Ultrasonographic indications for conservative treatment in pregnancy-related uterine arteriovenous malformations

Abstract

Background

Uterine arteriovenous malformations (AVMs) are known to spontaneously regress.

Purpose

To assess the predictive value of ultrasonography for patients requiring conservative treatment for pregnancy related to AVMs.

Material and Methods

Our prospective study included 75 patients (conservative management:therapeutic management = 45:30) with vaginal bleeding from pregnancy-related AVM. Clinical and ultrasonography examinations were reviewed, and the following information was gathered: complete blood count, AVM maximal diameter, AVM echogenicity, retained product of conception, number of blood vessels, and spectral Doppler (pulsatility index [PI], resistance index [RI], peak systolic velocity [PSV], time-averaged maximum velocity [TAMXV]). The Doppler criteria by Timmerman (mean PSV >70 cm/s: therapeutic management, mean PSV < 52 cm/s: conservative management) were used for the initial management selection. The association between experimental variables and outcomes was assessed to determine their usefulness for predicting conservative management.

Results

Features strongly associated with conservative management and their accuracy were PSV 89.6%, hemoglobin 84.7%, RI 83.1%, TAMXV 79.3%, and PI 78.6%. The overall accuracy for correct outcome classification was 64 (85.3%) of 75 patients. Most patients with conservative management had quicker improvement of symptoms and spontaneous regression at follow-up.

Conclusion

Ultrasonography can accurately predict selection of conservative management.

Keywords

Arteriovenous malformations (AVMs)uterus ultrasonography

Introduction

Uterine arteriovenous malformations (AVMs) can be either congenital or acquired. Congenital AVMs are rare, whereas acquired or traumatic AVMs are being diagnosed at increasing rates (1 –5). A prior dilation and curettage (D&C), therapeutic abortion, uterine surgery, and direct uterine trauma are commonly reported causes of AVMs (3,6 –8). Endovaginal ultrasonography with color Doppler is usually used for initial identification (“color mosaic” pattern) (9 –14). To control bleeding, aggressive therapeutic management such as transcatheter embolization of the uterine arteries or hysterectomy may be required (15 –19). However, a few authors have described regression of the AVM with conservative management or spontaneous resolution (20 –22). There are no clear consensus guidelines, and management of symptomatic AVMs is an open question. Few studies have assessed the criteria for and benefits of conservative management, and suggested that suspicious ultrasonographic features are helpful in deciding a course of management (23,24).

The purpose of our prospective study was to assess predictive values of ultrasonography for conservative management in AVMs.

Material and Methods

Our study protocol was approved by the local ethical committee and the respective institutional review board. Written informed consent was obtained from each patient.

Patients

In this prospective observation study between January 2009 and November 2011, 85 consecutive patients referred for our tertiary institution with symptomatic uterine AVMs were included for close follow-up. The diagnosis was identified based on ultrasonography performed at our institution. Ten patients had no relevant obstetric or gynecologic history, and we presumed these patients to represent cases of congenital AVMs. The remaining 75 patients had pregnancy-associated uterine AVMs owing to D&C for incomplete abortions, retained products of conception, or therapeutic abortions. Each of these patients had at least one episode of intermittent vaginal bleeding within 1–12 weeks after the procedure. Twenty-seven patients had a history of at least one D&C or therapeutic abortion before the procedure after which vaginal bleeding had occurred. No patient had a history of hypertension, anticoagulation therapy, or immunosuppressive medication. The patients ranged in age from 18 to 40 years (median, 25 years). Initial ultrasonography and clinical examination were performed between the first and third day after the first episode of vaginal bleeding. In patients with follow-up for conservative management, ultrasonography, and clinical examinations, including the presence and degree of vaginal bleeding, complete blood count, and serum beta human chorionic gonadotropin test (b-hCG), were performed at 2, 4, and 8 weeks. A b-hCG positive result suggested the presence of trophoblastic or molar tissues, and continuous rise of b-hCG helped to distinguish uterine AVM from these pregnancy-related gestational tumors.

