CardioVascular and Interventional Radiology

, Volume 29, Issue 3, pp 354–361

Balloon-Assisted Occlusion of the Internal Iliac Arteries in Patients with Placenta Accreta/Percreta

Authors

  • Leonard J. Bodner
    • Department of RadiologyUMDNJ–Robert Wood Johnson Medical School
    • Department of RadiologyUMDNJ–Robert Wood Johnson Medical School
  • Christopher Gribbin
    • Department of RadiologySt. Peter’s University Hospital
  • Randall L. Siegel
    • Department of RadiologyUMDNJ–Robert Wood Johnson Medical School
  • Stephanie Beale
    • Department of RadiologyUMDNJ–Robert Wood Johnson Medical School
  • William Scorza
    • Department of Obstetrics and Gynecology, Division of Maternal-Fetal MedicineSaint Peter’s University Hospital
Article

DOI: 10.1007/s00270-005-0023-2

Cite this article as:
Bodner, L.J., Nosher, J.L., Gribbin, C. et al. Cardiovasc Intervent Radiol (2006) 29: 354. doi:10.1007/s00270-005-0023-2

Abstract

Background

Placenta accreta/percreta is a leading cause of third trimester hemorrhage and postpartum maternal death. The current treatment for third trimester hemorrhage due to placenta accreta/percreta is cesarean hysterectomy, which may be complicated by large volume blood loss.

Purpose

To determine what role, if any, prophylactic temporary balloon occlusion and transcatheter embolization of the anterior division of the internal iliac arteries plays in the management of patients with placenta accreta/percreta.

Methods

The records of 28 consecutive patients with a diagnosis of placenta accreta/percreta were retrospectively reviewed. Patients were divided into two groups. Six patients underwent prophylactic temporary balloon occlusion, followed by cesarean section, transcatheter embolization of the anterior division of the internal iliac arteries and cesarean hysterectomy (n = 5) or uterine curettage (n = 1). Twenty-two patients underwent cesarean hysterectomy without endovascular intervention. The following parameters were compared in the two groups: patient age, gravidity, parity, gestational age at delivery, days in the intensive care unit after delivery, total hospital days, volume of transfused blood products, volume of fluid replacement intraoperatively, operating room time, estimated blood loss, and postoperative morbidity and mortality.

Results

Patients in the embolization group had more frequent episodes of third trimester bleeding requiring admission and bedrest prior to delivery (16.7 days vs. 2.9 days), resulting in significantly more hospitalization time in the embolization group (23 days vs. 8.8 days) and delivery at an earlier gestational age than in those in the surgical group (32.5 weeks). There was no statistical difference in mean estimated blood loss, volume of replaced blood products, fluid replacement needs, operating room time or postoperative recovery time.

Conclusion

Our findings do not support the contention that in patients with placenta accreta/percreta, prophylactic temporary balloon occlusion and embolization prior to hysterectomy diminishes intraoperative blood loss.

Keywords

Angiographic embolizationPlacenta accretePlacenta percreta

Placenta accreta is a disorder of placental implantation characterized by ingrowth of the placental villi into the uterine wall. It is subdivided into three distinct entities based on the depth of placental invasion. Placenta accreta is characterized by ingrowth of the placental villi through the thinned decidua basalis and the layers of Nitabuch. The term placenta increta is reserved for invasion of the placenta into the myometrium. Invasion through the myometrium reaching or penetrating the serosa is termed placenta percreta.

Patients with placenta accreta are at a significant risk for maternal and/or fetal mortality, massive blood loss, disseminated intravascular coagulation and infection. Antepartum diagnosis is critical to prevent life-threatening hemorrhage. Pelvic ultrasound [1, 2] and MRI [3] provide a means for antepartum diagnosis and treatment planning. The standard treatment of placenta accreta/percreta is hysterectomy [4].

DuBois [5] advocated prophylactic balloon occlusion of the anterior division of the internal iliac arteries followed by embolization and hysterectomy in patients with placenta accreta/percreta. Our study attempted to evaluate the efficacy of prophylactic catheter intervention and embolization prior to hysterectomy in this group of patients. Beginning in June 2000, all patients referred to the Section of Maternal Fetal Medicine with an antenatal diagnosis of placenta accreta/percreta were offered a treatment protocol of prophylactic balloon occlusion of the anterior division of the hypogastric artery followed by embolization of the anterior division if brisk bleeding persisted from the placental implantation site. Hysterectomy was performed after catheter intervention. We retrospectively compared this group with patients undergoing cesarean hysterectomy as the sole method of treatment during a similar time period in the same institution.

