US findings of adnexal torsion in children and adolescents: size really does matter
- First Online:
- Cite this article as:
- Linam, L.E., Darolia, R., Naffaa, L.N. et al. Pediatr Radiol (2007) 37: 1013. doi:10.1007/s00247-007-0599-6
Adnexal torsion is rare, and symptoms are nonspecific. Clinicians often rely on US examinations to evaluate girls with abdominal or pelvic pain.
To determine which sonographic findings can predict adnexal torsion by comparing pediatric and adolescent patients with surgically confirmed torsion (cases) to those without torsion (controls).
Materials and methods
Cases and controls were identified retrospectively by searching 7 years of medical record and radiology databases. An adnexal ratio was calculated as the volume of the affected adnexa divided by the volume of the unaffected adnexa.
We identified 61 menarchal subjects: 33 cases and 28 controls. Adnexal volume was larger in cases than in controls (185 vs. 37.8 ml, P < 0.001). A volume of >75 ml was more common in cases than in controls (64 vs. 15%, P < 0.001). No cases had an adnexal volume of <20 ml (P < 0.001). The adnexal ratio was larger in cases than in controls (16.1 vs. 6.7, P < 0.001). An adnexal ratio of >15 was seen in 40% of cases and in no controls (P = 0.08). Doppler US results were not predictive of torsion.
An adnexal volume of <20 ml is strong evidence against adnexal torsion in menarchal females. In this age group, the diagnosis of torsion cannot be established by US examination alone.
KeywordsTorsion Ovary Adnexa Doppler Adolescent
Adnexal torsion is a surgical emergency with potential life-long consequences for fertility. For these reasons, prompt diagnosis and intervention is important. However, the usual presenting symptoms such as lower abdominal pain and vomiting are nonspecific. Because these symptoms are common and nonspecific, and because of the difficulty in performing pelvic physical examination in children and young adolescents, clinicians rely extensively on imaging for the evaluation of the adolescent with abdominal/pelvic pain . Surgeons sometimes rely on this imaging to decide whether to operate for suspected torsion.
On US, absent Doppler arterial flow and ovarian enlargement are reported to be the hallmarks of torsion in women, but these findings are not always present [2, 3]. Some authors have reported that Doppler flow is 100% sensitive and 98% specific for ovarian torsion . Other predictive findings include visualization of a twisted pedicle on transvaginal US examination [5, 6], but transvaginal scans are rarely performed in children and adolescents. CT scans are often performed on these girls to evaluate the appendix, but CT findings that are suggestive of adnexal torsion have not been clearly characterized or defined.
Prior studies have been limited to adults or case series that lack control groups [3, 7, 8, 9, 10, 11]. Gray-scale and Doppler US findings of ovarian torsion in children have been reported [3, 7, 10, 11], but to our knowledge no large studies have been performed to determine the sensitivity and specificity of Doppler flow, absolute volume, or the ratio of the volume of torsed to nontorsed adnexa to predict torsion. The findings of torsion in children and adolescents have recently been described ; however, the study was not a case-control study, nor did the authors divide the patients by menarchal status. Normal ovarian volume increases with age and menarchal status, so it is important to separate those patients who have begun menses from those who have not in order to make a valid comparison of ovarian size. The authors also introduced the ovarian volume ratio as a predictor of the presence of an underlying mass in a torsed ovary ; to our knowledge this ratio has not been used to predict torsion.
The purpose of our study was to determine which sonographic findings are reliable predictors of adnexal torsion in a child and adolescents with pelvic pain. By comparing the US findings in patients with surgically confirmed torsion (cases) to those without torsion (controls) in our pediatric and adolescent population, we evaluated whether adnexal size or adnexal ratio predicts torsion, and whether Doppler flow is sufficiently sensitive and specific for torsion in this age group.
Materials and methods
A retrospective review was performed of all cases of suspected and confirmed adnexal torsion at a 400-bed children’s hospital between September 1998 and July 2005. The hospital’s Institutional Review Board granted approval for the study. Patients were identified using two search methods: the medical record database was searched for female patients with the ICD9 code for ovarian torsion, and the radiology database was searched for female patients with pelvic US examinations performed for the indication of “pain” and whose diagnosis included suspected or probable ovarian torsion. The medical record of each identified patient was reviewed to ascertain age, presenting symptoms, US findings, preoperative diagnosis, final diagnosis, and surgical management.