Assessment of AVM

The gynecologists were given the tentative diagnosis of uterine AVMs and a description of all ultrasonographic findings except for the results of spectral Doppler analysis. During a prospective follow-up with careful observation for each patient for a minimum period of the day after the initial diagnosis, only the patients who were anemic or hemodynamically unstable from persistent excessive vaginal bleeding were referred for therapeutic management such as embolization of the uterine arteries, or hysterectomy. Embolization was performed with polyvinyl alcohol (PVA) particles (Ivalon, Boston Scientific, Mississauga, Canada) and absorbable gelatin sponges (Gelfoam, Pharmacia-Upjohn, Kalamazoo, MI, USA). The size of the AVM was not measured on angiography, and therefore could not be correlated with the size on ultrasonography. In the ultrasonography findings and the discussions for clinical situations with referring clinicians, the possibility of careful conservative management and spontaneous disappearance of the lesions was highlighted.

Ultrasonography protocol

Endovaginal sonography with grayscale, color, and spectral Doppler imaging was performed using either EV-8CV endovaginal transducer (Sequoia 512, Siemens Healthcare, Erlangen, Germany) or C 10-3Vendovaginal transducer (IU22, Philips Healthcare, Bothell, WA, USA) in all patients. The examinations were performed and interpreted by the same radiologist (KSH) with experience in gynecologic ultrasonography for at least 10 years according to the following protocol. A transducer was used with an image of 140°. The filter was set on 50 Hz, and the Doppler sample volume was 2 mm. Each examination consisted of a standard grayscale ultrasonography of the uterus in both longitudinal and transverse planes. Color ultrasonography was performed to visualize vascularization of the uterus. Color gain adjustment was calibrated on the corresponding normal myometrium such that no signal was visible outside of the uterus. Transducer pressure was minimized in order to avoid compressing the vasculature. Spectral Doppler in both longitudinal and transverse planes was performed to examine vascular pulse curves. Different morphological criteria were assessed in all patients. The uterus was measured in three dimensions, in addition to the double endometrial thickness. The echogenicity of endometrium and overall appearance of myometrium were assessed. Special attention was paid to the maximal diameter of AVM, the presence of retained products of conception, and number of blood vessels with multidirectional flow on color Doppler. In all patients, spectral analysis of blood flow was performed. Pulsatility index (PI), resistance index (RI), peak systolic velocity (PSV, cm/s), and time-averaged maximum velocity (TAMXV, cm/s) were calculated based on the spectral Doppler curve. These values were not included in the ultrasound report but stored on transparent images. If different measurements had been obtained within the same lesion then the set with the highest PSV and matching TAMXV, PI, and RI were used for statistical analysis.

Statistical analysis

The Kolmogorov-Smirnov test was used to determine whether values were normally distributed. The statistical differences in ultrasonography and clinical features between conservative and therapeutic managements were tested by Student’s t-test, the Fisher exact test, and the Mann-Whitney U test. Guidelines by Timmerman et al. (22,23) and hemodynamic status assessed by the clinician are applied to initial selection of management. A mean PSV of 70 cm/s was associated with likelihood of further therapeutic management, whereas conservative management patients had a mean PSV of 52 cm/s (no problems).

A stepwise logistic regression model was used to determine the best predictors of differential diagnosis between conservative and therapeutic managements. With this information, we determined the diagnostic performance of ultrasonography for differentiating conservative and therapeutic managements. All statistical analyses were performed using SAS 9.2 software (SAS Institute, Cary, NC, USA). P values ≤0.05 were considered statistically significant.

