• complications of lymphocele;
  • pelvic lymphoceles;
  • radical prostatectomy


  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. References

Background:  Lymphoceles (LC) represent a well-described rare complication post-radical prostatectomy (RP). Our aim was to determine risk factors and to develop possible prevention strategies for LC in a community-based study.

Methods:  Data from 1163 RP-patients from 67 clinics between January 2002 and December 2004 were retrospectively evaluated. Patients underwent pelvic imaging procedures/LC-management during 3 weeks of rehabilitation post-RP.

Results:  LC were identified in 304 patients (26%). Lymphadenectomy was carried out in 92% of patients (1001/1086 patients), from which 28% had LC (n = 277) versus 14% without lymphadenectomy (12/85, P = 0.007). Complications (lower limb edema, pain, thrombosis, infection and bladder compression) were observed in 9% of patients (28/304; 2.4% of total patients); necessitating therapy. LC therapy was carried out in 59 patients (5.9%) with pelvic lymph node dissection (PLND) and in no patients (0%) without PLND (P = 0.021). Risk factors included were patients' age, body mass index, prostate volume, TNM-classification, number of removed lymph nodes, previous surgery/therapy, heparin prophylaxis, surgical instruments and pelvic lymphadenectomy. Univariate analysis showed lymphadenectomy as the only significant risk factor for the development of LC post-RP (P = 0.007). When applying multivariate analyses using stepwise logistic regression, only lymphadenectomy was associated with a significant risk for lymphoceles (odds ratio = 2.6, 95% CI = 1.3–4.9, P = 0.004). Adjusting for other factors, no other factor came close to being significant (P < 0.05). All symptomatic LC were successfully treated without further sequelae.

Conclusions:  Subclinical LC post-RP are more common than thought, and rarely necessitate intervention. Pelvic lymphadenectomy represents the only significant factor contributing to LC-development. Because of this, prevention remains difficult.


  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. References

A lymphocele (LC) is a collection of lymphatic fluid as a consequence of surgical dissection of afferent lymphatic vessels. It is a well-documented complication after pelvic surgery, with an incidence up to 27%.1 Nowadays, it is the most frequent non-functional complication of radical prostatectomy and pelvic lymph node dissection (PLND).

LC-development is a problem for patients, because of its health-related sequelae, such as secondary infection and thromboembolic events as a result of compression of pelvic vessels. A correlation between LC and deep venous thrombosis (DVT)/pulmonary embolism was shown.2 Furthermore, approximately 50% of all re-interventions carried out in patients post-radical prostatectomy (RP) were for LC-management.3

A positive association between the number of removed lymph nodes (LN) and age at RP with the risk of LC development was reported. The most informative predictive thresholds were 65 years-of-age and 20 LN.4

The exact mechanism (s) of LC-formation and risk factors have not been completely studied. In the present retrospective community representative study, tried to obtain more answers about the risk factors, development and possible prevention strategies for this problem.


  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. References

Between January 2002 and December 2004, data from 1163 patients after RP were retrospectively examined for postoperative LC incidence, risk factors and therapeutic management. RP were carried out by various surgeons from 67 urological departments (35% of patients were from five high-volume centers with ≥250 RP annually). There were no selection criteria except extravasations in cystography. All patients underwent 3 weeks of rehabilitation in a specialized hospital where they were recruited and where all examinations were either carried out or organized. Every patient underwent at least one pelvic ultrasound (US) at the beginning of hospitalization to diagnose/exclude any complications. LC were diagnosed clinically, during regular US carried out by a urology team and then by an experienced radiologist, as well as postoperative abdominal computed tomography (CT). Patients who had LC were controlled before discharge from hospital. The LC volume was measured by 3-D volume measurement during US.

If the patient had LC-related symptoms (e.g. pain, edema etc.), it was considered clinically symptomatic/complicated LC, otherwise it was considered as subclinical LC (incidental diagnosis).