Materials and Methods

Investigational review board approval was obtained for this retrospective review.

A review of the computerized database at a university affiliate teaching hospital showed that between June 2000 and December 2002, 28 patients were discharged with a diagnosis of placenta accreta/percreta. Twenty-seven of these patients were treated with a hysterectomy. One patient whose bleeding was felt to be adequately controlled by prophylactic balloon occlusion and embolization had uterine curettage. In 10 patients, second trimester ultrasound examination disclosed findings indicative of the presence of placenta accreta/percreta. MRI was performed in one case in which ultrasound raised a question of bladder invasion. Seven of these 10 patients were referred to the Division of Maternal Fetal Medicine for delivery. The Interventional Radiology service was then consulted for consideration of intraoperative balloon occlusion and embolization of the anterior division of the internal iliac arteries.

Six of these seven patients were brought to the interventional radiology suite on the day of delivery. The obstetric team accompanied the patient to the angiography suite. Continuous fetal monitoring was performed during the course of the procedure. Through 7 Fr sheaths inserted in both groins, access to both internal iliac arteries by a contralateral approach was achieved. Seven French, 11.5 mm occlusion balloons (Boston Scientific, Natick, MA, USA) were placed into the anterior division of the internal iliac arteries bilaterally. Fluoroscopy time was recorded in 3 cases. The sheaths and catheters were connected to saline flush solutions and the patients transported to an operating room with fluoroscopic capability. Each infant was delivered by cesarean section. Occlusion balloons were inflated at the time of cord clamping. The balloons were inflated until they assumed a slightly elliptical shape indicating apposition against the wall of the hypogastric artery. A small test dose of contrast medium confirmed flow arrest. The uterus was packed and embolization was performed with 2 mm × 2 mm Gelfoam pledgets (Pharmacia and Upjohn, Kalamazoo, MI, USA). The balloons were intermittently deflated during embolization to determine whether embolization was complete. Embolization was completed when flow arrest was evident fluoroscopically with the balloons deflated. With embolization completed, the hysterotomy was closed and hysterectomy or uterine curettage was performed. Arterial sheaths were removed in the recovery suite and the patients were subsequently transferred to the intensive care unit (ICU) for a minimum of 24 hr. A similar protocol was prearranged for the seventh patient; however, the sudden onset of antepartum hemorrhage necessitated immediate surgical intervention without endovascular intervention.

During the same period of time a control group of 22 patients with a postpartum diagnosis of placenta accreta/percreta were treated with cesarean hysterectomy but without endovascular intervention. One of these 22 patients was scheduled for prophylactic balloon occlusion and embolization, but, as described above, required immediate surgical intervention. Chart review of all 28 patients was performed and the two groups were compared with regard to the following variables: patient age, gravidity, parity, gestational age at delivery, days in the ICU after delivery, total hospital days, volume of transfused blood products (packed red blood cells), volume of fluid replacement intraoperatively, operating room time, estimated blood loss, and postoperative morbidity and mortality.

A two-sample, one-tailed Student’s t-test was performed for the comparison of each parameter. The null hypothesis for each test was that the mean value of the parameter was equal for the two groups. The statistical test was performed on Microsoft Excel.

Results

The study compares two nonrandomized groups of patients. The two groups were divided by referral patterns. Demographic characteristics and clinical outcomes in the embolization-assisted hysterectomy group are presented in Tables 1 and 2. Intraoperative data in the embolization-assisted hysterectomy group are presented in Table 3. Demographic characteristics and clinical outcomes in the control group are presented in Tables 4 and 5 and intraoperative data for this group are presented in Table 6.
Table 1

Demographic data of the embolization group

Patient no.

Maternal age (years)

Gravida/Para

Gestational age (weeks)

1

39

7/5

28

2

42

6/3

34

3

35

4/1

36

4

24

4/1

30

5

43

2/2

31

6

29

3/1

36

Mean

35.3

4.3/2.2

32.5

Table 2

Clinical data of the embolization group

Patient no.