Results

Univariate analysis

Forty-five of the 75 patients were included in the conservative management group, and 30 patients in the therapeutic management group. Ultrasonography features were analyzed in these patients. The average period between the D&C or therapeutic abortion and the first ultrasonography examination was 3.8 weeks, with a minimum of 2 days and a maximum of 12 weeks. The different clinical and ultrasonography features observed in patients with conservative or therapeutic management are summarized in Table 1. Patient age and platelet count were not significantly different between the groups. Fifteen (33.3%) of 45 patients in the conservative management group had more than one prior D&C or therapeutic abortion, which was not significantly different to the 12 (40.0%) of 30 patients in therapeutic management. The previous history of D&C or therapeutic abortion did not differ significantly between conservative and therapeutic management. The mean hemoglobin of patients with conservative management was significantly higher (11.2 ± 1.71, mean ± standard deviation) than of those with therapeutic management (9.40 ± 2.2, P <0.05).

Table 1.

Clinical and ultrasonography features by management type.

Variable	Conservative management (n = 45)	Therapeutic management (n = 30)	P value
Age (years)	28.2 ± 12.8	30.2 ± 11.2	0.295
D&C or therapeutic abortion Hx	0.105
First episode	n = 30	n = 18
Over first episode	n = 15	n = 12
Complete blood count
Hemoglobin (g/dl)	11.2 ± 1.7	9.40 ± 2.2	0.015
Platelet (10⁻³/uL)	252.00 ± 13.1	235.00 ± 14.5	0.300
AVM maximal diameter (mm)	2.7 ± 1.2	3.0 ± 1.3	0.209
AVM echogenicity	0.03
Hypoechoic	n = 18	n = 18
Iso to hyperechoic	n = 27	n = 12
Retained product of conception	n = 15	n = 20	0.03
Number of blood vessel	0.03
Single	n = 32	n = 18
Multiple	n = 13	n = 12
Doppler findings
Pulsatility index (PI)	0.43 ± 0.06 (0.27–0.52)	0.53 ± 0.08 (0.43–0.60)	0.02
Resistance index (RI)	0.40 ± 0.09 (0.30–0.55)	0.28 ± 0.05 (0.22–0.42)	0.01
Peak systolic velocity (PSV, cm/s)	35.8 ± 13.2 (18–68)	76.2 ± 14.2 (53–103)	0.005
Time-averaged maximum velocity	34.2 ± 8.8 (19–52)	51.8 ± 11.0 (41–91)	0.02
(TAMXV, cm/s)

Values are mean ± standard deviation, and values in parentheses are ranges rounded to nearest whole number.

The maximal diameter of AVM was not significantly different between the groups. Hypoechoic AVM echogenicity along the myometrium was more common in therapeutic management than in conservative management (40.0% vs. 60.0%, P <0.05), as was the presence of retained products of conception (28.9% vs. 40.0%, P <0.05), and multiple collateral vessels (33.3% vs. 66.6%, P <0.05). At initial ultrasonography, 15 patients had images suggestive of retained products of conception, which were pathologically confirmed to be trophoblastic tissue in four patients and molar tissue in an additional four patients. Also, three patients had undergone more than one D&C owing to the presence of retained chorionic villi. The spectral Doppler identified significant difference between the therapeutic management group and the conservative management group on several parameters. In the therapeutic management group, PSV, PI, and TAMXV were significantly higher, with significantly lower RI (Fig. 1). The results of spectral Doppler were significantly linked to selection of management. Interestingly, PSV and RI were highly specific (90.6% vs. 88.4%) and reasonably accurate (89.6% vs. 83.1%). High PSV and low RI are expected in patients with conservative management, but the number of such patients was small.

Fig. 1.

A 28-year-old woman who underwent therapeutic embolization. (a) Sagittal endovaginal image of the uterus shows intracavitary lesions in anterior myometrium. (b) Color and spectral Doppler image shows multiple tortuous vessels with multidirectional high-velocity, low-resistance flow producing a color mosaic pattern (PSV, 95 cm/s; TAMXV, 85 cm/s; PI, 0.51; RI, 0.25).

