Objective
The purpose of this study was to report the morbidity of nonemergent hysterectomy for suspected placenta accreta.
Study Design
This was a retrospective study of all patients who underwent nonemergent hysterectomy for placenta accreta at Tampa General Hospital from June 1, 2003 to May 31, 2009.
Results
Twenty-nine patients were identified. Diagnosis was suspected on ultrasound scanning in 26 women (6 women also underwent magnetic resonance imaging) and on direct vision at repeat cesarean section delivery in 3 women. All of the women were multiparous, and 18 women had undergone ≥2 cesarean section deliveries. Twenty-one women had a placenta previa, and 8 women had a low anterior placenta. Final pathologic findings revealed accreta (20 specimens), increta (6 women), and percreta (3 women). Mean total operative time was 216 minutes; blood loss was 4061 mL. Two women had ureteral transection (1 was bilateral); 3 women had cystotomy, and 3 women had partial cystectomy. Postoperative hemorrhage occurred in 5 women; 1 hemorrhage resolved after catheter embolization, and the other 4 hemorrhage required reoperation.
Conclusion
Nonemergent hysterectomy for placenta accreta is associated with significant morbidity in the forms of hemorrhage and urinary tract insult.
Placenta accreta, a rare but increasingly common complication of pregnancy, has been associated with significant maternal morbidity. The purpose of this article is to report our experience with nonemergent hysterectomy for suspected placenta accreta.
Materials and Methods
This was a retrospective, institutional review board–approved, chart review of all patients who underwent nonemergent hysterectomy for suspected placenta accreta from June 1, 2003 through June 30, 2009. Cases were identified through a medical records department and labor and a delivery records search with ICD-9 codes and search words (placenta accreta, cesarean hysterectomy). Patients whose surgery was precipitated by bleeding that was nonemergent in nature were included. Cesarean section deliveries were done by an obstetrician and a senior resident. The hysterectomies were done by 2 senior residents who were assisted by the obstetrician or a gynecologic oncologist. Prophylactic arterial catheters and/or ureteral stents were placed at the discretion of the attending physician.
For this study, major morbidity was defined as any of the following events: intra- or postoperative transfusion of ≥4 units of blood, clinical and laboratory documented coagulopathy that required blood products, ureteral injury, reoperation, major infection (sepsis syndrome, drainage of intraabdominal abscess, necrotizing), thromboembolism, fistula, or arterial catheter complications.
Results
Thirty-seven patients were identified in the database. Eight surgeries that were clearly emergent were excluded, which left 29 patients whose data were available for analysis. Patient characteristics are given in Table 1 . Diagnosis was suspected on ultrasound scanning in 26 women (6 diagnoses were magnetic resonance imaging) and on direct vision at repeat cesarean section delivery in 3 women. Nine of the 29 patients were referred from the region based on the suspected diagnosis. Nine of the women were hospitalized for bleeding before the planned delivery date. Although cesarean section delivery was done at an earlier date than originally planned (median, 4 weeks), in no case was it emergent. In 7 cases the surgery was performed on an urgent basis for bleeding (3 women), bleeding and premature labor (2 women), and premature rupture of membranes with suspected chorioamnionitis (2 women). Although a statistical analysis was not performed, a review of the data did not reveal any major differences in outcome between these and the truly elective cases.
| Characteristic | Measure |
|---|---|
| Mean age, y a | 33.4 (23–43) |
| Mean parity, n a | 3.2 (1–8) |
| Mean gestational age, wk a | 33.2 (26–39) |
| Mean previous cesarean delivery, n a | 2.3 (1–6) |
| ≥2 previous cesarean deliveries, n | |
| Placental location | 18 |
| Complete previa | 21 |
| Low anterior | 8 |
Fourteen of the 29 patients had prophylactic placement of bilateral iliac artery catheters (6 common iliac, 8 internal iliac) for the purpose of controlling hemorrhage. Use of the catheters was highly variable: the balloons were never inflated in 7 women, inflated prophylactically in 2 women, and inflated on the onset of hemorrhage in 5 women. Because of the inconsistent placement, location, and use of these catheters, it is not reasonable to attempt analysis of their effectiveness in the present study.
Simultaneously with cesarean section delivery, prophylactic ureteral stents were placed successfully bilaterally in 10 women, unilaterally in 2 women, and unsuccessfully in 2 women. There was no morbidity or additional operative time and minimal cost associated with the placement of these stents. As will be reported later, ureteral transections occurred in patients with and without stents. However, there is not a reasonable analysis of the data that can be done in terms of the relative benefit of the stents.