Total no. of hospital days

Hospital days after delivery

ICU days after delivery

Volume of transfused blood products (units packed RBCs)

Volume of fluid replacement (l)

DIC

Morbidity

Position of placenta

1

65

4

1

17

5.0

Yes

None

Previa + focal accreta

2

17

5

2

6

8.5

No

None

Previa + increta

3

9

7

4

9

9

No

None

Previa + accreta

4

11

10

3

7

3.2

No

Pelvic hematoma

Previa + percreta

5

27

5

3

0

45

Yes

None

Previa + accreta

6

8

6

2

0

4.5

No

None

Previa +percreta

Mean

23

6.2

2.5

6.5

5.8

   

ICU, intensive care unit; RBCs, red blood cells; DIC, disseminated intravascular coagulation

Table 3

Intraoperative data of the embolization group

Patient no.

OR time (min)

EBL (l)

OR procedure

1

215

2.2

Balloon occlusion + embolization anterior division + hysterectomy

2

405

2.5

Balloon occlusion + embolization anterior division + hysterectomy

3

551

7.0

Balloon occlusion + embolization anterior division + hysterectomy; cystostomy + bladder repair

4

380

2.5

Balloon occlusion + embolization anterior division + hysterectomy; cystotostomy + bladder repair

5

225

1.0

Balloon occlusion +embolization anterior division + curettage of lower uterine segment

6

254

1.5

Balloon occlusion + embolization anterior division + hysterectomy

Mean

338.3

2.8

 

OR, operating room; EBL, estimated blood loss

Table 4

Demographic data of the control group

Patient no.

Maternal age (years)

Gravida/Para

Gestational age (weeks)

1

38

6/3

40

2

34

4/1

40

3

39

4/1

38

4

28

2/0

37

5

33

2/1

38

6

34

2/2

38

7

35

4/2

27

8

40

4/2

32

9

34

3/2

38

10

37

3/2

38

11

37

2/1

38

12

38

9/4

35

13

36

2/1

40

14

31

5/2

40

15

36

5/2

40

16

37

5/2

34

17

29

2/1

37

18

40

3/0

36

19

38

3/1

37

20

36

2/1

39

21

26

3/2

28

22

40

2/1

35

Mean

35.3

3.5/1.5

36.5

Table 5

Clinical data of the control group

Patient no.

Preop. diagnosis of placenta accreta

Total no. of hospital days

Hospital days after delivery

ICU days after delivery

Volume of t ransfused blood products (units packed RBCs)

Volume of fluid replacement (l)

DIC

Morbidity

Position of placenta

1

No

4

4

0

0.0

3.6

No

None

Accreta + previa

2

No

8

7

4

17.0

15.5

No

None

Accreta + previa

3

No

9

9

5

18.0

8.0

Yes

Hemorrhagic shock POD 0, hypothermia, hemoperitoneum, laparotomy POD 1

Increta + previa

4

No

9

9

5

11.0

0.7

No

Intra-abdominal hemorrhage, laparotomy, bradycardia, cardiac resuscitation

Accreta + previa

5

No

4

3

1

4.0

4.6

No

None

Accreta + previa

6

No

5

4

2

3.0

3.2

No

None

Accreta + previa

7

Yes

8

7

4

19.0

11.0

Yes

None

Accreta + previa

8

No

17

6

0

7.0

3.5

No

None

Accreta + previa

9

No

5

5

2

6.0

5.5

No

None

Accreta + previa

10

No

6

6

0

7.0

4.0

No

None

Accreta + previa

11

No

7

6

2

6.0

6.0

No

None

Accreta + previa

12

No

10

9

4

6.0

7.0

No

None

Accreta + previa

13

No

5

5

0

0.0

4.5

No

None

Accreta + previa

14

No

5

4

1

2.0

3.5

No

None

Accreta + previa

15

No

7

6

1

4.0

5.3

No

None

Accreta + previa

16

No

13

5

0

4.0

4.7

No

None

Accreta + previa

17

Yes

8

8

3

7.0

7.0

No

None

Accreta + previa

18

No

5

5

1

6.0

7.0

No

Uterine atony

Accreta + previa

19

No

39

5

0

2.0

1.8

No

Fever

Accreta + previa

20

No

5

5

0

2.0

2.7

yes

None

Increta + previa

21

Yes (suspect percreta)

9

5

1

5.0

2.0

Yes

None

Percreta + previa

22

Yes

6

6

1

3.0

5.0

No

Wound infection

Accreta + previa

Mean

 

8.8

5.9

1.6

6.3

5.3

   

ICU, intensive care unit; RBCs, red blood cells; DIC, disseminated intravascular coagulation; POD, postoperative day

Table 6

Intraoperative data of the control group

Patient no.

