• robotics;
  • lymphocele;
  • pelvic lymph node;
  • lymphodeveatary complications


  1. Top of page
  2. Abstract

Study Type – Therapy (case series)

Level of Evidence 4

What’s known on the subject? and What does the study add?

Lymphocele formation after open pelvic lymph-node dissection is a known complication. However, reported incidences using the robotic approach are unclear and likely underestimated. The present study aims to better understand the true incidence of lymphocele formation after RARP.


• To determine the incidence and predictive factors of lymphocele formation in patients undergoing pelvic lymph node dissection (PLND) during robot-assisted radical prostatectomy (RARP).


• Between April and December 2008, 76 patients underwent PLND during RARP for ≥cT2c, prostate-specific antigen level ≥10, Gleason score ≥7 prostate cancer.

• All patients were prospectively followed up with pelvic computed tomography 6–12 weeks after the procedure.

• All patients received s.c. heparin preoperatively and postoperatively. PLND was limited to zones 1 and 2 as defined by Studer.

• Plasma-kinetic bipolar forceps were used for haemostasis during PLND.


• At a mean follow-up of 10.8 weeks, 51% (39/76) of patients had developed a lymphocele. Of these 39 lymphoceles 32 (82%) were unilateral and seven (18%) were bilateral.

• The mean (range) lymphocele size was 4.3 × 3.2 (1.5–12.3) cm; 41% of lymphoceles were <4 cm, 53.9% were 4–10 cm, and 5.1% were >10 cm in diameter. Six of the 39 lymphoceles (15.4%) were clinically symptomatic. The symptoms were as follows: pelvic pressure in five patients, abdominal distension with ileus in three patients, leg pain/weakness in one patient and costovertebral tenderness in one patient. Two lymphoceles required intervention.

• On the logistic regression model the presence of nodal metastases, tumour volume in the prostate specimen and extracapsular extension (ECE) were independent risk factors for the development of a lymphocele.

• There was no correlation between estimated blood loss, body mass index, pathological Gleason score or number nodes dissected and the presence of lymphocele.


• The incidence of lymphoceles was higher than anticipated given the believed protective effect of the transperitoneal approach against lymphocele formation.

• The risk of lymphocele seemed to increase linearly with the presence of more extensive disease, particularly ECE and nodal involvement.

• The benefit of PLND during RARP should be weighed against the elevated risk of lymphocele formation and its potential complications.


pelvic lymph node dissection


robot-assisted radical prostatectomy


extracapsular extension


prostate cancer




Jackson Pratt


postoperative day


estimated blood loss


body mass index


  1. Top of page
  2. Abstract

The introduction of PSA screening and increased patient awareness of prostate cancer (CaP) has led to a downward stage migration in the disease, leading in turn to a decreased probability of lymph node metastasis [1,2]. Indeed, based on contemporary nomograms, patients with low-risk CaP features (PSA <10 ng/mL, Gleason score ≥6, clinical stage ≤T2a) have <2% risk of node-positive disease [3]. Thus, routinely performing pelvic lymph node dissection (PLND) during robot-assisted radical prostatectomy (RARP) does not seem justified given its potential complications, costs and the extra surgical time associated with RARP. Instead, patients with intermediate- or high-risk clinically localized CaP features (Gleason score ≥7, PSA ≥10 ng/mL, cT2b) are still candidates for PLND as the yield of finding node-positive disease increases between 7–20%, depending on patient variables [4,5]. Although the required extent of PLND is debatable, PLND is considered to be the most accurate and reliable staging procedure for detecting lymph node invasion and to determine lymph node status [6–8]. The rationale for performing PLND in these patients is that early detection of micrometastatic disease can allow prompt initiation of adjuvant therapy, which is known to reduce disease progression and provide a survival benefit to these patients [9]. Furthermore, some studies show a potential curative role for PLND among patients with micrometastatic disease [6,10].

