Induced abortion, the most common surgical procedure among reproductive-aged women, has low morbidity and mortality.1 The most common serious complication of abortion—hemorrhage, as defined by an estimated blood loss of greater than 500 mL or bleeding requiring a transfusion—occurs in approximately 0.82 per 100,000 abortions, with risk increasing with gestation.2 Usually hemorrhage is initially managed with conservative measures such as uterotonics, laceration repair, bimanual uterine massage, and tamponade with an intrauterine pack or Foley bulb. When these methods fail, hemorrhage leads to hysterectomy in approximately 1.1–1.4 per 10,000 abortions.3,4
Uterine artery embolization has been in use for decades to treat postpartum hemorrhage and pelvic trauma with the goal of avoiding hysterectomy and preserving fertility.5–8 Uterine artery embolization has less commonly been used for postabortion hemorrhage. Three published case series9–11 and three case reports12–14 describe a total of 21 patients who underwent embolization after abortion. Uterine artery embolization was a successful therapy in the vast majority (91%) of the published cases.9–14 However, in each of these six studies, the number was low, and none included more than four women from the same institution. The most common causes of postabortion hemorrhage in these papers were uterine atony and perforation. Only three of these cases are patients with postabortion hemorrhage caused by probable placenta accreta, and all three were successfully treated with embolization.10
The limited availability of uterine artery embolization would make it unlikely that a low-risk patient having an abortion in an outpatient center would have the option of embolization. However, women who are at higher risk of hemorrhage, such as those with prior cesarean deliveries or abnormal placentation, may be referred for an in-hospital abortion and, therefore, have access to embolization. In this study we review a series of patients who underwent uterine artery embolization for treatment of refractory hemorrhage after an abortion.
MATERIALS AND METHODS
The Women’s Options Center is a hospital-based outpatient clinic at San Francisco General Hospital providing 2,200 abortions per year. Forty-three percent are in the first trimester, 22% are between 140/7 and 196/7 weeks, and 35% are between 200/7 and 231/7 weeks. The clinic receives referrals of patients in the second trimester because, other than the University of California, San Francisco, and San Francisco General Hospital, no outpatient abortion clinic in Northern California accepts state insurance for abortions after 18 weeks, and many other clinics do not care for high-risk patients in the second trimester. Our clinic, therefore, has a disproportionate representation of women with prior cesarean deliveries, suspected placenta accreta, and diseases such as obesity, hypertension, and diabetes.
At the Women’s Options Center, if a patient has one or more previous cesarean deliveries, the location of the placenta is determined by clinic or formal ultrasonography before placing cervical laminaria. If she is viewed as high risk for abnormal placentation, the Department of Radiology is contacted before surgery to ensure availability of an Interventional Radiology suite in the case of a refractory hemorrhage. All terminations after 14 weeks of gestational duration are done under ultrasound guidance. At the time of a hemorrhage, the usual diagnostic and treatment algorithm begins with inspection of the cervix for lacerations and examination of the uterus for evidence of atony. If the uterus is atonic, uterotonics are administered, usually beginning with methylergonovine. If the hemorrhage does not appear to be caused by atony or cervical laceration, we attempt to identify the source of bleeding by inserting a cannula to the fundus and slowly withdrawing it to the os, observing at what location within the uterus bleeding increases.15 In the event that uterotonics do not control the bleeding and no other source is found, an intrauterine Foley bulb is inflated with 40–60 mL of fluid to tamponade the lower uterine segment. The balloon is left in place as the patient’s bleeding is monitored. If the balloon stops the patient’s bleeding, she is then observed for at least several hours, depending on clinical concern, and after removal of the balloon she is discharged, if stable. If bleeding continues despite all measures, the attending physician determines if the patient is transferred to Interventional Radiology for embolization.
We reviewed all cases of uterine artery embolization from January 2000 to August 2007 after abortion. These cases were identified by searching the Women’s Option Center complication log and patient database using the terms “post-abortion hemorrhage” and “uterine artery embolization.” The cases were then cross-referenced with the San Francisco General Hospital Department of Radiology case log for “uterine artery embolization” and “post-abortion hemorrhage.” Using these methods we identified 42 women who underwent uterine artery embolization for postabortion hemorrhage.
