Drain Versus No-Drain in Patients After Surgery for Early Stage Gynecologic Malignancies: A Randomized Controlled Study

Introduction

The majority of early stage gynecologic malignancies are treated with surgery. Apart from total abdominal hysterectomy and bilateral salpingo-oophorectomy, retroperitoneal lymph-node dissection (rLND) plays a vital role in managing these tumors.^1,2 From the surgical perspective, lymphadenectomy is associated with lymphorrhea, leading to excessive production of lymph fluid that can accumulate in the abdominal cavity resulting in several complications. To prevent postoperative fluid accumulations, seromas, hematomas, and lymphoceles, drains are placed in the peritoneal cavity. ³

Several retrospective, prospective randomized studies have not shown any differences in postoperative morbidity between use and nonuse of drains. Some studies have even suggested worse morbidity occurs with the use of drains.^4–6 The current authors have been using pelvic drains routinely following these surgeries, thus, most times, requiring longer patient stays in the hospital until these drains were removed. Hence, the current authors wanted to conduct a randomized study on the effects of draining the abdominal cavity. The goal was to improve practice for patients undergoing bilateral pelvic lymph-node dissection (PLND) with or without para-aortic lymph-node dissection (LND). This study was performed to help to establish part of a definitive surgical procedure for managing early stage gynecologic malignancies (i.e., carcinomas of the endometrium, cervix, and ovaries).

Materials and Methods

This was a prospective randomized study performed to evaluate postoperative morbidity associated with the use of drains versus nonuse of drains in patients undergoing bilateral PLND with or without para-aortic LND in patients with gynecologic malignancies of the endometrium, cervix, and ovaries over a 3-year duration. The protocol was approved by the institutional ethics committee of the Sri Venkateswara Institute of Medical Sciences, in Tirupati, Andhra Pradesh, India. This trial was registered with the clinical trial registry of India (registration number: CTRI/2017/09/009592).

All consenting patients, ages 18–70, were included if they had histologically proven cancers of the endometrium, cervix, and/or ovaries, and underwent bilateral PLND, with or without para-aortic LND, as part of their standard treatment. These patients were randomized to 1 of 2 groups (1) a drain group or (2) a no-drain group, via an online computer-generated random-number table. ⁷ * Block randomization, in blocks of 10, were used to avoid heterogeneous accumulation of patients in both groups.

Each participant was evaluated, which included a detailed history, a physical examination, a biopsy or frozen section (as needed), a staging work-up, and preoperative investigations per standard guidelines. Patients with preoperative ascites and/or advanced disease with peritoneal and omental large deposits were excluded. If any patient was found during her surgery to have multiple large omental or peritoneal deposits or ascites >1 L, that case was also excluded. Other exclusions were patients who needed bowel resection and anastomosis; patients with intraoperative vascular or urinary-bladder injuries; or patients with troublesome hemostasis.

The protocol allowed the operating surgeon to decide whether or not draining the peritoneal cavity was in the best interest of each patient. If a drain was placed in violation of the randomization, that patient was then added to the drain group. Each patient randomized to the drain group had a single, closed, passive tube drain placed in her pelvis. No attempt was made to close the pelvic peritoneum any of the patients regardless of group assignments, but the vaginal cuffs were closed in all patients routinely. In the drain group, a subcutaneous suction drain was placed, depending on each patient's body habitus, at the discretion of the operating surgeon. Drains were removed when the daily output was <100 mL for 2 consecutive days.

Prophylactic antibiotics and pharmacologic prophylaxis from postoperative day (POD) 2 for deep-vein thrombosis (DVT) were given routinely to all of the patients, except when there was concern regarding immediate postoperative hemostasis. Whenever pharmacologic prophylaxis was delayed, mechanical DVT prophylaxis was provided.

A record was made of the histologic diagnosis, International Federation of Gynecologists and Obstetricians (FIGO) stage, pathologic TNM, and final stage.