Multivariable analysis and diagnostic performance

For each clinical and ultrasonography feature, accuracy in the detection of conservative management were calculated and summarized in Table 2. Stepwise logistic modeling revealed that although no single feature had an accuracy >90%, PSV of the involved AVM was a strong differentiating variable, with an odds ratio (OR) of 28.5 (95% confidence interval [CI]: 5.5–220.7). Hemoglobin and RI were close alternatives, with an OR of 17.8 (95% CI: 5.8–162.1) and 20.7 (95% CI: 6.6 - 185.0), respectively. PI and TAMXV were also useful, with ORs of 10.8 (95% CI: 3.1–60.5) and 12.1 (95% CI: 6.5–78.8). Therefore, spectral Doppler parameters had high diagnostic accuracy for selection of conservative management. Our study might have placed too much weight on one OR of spectral Doppler parameters being higher than another, but the differences were often not significant. For example, compared to both PSV and PI, PSV appears more important, with a higher OR, but the confidence intervals were wide enough that comparisons of ORs were not statistically different.

Table 2.

Diagnostic performance of clinical and ultrasonography features predicting conservative management.

Variable	Sn (%)	Sp (%)	PPV (%)	NPV (%)	Accuracy (%)	Odds ratio	95% CI
Hemoglobin	79.5	85.7	65.6	85.0	84.7	17.8	5.8–162.1
AVM echogenicity	52.9	53.5	58.8	75.3	49.2	1.2	1.0–7.6
Hypoechoic
Iso to hyperechoic	80.8	65.1	75.7	72.8	55.8	2.5	1.4–11.1
Retained product of conception	42.3	78.2	50.0	77.8	45.1	3.1	1.9–18.4
Number of blood vessels
Single	78.5	69.2	73.3	72.1	68.2	3.2	1.5–21.7
Multiple	43.3	52.7	49.5	79.4	38.1	1.1	0.3–12.0
Doppler findings
Pulsatility index (PI)	82.8	80.7	68.9	85.2	78.6	10.8	3.1–60.5
Resistance index (RI)	83.1	88.4	64.3	92.2	83.1	20.7	6.6–185.0
Peak systolic velocity (PSV, cm/s)	85.8	90.6	86.1	85.3	89.6	28.5	5.5–220.7
Time-averaged maximum velocity (TAMXV, cm/s)	86.5	79.2	62.4	82.8	79.3	12.1	6.5–78.8

95% CI, 95% confidence interval; NPV, negative predictive value; PPV, positive predictive value; Sn, sensitivity; Sp, specificity.

Iso-to-hyperechoic AVM echogenicity and the presence of single blood vessel revealed low specificity (65.1%, 69.2%) and relatively high positive predictive values (75.7%, 73.3%), regardless of sensitivity. Conversely, retained product of conception, hypoechoic AVM echogenicity, and the presence of multiple blood vessels revealed relatively high negative predictive values (77.8%, 75.3%, 79.4%).

When diagnostic predictive values were calculated by values of spectral Doppler and hemoglobin, patients were correctly identified as undergoing conservative management in 39 (86.7%) of 45 patients, and as undergoing therapeutic management in 24 (80.0%) of 30 patients. Considering all of the clinical and ultrasonography features, the overall accuracy was 85.3%, with a correct classification in 64 of the 75 patients.

Follow-up ultrasonography and clinical investigation

Of the 45 patients treated with conservative management, four (8%) were hemodynamically stable with nearly normal hemoglobin values, but had residual recurrent vaginal bleeding ranging from 1 week to 4 weeks; embolization was performed in all patients. The remaining 41 patients with conservative management had gradual improvement of vaginal bleeding and spontaneous regression of AVMs at follow-up ultrasonography, by an average of 6 weeks. In patients with conservative management, initial spectral Doppler values recorded were PSV 35.8 ± 13.2 cm/s; TAMXV 34.2 ± 8.8; and, RI 0.40 ± 0.09; PI 0.43 ± 0.06. Spectral Doppler intervals at 8 weeks were PSV 14.3 ± 6.2 cm/s; TAMXV 20.8 ± 5.8; RI 0.67 ± 0.05; and PI 0.30 ± 0.05 (Fig. 2). Therefore, spectral Doppler values gradually returned to normal ranges in patients with conservative management (Fig. 3). These changes are summarized in Table 3.

