Outline
Adnexal Causes of Pelvic Pain, 884
Adnexal Torsion, 884
Ruptured or Hemorrhagic Ovarian Cyst, 887
Pelvic Inflammatory Disease, 891
Endometriosis, 892
Peritoneal Inclusion Cyst, 897
Uterine Causes of Pelvic Pain, 899
Leiomyomas, 899
Malignant Uterine Tumors, 903
Adenomyosis, 904
Malpositioned Intrauterine Device, 906
Vascular Causes of Pelvic Pain, 906
Nongynecologic Causes of Pelvic Pain, 907
Acute Appendicitis, 907
Acute Diverticulitis, 908
Other Inflammatory Conditions of the Bowel and Mesentery, 910
Obstructive Uropathy, 912
Conclusions, 912
Summary of Key Points
- •
Pelvic pain in the reproductive age patient often presents a diagnostic challenge as there are many causes, including both gynecologic and nongynecologic disorders with overlap in clinical symptoms.
- •
Pregnancy status must always be determined to exclude a pregnancy-related complication, particularly ectopic pregnancy.
- •
Pelvic sonography is the well-established first-line imaging method of choice and may require both a transabdominal and transvaginal approach to increase diagnostic accuracy.
- •
Computed tomography (CT) and magnetic resonance imaging (MRI) may be of added value in evaluation of pelvic pain when ultrasound findings are not definitive.
- •
Adnexal torsion cannot be excluded in the presence of preserved ovarian blood flow, and gray-scale features along with the clinical picture are more reliable in making this diagnosis.
- •
Hemorrhagic ovarian cysts demonstrate a wide spectrum of sonographic appearances that may overlap with those of other complex cystic masses, and if greater than 5 cm these cysts may require follow-up imaging to confirm resolution.
- •
Endometriosis may demonstrate variable sonographic findings and cause both chronic and acute pelvic pain if there are associated complications.
- •
Most leiomyomas are benign; however, atypical imaging features should prompt further evaluation to exclude the possibility of a malignant uterine neoplasm.
- •
Focal adenomyosis may mimic myomas sonographically, and MRI can be helpful for more definitive characterization.
- •
Many nongynecologic causes of pelvic pain, including gastrointestinal and genitourinary disorders, may be identified when ultrasound examination is performed for suspected pelvic disease, and recognition of these disorders may enable prompt diagnosis and eliminate the need for further imaging.
Pelvic ultrasound imaging is the well-established first-line imaging method of choice for evaluation of pelvic pain in the reproductive age patient when a gynecologic or obstetric disorder is suspected. Ultrasound imaging is widely available; allows for high-resolution, dynamic real-time evaluation of the uterus, adnexa, and adjacent structures; is relatively low cost; and, importantly, involves no exposure to ionizing radiation. A transvaginal approach should be utilized whenever possible because it yields improved visualization of pelvic anatomy. This approach should be supplemented with transabdominal imaging when uterine or adnexal structures are beyond the field of view of the transvaginal probe or if an abnormality outside the pelvis is suspected. Occasionally, a transvaginal examination may not be tolerated and only a transabdominal scan can be performed. Duplex and color or power Doppler sonography are essential adjuncts to gray-scale imaging. Although CT may be more useful for evaluation of gastrointestinal or genitourinary disease, because of considerable overlap in clinical symptoms and laboratory findings, these disorders may not initially be suspected, and pelvic sonography should be the first imaging study performed. However, many of these nongynecologic disorders can also be recognized by sonography, and if a diagnosis is established sonographically, further imaging may not be necessary. In the pregnant patient, ionizing radiation should be avoided when possible, and sonography plays an even greater role.
When a patient complains of pelvic pain, it is critical to establish her pregnancy status as this information will help to narrow the differential diagnosis and direct an appropriate imaging workup. In the pregnant patient with pelvic pain, ectopic pregnancy is often the leading concern, and this important topic is covered in Chapter 33 . The primary focus of this chapter is the ultrasound evaluation of pelvic pain in the nonpregnant patient, although the impact of pregnancy on some disorders that cause pelvic pain is discussed when appropriate. Both gynecologic and nongynecologic causes of pelvic pain are reviewed. Disorders that may cause acute or chronic pelvic pain (or both), and conditions that can be medically managed as well as those that require urgent surgical intervention are included. The added value of other imaging modalities in the evaluation of pelvic pain, including CT and MRI, is also addressed.