LC were defined as any clear fluid collection anatomically associated with the pelvic sidewall without suspected anastomotic insufficiency (after cystograms) or resolving hematomas. In case of suspicion, CT was carried out. Aspiration or tube drainage in symptomatic cases confirmed the diagnosis. PLND was defined as excision of all fibro-fatty tissue along the external iliac vein, including the bifurcation of the common iliac artery together with fibro-fatty tissue within the obturator fossa. Sentinel PLND was carried out under radioguided surgery (with 250 MBq 99 m technetium nanocolloid). These were found in internal iliac LN (37.7%), together with the external iliac vein (19%) or with the obturator fossa (9.2%) as well as presacral, pararectal and paravesical (16%).5 All demographic, surgical and follow-up parameters were included in a univariate, as well as multivariate analysis seeking any significance of LC-development. All variables were considered as continuous variables without any cut-off values or size limits. Furthermore, risk factors were considered without subclassifications (e.g. PLND, if extended or not, which technique etc.), selection or exclusion (whether it seems logically applicable or not). The present study was intended to be community-based without adjustment for surgeon or department. Statistical analysis was only directed to find answers about the possible risk factors in general.

RP was carried out in 89.7% of patients (1024/1141), laparoscopic extraperitoneal prostatectomy (EERPE) was carried out in 6.5% of patients (74/1141) and perineal prostatectomy in 3.8% of patients (43/1141). PLND was carried out in 95.8% (931/972) of the RP patients, 78.9% (56/71) of EERPE patients and 22.9% (8/35) of perineal prostatectomy patients (carried out perineally with the same yield of LN [median 10]; P < 0.001). Information about operation technique and PLND was missing in 22 and 77 patients, respectively. All patients had one 24-Fr closed drainage system crossing the extraperitoneal space and fixed in the right lateral lower abdomen, postoperatively. This was removed when the amount of fluid was ≤30 mL/24 h. Prophylactic subcutaneous abdominal or upper arm low molecular weight heparin (LMWH) was given starting 1 day before surgery for the whole hospital stay.

Furthermore, extensive search of the PUBMED database was carried out and all previous experiences were used to explain and optimize the developed concepts.

Statistical analysis

For comparison of continuous variables (e.g. age, body mass index) between different patient groups (e.g. with/without lymphocele) the Mann–Whitney U-test was used. Categorized data (e.g. tumor classification, operation technique) was analyzed with the χ2-test. Logistic regression models were used to analyse potential factors for LC-occurrence. Because of the retrospective character of the study, there were some missing data in several variables, resulting in different total case numbers for the various statistical reports. P-values below 0.05 were regarded as significant. All calculations were carried out using software STATISTICA (release 8, StatSoft, Tulsa, OK, USA).


  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. References

Patients and tumor characteristics

Table 1 shows the relationship of age, body mass index, pathology obtained prostate volume and preoperative prostate-specific antigen level and the risk of LC development.

Table 1.  Correlation of continuous variables with lymphocele formation in 1163 patients
 LCNo LCP-value
  1. BMI, body mass index; LC, pelvic lymphoceles; LN, lymph nodes.

Prostate volume5015–1804210–1800.070
Number of removed LN101–3191–360.589

LC occurred in 27% of patients (201/741) who had pT2 tumors, in 25% (85/346) with pT3 and in 19% (6/31) with pT4 tumors (P = 0.459). Furthermore, 27% of patients (12/45) with missing tumor stage data had LC.

LC was found in 22% of patients with Gleason score ≤5 (38/176), in 26% (84/321) with Gleason score 6, in 30% (97/323) with Gleason score 7, in 23% (17/74) with Gleason score 8 and 17% (10/60) with Gleason score ≥9 (P = 0.107). 28% of patients (58/209) with missing Gleason score data had LC.

In addition, LC occurred in 22% of grade 1 (G1) tumors (13/60), in 26% of G2 (181/700), in 27 of G3 tumors (71/265; P = 0.715) and in 28% of patients (39/138) with missing G data.

LC developed in 29% of patients (19/65) with LN involvement (N+) versus 27% of patients (264/964) with free nodes (P = 0.747). Of the patients with missing LN status, 16% (21/132) developed LC.