Fig. 2.

A-32-year-old woman who underwent conservative management. Sagittal endovaginal color (a) and spectral Doppler (b) images show a tangle of multiple vessels with multidirectional flows in anterior myometrium (PSV, 39 cm/s; TAMXV, 29 cm/s; PI, 0.41; RI, 0.33). The follow-up sagittal endovaginal color (c) and spectral Doppler (d) images after 8 weeks show a gradual and spontaneous regression in AVM dimensions (PSV, 30 cm/s; TAMXV, 22 cm/s; PI, 0.35; RI, 0.61).

Fig. 3.

(a, b) Serial changes in mean values of PI, RI, PSV, and TAMXV in patients with conservative management.

Table 3.

Spectral Doppler components for patients in conservative management.

Values	Initial	2-week follow-up	4-week follow-up	8-week follow-up
PSV (cm/s)	35.8 ± 13.2 (18–68)	30.6 ± 9.2 (15–43)	22.8 ± 7.9 (13–37)	14.3 ± 6.2 (11–35)
TAMXV (cm/s)	34.2 ± 8.8 (19–52)	29.6 ± 7.4 (9–42)	25.3 ± 6.2 (8–25)	20.8 ± 5.8 (7–28)
RI	0.40 ± 0.09 (0.30–0.55)	0.53 ± 0.07 (0.43–0.64)	0.62 ± 0.05 (0.42–0.63)	0.67 ± 0.05 (0.45–0.73)
PI	0.43 ± 0.06 (0.27–0.52)	0.38 ± 0.06 (0.32–0.57)	0.35 ± 0.04 (0.30–0.41)	0.30 ± 0.05 (0.28–0.39)

Values are mean ± standard deviation.

Values in parentheses are ranges rounded to nearest whole number.

PI, pulsatility index; PSV, peak systolic velocity; RI, resistance index; TAMXV, time-averaged maximum velocity.

Of the 30 patients with therapeutic management, 25 patients underwent embolization to control bleeding. No major complication relating to the embolization procedure was encountered in any patient. In 23 (92.0%) of 25 patients, embolization was performed between days 0–25 after the initial ultrasonography diagnosis. Five patients with critical hemodynamic instability or prolonged vaginal bleeding after initial embolization were treated with hysterectomy. All 25 patients with therapeutic embolization had a complete recovery and near disappearance of AVM at follow-up ultrasonography at 48 h after the procedure.

Discussion

The guidelines for management of acute vaginal bleeding associated with uterine AVMs are well established. Many patients will remain asymptomatic, suggesting that AVMs do spontaneously regress. Therefore, it is important to establish a management approach for the greater number of uterine AVM patients with milder presentations. In general, the decision to undergo management is based entirely on clinical status of a given patient.

Our study assessed the values of color and Doppler ultrasonography in predicting which patients might require conservative or therapeutic management. Timmerman et al. concluded that conservative management was possible in more than two-thirds of patients with uterine AVMs diagnosed by color and spectral Doppler, and 41 (91.1%) of 45 patients with conservative management achieved symptomatic improvement and spontaneous regression based on a mean PSV of 52 cm/s (23,24). In our study, the mean PSV in patients with conservative management was 35.8 cm/s, in the range of 18–68 cm/s. This mean value is lower than those generally found in the literature. Despite a considerable overlap, high PSV values were associated with a higher probability of further therapeutic management. Therefore, our study suggests that PSV values <76.2 m/s may indicate less dangerous AVMs and that PSV values >35.8 m/s appear safe. However, these findings are derived from this limited study group of patients and should be interpreted with caution.