Adnexal Causes of Pelvic Pain
Adnexal Torsion
Adnexal torsion occurs when there is partial or complete rotation of the adnexal structures around their vascular pedicle with associated obstruction of venous outflow and arterial inflow. The true incidence of this disorder is unknown as the diagnosis is made definitively only during surgery. It is estimated that up to 3% of female patients with acute pelvic pain who come to the emergency department have adnexal torsion. Adnexal torsion may involve either the ovary or the fallopian tube or both when they twist around the infundibulopelvic ligament and tubo-ovarian ligament. Concomitant ovarian and tubal torsion has been shown to occur in up to 67% of cases. Twisting of the vascular pedicle initially leads to compromise of lymphatic and venous outflow, resulting in diffuse ovarian edema and enlargement. Arterial inflow may initially be sustained because the arteries have thicker, muscular walls and are less collapsible. If left untreated, arterial thrombosis, ischemia, and ultimately infarction and necrosis of the ovary occur. Early diagnosis and surgery are essential to salvage the viable ovary. Predisposing factors for adnexal torsion include an ipsilateral adnexal mass (particularly if larger than 5 cm), pregnancy, ovulation induction, polycystic ovary syndrome, prior pelvic surgery (including tubal ligation), and hypermobility of the adnexal structures (more common in children and adolescents).
Adnexal torsion may occur at any age, although it occurs predominantly in reproductive age women, most likely because of increased incidence of both physiologic and pathologic ovarian masses. However, up to 24% of cases occur in postmenopausal women. There is a reported higher frequency on the right, perhaps because the sigmoid colon is a relatively fixed structure occupying the left pelvis, which helps to prevent torsion of the left ovary and adnexa. An adnexal mass is reported to be present in 22% to 73% of cases. Large simple cysts and cystic neoplasms such as benign cystic teratomas, hemorrhagic cysts, and cystadenomas are the most commonly associated masses. Large ovaries in the setting of ovarian hyperstimulation are another predisposing factor. The absence of associated ovarian abnormality is more common in premenarchal girls, with 46% of cases of torsion involving normal-appearing ovaries. Potential causes in these patients include increased mobility of the fallopian tubes and mesosalpinx, elongated pelvic ligaments, fallopian tube spasm, strenuous exercise, or abrupt changes in intra-abdominal pressure. Approximately 10% to 25% of cases of adnexal torsion occur during pregnancy, with an incidence of 1 : 1000 pregnancies, most commonly in the first trimester. Torsion is less likely to occur in the setting of pelvic inflammatory disease (PID), endometriosis, or malignant neoplasms because in these conditions associated adhesions serve to fix the pelvic structures in place.
The clinical presentation of adnexal torsion is variable and may overlap with other causes of acute abdominal or pelvic pain, such as a ruptured ovarian cyst or appendicitis, leading to a diagnostic challenge. Classically, patients present with sudden acute onset of intense lower abdominal and pelvic pain, with a palpable painful adnexal mass and peritoneal signs. Additional symptoms may include nausea, vomiting, flank pain, and fever with a slight leukocytosis. A pregnancy test should be performed to exclude the possibility of a pregnancy-related complication such as ectopic pregnancy. The pain may be constant or intermittent, due to intermittent torsion and detorsion of the adnexa, and the intensity of pain experienced may vary with the degree of torsion or traction on the twisted pedicle. When symptoms are intermittent, the diagnosis is even more challenging. A high degree of clinical suspicion is necessary to avoid treatment delay and irreversible damage to the adnexa. Pain that lasts for more than 10 hours is associated with increased likelihood of adnexal necrosis at surgery.