The rate of perioperative complications after PLND has been reported to be 4–20%[7,11,12]. The most frequent complication is lymphocele formation, although other complications have been described such as vascular injuries, iliac vein thrombosis, neuropathy and development of lymphoedema [7,11–14]. Furthermore, although most lymphoceles appear to be subclinical and therefore require no intervention, some can lead to symptoms and secondary complications such as abdominal pain, leg pain, lower limb oedema, constipation, urinary frequency, deep vein thrombosis and infection or sepsis [7,11]. Some of these cases may require surgical intervention such as guided drainage or surgical unroofing [15]. The reported incidence of clinically detected lymphoceles after open PLND ranges from 2 to 61%[13,16] and depends on several factors, including the extent of dissection, lack of peritoneal communication and the use of thromboembolic prophylactic medication. In a small series of patients followed up using either pelvic sonography or CT, the frequency of subclinical postoperative lymphoceles was found to be 27 and 30%, respectively [11–14].

As most RARP procedures are currently performed using a transperitoneal approach, it is thought that the incidence of lymphocele formation after RARP may be decreased when compared with the open extraperitoneal PLND approach during Retropubic Radical Prostatectomy (RRP). The premise is that open lymphatic channels would drain freely into the abdominal cavity while the lymphatic fluid is reabsorbed by the peritoneum. The aim of the present study was to evaluate the true incidence of lymphocele formation in patients undergoing PLND during RARP and to identify possible predictive factors associated with it.


  1. Top of page
  2. Abstract

The data were prospectively collected and analysed using our Prostate Cancer Database. Patients eligible for the present study were those with biopsy-proven CaP, who had intermediate- or high-risk features according to D’Amico’s risk stratification (digital rectal examination ≥cT2b, preoperative PSA ≥10 ng/dL, or Gleason score from biopsy ≥7), and were therefore candidates for concurrent PLND [17]. All PLND cases were bilateral and performed by a single surgeon (V.R.P.) after RARP was completed. Preoperative evaluation using CT, abdomen and pelvis and bone scans, was performed on a case-by-case basis. Between April and December 2008, 466 consecutive patients underwent RARP at our institution for CaP. Of these, 80 patients fulfilled the eligibility criteria and were asked to participate in the study. Four patients declined to participate, and 76 patients agreed and gave written informed consent.

After general anaesthesia, all patients received 5000 IU of s.c. heparin before surgery began. After the field was prepared and draped and the robot docked, patients underwent RARP using a six-port technique as previously described [18], followed by PLND. The boundaries of dissection included removal of lymphatic tissue from the level of the external iliac vein laterally and anteriorly, to the obturator nerve posterior, extending proximally to the common iliac artery bifurcation and distally to the proximal femoral canal to include the node of Cloquet (Fig. 1). Monopolar scissors on the right and the plasma-kinetic (PK) bipolar forceps (Gyrus Medical Ltd, Maple Grove, MN, USA) on the left arm were used for haemostasis during PLND. No surgical clips were used. The procedure was repeated on the contralateral side using the same landmarks. At the end of the procedure, a single Jackson Pratt (JP) drain was routinely left in place.


Figure 1. View of PLND.

Download figure to PowerPoint

Postoperative management in the patients in the present study did not differ from patients who would have not undergone PLND. Patients were continued on every 12-hour (BID) regimen of 5000 IU s.c. heparin until discharge from the hospital. On postoperative day (POD) ♯0 patients were instructed to ambulate and given a clear liquid diet. On POD ♯1, diet was advanced and patients were released home if discharge criteria were met (i.e. vital signs were stable, patient was able to ambulate, tolerate a regular diet and pain was controlled with oral medications). The JP drain was removed if output was <100 nL/8 h. shift. In those patients with increased JP drain output, fluid was checked to rule out the presence of a urine leak and removed before the patient was discharged if fluid creatinine levels from the JP drain were comparable with serum levels. Patients returned to the office 4–6 days after surgery for follow-up, routine cystogram and Foley catheter removal if the study was negative for leak. All patients were prospectively followed with an abdomen–pelvis CT scan with i.v. contrast at 6–12 weeks after surgery, and the images were reviewed by an experienced radiologist.

The present study was performed with the approval of our institutional review board and ethics committee and all patients gave informed written consent before participating in the study. Additionally, patients were informed that follow-up CT scan would be performed only in the context of the study and was not part of the standard of care. The potential risks of i.v. dye use and long-term risks of radiation exposure were also discussed.