All patients had a vertical skin incision. Cases were done in the main operating room when available because of more experienced staff and greater accessibility to major surgical equipment. The operative team varied considerably among the cases. Residents and fellows were involved in all cases. A gynecologic oncologist participated in most of the operations but to a highly variable degree. Because of these variations, it is not possible to analyze the potential effects of level of training or specialty on outcome.
Three patients in whom placenta accreta was suspected based on ultrasound results and despite the desire for sterilization and consent for hysterectomy underwent the attempted removal of the placenta immediately after cesarean delivery. This resulted in immediate severe hemorrhage in 2 women, and hysterectomy was performed promptly in all 3 women. Estimated blood loss for these cases was 700, 2500, and 3000 mL.
During cesarean delivery in a patient with a low anterior placenta and suspected accreta, the placenta was transected extensively. Massive hemorrhage ensued; by the time the uterus was removed, the patient had received 60 units of blood and became severely coagulopathic. The pelvis was packed, and the patient was taken to interventional radiology. Embolization of bleeding vessels was performed, but persistent hemorrhage required reoperation the next day. This patient then had to be kept in the intensive care unit for several days but subsequently did well.
During hysterectomy, 1 patient with placenta percreta had bilateral distal ureteral transection (stents had been placed); in another patient with accreta (but suspected percreta), the distal left ureter was transected (no stents). In both cases, there was complete placenta previa that involved the cervix. In these cases, attempted mobilization of the urinary tract was postponed until the last step of the hysterectomy in hopes of devascularizing, mobilizing, and better defining the area of potential percreta. In both cases, the ureters had been dissected to the ureteric tunnels–but not beyond–because of concerns regarding inciting hemorrhage. The planned line of transection of the cardinal ligaments was deliberately kept lateral to the large vascular mass to prevent massive hemorrhage (all 3 ureteral transections occurred during this part of the operation). In both cases, we attempted to open the paravesical spaces medial to the umbilical ligaments; however, the large cervical-placental mass made this difficult. It was understood during both hysterectomies (the same technique was used for many of the other hysterectomies in this series as well) that the ureters were at risk, but it was believed that the benefit of the planned approach (in terms of avoiding massive hemorrhage) outweighed this risk. All 3 transections were recognized intraoperatively and repaired successfully by ureteroneocystotomy. The placenta of the patient with suspected percreta invaded to within 1 mL of the anterior cervical margin (increta on pathology) but did not seem to involve the bladder, which was separated without hemorrhage.
One patient who was suspected of having at least placenta increta preoperatively was noted at the time of surgery to have obvious percreta. The placenta extended grossly into the right broad ligament, which necessitated radical hysterectomy.
Incidental cystotomy occurred in 6 patients, all during attempts at careful (to avoid hemorrhage) mobilization of the bladder off a large cervical-placental mass. Three of the cystotomies were unintentional. In the other 3 patients, the bladder muscularis (midline posterior fundus) appeared to be invaded, and an approximately 3-5 cm disc of bladder was resected with the uterus. One of these 3 patients was noted to have placental invasion of the bladder during cystoscopic stent placement. Final pathologic findings confirmed percreta in 2 of the 3 cases and increta in the third case. All 6 bladders were repaired primarily and healed uneventfully.
Mean operative time for the cesarean hysterectomies was 216 minutes (range: 100-400 minutes). Six patients had a successful supracervical hysterectomy. In all of these patients, the placenta was clearly above a well-formed cervix. Twenty other patients had a planned total hysterectomy. Three patients had an initial attempt at a supracervical hysterectomy. However, significant bleeding required the immediate removal of the cervix in 2 patients and removal several hours later in the third patient. In addition to the 2 patients, 2 other patients required reoperation within 24 hours for postoperative bleeding.
Mean and median estimated blood loss for the cesarean hysterectomies was 4061 and 3000 mL, respectively (range, 500–30,000 mL). Twenty-one patients (72%) received a blood transfusion during or after surgery. Twelve of these patients received ≥4 units of blood. None of the patients experienced major infectious morbidity. Severe coagulopathy that required component therapy intra- and/or postoperatively developed in 6 patients (all after large volume blood transfusion). Component therapy included recombinant factor VIIa in 3 women. Two of these women who received recombinant factor VIIa experienced subsequent thromboembolism (both received 2 doses). One of 14 patients (7%) had a catheter/balloon–related complication, which was asymptomatic bilateral internal iliac artery stenosis (almost complete occlusion on 1 side). In total, 7 patients (24%) experienced major morbidity ( Table 2 ).