OR time (min)

EBL (l)

OR procedure

1

125.0

1.0

Cesarean hysterectomy

2

650

4.0

Cesarean hysterectomy, cystoscopy, lysis of adhesions

3

400

4.0

Cesarean hysterectomy

4

210

2.5

Cesarean hysterectomy; reoperation for intraperitoneal bleed

5

150

2.5

Cesarean hysterectomy; bilateral uterine artery ligation

6

130

2.0

Cesarean hysterectomy, left salpingoophorectomy

7

395

6.5

Cesarean hysterectomy

8

160

1.9

Cesarean hysterectomy, bilateral uterine artery ligation

9

240

4.5

Cesarean hysterectomy

10

125

1.5

Cesarean hysterectomy

11

235

3.0

Cesarean hysterectomy

12

300

2.5

NSVD, hysterectomy

13

105

3.0

Cesarean hysterectomy, repair of bladder laceration, right ureter stent

14

125

0.4

Cesarean hysterectomy

15

325

2.5

NSVD, hysterectomy

16

120

2.0

Cesarean hysterectomy

17

275

2.5

Cesarean hysterectomy, bilateral uterine artery ligation

18

307

3.0

Cesarean hysterectomy, bilateral uterine artery ligation, right hypogastric artery ligation

19

160

1.8

Cesarean hysterectomy, bilateral uterine art ligation, right hypogastric and ovarian art ligation, left salpingoophorectomy

20

133

2.5

Cesarean hysterectomy

21

144

2.0

Cesarean hysterectomy

22

210

3.0

Cesarean hysterectomy, left hypogastric artery ligation

Mean

228

2.6

 

OR, operating room; EBL, estimated blood loss; NSVD, normal spontaneous vaginal delivery

On the basis of gray-scale and color Doppler ultrasound examinations performed in the second or third trimester, placenta previa was detected in all 28 patients and, therefore, all patients were scheduled for cesarean section and possible hysterectomy. Gray-scale and color Doppler ultrasound suggested the presence of placenta accreta/percreta in 4 of 22 patients in the group treated without endovascular intervention. Placenta accreta/percreta was identified at the time of delivery in the remaining 18 patients in the latter group. Two patients in the group treated without endovascular intervention with a preoperative diagnosis of placenta accreta/percreta were scheduled to undergo cesarean hysterectomy. The other 2 patients with preoperative ultrasound findings suggestive of placenta accreta/percreta expressed a desire for an attempt at uterine-sparing surgery. Due to uncontrolled hemorrhage at the placental implantation site, hysterectomy was performed in the latter 2 patients as well as in the 18 patients in whom placenta accreta/percreta was discovered intraoperatively. Placenta accreta/percreta was suspected in all 6 patients in the embolization group. Therefore, all patients in the embolization group were scheduled for cesarean section and possible hysterectomy. Due to uncontrolled hemorrhage, all but 1 patient required hysterectomy.

Demographic data in the two groups were similar (Tables 1, 4). Two variables demonstrated significant differences between the embolization group and the control group: (1) patients in the embolization group were delivered at an earlier gestational age than those in the control group (32.5 weeks vs. 36.5 weeks, p = 0.019), and (2) patients in the embolization group were hospitalized for a longer period of time prior to delivery than patients in the control group (16.7 days vs. 2.9 days) and, therefore, the embolization group spent significantly more days in the hospital than the control group (23 days vs. 8 days, p = 0.007). There were no significant differences in blood loss (p = 0.40), units of packed red blood cells transfused through the entire hospitalization (p = 0.47), days spent in the ICU after delivery (p = 0.14), or the volume of fluid replacement required to maintain hemodynamic stability (p = 0.28). Though the average operating room time was greater in the embolization group (338 min vs. 228 min), the difference was not significant (Table 7).
Table 7

Statistical comparison of the prophylactic embolization group versus the control group

