Introduction

Surgical radical lymph node dissection with either curative or palliative intent may be a treatment option in melanoma patients with regional metastases to the axillary lymph node basin. Whereas 20 years ago the prophylactic lymphadenectomy of the regional lymph node basin was frequently performed even if a negative sentinel node biopsy result was obtained, nowadays this is discussed controversially. In contrast to this, for proven lymphatic metastases the radical surgical removal of the lymph nodes is strongly recommended [1–3]. The surgical procedure in the axilla can be associated with considerable morbidity such as postoperative lymphedema, prolonged lymphatic wound secretion, wound necrosis, and sensory or motor neurologic deficit [4–7]. The extent of surgical dissection of the axilla is a point of controversy. Many surgeons generally include the lymph nodes of levels I, II, and III, whereas others dissect only levels I and II routinely and nodes of level III only when suspicious nodes are detected during the operative procedure. The latter less invasive procedure may lead to a reduced rate of clinically relevant postoperative lymphedema and injuries of the axillary nerve. To date no clear advantage in the time of survival with respect to the extent of surgical dissection has been reported. According to Kretschmer and Preusser [8], a standardized surgical technique of the axillary dissection improves local control but not the survival of the patients. It has been advocated that the use of fibrin glue may reduce the incidence of postoperative lymphatic secretion. Waclawizcek and coworkers [9] found a reduced incidence of postoperative lymphatic fistulas when they applied fibrin glue at the end of a radical axillary lymph node dissection (RALND) operation. However, other authors did not have the same results after application of fibrin glue after radical surgical lymph node removal of the axilla [10, 11]. Many variations of the surgical technique and a different combination of fibrin sealants were investigated in an attempt to reduce postoperative morbidity rates secondary to RALND. However, the different techniques and the resulting complication rates are difficult to compare due to different study designs, patient populations, and selection criteria. To investigate the effect of the intraoperative application of 2 ml of fibrin glue following a standardized operative RALND, a prospective randomized study of patients with malignant melanoma metastatic to the axilla was performed.

Methods

Following approval of the Institutional Ethics Committee of the Medical Faculty of Humboldt University of Berlin, Germany, 58 patients suffering from malignant melanoma with proven lymph node metastasis of the axilla were randomized into two groups consecutively between September 2003 and July 2005. Signed informed consent was obtained from all patients before enrollment in the study and the research respected the rules of the Declaration of Helsinki. Exclusion criteria were known allergy to bovine proteins, ASA (American Society of Anesthesiology) class 5, mental disorders resulting in disability to sign the informed consent, and loss to follow-up within the first three postoperative months. Following adequate staging according to AJCC [12], all patients underwent RALND by a single consultant surgeon (JWM) for therapeutic indication when they had proven lymphatic metastatic disease in the axilla by either sentinel node biopsy or fine-needle aspiration cytology before surgery.

The operation was carried out in identical fashion in all 58 patients according to an accepted standard of surgical oncologic practice. Preoperatively the patients received a single shot of third-generation cephalosporine and low-molecular-weight heparin the evening before the operation. Following a longitudinal 8-cm-long incision that usually incremented the scar of the previous sentinel node biopsy (when applicable), the dissection included the en-bloc resection of the fascia of the pectoralis minor and the interpectoral lymph nodes as well as the removal of the lateral chest wall nodes until the fascia of the serratus anterior muscle. The pectoralis minor muscle itself was not routinely resected, only when tumor involvement was diagnosed intraoperatively. All axillary lymph nodes around the axillary vein were then removed, including complete dissection of the vein during the course through the axilla until the chest wall, and the fascia of the latissimus dorsi muscle was exposed completely (Fig. 1). Above the axillary vein the level III node was dissected. In addition, the supra-axillary fat pad was routinely resected. The long thoracic and the thoracodorsal nerves were preserved in all cases, whereas the intercostobrachial nerves were resected when necessary. The specimen was taken to the Pathology Department for histologic evaluation. Following completion of the dissection of levels I, II, and III, the wound was irrigated with Ringer`s solution and, according to randomization, 29 patients received 2 ml of fibrin glue (Tissuecol®, Baxter, Germany) via spray applicator (Fig. 2). Two closed suction drains were placed in the wound and the closure took place in a three-layer fashion. Because it has been described in the literature that immediate wound closure after applying the fibrin sealant into the wound is important for the procedure to be beneficial [13], we placed two layers of sutures and the drains before the fibrin application and tied them immediately after application of the fibrin. The wound edges were then compressed for 2 min to allow the fibrin glue to completely adhere. Although it was felt from our previous clinical experience that one drain would have been sufficient for drainage of the wound, we decided to place two drains for our study to determine exact fluid collection from the wound. At the end of the operation a breast tape with dressings in the axilla was applied to minimize the created space in the dissected axilla. Before the operation, all patients had an exactly measured arm compression stocking prescribed by an orthopedic technician to reduce the incidence of postoperative lymphedema. The amount of fluid from the closed suction drains was recorded daily and when the output decreased to 50 ml or less over 24 h the drain or drains were removed. All patients were assessed twice daily, including wound inspection, general inspection of the patient, and measurement of the each arm’s circumference 10 cm above and below the elbow to detect lymphedema. We considered lymphedema present when the difference in circumference between the operated arm and the healthy arm was 2 cm or greater. In case of seroma formation, puncture of the secretion was performed under sterile conditions.