If a patient presents with pelvic pain, and adnexal torsion is suspected, prompt emergent pelvic sonography, including gray-scale, color, and spectral Doppler, is the imaging study of choice. However, as with the clinical diagnosis, the sonographic diagnosis may be challenging because the findings vary with degree and duration of torsion, involvement of the fallopian tube, and the presence of an associated mass. The most consistently reported gray-scale ultrasound finding is the presence of a unilaterally enlarged ovary, usually larger than 4 cm, with or without an associated mass ( Figs. 29-1 and 29-2 ). Houry and Abbott found that in a series of 87 cases of torsion, the mean ovarian size was 9.5 cm, with 89% measuring greater than 5 cm. In a study by Chiou and colleagues, an underlying mass was found in 65% of 34 cases of adnexal torsion and a large ovary without a mass in 32% of cases. An additional ultrasound finding is the presence of multiple peripherally located small follicles in the enlarged hypoechoic ovary, the “string of pearls” sign (see Fig. 29-1 ), which is thought to occur as a result of peripheral displacement of the ovarian follicles secondary to stromal edema and venous congestion, although the reported incidence of this finding varies. In more longstanding cases, the ovarian stroma may become more heterogeneous in echotexture because of foci of hemorrhage or necrosis. Comparison with the asymptomatic ovary is often helpful in recognizing these findings. An abnormal position of the ovary, which may be located in the midline either superior to the uterus or inferiorly in the pouch of Douglas, or displacement to the contralateral side of the pelvis is an important observation (see Figs. 29-1 and 29-2 ). The twisted and congested vascular pedicle may appear as an ill-defined adnexal mass adjacent to the torsed ovary. In cross section, it may demonstrate a target appearance with alternating hyperechoic and hypoechoic bands, the so-called “whirlpool” sign (see further discussion later). The amount of associated free fluid around the ovary or in the cul-de-sac is variable, but free fluid is observed in up to 87% of cases.
A more recently described gray-scale sign of ovarian torsion is the follicular ring sign . A follicular ring is defined as a perifollicular hyperechoic rim, 1 to 2 mm in thickness, seen surrounding the small (3 to 7 mm) peripheral antral follicles of the torsed ovary ( Fig. 29-3 ). In a series of 15 patients with surgically proven torsion, Sibal observed the follicular ring sign in 12, the whirlpool sign in 7, and absent ovarian blood flow in 7. The follicular ring sign was the only sonographic finding in 4 patients. On microscopic examination, the follicular rings were found to be areas of hemorrhage and edema surrounding the antrum of small follicles. The authors caution that this sign may not be seen if the ovary is necrotic or if a large cyst or mass compresses the follicles.
In addition to gray-scale findings, color and spectral Doppler evaluation is an important component of the pelvic ultrasound examination in the assessment of adnexal torsion. As with gray-scale imaging, Doppler findings vary with the degree of torsion, time course, and vascular compromise. Absence of detectable blood flow in the affected ovary allows for a confident diagnosis, with a positive predictive value (PPV) of 94% in one series. However, this is a late finding and suggests that there has already been infarction and that the ovary is no longer viable ( Fig 29-4 ). It may be difficult to detect flow if the ovary is mostly replaced by a large mass, and in this setting an attempt should be made to identify and search for flow in the rim of ovarian parenchyma around the mass. Many reports in the literature have shown that the detection of flow in an ovary on color and spectral Doppler sonography cannot exclude the presence of torsion. The dual arterial supply to the ovary may allow for early preservation of arterial flow, with initial loss of venous flow, although if the torsion is early, intermittent, or only partial, both arterial and venous flow may be preserved (see Figs 29-1 and 29-2 ). In a study by Mashiach and associates 13% of women with laparoscopically proven ovarian torsion had normal ovarian blood flow on Doppler. In a study by Chiou and colleagues arterial and venous flow was considered normal in 19% of cases of proven torsion. Bar-On and coworkers reported abnormal ovarian flow on Doppler sonography in only 43.8% of ovarian torsion cases, with a specificity of 91.7%. Shadinger and associates reported preserved arterial flow in 54% and preserved venous flow in 33% of cases in a series of 39 patients with pathologically proven ovarian torsion. Other reports have also demonstrated that Doppler findings can be normal in 45% to 61% of cases of ovarian torsion. However, analysis of spectral Doppler waveforms may increase sensitivity for the diagnosis of torsion. An arterial waveform with reversal of diastolic flow, a high resistance pattern, may suggest the diagnosis. Recently, an abnormal discontinuous pattern of venous flow has been suggested as a clue to the presence of ovarian torsion when arterial and venous flow is preserved, although this result has not been reproduced in larger studies. In general, the gray-scale findings as well as the clinical picture are thought to be more reliable than Doppler in the diagnosis of adnexal torsion.