Continuous variables were reported as mean sd. When indicated, continuous variables were compared using the Student’s t-test. A univariate analysis was performed using a logistic regression model to determine those factors that predicted lymphocele formation. Subsequently, a multivariate stepwise logistic regression model was constructed to assess whether the variables associated with lymphocele formation in the univariate analysis were independent predictors. Variables with more than two categories included in the univariate and multivariate analysis were bifurcated into binary parameters so that they could be included in the analysis. The information was processed with SigmaStat v.3.5 software (Systat Software, San Jose, CA) and a double-sided P value of <0.05 was considered to be statistically significant.


  1. Top of page
  2. Abstract

Demographic data are shown in Table 1. The mean (range) patient age was 60.9 (45–80) years and the mean (range) PSA level was 6.7 (1.4–28) ng/mL. As expected, the majority of patients had a Gleason score ≥7 in the postoperative pathology (65/76 or 85.6%) and had pathological stage ≥T2b (69/76 or 90.8%). The mean (range) time to performing PLND dissection bilaterally was 12.7 (9–17) min. There were no intraoperative complications including vascular, gastrointestinal or nerve injuries. The mean (range) estimated blood loss (EBL) for the complete surgical procedure (RARP with PLND) was 108.1 (20–250) mL.

Table 1.  Study population demographics
Patient characteristicValue
Mean (range) age, years60.9 (45–80)
Mean (range) PSA, ng/mL6.7 (1.4–28)
Mean (range) BMI, kg/m228.2 (21–41)
Preoperative Gleason score, n (%)
 67 (9.2)
 750 (65.8)
 813 (17.1)
 96 (7.9)
Clinical stage, n (%) 
 T1c37 (48.7)
 T2a29 (38.2)
 T2b8 (10.5)
 T32 (2.6)
Postoperative Gleason score, n (%)
 6 11 (14.4)
 750 (65.8)
 85 (6.6)
 910 (13.2)
Pathological stage, n (%) 
 T2a7 (9.2)
 T2b40 (52.6)
 T3a12 (15.8)
 T3b17 (22.4)

Five patients had increased JP drain output by POD♯1 (>100 mL/8-h shift). In three of these patients the output decreased to normal ranges by late POD♯1 and no further evaluation was performed before the removal of the drain and discharge from hospital. In the other two, the fluid was tested for creatinine levels and urine leak was ruled out in both patients. One patient was discharged on POD♯2 after fluid was tested and the JP drain was removed. The other patient developed postoperative ileus and was discharged on POD♯4 after ileus had resolved. The JP drain was removed before patients were discharged, with outputs ranging from 75 to 110 mL/8-h shifts. Notably, none of these patients ultimately developed a symptomatic lymphocele.

Although specific intraoperative EBL associated with the PLND portion of the procedure was not evaluated, this was negligible. Both mean hospital stay and mean time to catheter removal were higher within patients who developed a lymphocele, although these did not reach statistical significance (P= 0.22 and P= 0.18, respectively; Table 2)

Table 2.  Intraoperative data
Mean (range) operating time, min  
 Overall76.9 (60–120)
 PLND portion only12.7 (9–17)
Mean (range) EBL, mL108.1 (20–250)
Intraoperative complicationsnone
Mean (range) hospital stay, days  
 Overall1.35 (1–6)0.22
 Patients without lymphocele formation1.2 (1–2) 
 Patients with lymphocele formation1.46 (1–5) 
Mean (range) time to catheter removal  
 Overall6 (4–14)0.13
 Patient without lymphocele formation5.6 (4–7) 
 Patient with lymphocele formation6.3 (4–14) 

At a mean follow-up of 10.8 weeks (minimum follow-up of 6 weeks in all patients), the overall frequency of lymphocele formation was 51% (39/76). The mean (range) lymphocele size was 4.3 × 3.2 (1.5–12.3) cm with 41% lymphoceles <4 cm, 53.9% 4–10 cm, and 5.1% >10 cm in diameter. Unilateral lymphoceles were seen in 32/39 (82%) while seven patients (18%) developed bilateral lymphoceles. The overall incidence of positive lymph nodes in the series was 7.9% and the median (range) lymph node yield was six (2–12) nodes per side (Table 3).

Table 3.  Frequency and location of lymphoceles
Mean (range) follow-up, weeks10.8 (7–14)
  • *

    CT-guided placement of JP drain.