Fig. 1
figure 1

Intraoperative view after RALND

Full size image
Fig. 2
figure 2

Intraoperative view after RALND after application of 2 ml fibrin glue via spray applicator (Tissucol®, Baxter, Germany)

Full size image

According to the study protocol, patients were discharged when both drains were removed and no clinical sign of wound swelling or other signs of infection were detectable. Patients were seen on an outpatient basis routinely at the Department of Dermatology and Allergology on a three monthly term.

Statistical analysis

Statistical analysis was performed using SPSS 12.0G for Windows (SPSS Inc., Chicago, IL). The sample size was calculated before the beginning of the study. For the main outcome criterion of the study, a difference in the duration of drain placement of 1.5 days between the study and the control groups was considered as clinically relevant. The duration of drain placement in the wound in the control group was estimated to be 3 (0.5) days. Using a two-tailed t-test, α = 0.05 and β = 0.2, a difference of 1.5 days between the groups could be detected with 29 patients in each group.

All patients were blinded to the group allocation. Two surgeons (CR, GK) of our study group randomized the patients via computer program (Random A: Martin Kracht AG, Rostock, Germany) the day before operation. Stratification criteria were sex, age, and body mass index.

Continuous data were given as median (95% confidence interval) and were compared between groups using the Mann–Whitney U-test. Categorial data were compared using χ2 test or Fisher’s exact test. Graphical data were presented as box and whisker plots. p-Values less than 0.05 were considered as significant.

Results

Both groups were comparable with respect to age, sex, and body mass index (Table 1). The median age of the patients was 58 years (95% CI = 52–61) in group 1 (with fibrin glue) and 57 years (95% CI = 50–61) in group 2 (no fibrin glue) (p = 0.84). None of the patients died during the surgical procedure or within the first 30 days. There were no injuries of the long thoracic nerves or the thoracodorsal nerves. In addition, no allergic reaction to bovine fibrin glue was observed in the treatment group. According to Center for Disease Control and Prevention (CDC) classification, the wounds healed well in both groups; only two patients experienced an erythematous inflammation that was treated with oral antibiotics (Clindamycin) over 5 days leading to complete regression of the inflammation and an uneventful hospital course. Early lymphedema during the postoperative stay was diagnosed in two patients of group 1 (7%) and 5 in group 2 (p = 0.42). There was no clinical relevant lymphedema seen in the 3-month interval during the follow-up consultations (Table 2).

Table 1 Characteristics of patients
Full size table
Table 2 Postoperative complications
Full size table

Duration of drain placement

The duration of drain placement in the wound in both groups is shown in Fig. 3. The drains in group 2 (no fibrin glue) were in place for a median of 5 days (range = 3.6–5.7) and the drains in group 1 also were left in place for a median of 5 days (range = 4.1–8.5). The differences between treatment and control groups were not significant (p = 0.701). As an unexpected result, we did detect a slightly longer duration of drain placement in the treatment group, although this difference did not reach statistical significance.

Fig. 3
figure 3

Duration of the drainage with longer placement (days)

Full size image

Amount of postoperative fluid

The amount of collected fluid was even less in group 2, with a total of 410 cc (range = 362–727 cc), compared with that of the treatment group, with 503 cc (range = 369–1098 cc) without statistical difference (p = 0.605) between the two groups (Fig. 4). There was no difference in the quality of sanguineous versus serosanguineous secretion either. In addition, the amount of fluid was not different in both groups when compared during the first three postoperative days (group 1: 365 cc, range = 274–437; group 2: 285 cc, range = 259–448) (p = 0.708).