Data were abstracted by a trained clinician from the original hospital charts, operation notes, anesthesia notes, discharge summaries, nursing notes, and laboratory records. Microsoft Excel (Microsoft Corporation, Redmond, WA) and SPSS 15 (SPSS Institute Inc., Chicago, IL) were used to organize and describe the cases. Abstracted information included demographic, medical, and procedural data, as well as risk factors for abnormal placentation. Gestational age was determined by ultrasonographic evaluation of biparietal diameter for all patients.16 The study was approved by the Institutional Review Board of the University of California, San Francisco.
Detailed information about complications and treatments was abstracted. The cause of postabortion bleeding for each patient was identified as one of the following: uterine atony, cervical laceration, abnormal placentation (ie, placenta accreta, increta, or percreta), disseminated intravascular coagulation (DIC), uterine perforation, or “unknown” if records were not available. If a patient had a cervical laceration that was easily repaired and became hemostatic, the laceration was not listed as the primary cause. When patients were diagnosed with bleeding from the lower uterine segment without diagnosis of atony or cervical laceration, we considered the diagnosis to be lower uterine segment bleeding. “Formal” ultrasound scans were performed by registered ultrasonographers using both grayscale and color Doppler ultrasonography, and perinatal or radiological faculty interpreted all studies. “Clinic” ultrasonograms were done by nurse practitioners, registered nursing personal with ultrasound training, or obstetrics and gynecology residents using grayscale ultrasonography and interpreted by obstetrics and gynecology faculty.
Forty-two women received uterine artery embolization for treatment of postabortion hemorrhage from January 2000 to August 2007, including one woman who had prophylactic embolization (Tables 1 and 2). Of these women, 22 (52%) had hemorrhage from atony, seven (17%) from abnormal placentation, five (12%) from cervical laceration, three (7%) from perforation, two (5%) from lower uterine segment bleeding, and two (5%) from DIC. The mean estimated blood loss was 1,580 mL, and the median estimated blood loss was 1,150 mL and ranged from 400 mL to 4,000 mL. One patient underwent embolization after only 400 mL of blood loss because of high suspicion for abnormal placentation. Uterine artery embolization was successful in avoiding hysterectomy in 38 of 42 patients (90%). One patient received a prophylactic uterine artery embolization before her abortion and subsequently required a hysterectomy. All four embolization failures were in women who had histopathological confirmation of accreta, increta, or percreta. However, three of seven women (43%) with probable accreta diagnosed by formal ultrasonography were treated successfully with embolization. The median time from completion of abortion to transfer to the Interventional Radiology suite was 3 hours, 18 minutes (mean 5 hours, 6 minutes, 95% confidence interval 2 hours, 36 minutes to 7 hours, 36 minutes). Seven women (17%) were transferred emergently to the Interventional Radiology suite in less than 1 hour. In one patient with known DIC who had presented to Labor and Delivery with bleeding, the patient was transferred immediately after her abortion to Interventional Radiology in 6 minutes. The remaining 35 women (83%) were transferred after a period of stable observation, the longest of which was 25 hours (excluding patient 29 who is described below). All but two women had bilateral embolization with Gelfoam (Pfizer Inc., New York, NY). One woman had coil embolization in the left uterine artery and Gelfoam in the right uterine artery, and one woman received only a left uterine artery embolization due to localized extravasation from the left uterine artery, which became immediately hemostatic upon embolization. In 40 of 42 cases, Gelfoam was used alone as the embolic material. Two cases used a combination of Gelfoam and embolization coils to ensure hemostasis. Procedure length ranged from 35 to 315 minutes. The mean duration of the procedure was 2 hours (95% confidence interval 1 hour, 40 minutes to 2 hours, 20 minutes).
During chart review 10 cases warranted special attention, including six cases of accreta with postoperative uterine artery embolization, the one preoperative prophylactic embolization, two cases of uterine perforation and broad ligament hematomas, and one case of probable perforation.