US examinations performed at the primary institution were performed on Acuson Sequoia machines (Mountain View, Calif.); however, some girls in our study had US examinations at outside hospitals. US images of these patients were reviewed by two authors (L.L. and S.O.). Transverse and longitudinal images were obtained of the uterus and ovaries using a transabdominal approach with a full urinary bladder as an acoustic window. The bladder was distended by direct infusion of sterile saline via a bladder catheter or by oral or intravenous hydration. Both arterial and venous color and pulsed Doppler flow were recorded, as well as the dimensions and volumes of the uterus, each ovary, and any adnexal mass. Volumes were calculated on the US machines for the majority of the cases. Because this was a retrospective study, volumes in some cases were not calculated prospectively; in these cases, the volume was calculated by the authors using the formula for an ellipse (length × width × height × 0.523) . We defined the adnexal ratio as the volume of the affected (painful) ovary/adnexa divided by the volume of the unaffected (nonpainful) ovary/adnexa in the same patient.
Cases were defined as those girls with surgically confirmed torsion. Controls were those with surgically excluded torsion. For analysis, the girls were stratified by menarchal status (premenarchal or menarchal). Premenarchal status was determined from documented menstrual history or prepubertal configuration of the uterus and age <13 years.
Data were entered into a spreadsheet and analyzed using a statistical program (Stata version 8.0, Statacorp, College Station, Tx.). We examined arterial and venous Doppler flow using several approaches such as present vs. absent, and as a three-level variable (normal, decreased or peripheral only, absent), and found the best predictor was when decreased/absent flow was compared to normal flow. Adnexal volume and ratios were compared as continuous variables, then as dichotomous variables using various cut points to assess the cut point with the best predictive value for torsion. Continuous variables were examined using paired t-tests with unequal variances, and chi-squared and Fisher’s exact tests were used to assess statistical significance of dichotomous variables.
The medical record and radiology database search identified 107 girls with suspected adnexal torsion. Two neonates with torsion diagnosed antenatally were excluded. Of the remaining patients, 46 had surgically documented torsion (cases) of whom 13 (28%) were premenarchal and 33 (72%) were menarchal. Of the 59 possible controls, all of whom were menarchal, 30 did not have surgery, and one did not have a US examination. The remaining 28 patients with surgically excluded torsion were used as controls.
All 13 premenarchal girls in our series had surgically documented adnexal torsion. Six of these had torsion of the ovary alone; seven had torsion of both fallopian tube and ovary. Because we lacked a control group for this subset of patients, we present only descriptive data for these subjects. Subjects ranged in age from 6 to 14 years (mean 9.8 years). Three patients had CT scans only, and one went to the operating room without any imaging. Nine patients had US examinations, and both ovaries were visualized in seven of these nine. The volume of the affected ovary ranged from 2.4 ml to 513 ml (mean 80.1 ml, median 12.5 ml). The unaffected ovary volume ranged from 1.2 ml to 16.5 ml (mean 5.8 ml, median 4.9 ml). The mean adnexal ratio (affected/unaffected) for the seven patients in whom both ovaries were visualized was 11.2 (range 0.6–31), and two of the seven (28%) had an adnexal ratio greater than 15. Of the nine patients with US and Doppler US results, arterial flow was absent in three, decreased in three, and normal in three. Venous flow was absent in three, decreased in two, and normal in four.
US volumes in menarchal subjects, by torsion status
All subjects (n = 54)
Torsion cases (n = 28)
No-torsion controls (n = 26)
Adnexal volume (ml)
Adnexal ratio (affected/unaffected)
Volume, n (%)
Adnexal ratio >15, n (%)
Doppler US findings in menarchal patients with and without torsion
All subjects (n = 43)
Torsion cases (n = 24)
No-torsion controls (n = 19)
Arterial Doppler results, n (%)
Venous Doppler results, n (%)
Decreased or absent flow
Test performance of significant predictors of adnexal torsion, expressed as percentages
Volume >20 ml
Volume >75 ml
Ovarian ratio >15
Among menarchal subjects, pathologic findings did not differ between cases and controls. The most common pathologic findings were an ovarian or paratubal cyst (34%). No pathology was reported in 40% of patients; presumably these ovaries were normal. Twelve subjects had a paratubal cyst and one had a hydrosalpinx. There were no malignant tumors causing torsion, and only two patients (6%) had torsion of a dermoid cyst.