Fig. 4
figure 4

Amount of postoperatively collected fluid (drains 1 and 2 together)

Full size image

Length of postoperative hospital stay

The median of length of postoperative hospitalization was 7 days (range = 6–8.5 days) without any differences between the treatment group (7 days, range = 5.9–10.7) and the control group (6 days, range = 5.4–7) (p = 0.387). The fibrin sealant had no effect on the duration of the hospital stay after radical axillary lymph node dissection (Fig. 5).

Fig. 5
figure 5

Length of postoperative hospital stay (days)

Full size image

Numbers of lymph nodes on histologic examination

There were no differences in the numbers of excised lymph nodes between both groups (p = 0.25). In the treatment group, 16 lymph nodes (range = 12–21) and in the control group 15 lymph nodes (range = 11–24) were analyzed. The extent of operation in both groups was identical and the operation was performdd by the same surgeon (JWM).

Discussion

Radical axillary lymph node dissection (RALND) is a standard surgical procedure for the treatment of metastatic lymph node disease of the axilla in melanoma patients. A positive sentinel node is the classical indication for RALND in patients with a primary melanoma, with the axilla representing the first regional lymph node basin. Other indications may be palliative axillary lymph node dissection or, to a lesser extent, the preventative RALND. Even when this meticulous surgical technique is applied, RALND may lead to considerable postoperative morbidity [14, 15]. In addition to well known injuries to long thoracic nerves and/or thoracodorsal nerves resulting in permanent muscular deficits and subsequent physical deformities, prolonged lymphatic secretion also may lead to prolonged drainage of wound secretion, swelling, and finally infection of the wound. Despite the use of different surgical techniques such as the use of Ligasure in an attempt to minimize lymphatic leakage or the traditional use of ligation techniques of dissected lymphatic channels or endoscopic axillary dissection of lymph nodes, there is still a considerable percentage of patients who suffer from prolonged lymphorrhea after RALND postoperatively. This is the most common reason for a prolonged stay in the hospital and frequent visits to outpatient clinics. Numerous attempts have been made to overcome postoperative lymphorrhea. Some authors investigated tetracycline sclerotherapy, others instilled talcum poudrage, whereas others used a fibrin-coated collagen patch following RALND.

Several studies have investigated the effects of fibrin sealant after RALND with different results. Most authors studied patients with breast cancer who underwent radical mastectomy with additional RALND. Whereas some authors did not find a reduction in the amount of drainage or a shortened hospital stay, other investigators demonstrated reduced postoperative drainage after axillary lymphadenectomy following application of fibrin sealant at the end of the operation [16–19]. Finally, Mustonen et al. [10] reported their results of a randomized trial of patients with breast cancer who underwent radical axillary lymph node dissection with application of a combination of fibrin sealant and fibrinolysis inhibitor. Although they did not find a reduction in the incidence of postoperative seroma in the treatment group, they found the occurring seroma in the treatment group easier to treat with less need for postoperative punctures of the seroma (2 punctures vs. 3).

Most studies concerning the different amounts (1–4 cc per dose) and application techniques (spray technique vs. administration with of a syringe) of fibrin glue and the technique of wound closure (immediate vs. delayed wound closure) are difficult to compare. In patients with skin cancer who undergo axillary or inguinal radical lymphadenectomy, the results of fibrin glue administration have also been conflicting. In a retrospective analysis, Waclawicek and Pimpl [9] found a significant decrease in the amount and duration of drainage using fibrin sealant following RLND of the axilla and groin in 51 patients. However, Furrer et al. [20] could not reproduce these results when they studied 30 consecutive patients prospectively. They used only 1 cc of fibrin glue in patients undergoing RLND. In addition, a selective hyperthermic limb perfusion of the extremity was performed in 27 of 30 patients. In our study we could not reproduce the results of these investigators, detecting a reduced amount of drainage fluid or a reduced rate of seroma formation. Although we used a reliable technique to administer the fibrin sealant in a sufficient dosage of 2 cc, we did not find a reduction in either the amount of drainage or the length of the closed suction drain in place. We cannot speculate about the reasons for the conflicting reports on the effects of fibrin sealant following RALND. It may be caused partly by the fact that in many studies a mastectomy or a modified mastectomy with or without additional lymphadenectomy of the axilla was performed and we considered solely melanoma patients undergoing RALND. The different tumor biology of the two entities may contribute to the different results. Nevertheless, we conclude that it is not feasible to prevent postoperative lymphatic secretion by intraoperative application of 2 ml of fibrin sealant in patients with lymph node metastasis of malignant melanoma who undergo radical lymph node dissection of the axilla.