Abnormal Placentation Successfully Treated With Embolization
Patient 36 is a 34-year-old woman with two prior cesarean deliveries who presented for termination of an 18-week pregnancy. Formal ultrasonography showed evidence of placenta accreta. Because of an immediate brisk postoperative hemorrhage, the patient underwent uncomplicated uterine artery embolization (total estimated blood loss 400 mL) and had a normal postoperative course. Patient 37 is a 24-year-old woman with two prior cesarean deliveries. She presented for termination of a 22-week pregnancy, and formal ultrasonography suggested placenta accreta. She had moderate bleeding during uterine evacuation, which was controlled initially with placement of a Foley bulb. The patient was observed for 3 hours, at which time bleeding increased and she was transferred to the Interventional Radiology suite for embolization followed by an uncomplicated postoperative course (total estimated blood loss 1,500 mL). Patient 38 is a 35-year-old woman with one prior cesarean delivery who presented for termination of a pregnancy at 21 weeks. Formal ultrasonography confirmed placenta previa and showed no evidence of accreta. She had heavy bleeding during her procedure and was transferred to Interventional Radiology for embolization (total estimated blood loss 3,000 mL). Because of the high probability of placenta accreta, a methotrexate/leucovorin rescue protocol was initiated after the uterine artery embolization. This protocol entails alternating doses of methotrexate (1 mg/kg, intramuscularly on days 1, 3, 5, etc.) with doses of leucovorin (0.1 mg/kg intramuscularly on days 2, 4, 6, etc.) until four doses of methotrexate have been administered. Our patient received two doses of methotrexate (75 mg each dose) on postoperative days 1 and 3. On day 4, she was found to have elevated liver enzymes and a drop in β-hCG from 10,093 to 4,085, and methotrexate was discontinued. After discharge she was stable, and her β-hCG levels returned to normal.
Abnormal Placentation Unsuccessfully Treated With Embolization
Patient 39 is a 27-year-old woman with three prior cesarean deliveries who presented for termination of an 18-week twin gestation. Clinic ultrasonography detected a placenta previa with concern for accreta. During uterine evacuation a partial-thickness defect in the anterior lower uterine segment was noted on intraoperative ultrasonogram. Conservative measures including methylergonovine did not control her hemorrhage (total estimated blood loss 1,000 mL), and she received one dose of phenylephrine to control a brief episode of hypotension before she was transferred to the Interventional Radiology suite. After her embolization she developed a femoral artery embolus requiring embolectomy by vascular surgery. She spent 2 days in intensive care and was discharged on her sixth postoperative day and with full lower extremity function. She ultimately had a hysterectomy at another hospital for persistent bleeding. Patient 40 is a 29-year-old woman with two prior cesarean deliveries who presented for termination of a 22-week pregnancy for multiple anatomic abnormalities. Formal ultrasonography suggested placenta previa and possible accreta. The uterine evacuation was complicated by hemorrhage initially treated with uterine artery embolization. Heavy vaginal bleeding continued on her first postoperative day, and on hospital day 2 the patient underwent total abdominal hysterectomy, which confirmed placenta increta (total estimated blood loss 3,000 mL).
Patient 41 is a 34-year-old woman with two prior cesarean deliveries who presented for termination of a 21-week pregnancy. Clinic ultrasonography demonstrated a high anterior placenta. Because of refractory hemorrhage (total estimated blood loss 3,200 mL) she underwent uncomplicated uterine artery embolization. Subsequently, she had decreasing hematocrit measurements over the next 3 days, and magnetic resonance imaging (MRI) revealed evidence of placenta percreta. The patient did not desire future fertility and underwent a supracervical hysterectomy that confirmed placenta percreta by pathology. Patient 42 is a 20-year-old woman with two prior cesarean deliveries who desired termination of a 22-week pregnancy. A clinic ultrasonography showed placenta previa and likely increta, which was confirmed by subsequent formal ultrasonography. At the time, clinic protocol for suspected accreta was to perform prophylactic bilateral uterine artery embolization, which was uncomplicated. The evacuation proceeded without complication until removal of placenta, when she began to hemorrhage (total estimated blood loss 3,000 mL), and intraoperative ultrasonography showed almost no myometrium between the placenta and the bladder. A total abdominal hysterectomy with left salpingo-oophorectomy was ultimately necessary for hemostasis, and increta was confirmed on pathology.
