 2–25)0.5
  Amount of lymphorrhoea, mL967 (555; 30–7750)1237 (615; 90–7550)933 (545; 30–7185)0.05
Figure 1.

The relationship between lymphorrhoea and patient age (A) and between lymphorrhoea and the number of LNs removed (B).

Both linear and logistic multivariable regression analysis showed that the number of removed LNs and patients’ age were the only two statistically significant predictors of CSL and total amount of lymphorrhoea after RRP and PLND (both P < 0.01; Table 2). The odds ratio for patients’ age and number of LNs removed was 1.05. This means that the probability of developing a CSL increases by 0.05 (5%) for every year of age or every LN removed. For example, the removal of 30 LNs involves a two-fold higher risk of having CSL relative to the removal of 10 LNs (Table 2). Moreover, cubic splines analyses show that the risk of developing a CSL significantly increases in patients aged >65 years and when a lymphadenectomy with >20 LNs removed is performed (Fig. 2A,B). The number of removed LNs and patients’ age were borderline predictors of total days of drainage, as well. As the number of LNs removed resulted one of the most informative predictor of lymphorrhoea, we also tested whether the region of PLND (obturator vs external iliac) was associated with lymphorrhoea after RRP and PLND (Table 3). Univariable and multivariable linear regression analyses showed that the number of external iliac LNs removed was an independent predictor of the total amount of lymphorrhoea (P= 0.001). The number of obturator LNs removed was not associated with the total amount of lymphorrhoea after RRP and PLND (P= 0.1) (Table 3).

Table 2.  Univariable and multivariable linear regression models predicting the days of drainage and the total amount of lymphorrhoea after RRP and PLND. Univariable and multivariable logistic regression models predicting the presence of a CSL after RRP and PLND
VariablePrediction of days of drainagePrediction of lymphorrhoeaPrediction of CSL
UnivariableMultivariableUnivariableMultivariableUnivariableMultivariable
βPβPβPβPORPORP
  1. NA, categorical variables require multiple hazard ratios, not shown for the sake of simplicity as long as statistical significance is not achieved. OR, odds ratio.

Clinical            
 Age0.080.070.090.0422.8<0.00123.5<0.0011.050.021.050.02
 Body mass index−0.030.5−0.040.4−23.60.1−23.50.20.980.30.960.4
 Clinical stage0.030.5−0.010.844.80.512.80.9NA0.9NA0.6
 Prostate volume−0.010.9−0.010.92.60.21.20.50.990.90.990.7
 ASA score0.010.90.030.532.30.7−25.00.81.170.50.880.9
Surgical            
 No. of LNs removed0.090.040.080.0621.6<0.00121.6<0.0011.040.011.050.007
 Surgeon0.010.8−0.020.730.30.213.10.6NA0.4NA0.8
Pathological            
 pT (extraprostatic vs organ-confined disease)0.10.030.080.1−48.80.7−126.90.71.150.61.240.4
            
 pN (pN1 vs pN0)0.080.070.050.470.10.7 114.00.70.680.40.520.2
Figure 2.

The relationship between the risk of developing a CSL and patient age (A) and between CSL and the number of LNs removed (B).

Table 3.  Analyses focusing on region of lymphadenectomy as predictor of lymphorrhoea, CSL and days of drainage. Univariable and multivariable linear regression models predicting the total amount of lymphorrhoea and days of drainage after RRP and PLND. Univariable and multivariable logistic regression models predicting the presence of a CSL after RRP and PLND
VariablePrediction of days of drainagePrediction of lymphorrhoeaPrediction of CSL
UnivariableMultivariableUnivariableMultivariableUnivariableMultivariable
βPβPβPβPORPORP
  1. OR, odds ratio.

Age0.080.060.090.0522.80.00123.8<0.0011.050.021.060.001
No. obturator LNs removed0.090.090.090.0417.80.0814.50.11.060.70.990.8
No. external iliac LNs removed0.040.040.040.426.2<0.00127.20.0011.060.0011.070.001

Finally, all the other clinical, pathological and surgical variables (namely surgeon, ASA score, prostate volume, pathological T and N stage, clinical stage and BMI) did not show any association with CSL and lymphorrhoea in univariable or multivariable analyses.

DISCUSSION

Although controversies exist regarding the use of a routine bilateral PLND, especially in men with low- and intermediate-risk prostate cancer [1,14–19], PLND remains the most accurate staging procedure for LN invasion [1,20]. Moreover, it has been shown that when PLND is indicated and planned at the time of RRP, this should be extended [7,8]. However, PLND may also account for substantial perioperative morbidity and complications, including CSLs and lymphorrhoea [5]. A CSL and lymphorrhoea might lead to infection, nerve injury, prolonged drain stay and, consequently, hospital stay. Moreover, CSL and lymphorrhoea are associated with an increased risk of deep venous thromboembolism and pulmonary embolism.