Of the 33 menarchal patients who had documented torsion at surgery, 14 had torsion of the ovary alone, 8 had torsion of the ovary and Fallopian tube, 8 had torsion of the Fallopian tube alone, and 3 had a torsed paratubal cyst. Of the 11 without ovarian involvement, a separate normal ovary was identified sonographically in 3, but for the rest the adnexal structures could not be differentiated by US. Therefore, visualization of a normal ovary did not exclude torsion of other adnexal structures.
In girls with lower abdominal pain, it is often difficult to distinguish ovarian/adnexal causes of pain from other causes such as appendicitis on the basis of clinical signs and symptoms alone. Because of the difficulty in obtaining a history from children and the challenges of performing pelvic examinations in children and adolescents, imaging plays a major role in the diagnosis of torsion. Typically, transabdominal sonography or CT will be the chosen imaging modalities. US examination might be preferable, as it allows a detailed evaluation of the pelvic organs and can distinguish adnexal causes of pain from other causes such as appendicitis without using ionizing radiation .
Our data show that in young girls with pelvic/abdominal pain, US evaluation should include measurements of the volumes of bilateral ovaries or adnexa. The total adnexal volume and calculated adnexal ratio are the two most useful radiographic signs for the evaluation of adnexal torsion. The best negative predictor in menarchal girls was adnexal volume <20 ml, with a NPV of 100%. Parameters for ovarian enlargement were previously defined as 34 times larger than the average normal prepubertal ovarium and 15 times larger than the average adult ovarium . A recent study by Servaes et al.  evaluating girls with torsion found that an adnexal volume ratio of >20 was associated with a 70% likelihood of a mass within a torsed ovary. To our knowledge, however, the adnexal volume ratio has not previously been used for the prediction of torsion. We found that the adnexal ratio was higher in those with torsion compared to controls; however, even those without torsion had a marked degree of ovarian asymmetry as evidenced by a mean adnexal ratio of 6.7. Although others have suggested that a twofold difference in ovarian size suggests torsion, we found that an adnexal ratio >15 occurred in 40% of cases and in no controls.
In this retrospective case-control study, we demonstrated that Doppler findings have a low sensitivity and specificity. The best predictor in our study was decreased or absent venous Doppler flow, but this did not reach statistical significance and had limited PPV and NPV. Our results are in contrast to those of Ben-Ami et al. , who found that in adults, absent venous Doppler flow by transvaginal sonography was 100% sensitive and 98% specific, with a 94% PPV. However, the prevalence of torsion in that series was 23%. Cil et al.  recently suggested that color Doppler US is the most useful tool in the diagnosis of torsion. Several case reports and small case series without controls [2, 3, 7, 8, 10] have suggested that blood flow can be present in torsed ovaries but that absence of flow indicates torsion. In our data, one-third of normal controls had absent Doppler flow; thus, in the pediatric age group absent Doppler flow is a nonspecific finding.
Our results might have differed from those of prior studies for several reasons. Many previous studies only addressed the appearance of ovarian torsion and did not include torsion of other adnexal structures such as the Fallopian tube or a paraovarian cyst [3, 4, 7, 8, 10]. We included all cases of adnexal torsion rather than just ovarian torsion, because surgical management is indicated for these conditions as well. However, isolated torsion of the Fallopian tube or of paratubal cysts could not be distinguished sonographically from ovarian torsion in our population; blood flow to these areas differs from the blood flow to the ovaries, and thus the sonographic findings might be expected to differ as well. Studies comparing the findings of transvaginal US to those of transabdominal US for pelvic inflammatory disease [15, 16] and pelvic masses  have demonstrated that transvaginal US better delineates anatomy and provides additional diagnostic information in up to 71% of patients . The transabdominal US approach used in the menarchal age group undoubtedly limits the reliability of Doppler flow findings. In addition, in some patients with torsion the blood flow is preserved, resulting in healthy-appearing tissue upon surgical evaluation, while in other patients flow is compromised, resulting in necrosis and tissue damage. Therefore, it is not surprising that Doppler findings are insufficient to diagnose adnexal torsion.