An assessment of postoperative pain as perceived by each patient using a visual analogue scale (VAS) was recorded on the morning of POD 1 in both groups. The postoperative duration of drainage and hospital stay were also recorded. Notes were also kept on any postoperative morbidities, such as wound complications (infection, gaping or burst abdomen), fluid accumulation needing intervention (radiologic or surgical), hematomas, seromas, and lymphocysts. Abdominal imaging (ultrasound [US] or contrast-enhanced computed tomography [CT]) was done at the time of discharge in the no-drain group and was repeated as indicated by the clinical course of the patient. All patients were followed postoperatively for 30 days, clinically and radiologically (US or CT), to evaluate and compare the postoperative morbidity in both groups.

The sample size required for this study was calculated from sample-size determination Excel^® software, developed by Microsoft^® Excel, using 2 samples with a dichotomous-outcome model. Per Franchi et al., ⁸ the prevalence of symptomatic lymphocysts was 6.08% in their drain group and 0.85% their no-drain group is. Hence, the sample size for the current study was calculated at 285 patients in each group. However it was only possible to enroll a total of 144 patients over 33 months; thus there were 73 patients in the drain group and 71 patients in the no-drain group. Frequency of postoperative complications was calculated and expressed as a percentage. Comparison between 2 groups was performed with Pearson's χ ² test, using commercially available statistical software with a significance of p < 0.05.

Results

During the study, 185 patients underwent PLNDs, 144 patients were enrolled and 41 patients were excluded. A total of 73 patients were randomized to the drain group and 71 were randomized to the no-drain group. Three patients were converted from the no-drain group to the drain group intraoperatively due to doubtful hemostasis. Reasons for exclusions from the study included advanced disease (omental, peritoneal, and serosal deposits in 26 patients), ascites (in 5 patients), urinary-bladder injury (in 4 patients), bowel anastomosis (in 2 patients), vascular injury (in 2 patients) and age older than 70 (2 patients). A total of 7 patients (2 in the drain group and 5 in the no-drain group) were lost to follow-up.

The baseline clinicopathologic characteristics of the included patients are shown in Table 1. The most- common cancer site was uterine cervix followed by ovary and uterine body, in that order. Common histologies were as follows: cervical cancers, squamous-cell cancers (23 versus 26); endometrial cancers, serous cystadenocarcinomas (16 versus 15); and ovarian and endometrioid cancers, adenocarcinomas (13 versus 12)—in the drain versus the no-drain groups, respectively. The most-common FIGO stage in both groups was FIGO IA for endometrial and ovarian cancers, whereas for cervical cancers it was IB1 (Table 1).

Fifty eight percent of patients in the drain group underwent bilateral PLND alone, whereas 70% of patients in the no-drain group underwent bilateral PLND alone. There was no significant difference in intraoperative blood loss, number of LNs harvested, and number of positive lymph nodes in the 2 groups. There was no significant difference in VAS scores for pain on POD 1 between the 2 groups. Two patients in each group had surgical-site infections. One patient in the drain group had a pelvic abscess on that was seen on a US scan but that patient defaulted for follow-up. In the no-drain group, 1 patient developed gross ascites on POD 4, which required US-guided pigtail insertion to relieve her discomfort. Three patients in the drain group, compared to none in the no-drain group, developed paralytic ileus, which was managed conservatively with bowel rest and electrolyte supplementation. Average postoperative length of drainage in the drain group was 8 days. Average length of postoperative stay was significantly longer in the drain group (Table 2). Seven patients in the drain group and 5 patients in the no-drain group developed asymptomatic lymphocysts during the follow-up time; by 3-months, these cysts resolved spontaneously.