Median (range) lymph node yield6 (2–12)
Incidence of positive lymph nodes, n/N (%)6/76 (7.89)
Overall lymphocele incidence, n/N (%)39/76 (51)
Unilateral lymphoceles, n/N (%)32/39 (82)
Bilateral lymphoceles, n/N (%)7/39 (18)
Incidence of symptomatic lymphoceles, n/N (%)6/76 (7.9)
Symptomatic lymphoceles, n/N (%)6/39 (15.4)
 Pelvic pressure5
 Leg pain/weakness1
 CVA tenderness1
Lymphocele size, cm4.3 (1.5–12.3)
 Mean (range) 41
 <4 cm (%)53.9
 4–10 cm (%)5.1
 >10 cm (%) 
Incidence of intervention* required, n/N (%)1/76 (1.3)

Six patients developed clinically symptomatic lymphoceles (15.4% of the total number of patients who developed a lymphocele and 6.6% of the study group). Pelvic pressure was the most prevalent presenting symptom, occurring in five of the six patients. Abdominal distension with ileus requiring admission was seen in three patients, leg pain/weakness in one patient, costovertebral tenderness in one patient and fevers in one patient. Of these, only the patient who presented with fevers and additionally complained of pelvic pressure and leg pain required intervention, because of an infected lymphocele. The patient was managed successfully with antibiotics and CT-guided drainage of the fluid collection with drain placement. Fluid culture revealed gram-negative bacilli (Table 3). All three patients admitted for ileus were successfully managed conservatively with bowel rest and prokinetic medications. The remainder of patients presenting with pelvic pressure but no signs of infection were also successfully managed conservatively with pain control medication.

On the logistic regression model the presence of nodal metastases, tumour volume in the prostate specimen and ECE were independent risk factors for the development of a lymphocele. There was no correlation between EBL, body mass index (BMI), pathological Gleason score or number of nodes dissected and the presence of lymphocele (Tables 4,5)

Table 4.  Univariate analysis
VariableOdds ratio (Pt. Est.)Lower C.L oddsUpper C.L oddsP
  1. HGPIN, high grade prostatic intraepithelial neoplasia; SVI, seminal vesical invasion; PNI, percentage of nodal invasion; PSM, positive surgical margins.

Age ratio1.0360.9751.10.25
Clinical stage0.4450.1661.1900.11
Preoperative Gleason score0.7800.2752.2120.64
Pathological Gleason score1.0940.3303.6260.88
Percent tumour0.9640.9360.9940.02
Gland size1.0530.9961.1130.06
Operating time0.9810.9431.0210.34
Table 5.  Predictors for lymphocele formation after RARP
VariableOdds ratio (Pt. Est.)Lower C.L oddsUpper C.L oddsP
  1. Multivariate stepwise analysis to maximum yields.

SV involvement0.1180.0140.9930.05
Nodal involvement2.1590.4440.990.03
Gland size1.04111.0830.04


  1. Top of page
  2. Abstract

Pelvic lymph node dissection is considered to be the most accurate staging procedure for evaluating the presence of regional lymph node involvement in patients with CaP [6–8]. With the increased use of PSA screening and patients’ awareness of CaP, we have witnessed a remarkable tumour stage migration, raising the question of whether patients require routine PLND when undergoing RARP [1,3]. Detractors of the procedure have challenged its true therapeutic benefit in CaP, while reported series have shown increased associated costs and substantial morbidity, including intraoperative vascular and nerve injuries, increased risk of intraoperative bleeding and risk of postoperative formation of lymphoceles [7,11–16].

Our data show that, when performed by surgeon experienced in robot-assisted surgery, PLND during RARP is feasible and not associated with an increased risk of intraoperative complications. Although the need for extended PLND for prostate cancer is controversial, it was encouraging to learn that, even when performing a standard PLDN procedure, we were able to obtain a median of six lymph nodes per side. Furthermore, performing a PLND added a mean of only 12.7 min of overall surgical time to the procedure.