Servaes et al.  recently reviewed the sonographic findings of ovarian torsion in their pediatric population at Boston Children’s Hospital and found that the most common US finding in ovarian torsion is an enlarged heterogeneous ovary. However, there are many differences between their study and ours. First, their study examined the US findings in all pediatric and adolescent girls ranging in age from 0–21, regardless of menarchal status. We excluded neonatal cases because symptoms and management differ in these infants. We divided our patients by menarchal status because hormonal influences affect the size and function of the ovary. Additionally, we found that adnexal volumes ranged from 2.4 ml to 513 ml in the premenarchal patients, suggesting that size is not a good predictor of torsion in this group. Their study evaluated arterial and venous Doppler US together; we evaluated these separately and found that venous Doppler US might be more useful in the evaluation of torsion. Their study introduced the adnexal volume ratio as a predictor of underlying mass in torsed ovaries, while we used adnexal volume ratio to predict the likelihood of torsion. The greatest difference between the studies is the study design: we evaluated patients with surgically confirmed nontorsion causes of pelvic pain as a control group to determine what findings were different in patients with torsion from other patients with pain, while there was no control group in the study of Servaes et al.. This allows us to refute the oft-quoted statistic that absent Doppler flow is predictive of torsion.
Unless a normal ovary is seen separately from a pelvic mass, it can be difficult to distinguish adnexal structures sonographically. In 12 cases (26%) in our study, the Fallopian tube or a paraovarian cyst was the torsed structure, yet this was described as an adnexal mass following US examination and was sometimes described as ovary. Therefore, we chose to include all structures of the adnexa in our study and calculated “adnexal” volume and ratio.
In our study, all of the premenarchal patients evaluated for pelvic pain had adnexal torsion. Because there were no controls (that is, no patients were taken to surgery in whom adnexal torsion was not found), we were unable to determine sonographic predictors of torsion. In the premenarchal group, adnexal volume ranged from normal size (2.4 ml) to very large (513 ml). The premenarchal patients were equally distributed among normal, decreased, and absent Doppler flow.
Lee et al.  and Vijayaraghavan  have suggested that US visualization of a twisted vascular pedicle is a specific indicator of torsion when using a transvaginal approach in adults but not when using a transabdominal approach. We were unable to see the twisted vascular pedicle in any of our cases. In the pediatric and adolescent population, many of the patients are virginal, which is a relative contraindication to transvaginal scanning. Therefore, the twisted vascular pedicle is not a useful sign of torsion in this population due to the limitation of the transabdominal approach.
The limitations of our study are its small sample size and retrospective design. Because ovarian torsion is an uncommon condition, the patient population is small; however, this is one of the largest series reported and one of the few to include a control group. In addition, because of the retrospective design, our US data are incomplete. Several of the subjects had unilateral or no Doppler US evaluation or were missing a calculated ovarian/adnexal volume of the contralateral adnexa.
The intention of this study was to determine an imaging predictor of torsion in pediatric and adolescent girls with abdominal/pelvic pain. Although decreased or absent venous flow showed a trend toward predicting torsion, normal Doppler evaluation does not exclude torsion. An adnexal volume >75 ml should also increase suspicion for adnexal torsion but is a nonspecific finding with a high false-positive rate, as other causes of ovarian enlargement, such as bleeding into a corpus luteum cyst, can cause pain without torsion. Massive adnexal asymmetry (ratio >15) is insensitive but very specific for ovarian torsion. An ovarian volume <20 ml (normal size in the menarchal patient) is one of the few findings that reliably excluded adnexal torsion in our series. Prospective studies should be performed to further define the roles of clinical evaluation, laboratory findings and radiologic signs in the assessment of adnexal torsion in young girls with abdominal and pelvic pain. Our findings indicate that in this age group US examination might be helpful, but the diagnosis of torsion cannot be established by US alone, and the decision to operate cannot be based exclusively on sonographic findings.