Variable

Prophylactic embolization group (n = 6)

Control group (n = 22)

p value (Student’s t-test)

Maternal age

35.3

35.3

0.526

Gravidity

4.3

3.5

0.070

Parity

2.2

1.5

0.127

Gestational age at delivery (weeks)

32.5

36.5

0.019

ICU days after delivery

2.5

1.6

0.14

Total inpatient days

23

8.8

0.007

Volume of transfused packed RBCs (units)

6.5

6.3

0.47

Volume of intraoperative fluid replacement (l)

5.8

5.3

0.28

OR time (min)

338.3

228

0.052

Estimated blood loss (l)

2.8

2.6

0.40

ICU, intensive care unit; RBCs, red blood cells; OR, operating room

Two patients in the embolization group (33%) and 5 patients in the control group (23%) experienced disseminated intravascular coagulation. Two patients in the embolization group required cystostomy and bladder repair. One of these patients had a placenta percreta growing into the serosal surface of the bladder; the other patient had a placenta accreta involving the lower uterine segment. One patient in the control group with placenta accreta involving the lower uterine segment required cystostomy and bladder repair. In an unsuccessful attempt to stem intraoperative bleeding and spare the uterus, 5 patients in the control group required bilateral uterine artery ligation. Due to continued hemorrhage, one of these 5 patients had unilateral hypogastric artery ligation and a second patient had unilateral hypogastric and ovarian artery ligation. A sixth patient in the control group underwent unilateral hypogastric artery ligation. There was no need to ligate the uterine, hypogastric, or ovarian arteries prior to hysterectomy in the embolization group. Two patients in the control group were explored on postoperative day 1 for continuing hemorrhage and hemodynamic instability. One of these 2 patients survived an episode of cardiac arrest. None of the patients in the embolization group were hemodynamically unstable after delivery. There was no maternal or fetal mortality in either group.

Fluoroscopy time for placement of occlusion balloons into the anterior division of the hypogastric arteries was recorded in 3 cases. The average time was 4 min 20 sec. The maximum time recorded was 5 min 35 sec. The average estimated fetal radiation dose was 3.2 rads.

Discussion

Obstetric hemorrhage remains a leading cause of pregnancy-related mortality in the United States. Approximately 29% of maternal deaths are due to bleeding [6]. Placenta accreta is an important cause of obstetric hemorrhage and a leading cause of peripartum hysterectomy [7].

Placenta accreta is more common in multiparous women, especially those with a history of previous cesarean section or placenta previa. It is encountered in 0.04% or more of pregnancies and the incidence is increasing, likely reflecting the increasing frequency of cesarean section [8]. Placenta percreta, the most serious form of placental invasion, constitutes only 5% of cases of placenta accreta. The treatment of placenta accreta/percreta is hysterectomy [4]. Uterine-sparing surgery can be attempted dependent on the degree of placental invasion and the desire of the patient for future pregnancy.

The diagnosis of placenta accreta/percreta can be made early in pregnancy using ultrasound with color flow imaging and more recently with MRI. Combined gray-scale and color Doppler criteria for the diagnosis include thinning or loss of the hypoechoic retroplacental myometrial zone in the lower uterine segment, thinning or interruption of the hyperechoic uterine–bladder wall interface, and exophytic placental growth beyond the uterine serosa and lacunar flow, i.e., the presence of visible flow within intraplacental lacunar vascular spaces [1, 2]. Using these criteria, ultrasound has 93% sensitivity and 79% specificity for the diagnosis of placenta accreta [2].

With the diagnosis of placenta accreta/percreta established in the antepartum period, the obstetrician can make preparations for delivery of the neonate, and safe removal of the placenta with a minimal risk of postpartum hemorrhage. Frequently, this entails gathering an interdisciplinary team of anesthesiologists, vascular surgeons, urologists, and interventional radiologists.