Patient 28 is a 22-year-old woman who presented for termination of a 22-week pregnancy. Clinic ultrasonography was unremarkable. Blood loss after evacuation was average, but while in recovery she bled briskly and did not respond to conservative treatment (total estimated blood loss 4,000 mL). The patient was intubated and transferred to Interventional Radiology where DIC was treated with 4 units of packed red blood cells and 2 units of fresh frozen plasma. Embolization was uncomplicated and initially thought to be successful, but postoperatively her hematocrit continued to decrease. Computed tomography scan revealed a right pelvic hematoma and right hydronephrosis. Exploratory laparotomy lead to a right salpingo-oophorectomy, evacuation of the hematoma, repair of the right uterine laceration, and placement of a stent in the right ureter. Her hospital course was further complicated by prolonged intubation for pulmonary edema and subsequent pneumonia. After an 11-day hospitalization with 7 days in intensive care, she was discharged in stable condition and did not require a hysterectomy.
Patient 29 is a 20-year-old woman who requested a termination at 12 weeks of gestation. Formal ultrasonography revealed either a bicornuate or septate uterus. Her evacuation was complicated by bleeding (100 mL) that stabilized with uterotonics and massage. She was discharged and returned 2 hours later with pain and increased vaginal bleeding. Her uterus was reaspirated and she became hypotensive, so she was admitted for observation. Computed tomography scan on the first postoperative day suggested a left broad ligament hematoma. An exploratory laparotomy on the second postoperative day revealed 600 mL of hemoperitoneum and a large but stable left broad ligament hematoma. Surgeons elected left uterine artery embolization rather than evacuate the hematoma, which was successful in controlling bleeding (total estimated blood loss 900 mL, including hemoperitoneum). After the exploratory laparotomy, the patient was having low oxygen saturation, and by spiral computed tomography was diagnosed with a right lower lobe pulmonary embolism and started on a heparin protocol. She was discharged on postoperative day 13.
Patient 30 is a 30-year-old woman with three prior cesarean deliveries who requested a termination at 10 weeks of gestation. Her procedure was complicated by increased pain, nausea, and vomiting in the recovery room. She was reaspirated with ultrasonographic guidance, and 500 mL of bright red blood was evacuated. A partial-thickness false passage was identified in the anterior uterus. After the aspiration, a Foley bulb was placed in the uterine cavity and inflated because of continued bleeding. The collection bag filled with 300 mL of blood in 30 minutes, prompting uterine artery embolization which controlled her bleeding. On postoperative ultrasonography, there was no evidence of hemoperitoneum, but a fluid collection, suggestive of a hematoma, was noted in the anterior myometrium consistent with a partial perforation. The patient was observed overnight and discharged on postoperative day 2. She returned on postoperative day 6 with endometritis and was treated successfully with antibiotics.
Complications of Uterine Artery Embolization
There were two complications of uterine artery embolization in our series of 42 women. The first case involved patient 32, a 28-year-old woman who required uterine evacuation for intrauterine fetal demise at 20 weeks. She developed refractory hemorrhage (total estimated blood loss 800 mL) requiring embolization. Her procedure was complicated by an anaphylactic reaction to the contrast, which was treated with subcutaneous epinephrine and intravenous diphenhydramine. The embolization was otherwise successful and postoperative care was unremarkable. The second case involves patient 39, whose course was described previously. The patient’s embolization was complicated by embolus of the right femoral artery, requiring embolectomy. She had received methylergonovine and phenylephrine for management of atony and hypotension after her abortion.
Our series suggests that uterine artery embolization can safely control postabortion hemorrhage and avoid hysterectomy. In our 42 cases, the median time from completion of abortion to transfer to the Interventional Radiology suite was 3 hours, 18 minutes. This suggests that there is time to assemble the Interventional Radiology team, prepare the operating room, and transfer the patient if necessary, either urgently or after a period of observation.
As uterine fibroid embolization has become more common, the availability of interventional radiologists skilled in deep pelvic arterial embolization procedures has increased. The primary difference between embolization for uterine fibroids and for postabortion or postpartum hemorrhage is the agent used. In embolization for uterine fibroids, the most commonly used agents are polyvinyl alcohol particles and trisacryl microspheres, which are considered to be permanent. In women who desire future fertility and who undergo embolization for postpartum or postabortion hemorrhage, the most commonly used agent is Gelfoam, which is absorbed over 30–45 days and is considered temporary.17 There is, however, a subset of postabortion and postpartum hemorrhage patients in whom the bleeding may not respond to Gelfoam alone, and more extensive and permanent embolization with metallic microcoils may be required. Thus, most interventional radiologists trained in uterine fibroid embolization would be able to provide embolization in urgent settings such as pelvic trauma and postpartum and postabortion hemorrhage.