Lymphocele formation after PLND is by far the most common complication. Its incidence after PLND during open and minimally invasive radical prostatectomy has been evaluated previously by other authors. Freid et al.[14]. evaluated the frequency of subclinical lymphocele formation after laparoscopic PLND during radical prostatectomy and reported a 30.4% overall incidence. Similar findings were reported by Solberg et al.[13] who compared laparoscopic with open PLND. The authors found a 37% incidence, while the incidence among the open group was higher at 61%. In the present study we found a 51% incidence of lymphoceles. This figure is even more concerning given that we performed limited PLND. It is likely that the incidence of lymphocele formation would increase with more extended PLND. Additionally, our results with robot-assisted transperitoneal PLND are somewhat higher than previous reports. This discrepancy may be attributed to different factors. First, we have consistently used antithromboembolic medication in our patients. This is a known factor for increased risk of lymphocele formation. In a study by Tomic et al.[19], patients randomized to the placebo arm developed significantly fewer lymphoceles compared with those who received preoperative low molecular weight heparin. However, we believe that prevention of a potentially life-threatening event such as a pulmonary embolism prevails over the risk of lymphocele formation. Indeed, in our experience of over 3500 RARPs, the incidence of pulmonary embolism is only 0.5%[20]. A second reason that can, at least in part, explain a higher incidence of lymphocele formation in the present series is the type of haemostatic agents used during PLND. Lymphoceles occur as a result of lymph leakage from afferent lymphatic channels transected during dissection. As a result, their incidence should decrease significantly by attending to careful sealing of these channels using adequate haemostatic tools. Series in which clips or ligatures have been used for ligation of lymphatic channels have consistently reported a low incidence of lymphoceles [7,14], whereas the use of fibrin sealant agents does not seem to reduce the risk of lymphoceles [21]. In the present series we used a combination of monopolar energy and PK bipolar forceps. From our results it seems clear that the PK bipolar forceps do not adequately seal the lymphatic channel and prevent lymphocele formation. Owing to the exceedingly high lymphocele rates observed in our study, we have since switched to the use of Hemo-lock clips while performing PLND.

Despite the high incidence of lymphocele formation after PLND, most remain asymptomatic. In the present series, 7% of patients developed a symptomatic lymphocele (15.4% of those who developed a lymphocele), while only one required intervention. These findings are similar to other investigators [13,14]. Contrasting findings were recently reported by Naselli et al.[11]. The authors reported an overall 12.6% incidence of lymphocele formation as diagnosed by follow-up ultrasound. Of these 59% were symptomatic, corresponding to 7.6% of the total number of patients. The discrepancy between these and the present findings is probably attributable to the limitations of ultrasound imaging. In contrast to CT, ultrasound is operator-dependent. In addition, other variables, such as abdominal gas distension or obesity, can decrease the sensitivity of ultrasound in detecting intraabdominal/pelvic fluid collections.

The high frequency of lymphocele formation in the present series does not substantiate the theory of the protective effect of the transperitoneal approach against lymphocele formation. In this regard, it seems apparent that the preperitoneal space developed during a transperitoneal RARP rapidly walls off, so the potential risk of lymphocele formation still exists and is comparable to a preperitoneal approach. The present findings are also supported by others investigators. Zorn et al.[16] showed that the incidence of symptomatic lymphoceles was comparable between a cohort of patients who underwent PLND after RARP or after open RRP.

The present study has the inherent limitations of any retrospective study. Secondly, our results may not be extrapolated or compared with other series in which an extended PLND has been performed. It is likely that the incidence of both asymptomatic and symptomatic lymphocele formation would increase with more extended dissection. Finally, it has been our policy to remove the JP drain in the early postoperative period. It is unclear whether the incidence of lymphoceles could have been reduced by leaving the drain for an extended period of time, or on the contrary, whether the negative pressure given by the drain bulb could have potentially perpetuated an otherwise small lymph leak, leading to an increased incidence of lymphoceles.

In conclusion, our data show that PLND during RARP is safe and associated with low intraoperative morbidity. However, the incidence of postoperative lymphocele formation in our study was high. Thus, the benefit of PLND during RARP should be weighed against the elevated risk of lymphocele formation and its potential complications. Furthermore, surgeons performing robot-assisted PLND should be aware that the transperitoneal approach did not appear to prevent lymphocele formation in our patients and caution should be exercised in carefully ligating lymphatic channels during dissection.