Discussion

The present study looked at the practice of placing drains following surgery for gynecologic malignancies and found no advantage for such a practice. Despite patients undergoing extensive rLND, patients in the no-drain group did not have any adverse outcome, and could be discharged to go home early. There are few studies, including Cochrane reviews, on the topic. An updated version of the Cochrane Database review published in 2017 did not find any new randomized studies to add to its 2014 review. ⁹ The authors reviewed 4 studies with 571 subjects and concluded that there was no additional benefit with retroperitoneal drain placement in women undergoing surgery for gynecologic malignancies. This review noted that 3 of the 4 studies included patients with cancers of the endometrium and cervix and only 1 study included a patient with ovarian cancer. Hence, the current authors felt a need to study this area further, as the available data were limited and LNDs in ovarian cancer tend to be more extensive because the procedure includes para-aortic region more often than for the other 2 cancers.

Cancers of the cervix, endometrium, and ovaries are the most-common gynecologic malignancies that are treated with surgery. Pelvic lymphadenectomy is an important integral component of surgical staging, as the information derived from this procedure is of value, not only for staging, but also for prognostication as well as selection of adjuvant treatment. Para-aortic lymphadenectomy is an additional procedure used for early ovarian and high-risk endometrial cancers, which adds to the surgical morbidity of pelvic lymphadenectomy.

Systemic lymphadenectomy is associated with disruption of lymphatics, which would leak in the immediate postoperative time for several days to weeks, leading to fluid accumulation within the abdomen. Postoperative bleeding from raw areas within the abdomen may also contribute to fluid accumulation within the abdomen. This fluid accumulation can develop into a fibrous capsule, leading to lymphocyst formation. When lymphocysts are large, they can be symptomatic. Lymphocysts have been reported to form in 1%–29% of patients undergoing rLNDs. ¹⁰ The risk of developing lymphocysts is associated with prior irradiation, extent of nodal involvement, lymphadenectomy and use of heparin for prophylaxis against DVT. Large lymphocysts can also cause ureteral obstruction, edema of the legs, venous thrombosis, paralytic ileus, and abscesses as well as fistulization in the intestines or exterior areas. ⁸ To prevent lymphocyst formation, keeping active or passive drains within the retroperitoneal cavity or abdomen became popular as early as the 1960s.^11–13

As the practice of using drains was universal, several researchers started reporting problems caused by drainage, such as the drain itself, being a foreign body, potentially interfering with the absorptive function of the peritoneum and increasing the risk of developing lymphocysts. ¹⁴ It was also noted drains could be associated with increased local pain, problems caused by accidental dislodgement, omental or bowel herniation, and intestinal obstructions. Furthermore, most of the time, drains would get blocked due to fat or omental herniation and cease to function in the early PODs, leading to peritubal leakage and patients more discomfort. In line with the Cochrane review as well as other studies published in the field the current study did not demonstrate any benefit of adding drainage following gynecologic staging procedures. The current study is different from the majority of the earlier studies in having more patients with ovarian cancer and a greater proportion of patients who underwent para-aortic lymphadenectomy.

No special effort was made to close the peritoneum in each patient, which would have resulted in a free flow of fluid across the retroperitoneum and peritoneal cavity, with the fluid ultimately being absorbed. The current study looked for any difference in pain perception in the drain group versus the no-drain group—no difference was found. There was a significant difference in length of postoperative hospital stay favoring the no-drain group, which was consistent with several previous studies.^{4–6,10,15,16} The mean duration of drainage in the present study was 8 days, which was slightly higher compared to Bafna et al. ¹⁷ but less than Benedetti-Panici et al. ⁶ This duration of drainage was the main reason for the prolonged postoperative stays in the drain group. There was no increased surgical-site or pelvic infection risk with drainage, compared to no-drainage, as was noted in several previous studies.^4–6,8,15 While 1 patient in the current study's no-drain group needed intervention for fluid accumulation, other studies have reported more fluid accumulation necessitating interventions in those studies' drain groups.^5,6,15 This was thought to be due to the drainage tube itself acting as a foreign body and interfering with the reparative and absorptive functions of the peritoneum. ¹⁸

Some of the limitations of this study include the sample size being small and the study's single-institution character.

Conclusions

Despite the limitations of this study, the current authors believe that the findings are valid and that the routine practice of using drainage after extensive LNDs should revisited, particularly when no special attempt is made to close a patient's peritoneum.