The present results showed that, among all the possible predictors, only age at surgery and the number of removed LNs achieved the independent predictor status in predicting CSL and lymphorrhoea (Table 2) in men treated with RRP and PLND. Interestingly, every additional LN removed increased the risk and every additional year of age increased the risk of having a CSL by 5%. The most informative thresholds in predicting the development of a CSL were being 65 years of age and having 20 LNs removed (Fig. 2A,B). Moreover, also considering only young patients (aged <65 years) or patients treated with limited lymphadenectomy (<20 LNs removed), the extension of lymphadenectomy and patient age remain important predictors of CSL, respectively (Fig. 3A,B). Moreover, external iliac lymphadenectomy resulted in a higher risk of lymphorrhoea and CLS compared with obturator LN removal (P= 0.001 vs P= 0.1, respectively).

Figure 3.

The relationship between the risk of developing a CSL and the patient age only considering limited lymphadenectomy cases with <20 LNs removed (A) and between the number of LNs removed only considering younger men aged <65 years (B).

It is not surprising that the extension of lymphadenectomy was associated with higher risk of lymphorrhoea and CSL. Indeed, the removal of LNs is clinically frequently associated with leakage from lymph vessels as confirmed by previous reports [5,9]. It is of interest that LN status (pN1 vs pN0) was not associated with the risk of having lymphorrhoea or of developing a CSL (Table 2). This is probably because RRP candidates usually show a limited LN metastatic spread not extended enough to cause lymphatic leakage. Less intuitively and more interestingly, in the present study we showed that patients’ age is also a strong predictor of CSL and lymphorrhoea. This second relationship might be explained by the fact that mature age is generally associated with remarkable changes in tissue healing aptitude and fluid components characteristics, namely, e.g. hypo-albuminaemia and hypo-proteinaemia that may explain the rationale for lymphorrhoea in older patients.

To our knowledge, the present prospective study represents the first report assessing clinical and pathological variables that may predict a CSL and lymphorrhoea after RRP and PLND. To foresee such complications has important implications at three stages during the management of patients with prostate cancer:

  • • treatment decision-making,
  • • intraoperative technique,
  • • and managing of the postoperative period.

First, predicting an increased risk of lymphorrhoea may help clinicians during treatment decision-making. Clinical choices require a difficult balance of benefits, life expectancy, co-morbidities and potential treatment-induced complications. The present results clearly suggest that patients’ age and lymphadenectomy indication should be contemplated before clinical decision, to maximize the advantages for the patient.

The second clinical implication in predicting lymphorrhoea resides in the intraoperative management. In the last decade, several measures have been suggested to prevent lymph leakage: closed suction drain, sclerotherapy, talc poudrage, fibrin sealant, i.m. injection of octeotride, pressure dressings, careful ligation of lymphatic vessels and avoiding the use of electrocautery [21–29]. The use of such measures might be pushed by the consciousness that a CSL and lymphorrhoea should be expected after surgery.

Thirdly, also postoperative management can be affected by our ability to predict a CSL and lymphorrhoea. Indeed, abdominal imaging should be considered in patients at high risk of lymphocoeles and lymphorrhoea, to exclude the presence of deep thromboses or simply to diagnose such conditions before their symptomatic presentation. Moreover, the decision about the anti-thrombotic prophylaxis regimen should be taken also considering the likelihood of a CSL or lymphorrhoea, as the use of low-dose heparin for thromboembolism prophylaxis has been associated to an increased risk of lymphorrhoea and lymphocoele formation [30–33]. Unfortunately, in the present study the effect of heparin on CSL and lymphorrhoea could not be addressed, as all patients received the same heparin prophylaxis.

Above and beyond those important clinical considerations, the present study has some limitations. First, although we excluded the presence of peri-anastomotic urine leakage as described above, we cannot exclude that the total amount of lymphorrhoea might be overestimated by the contemporary drainage of urine or blood. Moreover, the amount of lymphorrhoea that was not drained by the drains, namely, e.g. the amount of lymph collected in lymphocoeles, could not be calculated in the present study.

Although in most of the patients (94%) at least 10 LNs were removed, in some cases (6%) PLND resulted in fewer LNs being removed. In those cases, although an anatomically extended PLND, the individual variability in the number of LNs in the fibro-fatty tissue may have affect the total number of removed LNs. As previously published, in some individuals the fibro-fatty tissue may contain substantially more LNs than in others [34].

Finally, as all patients received the same standard heparin dosage, no conclusions regarding heparin use and the risk of developing lymphorrhoea and a CSL can be made. Besides these and others possible limitations, we consider the present study a novel and significant piece of information that may help clinicians in daily clinical practice.

In conclusion, the number of LNs removed and age at RRP showed a positive association with the amount of lymphorrhoea and the risk of developing a CSL. The most informative thresholds for predicting the development of a CSL were 65 years of age and 20 LNs removed. External iliac lymphadenectomy resulted in a higher risk of lymphorrhoea and CLS compared with obturator LN removal.

CONFLICT OF INTEREST

None declared.

Ancillary