Several surgical options have been investigated to minimize blood loss in the high-risk obstetric patient. Proximal ligation of the internal iliac artery results in the control of obstetric hemorrhage with success in less than 50% of patients [9, 10]. While ligation of the internal iliac arteries appears to be effective for bleeding due to uterine atony it is less effective for placenta accreta [6]. The efficacy of ligation is limited by rapid recruitment of an extensive collateral system in the pelvis [11]. AbdRabbo described a series of surgical procedures designed to gradually devascularize the uterus, beginning with unilateral uterine artery ligation followed sequentially by bilateral uterine artery ligation, low uterine artery ligation and ovarian artery ligation [12]. The technical success rate was 100% and hysterectomy was avoided in each case. Nonetheless, this system of stepwise uterine devascularization is often deferred in favor of hysterectomy. Suture ligation of the uterine artery as it ascends on the uterine wall has also been described to control obstetric hemorrhage when the artery is accessible to the surgeon.

Transcatheter embolization has proven to be a valuable tool for the treatment of pelvic hemorrhage from trauma, malignancy, and obstetric and gynecologic causes. Embolization for the treatment of obstetric hemorrhage was first reported in 1979 [13]. It has been described for the treatment of hemorrhage from uterine atony, uterovaginal lacerations, uterine vascular malformations, ectopic pregnancy, and abnormalities of placental implantation.

The value of internal iliac artery or uterine artery embolization in the treatment of obstetric hemorrhage is supported by case reports and several small series. In a review of 138 cases reported in the literature between 1979 and 1999, Badawy et al. described procedural success in 95% of patients [14] as judged by the avoidance of hysterectomy or comparison of blood loss with that in historical controls.

In some series, embolization was employed without prior balloon occlusion and in others, balloon occlusion preceded embolization. We preferred the latter method of treatment believing that balloon occlusion would allow temporary control of hemorrhage. Once the balloons were deflated, bleeding would likely occur through pelvic collaterals as reported in 1 patient [5]. With the balloons inflated, hemodynamic stability would be maintained while embolization proceeded. We elected to inflate the occlusion balloons in the anterior division of the hypogastric arteries and embolize from there. We felt that placing occlusion balloons in the uterine arteries prior to delivery would likely induce spasm, making distal embolization impossible and possibly creating some degree of uteroplacental insufficiency. We opted not to try balloon occlusion alone without embolization. We felt that balloon occlusion would be no more efficacious than hypogastric artery ligation.

Embolization of the internal iliac arteries or superselective embolization of the uterine arteries has been utilized both on an emergent basis and as prophylactic treatment for placenta accreta. Mitty et al. [15] described five cases of placenta accreta treated with uterine artery embolization: 3 emergently and 2 electively. None of these patients required hysterectomy for hemostasis. Blood loss in the emergent treatment group was 2500 and 4100 ml, and in the elective group was 100 to 1000 ml. None of these patients had placenta increta or percreta. Hansch et al. [16] reported embolization for obstetric hemorrhage in 6 patients, 2 of whom underwent prophylactic embolization of the internal iliac and uterine arteries at the time of cesarean section for placenta accreta. Catheters were placed in the internal iliac arteries prior to cesarean delivery. In 1 patient, occlusion balloons were placed in the anterior division of the internal iliac artery prior to embolization. The second patient was treated by embolization alone. In the patient treated by embolization without balloon occlusion, bleeding could not be controlled and hysterectomy was required. The volume of blood loss was 21 in the patient treated with balloon occlusion and embolization and 41 in the patient treated with embolization and subsequent hysterectomy. Chou et al. [17] reported prophylactic embolization of 6 patients undergoing abortion/hysterectomy for placenta accreta at a mean gestational age of 19.7 weeks. On pathologic examination 1 of the patients had placenta accreta, 3 had placenta increta, and 2 had placenta percreta. Mean blood loss was reported as 1767 ml (range 300–3000 ml) with a transfusion requirement of 4.33 units and hospital stay of 8.17 days. Dubois et al. [5] prophylactically embolized 2 patients with placenta percreta prior to hysterectomy. The estimated blood loss during surgery in these 2 cases was 1500–2000 ml. The estimated blood loss in 5 of our 6 patients was similar to the amounts reported in the aforementioned reports (1000–2500 ml). Despite embolization, 1 patient in our series lost 71 of blood during surgery.