RI is also important for management selection because patients with lower RI values are more likely receive a therapeutic management. High velocity or low resistance index would be associated with a therapeutic management. The intervals of RI and PI were from 0.22 to 0.42 and from 0.43 to 0.60, respectively, and our study was in agreement with previous reports (22,24). RI revealed relatively high specificity and negative predictive values. PI and TAMXV were significantly higher in patients with therapeutic management than in those with conservative management. However, PI and TAMXV revealed relatively lower specificity and negative predictive values, compared PSV and RI. The clinical presentation and ultrasonographic findings of lesions, turbulent flow, on color flow mapping plus low-resistance, and high-velocity pulsating flow on spectral Doppler analysis were the main features predicting the correct differential diagnosis and choice of management for AVMs.

On the basis of the data presented in our study, we attribute the relatively higher rate to accurate selection of patients for management. We suggest conservative management in patients presenting with Doppler indices in the presumed safe ranges. We prospectively tried to determine critical cut-off values for distinguishing between probable spontaneously regressive AVMs and other AVMs potentially requiring. AVMs have been reported to resolve spontaneously in several cases with increased PSV values (24). Hence, it is possible that using PSV values for management decisions may lead to over intervention.

Our study reveals that the most suspicious clinical and ultrasound features were PSV (OR, 28.5), RI (OR, 20.7), hemoglobin value (OR, 17.8), TAMXV (OR, 12.1), and PI (OR, 10.8). Of these, PSV was more accurate than RI, PI, or TAMXV in spectral Doppler. When hemodynamic stable patients developed high PSV at follow-up ultrasonography, bleeding could continue and gradually increase. At follow-up ultrasonography in the conservative management group, our graphs clearly revealed AVMs from the area of high PSV and low resistance values to the area of a low PSV and high resistance values, along with morphological regression.

We suggest that additional combinations of ultrasonography characteristics require management selection: AVM echogenicity, retained product of conception, and number of blood vessels. Our results included iso-to-hyperchoic AVM (OR, 2.5), absence of retained product of conception (OR, 3.1), and the presence of single blood vessel (OR, 3.2). Each of these ultrasonography features had lower diagnostic accuracy. AVMs associated with gestational trophoblastic disease are found to be iso-to-hyperechoic, and generally disappear after medication (21,24). The radiologist can combine all suspicious characteristics into a global assessment rather than relying on a single feature, making a management decision by ultrasonographic interpretation more accurate.

Our study has some limitations. The most important limitation, perhaps, is the prospective design and selection of patients only by Timmerman’s criteria, which meant that selection bias was inevitable. It is also not clear whether study patients had true AVM, which was not confirmed by angiography. Mean values were mainly used to distinguish between conservative and therapeutic management. Future studies should assess the generalizability of proposed cut-off values. Finally, there was the potential for inter-observer variation in specifying ultrasonography features.

In conclusion, for pregnancy-related AVMs diagnosed by color and spectral Doppler, initial conservative management may be appropriate for patients with suitable criteria, and therapeutic management should be recommended if there is no symptomatic improvement and spontaneous regression at follow-up exams.

Footnotes

Funding

This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.

References

O’Brien

Neyastani

Buckley

. Uterine arteriovenous malformations from diagnosis to treatment. J Ultrasound Med 2006; 25: 1387–1392.

Kido

Togashi

Koyama

. Retained products of conception masquerading as acquired arteriovenous malformation. J Comput Assist Tomogr 2003; 27: 88–92.

Halperin

Schneider

Maymon

. Arteriovenous malformation after uterine curettage: a report of 3 cases. J Reprod Med 2007; 52: 445–449.

Grivell

Reid

Mellor

. Uterine arteriovenous malformations: a review of the current literature. Obstet Gynecol Surv 2005; 60: 761–767.