A twisted vascular pedicle or “whirlpool sign” is an additional gray-scale and color Doppler finding useful for the diagnosis of adnexal torsion (see Figs. 29-1 and 29-2 ). The twisted vascular pedicle is the rotation site of the pedicle on itself and appears as a round or beaked mass with concentric alternating hyperechoic and hypoechoic circles or rings on gray-scale sonography. The rings are composed of the components of the pedicle, including the broad ligament, fallopian tube, and branches of the ovarian artery and vein. At color Doppler, the twisted vessels in the pedicle appear to be coiled or swirling, as a whirlpool. This finding is best detected when the probe is moved back and forth in a transverse plane along the central axis of the twisted pedicle, which may be located either medial or lateral to the ovary. In a study by Lee and colleagues this finding was observed in 88% of cases of torsion. This finding may be a more direct sign of adnexal torsion and may be particularly useful in equivocal cases because it represents the actual site of torsion rather than the secondary effects on the ovary. In a study by Valsky and coworkers there was an increase in true positive cases from 55% to 90% when this sign was observed. The presence of flow in the whirlpool is reportedly a useful predictor of ovarian viability.
Although sonography is the primary modality of choice for evaluation of clinically suspected ovarian torsion, CT may be the first study performed in the emergency room setting if ovarian torsion is not initially considered. CT findings have been well described and parallel those observed sonographically, including an enlarged ovary with or without an associated mass, ovarian stromal edema with peripheral follicles, inflammatory stranding in the periovarian fat, a twisted vascular pedicle, a thickened fallopian tube, pelvic free fluid, midline position of the ovary, and deviation of the uterus to the side of the twist ( Fig. 29-5 ). Lack of enhancement or hematoma may suggest hemorrhagic infarction. In the author’s experience, the twisted pedicle may be better appreciated in a nonaxial imaging plane, and multiplanar reformatted images should be reviewed ( Fig. 29-6 ). Recently, in 2014, it was suggested that CT may be more accurate than ultrasound imaging in detection of ovarian torsion. In one series, CT was performed prior to pelvic ultrasound examination in 70% of cases. In this study, ultrasound imaging was 80% sensitive and 85% to 95% specific for ovarian torsion, whereas CT had a sensitivity of 90% to 100% and specificity of 85% to 90%; however, this was a retrospective study and the results have not been duplicated in larger prospective series. Hence, pelvic sonography remains the first-line imaging test of choice and, importantly, does not expose the patient to radiation or intravenous iodinated contrast agents.
MRI features of ovarian torsion have also been well described and parallel those seen by ultrasound imaging ( Fig. 29-7 ). MRI may serve as an adjunct to sonography in equivocal cases, enabling better visualization of the twisted pedicle and ovarian edema. MRI is particularly helpful in evaluating the patient with hyperstimulated ovaries in whom the diagnosis of torsion can be particularly challenging because of underlying ovarian enlargement. However, MRI may not be readily available in the emergency setting, and inability to obtain an MRI scan should not result in treatment delay.
Isolated tubal torsion is very rare and usually associated with tubal disease. Clinical symptoms are similar to those for ovarian torsion, although imaging features may vary. The most common sonographic appearance is a tubular, dilated fluid-filled structure, often with pointed or “beaked” ends where the twist occurs, in an unusual position in the pelvis and without demonstrable flow in the wall, in association with a normal-appearing ovary ( Fig. 29-8 ). The whirlpool sign may also help with this diagnosis.
Ruptured or Hemorrhagic Ovarian Cyst
Ovarian cysts are common in women of reproductive age, and most are physiologic (functional) cysts. Ovarian cyst rupture and hemorrhage are common physiologic events during the ovarian cycle, involving the follicle or corpus luteum. Although follicular rupture occurs unnoticed in most women, the term mittelschmerz (middle pain) is used to describe pain resulting from the release of fluid into the peritoneum with rupture of the normal follicle during ovulation. Hemorrhage into the corpus luteum is common and is due to the increased vascularity of this structure. As the maturing graafian follicle enlarges and the surrounding stromal cells undergo luteinization, the luteinized theca cells become more vascular. During ovulation at midcycle when the follicle ruptures and expels the oocyte, the corpus luteum is formed and the granulosa layer becomes vascularized. The vessels within the wall rupture easily, giving rise to a hemorrhagic cyst. Usually, symptoms are mild and self-limited. However, patients may present to the emergency department with more significant pain if there has been a greater degree of hemorrhage into the cyst or into the peritoneal cavity. Pain is usually sudden in onset, localized to one side of the pelvis, and gradually improves. Peritoneal signs are variable in intensity depending upon the amount of fluid and hemorrhage causing peritoneal irritation. There may be considerable overlap in clinical presentation with other gynecologic and nongynecologic disorders including ovarian torsion, ectopic pregnancy, and acute appendicitis. In one recent series of reproductive age women who presented with right lower quadrant pain and clinical suspicion for appendicitis, 12.8% had gynecologic causes, including 7.2% with cyst rupture, 4.2% with ruptured hemorrhagic corpus luteum, and 1.4% with adnexal torsion. Most of the time, the process is self-limiting, although severe life-threatening hemorrhage may occasionally occur and require urgent surgical intervention. Women with clotting disorders are at a higher risk for severe hemorrhage.