The increasing rate of primary and repeat cesarean deliveries in the United States has increased the incidence of abnormal placentation.18 Uterine artery embolization can be considered a possible treatment for hemorrhage when accreta is suspected. In the setting of postpartum hemorrhage, abnormal placentation accounts for more than 50% of embolization failures.19 The success of the procedure in our patients with accreta (n=7) was 43%. Of the three successful cases, one woman received further treatment with methotrexate, and two were in women with accreta suspected on ultrasonography without MRI or pathologic confirmation, thus, making it possible that this estimate of success is overly optimistic.
Ultrasonography is considered the primary tool in screening women at risk of placenta accreta. However, results vary widely in its ability to correctly identify abnormal placentation or its absence,20 with a sensitivity ranging from as low as 33% to as high as 86%.18,21,22 It has also been suggested that MRI be used in cases where ultrasonography is inconclusive18,22 or as an aide in surgical planning,23 but methodological concerns about studies, and expense and availability of MRI limits its widespread use. Our clinic protocol at the beginning of this case series was to do prophylactic embolization when there was high suspicion of accreta, as has been described in a previous case series.24 We changed this protocol after considering the poor predictive ability of ultrasonography for accreta, particularly the high false-positive rates of 54.5%,18,20,25 as well as our experience with the patient who required hysterectomy after preoperative embolization. Further, embolization in the emergent setting may be more successful than prophylactic procedures because active bleeding can be seen under fluoroscopy, allowing embolization in the region of the hemorrhage.26,27
When counseling patients with suspected accreta about the efficacy of uterine artery embolization, they should be informed about the risk of requiring subsequent medical or surgical treatment. Women desiring fertility are likely to choose embolization first. Although the effect of embolization on fertility is undetermined, there have been a few studies of women who have had pregnancies after embolization to treat uterine fibroids,28 and two of the patients in our embolization series presented subsequently for abortions, which were uncomplicated. Women with probable accreta who do not desire future fertility may opt for hysterectomy instead of embolization in the setting of refractory hemorrhage, and they may choose to try embolization with a 43% chance of avoiding surgery. If embolization fails to treat the hemorrhage completely, it may decrease intraoperative blood loss during the subsequent hysterectomy.
In conclusion, using uterine artery embolization to treat refractory postabortion hemorrhage avoids hysterectomy, especially when hemorrhage is caused by uterine atony or cervical laceration. By avoiding hysterectomy in most cases, embolization of the uterine arteries not only maintains potential fertility but also decreases further operative morbidity after abortion complicated by hemorrhage.
1. Bartlett LA, Berg CJ, Shulman HB, Zane SB, Green CA, Whitehead S, et al. Risk factors for legal induced abortion–related mortality in the United States. Obstet Gynecol 2004;103:729–37.
2. Ferris LE, McMain-Klein M, Colodny N, Fellows GF, Lamont J. Factors associated with immediate abortion complications. CMAJ 1996;154:1677–85.
3. Grimes DA, Flock ML, Schulz KF, Cates W Jr. Hysterectomy as treatment for complications of legal abortion. Obstet Gynecol 1984;63:457–62.
4. King TM, Atienza MF, Burkman RT. The incidence of abdominal surgical procedures in a population undergoing abortion. Am J Obstet Gynecol 1980;137:530–3.
5. Abu-Judeh H, Min R, Khilnani N, Kandarpa K. Uterine artery embolization. Applied Radiol 2002;31 suppl:29–37.
6. Descargues G, Douvrin F, Degre S, Lemoine JP, Marpeau L, Clavier E. Abnormal placentation and selective embolization of the uterine arteries. Eur J Obstet Gynecol Reprod Biol 2001;99:47–52.
7. Ledee N, Ville Y, Musset D, Mercier F, Frydman R, Fernandez H. Management in intractable obstetric haemorrhage: an audit study on 61 cases. Eur J Obstet Gynecol Reprod Biol 2001;94:189–96.
8. Wang H, Garmel S. Successful term pregnancy after bilateral uterine artery embolization for postpartum hemorrhage. Obstet Gynecol 2003;102:603–4.