Cura

Martinez

Cura

. Arteriovenous malformations of the uterus. Acta Radiol 2009; 50: 823–829.

Kelly

Belli

Campbell

. Arteriovenous malformations of the uterus associated with secondary postpartum hemorrhage. Ultrasound Obstet Gynecol 2003; 21: 602–605.

Jain

Fogata

. Retained products of conception mimicking a large endometrial AVM: complete resolution following spontaneous abortion. J Clin Ultrasound 2007; 35: 42–47.

Nasu

Yamaguchi

Yoshimatsu

. Pregnancy complicated by asymptomatic uterine arteriovenous malformation: a case report. J Reprod Med 2007; 52: 335–337.

Valsky

Hamani

Verstandig

. The use of 3D rendering, VCI-C, 3D power Doppler and B-flow in the evaluation of interstitial pregnancy with arteriovenous malformation treated by selective uterine artery embolization. Ultrasound Obstet Gynecol 2007; 29: 352–355.

10.

Polat

Suma

Kantarcy

. Color doppler US in the evaluation of uterine vascular abnormalities. Radiographics 2002; 22: 47–53.

11.

Brown

. Pelvic ultrasound in the postabortion and postpartum patient. Ultrasound Q 2005; 21: 27–37.

12.

Sugiyama

Honda

Kataoka

. Diagnosis of uterine arteriovenous malformation by color pulsed Doppler ultrasonography. Ultrasound Obstet Gynecol 1996; 8: 359–360.

13.

Jain

Jeffrey

Jr Sommer

. Gynecologic vascular abnormalities: diagnosis with Doppler US. Radiology 1991; 178: 549–551.

14.

Abu Musa

Hata

. Pelvic arteriovenous malformation diagnosed by color flow Doppler imaging. Am J Roentgenol 1989; 152: 1311–1312.

15.

Ghai

Rajan

Asch

. Efficacy of embolizations in traumatic uterine vascular malformations. J Vasc Interv Radiol 2003; 14: 1401–1408.

16.

Delotte

Chevallier

Benoit

. Pregnancy after embolization therapy for uterine arteriovenous malformation. Fertil Steril 2006; 85: 228–228.

17.

Yokomine

Yoshinaga

Baba

. Successful management of uterine arteriovenous malformation by ligation of feeding artery after unsuccessful uterine artery embolization. J Obstet Gynaecol Res 2009; 35: 183–188.

18.

Lim

Agarwal

Seckl

. Embolization of bleeding residual vascular malformations in patients with treated gestational trophoblastic tumours. Radiology 2002; 222: 640–644.

19.

Degani

Leibovitz

Shapiro

. Expectant management of pregnancy-related high-velocity uterine arteriovenous shunt diagnosed after abortion. Int J Gynaecol Obstet 2009; 106: 46–49.

20.

Forssman

Lundberg

Scherstén

. Conservative treatment of uterine arteriovenous fistula. Acta Obstet Gynecol Scand 1982; 61: 85–87.

21.

Timmerman

Wauters

Van Calenbergh

. Color Doppler imaging is a valuable tool for the diagnosis and management of uterine vascular malformations. Ultrasound Obstet Gynecol 2003; 21: 570–577.

22.

Timmerman

Van der Bosch

Peeraer

. Vascular maformations in the uterus; ultrasonographic diagnosis and conservative management. Eur J Obstet Gynecol Reprod Biol 2000; 92: 171–178.

23.

Van den Bosch

Van Schoubroeck

. Color Doppler and gray-scale ultrasound evaluation of the postpartum uterus. Ultrasound Obstet Gynecol 2002; 20: 586–591.

24.

Van Schoubroeck

Van den Bosch

Scharpe

. Prospective evaluation of the blood flow in the myometrium and in the uterine arteries after term pregnancy. Ultrasound Obstet Gynecol 2004; 23: 378–381.