If ovarian cyst rupture or hemorrhage is suspected, sonography is the study of choice. The cyst may not be visualized if it has ruptured and completely decompressed; however, there will generally be free pelvic fluid. Hence, ovarian cyst rupture is often a diagnosis of exclusion if no other cause for the patient’s pain is identified. Although variable in appearance, the ruptured or leaking corpus luteum classically appears on ultrasound imaging as a small cystic lesion with a crenulated homogeneously hypoechoic wall and low-level internal echoes during the luteal phase of the menstrual cycle. Size is variable, but is typically less than 3 cm. In most cases, there is a peripheral rim of increased vascularity, the “ring of fire” appearance ( Fig. 29-9 ). The sonographic appearance of a hemorrhagic cyst is also quite variable, depending upon when in the evolution of the hemorrhage the patient undergoes imaging. There is a continuum, from acute hemorrhage to clot formation followed by clot retraction. Because of the wide spectrum of appearance, hemorrhagic cysts may mimic other processes and have been referred to as the “great imitator.” However, the characteristic sonographic features have been well described ( Fig. 29-10 ). Size may vary considerably from 2.5 to 10 cm. Acutely, hemorrhage into a cyst has a sonographic pattern of diffuse, homogeneous low-level echoes. A fluid-fluid level may be observed with the dependent echogenic component corresponding to blood. Over time, as red blood cells lyse and fibrin strands form, a reticular pattern of internal echoes is observed (also described as fishnet, cobweb, spider web, or lace-like) ( Fig. 29-10B ). Fibrin strands produce a network of fine linear echoes and are distinct from true septations in that they are usually innumerable, thinner, and discontinuous, with an irregular fine branching pattern; lack vascularity; and do not extend from wall to wall as a true septation would. An acute clot filling the cyst may appear as dense internal echoes that simulate a solid mass, although lack of blood flow on Doppler imaging and demonstration of increased posterior through transmission are helpful features in suggesting the diagnosis. As the clot retracts and lyses, it may appear on gray-scale sonography as an echogenic mural component within a cystic lesion and thus may simulate a solid neoplastic mural nodule. However, the retracting clot will often demonstrate concave or straight margins with acute angles, is often triangular, and may have a moth-eaten echotexture ( Fig. 29-10C ). Lack of Doppler flow is an important feature suggesting mural thrombus as opposed to a tumor nodule, although low-levels of flow in solid tissue may not be detectable by Doppler sonography. Hence, absence of Doppler-detected blood flow is not an entirely reliable feature for a clot in a hemorrhagic cyst. The diagnostic performance of sonography in discriminating hemorrhagic ovarian cysts from other adnexal lesions, including malignancies, has been investigated in a study by Patel and associates, who reported that the reticular or fishnet pattern had a sensitivity of 90%, specificity of 98%, and positive likelihood ratio (LR) of 40 for a hemorrhagic cyst; a retracting clot (solid echoes with a concave margin) was found to have a higher LR (>67) and specificity (100%) but lower sensitivity of 30%, although the combination of fibrin strands, no septations, and smooth wall was found to have an LR of 200, with sensitivity of 90% and specificity of 100%. Approximately 90% of hemorrhagic ovarian cysts will exhibit these features.