9. Borgatta L, Chen AY, Reid SK, Stubblefield PG, Christensen DD, Rashbaum WK. Pelvic embolization for treatment of hemorrhage related to spontaneous and induced abortion. Am J Obstet Gynecol 2001;185:530–6.
10. Hong TM, Tseng HS, Lee RC, Wang JH, Chang CY. Uterine artery embolization: An effective treatment for intractable obstetric haemorrhage. Clin Radiol 2004;59:96–101.
11. Liu X, Fan G, Jin Z, Yang N, Jiang Y, Gai M, et al. Lower uterine segment pregnancy with placenta increta complicating first trimester induced abortion: diagnosis and conservative management. Chin Med J (Engl) 2003;116:695–8.
12. Elia G, Counsell C, Singer SJ. Uterine artery malformation as a hidden cause of severe uterine bleeding: a case report. J Reprod Med 2001;46:398–400.
13. Haseltine FP, Glickman MG, Marchesi S, Spitz R, Dlugi A, DeCherney AA. Uterine embolization in a patient with postabortal hemorrhage. Obstet Gynecol 1984;63:78S–80S.
14. Imbar T, Bloom A, Ushakov F, Yagel S. Uterine artery embolization to control hemorrhage after termination of pregnancy implanted in a cesarean delivery scar. J Ultrasound Med 2003;22:1111–5.
15. Darney PD, Horbach NS, Korn AP. Protocols for office gynecologic surgery. Cambridge (MA): Blackwell Science; 1996.
16. Hadlock FP, Harrist RB. Determination of gestational age. In: Harshberger SE, editor. Ultrasound in obstetrics and gynecology. St. Louis (MO): Mosby; 1981. p. 23–26.
17. Siskin GP, Englander M, Stainken BF, Ahn J, Dowling K, Dolen EG. Embolic agents used for uterine fibroid embolization. AJR Am J Roentgenol 2000;175:767–73.
18. Warshak CR, Eskander R, Hull AD, Scioscia AL, Mattrey RF, Benirschke K, et al. Accuracy of ultrasonography and magnetic resonance imaging in the diagnosis of placenta accreta. Obstet Gynecol 2006;108:573–81.
19. Descargues G, Mauger Tinlot F, Douvrin F, Clavier E, Lemoine JP, Marpeau L. Menses, fertility and pregnancy after arterial embolization for the control of postpartum haemorrhage. Hum Reprod 2004;19:339–43.
20. Comstock CH, Love JJ Jr, Bronsteen RA, Lee W, Vettraino IM, Huang RR, et al. Sonographic detection of placenta accreta in the second and third trimesters of pregnancy. Am J Obstet Gynecol 2004;190:1135–40.
21. Lam G, Kuller J, McMahon M. Use of magnetic resonance imaging and ultrasound in the antenatal diagnosis of placenta accreta. J Soc Gynecol Investig 2002;9:37–40.
22. Levine D, Hulka CA, Ludmir J, Li W, Edelman RR. Placenta accreta: evaluation with color Doppler US, power Doppler US, and MR imaging. Radiology 1997;205:773–6.
23. Palacios Jaraquemada JM, Bruno CH. Magnetic resonance imaging in 300 cases of placenta accreta: surgical correlation of new findings. Acta Obstet Gynecol Scand 2005;84:716–24.
24. Soncini E, Pelicelli A, Larini P, Marcato C, Monaco D, Grignaffini A. Uterine artery embolization in the treatment and prevention of postpartum hemorrhage. Int J Gynaecol Obstet 2007;96:181–5.
25. Comstock CH. Antenatal diagnosis of placenta accreta: a review. Ultrasound Obstet Gynecol 2005;26:89–96.
26. Brown BJ, Heaston DK, Poulson AM, Gabert HA, Mineau DE, Miller FJ Jr. Uncontrollable postpartum bleeding: A new approach to hemostasis through angiographic arterial embolization. Obstet Gynecol 1979;54:361–5.
27. Heaston DK, Mineau DE, Brown BJ, Miller FJ. Transcatheter arterial embolization for control of persistent massive puerperal hemorrhage after bilateral surgical hypogastric artery ligation. AJR Am J Roentgenol 1979;133:152–4.
28. Usadi RS, Marshburn PB. The impact of uterine artery embolization on fertility and pregnancy outcome. Curr Opin Obstet Gynecol 2007;19:279–83.