  1. Top of page
  2. Abstract
  • 1
    Palapattu GS, Allaf ME, Trock BJ, Epstein JI, Walsh PC. Prostate specific antigen progression in men with lymph node metastases following radical prostatectomy: results of long-term followup. J Urol 2004; 172: 18604
  • 2
    Daneshmand S, Quek ML, Stein JP et al. Prognosis of patients with lymph node positive prostate cancer following radical prostatectomy: long-term results. J Urol 2004; 172: 22525
  • 3
    Crawford ED, Batuello JT, Snow P et al. The use of artificial intelligence technology to predict lymph node spread in men with clinically localized prostate carcinoma. Cancer 2000; 88: 21059
  • 4
    Kawakami J, Meng MV, Sadetsky N et al. Changing patterns of pelvic lymphadenectomy for prostate cancer: results from CaPSURE. J Urol 2006; 176: 13826
  • 5
    Ward JF, Slezak JM, Blute ML, Bergstralh EJ, Zincke H. Radical prostatectomy for clinically advanced (cT3) prostate cancer since the advent of prostate-specific antigen testing: 15-year outcome. BJU Int 2005; 95: 7516
  • 6
    Bader P, Bukhard FC, Marwalder R, Studer UE. Disease progression and survival of patients with positive lymph nodes after radical prostatectomy. Is there a chance of cure. J Urol 2003; 169: 84954
  • 7
    Heidenreich A, Varga Z, von Knobloch R. Extended pelvic lymphadenectomy in patients undergoing radical prostatectomy: high incidence of lymph node metastasis. J Urol 2002; 167: 16816
  • 8
    Allaf ME, Palapattu GS, Trock BJ, Carter HB, Walsh PC. Anatomical extent of lymph node dissection: impact on men with clinically localized prostate cancer. J Urol 2004; 172: 18404
  • 9
    Messing EM, Manola J, Yao J. Immediate versus deferred androgen deprivation treatment in patients with node-positive prostate cancer after radical prostatectomy and pelvic lymphadenectomy. Lancet Oncol 2006; 7: 4729
  • 10
    Joslyn SA, Konety BR. Impact of extent of lymphadenectomy on survival after radical prostatectomy for prostate cancer. Urology 2006; 68: 1215
  • 11
    Naselli A, Andreatta R, Introini C, Fontana V, Puppo P. Predictors of symptomatic lymphocele after lymph node excision and radical prostatectomy. Urology 2010; 75: 6305
  • 12
    Briganti A, Chun FK, Salonia A et al. Complications and other surgical outcomes associated with extended pelvic lymphadenectomy in men with localized prostate cancer. Eur Urol 2006; 50: 100613
  • 13
    Solberg A, Angelsen A, Bergan U, Haugen OA, Viset T, Klepp O. Frequency of lymphoceles after open and laparoscopic pelvic lymph node dissection in patients with prostate cancer. Scand J Urol Nephrol 2003; 37: 21821
  • 14
    Freid RM, Siegel D, Smith AD, Weiss GH. Lymphoceles after laparoscopic pelvic node dissection. Urology 1998; 51: 1314
  • 15
    Musch M, Klevecka V, Roggenbuck U, Kroepfl D. Complications of pelvic lymphadenectomy in 1,380 patients undergoing radical retropubic prostatectomy between 1993 and 2006. J Urol 2008; 179: 9238
  • 16
    Zorn KC, Katz MH, Bernstein A et al. Pelvic lymphadenectomy during robot-assisted radical prostatectomy: assessing nodal yield, perioperative outcomes, and complications. Urology 2009; 74: 296302
  • 17
    D’Amico AV, Whittington R, Malkowicz SB et al. Biochemical outcome after radical prostatectomy, external beam radiation therapy, or interstitial radiation therapy for clinically localized prostate cancer. JAMA 1998; 280: 96974
  • 18
    Orvieto MA, Patel VR. Evolution of robot-assisted radical prostatectomy. Scand J Surg 2009; 98: 7688
  • 19
    Tomic R, Granfors T, Sjödin JG, Ohberg L. Lymph leakage after staging pelvic lymphadenectomy for prostatic carcinoma with and without heparin prophylaxis. Scand J Urol Nephrol 1994; 3: 2735
  • 20
    Coelho RF, Palmer KJ, Rocco B et al. Early complication rates in a single-surgeon series of 2500 robotic-assisted radical prostatectomies: report applying a standardized grading system. Eur Urol 2010; 57: 94552
  • 21
    Scholz HS, Petru E, Benedicic C, Haas J, Tamussino K, Winter R. Fibrin application for preventing lymphocysts after retroperitoneal lymphadenectomy in patients with gynecologic malignancies. Gynecol Oncol 2002; 84: 436