During the same period in which our team performed embolization-assisted hysterectomy, other practices treated placenta accreta by cesarean hysterectomy alone. This afforded us the opportunity to compare cesarean hysterectomy or uterine curettage combined with prophylactic balloon occlusion and embolization with cesarean hysterectomy alone for the treatment of placenta accreta. Our study is not randomized. The two groups of patients were divided by referral patterns. Randomization of these two groups, while providing an ideal study, would undoubtedly have been exceedingly difficult to accomplish. The sudden onset and massive hemorrhage that can complicate placenta accreta/percreta frequently makes it impossible to assemble an interventional radiology team prior to surgery. Furthermore, whereas the diagnosis of placenta previa is readily apparent on antepartum sonography, the findings of placenta previa with placenta accreta/percreta are far more subtle. Those who treat high-risk obstetric patients are more likely to recognize the findings of placenta accreta/percreta than community obstetricians who do not deal with this condition on a frequent basis. This may account for the lack of a preoperative diagnosis of placenta accreta/percreta in 18 of 22 patients in the group treated without endovascular intervention. Our study, therefore, compares a group of patients with placenta accreta/percreta treated with prophylactic embolization and a group of historical controls within the same community treated during the same time interval. The lack of randomization introduced a referral bias in which patients with the prenatal diagnosis of placenta accreta, and especially those with a more complicated prenatal course, fell into the embolization group. This is reflected in the fact that in the embolization group the duration of hospitalization, primarily reflecting the prenatal hospitalization period, was statistically longer and the gestational age at delivery was younger. Despite this potential referral bias, the most severe forms of abnormal placental implantation, placenta increta and percreta, were present in both groups (2 cases in the embolization group and 3 cases in the surgical group). When comparing variables including estimated blood loss, days spent in the ICU after delivery or transfusion requirements we found no convincing difference between the two groups. It is, therefore, not possible to recommend prophylactic embolization of the anterior division of the uterine arteries prior to cesarean hysterectomy for placenta accreta. The limited number of cases of placenta increta/percreta in both groups provides no information on the place of embolization in the more advanced stages of placental invasion.

Levine et al. reported a similar experience to ours with a slightly different approach. When they compared prophylactic balloon occlusion of the anterior division of the internal iliac artery followed by cesarean hysterectomy with cesarean hysterectomy alone for the treatment of placenta accreta, they found no difference in estimated blood loss, transfusion requirements or length of hospitalization [18].

When considering prophylactic uterine artery or anterior division internal iliac artery embolization for placenta accreta one needs to consider the safety of the procedure, the effect on future childbearing as well as the potential effects of radiation on the fetus. The complication rate of embolization for postpartum and post-cesarean bleeding is approximately 6–7% with no reports of maternal mortality [19]. More recently, the safety of uterine artery embolization has been demonstrated in large series of embolizations for uterine fibroids in which 1% of patients experienced serious complications such as endometritis or pulmonary embolus [20]. Of note are two ischemic complications reported by Ornan et al. in patients embolized for control of postpartum hemorrhage. One patient experienced buttock claudication and one suffered bowel infarction [21]. Each of these patients underwent ligation of the internal iliac arteries prior to embolization and as a result had limited capacity for collateral flow in the pelvis. The capacity for collateral flow should be a consideration prior to undertaking internal iliac artery embolization in patients with previous surgery or vascular disease. Other severe but rare ischemic complications include bladder gangrene [22] and lower extremity paresis due to ischemia of the sciatic and femoral nerves [23]. We chose Gelfoam as the embolic agent as it is absorbed in 30–60 days and can be cut into large particles to achieve a proximal occlusion and not compromise collateral revascularization in the short term. Ornan et al. reviewed the implications of pelvic embolization on fertility in their group of 28 consecutive embolizations and reported 6 patients with subsequent uncomplicated pregnancy [21].

The average fetal radiation dose in our patients was 3.2 rads. The median fetal dose in the study by Levine et al. was 6.1 rads [18]. It has been estimated that there is a 99.5% likelihood of a child never developing childhood cancer following fetal radiation doses up to 10 rads [24].

Other approaches to the treatment of placenta percreta include reports of embolization followed by methotrexate administration and delayed delivery of the placenta [2527]. Further investigation is necessary to determine whether there is any role for prophylactic embolization in treating patients with placenta percreta where the placenta is left in situ.

In summary, based on our experience and comparison of our results with a historical population of cesarean hysterectomies performed at our institution during the same time period, we question the value of prophylactic embolization of the anterior division of the internal iliac arteries in conjunction with cesarean hysterectomy or uterine curettage in patients with placenta accreta/percreta.

Copyright information

© Springer Science+Business Media, Inc. 2006