Similarly, LC occurred in 25% of patients (61/240) with positive surgical margins (R+) versus 25% (151/611) (P = 0.831) with free margins (R0) and in 29% of patients (92/312) with missing surgical margin status.

There was no significant difference in LC-incidence between the patients with and without preoperative anti-androgen therapy, 29% (31/107) versus 26% (273/1056) (P = 0.484). No significant difference in incidence was found between patients with versus without postoperative radiotherapy, (22% [17 from 79]vs 26% [287 from 1084]; P = 0.333).

There was no significant difference between university, teaching, tertiary and private hospitals (31 vs 25 vs 29 vs 31%, respectively).

Surgical factors

As shown in Table 2, LC were found in 27% of patients (275/1024 patients) after RP, in 30% (22/74) after EERPE and in 16% (7/43) after perineal prostatectomy, showing no significant difference in incidence (P = 0.253). Many cutting and/or coagulation devices were used during the operations, such as bipolar coagulation or ultrasonic shears.

Table 2.  Correlation of categorical variables with lymphocele formation in 1163 patients
 % LCn LC/n totalP-value
  1. EERPE, extraperitoneal endoscopic radical prostatectomy; N+, lymph node metastasis; N0, tumor free lymph nodes; perineal RP, perineal radical prostatectomy; PLND, pelvic lymph node dissection; pT, pathological stage; R+, positive surgical margins; R0, free surgical margins; RRP, retropubic radical prostatectomy; SLN, sentinel lymph node dissection.

Gleason ≤625122/4970.059
Gleason 73097/323
Gleason 8–102027/134
Anti-androgen therapy2931/1070.484
No anti-androgen therapy26273/1056
No radiotherapy26287/1084
Type of hospital  0.132
Operation technique  0.253
Wide excision2228/182
No LAE1412/85
No SLN26291/1112

LC developed in 28% of patients (53/192) using a nerve-sparing technique versus 22% (28/128) using a wide excision (P = 0.248). The information about nerve-sparing technique was missing for 843 patients. Of these patients, 26% (223/843) developed a LC. Lymphadenectomy was carried out in 92% of patients (1001/1086 patients with available data about PLND status), from which 28% had LC (n = 277) versus 14% without PLND (12/85; P = 0.007). Of the patients with missing data about PLND, 19% (15/77) showed a LC. Logistic regression analysis (Table 3) showed PLND as the only significant risk factor for LC (odds ratio = 2.6; 95% confidence interval 1.3–4.9; P = 0.004). There was no significant difference in regard to the number of resected LN between the two groups: median 10 (range 1–31) LN in patients with LC. Median 9 (range 1–36) LN in patients without LC (P = 0.589). Furthermore, 25% (13/51) of the patients with sentinel LN dissection and 26% (291/1112) of the patients without sentinel LN dissection had LC (P = 0.914). Therapeutic intervention was carried out in 59 LC-patients. There were 5.9% (59/1001) LC requiring intervention in patients after PLND compared with no therapy in patients without PLND (P = 0.021). There was no correlation between the number of removed LN and the need for intervention therapies: median 9 (range 1–36) LN in patients without therapy versus median 10 (range 2–27) LN in patients with therapy, respectively (P = 0.476). There was no correlation between other parameters and LC requiring therapy.

Table 3.  Multivariate analysis (logistic regression model) of factors that were significantly correlated with lymphocele formation in univariate analysis
 Odds ratio95% CIP-value

Lymphocele characteristics

LC were diagnosed in 304 patients (26%) through postoperative abdominal US with a median volume of 83 mL (range 20–1800 mL). There was no correlation between the localization of the LC (26% left side, 57% right side, 11% bilateral, 6% paravesical) and the operation technique (P = 0.404). A total of 83% of the LC were unilateral.

Complications were observed in 9% (28/304) of LC-patients: lower limb edema occurred in 4.3% (13 patients), pain in 3.0% (9 patients) thrombosis in 1.3% (4 patients), infection in 1.3% (4 patients) and compression of the bladder (progressively increasing incontinence) in 0.3% (1 patient). There was no correlation between the incidence of complications and the surgical approach (P = 0.393).