For some hemorrhagic ovarian cysts, there is overlap in imaging features with other complex cystic masses, including endometriomas, tubo-ovarian abscesses (TOAs), and cystic ovarian neoplasms, and definitive diagnosis may not be possible at the time of the initial sonogram. The clinical setting is often helpful in focusing the differential diagnosis. However, if a hemorrhagic cyst is of primary consideration based on imaging features, short-interval follow-up sonography to confirm resolution is recommended because most hemorrhagic cysts typically resolve within 8 weeks. This approach will prevent unnecessary surgical intervention. However, for a classic-appearing hemorrhagic cyst, defined as demonstrating either the reticular fibrin strand or retractile clot patterns, the Society of Radiologists in Ultrasound Consensus Conference Statement for management of asymptomatic ovarian cysts offers the following guidelines: In women of reproductive age, hemorrhagic cysts 3 cm or less need not be described in the sonogram report, and no follow-up is necessary. Hemorrhagic cysts between 3 and 5 cm should be described in the report, but require no follow-up. Hemorrhagic cysts more than 5 cm should undergo short-interval follow-up sonography (6-12 weeks) with imaging optimally performed on days 3 to 10 of the menstrual cycle. In early postmenopause, all suspected hemorrhagic cysts should be described in the report, and short-interval follow-up sonography recommended in 6 to 12 weeks. Because women in late menopause should never develop hemorrhagic ovarian cysts, suspicion for neoplasm is heightened in this setting and surgical evaluation or additional imaging evaluation should be considered.
If a hemorrhagic ovarian cyst ruptures, the cyst may decompress and might not be identified as a discrete finding at sonography. Color Doppler sonography may identify the residual wall of the cyst ( Fig. 29-11 ). A sentinel clot is sometimes identified in the adnexa as a complex mass of mixed echogenicity without blood flow. A variable amount of hemoperitoneum will be observed as free fluid containing low-level echoes, primarily within the pelvis. However, hemoperitoneum may also be noted in the upper abdomen if there has been a large amount of bleeding. Thus, the Morison pouch and the left upper quadrant should be surveyed to assess the amount of intraperitoneal blood. A ruptured ectopic pregnancy could present with similar sonographic and clinical findings; therefore, a human chorionic gonadotropin (hCG) level should always be measured to exclude this possibility. Infected fluid could also mimic the sonographic appearance of hemoperitoneum; therefore, clinical correlation and laboratory findings are important.
CT may occasionally be requested if a ruptured hemorrhagic ovarian cyst is not initially suspected when a patient presents with diffuse abdominal pain. CT findings include a high attenuation adnexal mass compatible with clot and high attenuation peritoneal fluid compatible with hemoperitoneum ( Fig. 29-12 ). A rim-enhancing ovarian cyst may be identified, and even if the cyst has ruptured, a remnant of the cyst wall may remain visible. If there is active bleeding at the time of the scan, extravasation of contrast material with pooling in the pelvis will be observed.
Other cysts may also rupture, including cystic neoplasms, most commonly mature cystic teratomas. Rupture of ovarian teratomas has been reported to occur in 1% to 4% of cases. Leakage of the liquefied sebaceous contents into the peritoneal cavity results in chemical peritonitis. Chronic leakage from the cyst may result in chronic granulomatous peritonitis. If an adnexal mass is identified that exhibits classic sonographic features of a mature cystic teratoma and the patient has pain, either torsion or rupture of the mass should be considered. A large volume of intraperitoneal fluid, distorted or flattened shape of the mass, and discontinuity of the cyst wall suggest rupture. CT may be helpful for further evaluation, as it may show scattered fat droplets outside the mass, peritoneal fluid, and findings of peritonitis ( Fig. 29-13 ). The peritoneal inflammatory changes may mimic peritoneal carcinomatosis or tuberculous peritonitis, possible diagnostic pitfalls.
Pelvic Inflammatory Disease
PID refers to a spectrum of disease that occurs when microorganisms ascend from the lower genital tract to infect the uterus, fallopian tubes, and ovaries and is a common cause of emergency room visits and hospitalizations for acute gynecologic disorders. The continuum of infection begins with cervicitis and progresses to endometritis, salpingitis, pyosalpinx, tubo-ovarian complex (TOC), and, ultimately, tubo-ovarian abscess (TOA). One third to one half of cases are due to Chlamydia trachomatis or Neisseria gonorrhoeae. However, PID is most commonly a polymicrobial infection, and a substantial proportion of cases are nongonococcal and nonchlamydial in origin, involving vaginal flora, anaerobic gram-negative rods, and Mycoplasma bacteria. The adnexa may also become secondarily infected from other inflammatory processes, usually gastrointestinal in origin, including appendicitis and diverticulitis. The clinical diagnosis may be challenging, as symptoms and signs vary widely and overlap with other processes including endometriosis, appendicitis, and ectopic pregnancy. Pelvic pain is the most common presenting symptom, although it may be absent or mild in some patients. On physical examination, cervical motion tenderness as well as uterine and adnexal tenderness are classic findings. There may be associated mucopurulent vaginal discharge, white blood cells on saline microscopy of vaginal secretions, elevated erythrocyte sedimentation rate, and C-reactive protein, leukocytosis, and fever. A delay in treatment may result in significant reproductive and gynecologic morbidity such as infertility, increased risk of ectopic pregnancy, chronic pelvic pain, and recurrent infection. Most patients can be effectively treated as outpatients with broad-spectrum antimicrobial therapy. However, hospitalization may be required for severe cases (including TOAs) or if the diagnosis is in doubt.
Imaging evaluation of patients with suspected PID may be necessary if the symptoms are nonspecific and diagnosis is uncertain, if the patient is not responding to treatment, or if complications such as abscess formation are suspected. If a TOA has formed, imaging may aid in determining the most appropriate method of treatment, with possible options including percutaneous drainage and surgery.
Pelvic ultrasound examination is the first-line imaging method of choice for the evaluation of PID. Ultrasound examination findings vary depending upon the stage of infection, and if PID is early or mild, findings may be nonspecific. Patients with endometritis may not exhibit any sonographic imaging findings. A fluid-filled endometrial cavity is suggestive in the clinical setting of fever, vaginal discharge, and uterine tenderness on physical examination ( Fig. 29-14 ). However, intrauterine fluid is a nonspecific finding. Gas in the endometrial cavity will appear on ultrasound imaging as foci of increased echogenicity with associated posterior acoustic shadowing. Although the presence of gas increases concern for possible infection, gas in the endometrial cavity may also be observed postpartum or post instrumentation. Free fluid in the cul-de-sac is also a nonspecific finding, although echoes within the fluid raise suspicion for infection (or hemorrhage). Enlargement and hyperemia of the ovaries with multiple small cysts indicate oophoritis.
Findings involving the fallopian tubes are more specific for PID. Although isolated salpingitis may not always be recognizable by sonography, infected fallopian tubes may demonstrate wall thickening and hyperemia. Obstruction at the fimbrial end of the tube due to inflammation will result in a pus-filled dilated tube or pyosalpinx, which appears as a tubular structure distended with echogenic intraluminal fluid and debris, sometimes with layering echoes indicating a fluid-debris level, and thick hypervascular walls ( Fig. 29-15A ). Thickened and inflamed endosalpingeal folds may suggest small mural nodules, which give the appearance of a “cogwheel” when imaged in cross section. This finding helps differentiate a pyosalpinx from an inflamed appendix, which is blind-ending and does not have endosalpingeal folds ( Fig. 29-15B ). If untreated, PID progresses to form a TOC as the inflammation spreads to the adjacent ovary, which becomes enlarged with indistinct contours and increased vascularity, although the ovary remains identifiable as a discrete structure. The inflamed fallopian tube is often adjacent or adherent to the ovary. The final stage is TOA, in which the normal ovary is no longer delineated and an inflammatory mass replaces the ovary and fallopian tube. This appears as a complex multiloculated cystic adnexal mass with internal echoes, layering debris, and thick septations ( Fig. 29-16 ). The mass will be markedly vascular on color Doppler interrogation. The presence of gas in a TOA is very rare. It is estimated that TOAs complicate approximately 10% to 15% of cases of PID. TOA rupture may result in septic shock.
PID most often involves both adnexa, but may occasionally be unilateral. If the process is right-sided, a perforated appendicitis with abscess formation could have a similar appearance, and if left-sided, perforated diverticulitis may give a similar picture. Assessment of patient risk factors for PID is very important in this setting. CT may be required to exclude appendicitis or diverticulitis as the cause of the abscess ( Fig. 29-17 ). CT also may be helpful in detecting complications such as extension of inflammation out of the pelvis and involvement of adjacent structures including the bowel or ureter. A blood-filled fallopian tube (hematosalpinx) in the setting of endometriosis may mimic pyosalpinx and is another diagnostic pitfall. A complex or infected endometrioma may mimic a TOA (see further discussion later). MRI can serve as a useful complementary imaging technique because signal intensity characteristics compatible with chronic blood products will be observed in endometriosis but not in PID. One series evaluating patients with both ultrasound imaging and MRI reported the sensitivity and specificity of MRI for laparoscopically proven PID to be 95% and 89% and of ultrasound imaging to be 81% and 78%, respectively.