Data about the complete progress of LC-volume over the whole follow-up period were complete in 48 cases. The median LC-volume was higher in patients with complications than in patients without (median volume 120 mL [range 20–1800 mL]vs 83 mL [range 20–500 mL], respectively; P = 0.258). Patients with <100 mL LC-volume had 23% incidence of complications versus 27% in patients with ≥100 mL LC-volume (P = 0.738). Patients with LC therapy had a significantly higher LC volume than patients without therapy (median 110 mL vs 48 mL, P = 0.049). LC therapy was carried out prophylactically in some patients with large LC to prevent complications.


  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. References

The present patient cohort from 67 clinics represents a non-selected group of men who underwent pelvic imaging procedures during standard 3-week rehabilitation post-RP. The present cross-sectional study represents the community in Bavaria Governorate, because it corresponds to a randomly selected group of patients from all patients who underwent radical prostatectomy in the given period (3 years). Furthermore, the patients included represent small and high-volume clinics, which is a further aspect of a community representative study.

The present results confirm the published data that subclinical-LC occur with higher frequency.1,6 These resolve spontaneously and rarely become symptomatic requiring treatment. Diagnosis is mostly with US, although small LC can be obscured necessitating abdominal-CT in unclear/complicated cases. Meanwhile, clinically relevant LC might reach large sizes and/or cause many complications.

The strict extraperitoneal access seems responsible for the absence of lymph drainage causing 2.4% symptomatic-LC (28/1163 patients). Interestingly, the reported total complication rate after EERPE is 9.68%, which renders the symptomatic-LC as the most frequent post-RP-complication.7

There is a wide range of differences in LC-incidence between studies, suggesting differences in the thoroughness of operation techniques, different identification methods or diagnostic imaging, but the actual etiological and preventive mechanism(s) of this phenomenon remain speculative.

Several surgical devices have been investigated, hoping to achieve better hemostasis and reduce seroma formation. Some of our surgeons speculated an increase in LC-incidence after excessive electrocauterization of lymph vessels (LV), rather than ligation. This is consistent with the findings of Porter et al.8 Furthermore, no significant effect of ultrasonic scalpel was reported,9 which suggests ligation of lymphatics as an important aspect of PLND-technique. Meanwhile, 14% of patients had LC without PLND, which could suggest rough surgical manipulation, abnormal lymph pathways or pelvic collections (e.g. because of pelvic drain). These were all asymptomatic and necessitated no treatment.

Interestingly, we observed that only PLND, as such, significantly affected the rate of LC-formation regardless of the number of LN, which opposes Naselli et al.10 The definition of an extended LN-dissection is controversial. LN-count has been suggested by some authors to differentiate an extended from limited dissection.11 However, as LN-count is not available to the surgeon during surgery, it is wise to define this solely on the template excision.12 The current study showed no increased LC-incidence with increasing number of LNs, supporting our theory that LC-formation depends solely on the technique with adequate ligation of lymphatics. Furthermore, there was no correlation between the number of LN and the need for intervention therapies for LC (P = 0.476). Capitanio et al.4 reported, in a patient collection from an academic center, an informative threshold of 20 LN to increased LC risk, which was not found in the present community-based study.

We believe that every surgeon dealing with pelvic-LV will produce his own rate of LC. Sansalone et al.13 reported a similar observation that the injury of the lymphatic network close to the external iliac arteries is the main route for lymph collection after kidney transplantation and that this can be greatly decreased by using the common iliac axis as the site for vascular anastomoses away from this lymphatic injury.

Positive-LN were suspected to increase LC-formation after gynecological operations.14 This was not found in the current study and could be explained by lymphatic obstruction. Similarly, the reason that radioguided surgery often fails to detect all positive nodes was suggested to be its high false-positive rate, explained by the fact that tumor cells obstruct the LV, preventing tracer transport to all “sentinel nodes”.15

Surgeons should be more aware of this postoperative complication, particularly with high molecular weight heparin (HMWH) prophylaxis.16 Prophylactic heparin has been implicated as a possible risk factor of LC.17–19 In the current study, LMWH injection in the abdominal wall or arm was used as a perioperative standard prophylaxis for all patients. There was no observed increased incidence of LC-formation compared with the literature. The same results were stated by Koch and Smith,20 encouraging the use of prophylactic LMWH injections in the arm rather than the lower limbs. Furthermore, studies evaluating prophylaxis starting postoperatively showed little apparent compromise in efficacy compared with when prophylaxis was initiated before surgery.20 Thus, our concept of starting LMWH preoperatively and continuing throughout the hospital stay could be recommended.

Araki et al. showed in a large RP series that a drain might not be necessary in all cases.21,22 This was consistent with recent studies in other surgical specialties.23 There was no decrease in LC-incidence alleviating the role of the drainage in reducing the risk of LC. In contrast, there were reports favoring drainage.14,24 Meanwhile, it is wise to use a postoperative drain after EERPE where there are already punctures.

Furthermore, leaving pelvic drains in situ for some days was reported to impair the reparative and absorptive capacities of peritoneum, contributing to the problem that its use was intended to prevent.14,25 Consequently, some intend to leave the peritoneum open23 or to use an omental J-flap26 to avoid pelvic collections. The former concept was recently reported to reduce the rate of post-RP-LC.7

Similar to axial seroma after axillary lymphadenectomy,27,28 there was a progressive increase in the incidence and size of LC during the first postoperative month as the patients resumed their normal activities. These LC were soon stable in size, requiring intervention in just 13.4% of cases. Contemporary to axial seroma, prophylactic postoperative pressure bandage is not applicable for pelvic-LC, which makes this observation irrelevant for LC-prophylaxis, but it might be relevant for the timing of treatment. Though, following up LC during the postoperative month before planning interventional therapy, in the absence of complications, seems appropriate.

Obesity has been identified as a LC risk factor after kidney transplantation,29 which was not found in current study. Similarly, adjuvant therapies, such as hormonal ablation, were not a risk factor. The small, but significant, difference in age regarding LC-incidence (univariate analysis) in the present study was probably a result of the high number of patients and younger patients compared with the published informative threshold (63 vs 65 years, respectively).4

The most important issue when dealing with lymphoceles remains its prevention. This is difficult, because the actual etiological factors are not definitely described in the literature, but current multivariate analysis showed PLND as a single independent risk factor. Injury or division of efferent LV during PLND might cause subsequent accumulation of lymph fluid in the retroperitoneal space. Pelvic operations require thorough knowledge about the normal anatomy of lymphatics and its multiple variants. Careful ligation of lymphatics at all margins of resection seems a logical tool to prevent subsequent LC-formation. Another way is prophylactic opening of the peritoneum at the operation end.7 As mentioned, it is advisable to avoid the administration of heparin in the lower limbs. Application of fibrin sealant was recently published as a preventive measure.30 Isolation and ligation of leaking LV reported for recurrent symptomatic inguinal LC is difficult to apply to pelvic lymphatics, which have complex anatomy.

Last, several publications have shown that lymph, like plasma, contains coagulation factors, and delayed coagulation and wound closure in dissected LV because of specific characteristics in their cellular structure and the composition of clotting factors in lymph fluid could result in LC.2,31

The present study has some limitations. First, its retrospective character, but a prospective study will have ethical and technical problems. The patient number is enough to develop a generalized concept, but our intention was to provide a frame that could hopefully be compared with the other experiences. Also, dealing with the data as community-based without specifications/exclusion criteria, for example, surgeon/approach related, was intended to generalize the concept. This could be seen as critical by some authors. Last, as a rare complication, it is difficult to describe LC-etiology and progress precisely, but hopefully accumulating experiences could help in this matter.

PLND is the only independent significant factor contributing to the development of post-RP-LC, which makes prophylaxis difficult. Complications are rare and interventions are rarely necessary, but high suspension is advisable.


  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. References