Chronic PID may result in obstruction of the ampullary segment of the fallopian tube, causing hydrosalpinx with fluid-filled dilatation of the fallopian tube. Although PID is the most common cause of hydrosalpinx, other causes include tubal ligation, hysterectomy without salpingo-oophorectomy, endometriosis, prior surgery, and malignancy. Hydrosalpinx may be a cause of pelvic pain and infertility. At sonography, a dilated fallopian tube in the absence of infection appears as a C-, U-, or S-shaped anechoic, avascular tubular structure, generally with a thin wall less than 5 mm thick ( Fig. 29-18 ). Small 2 to 3 mm nodules representing the remnants of the endosalpingeal folds may appear as “beads on a string.” In one study, hydrosalpinx was diagnosed with the highest likelihood when a tubular mass with small round mural projections or a waist sign was visualized. The incomplete septation sign, due to the distended tube folding on itself, may also be observed but is a less reliable finding. If these features are not observed, hydrosalpinx may be difficult to distinguish from a cystic ovarian neoplasm. Three-dimensional (3D) sonography may increase specificity. MRI may be helpful by demonstrating the tubular nature of the mass and intraluminal fluid content. PID may also result in formation of a peritoneal inclusion cyst (discussed later).
Endometriosis
Endometriosis is defined as the presence of endometrial glands and stroma in ectopic locations outside the uterus. Neoangiogenesis and capillary recruitment are associated with endometriotic lesions at laparoscopy, and there is an associated inflammatory response ultimately progressing to fibrosis. This disorder is associated with a wide variety of symptoms including dysmenorrhea, dyspareunia, chronic pelvic pain, and dysfunctional uterine bleeding. Although pain is usually chronic, complications of endometriosis may result in a more acute presentation. Infertility is an important consequence of endometriosis, due to associated anatomic distortion of the pelvic structures and obstruction of the fallopian tubes. Endometriosis affects up to 10% of women of reproductive age, although in women with pelvic pain, infertility, or both, the prevalence is estimated to be as high as 35% to 50%. Because endometriosis is hormonally responsive, symptoms are often cyclic in nature, and there is repetitive hemorrhage into these lesions. The pathogenesis of endometriosis is complex and remains the subject of debate. Proposed theories include origin of implants from the uterine endometrium, possibly from lymphatic or hematogenous dissemination of endometrial cells or retrograde menstruation. Other theories propose that implants arise from tissues outside the uterus, for example, coelomic metaplasia with transformation of peritoneal tissue to ectopic endometrial tissue or embryonic müllerian rests that develop into endometriotic lesions under the influence of estrogen. A more recent theory suggests that extrauterine stem/progenitor cells originating from bone marrow may differentiate into endometriotic tissue, and this topic is currently an active area of investigation.
The most common sites of endometriotic implantation include the surface of the ovary, uterine suspensory ligaments, uterus or fallopian tube, and the peritoneal surfaces of the pouch of Douglas. Less common sites of implantation include vagina, bladder, cervix, intestine, cesarean delivery scars, abdominal scars, or the inguinal ligament. Deep pelvic endometriosis is defined as invasive tissue that infiltrates structures at a depth of more than 5 mm from the peritoneal surface and is associated with fibrosis and muscular hyperplasia. This preferentially affects the dependent portions of the posterior peritoneal spaces, most commonly the uterosacral ligaments, torus uterinum, rectovaginal pouch, rectum, and rectovaginal septum, with anterior involvement less common. Current treatment options for endometriosis include both medical (primarily hormonal) and surgical management. Diagnostic laparoscopy remains the reference standard for diagnosis and staging of endometriosis, although the role of imaging in endometriosis has continued to evolve. The ovaries are the most common sites of endometriosis and are frequently involved with multiple and bilateral lesions. The classic sonographic appearance of an endometrioma, often referred to as a chocolate cyst because of the presence of thick, dark, degenerated blood products from repetitive cyclic episodes of bleeding, is that of a homogeneous, hypoechoic lesion with low to medium level echoes and no internal vascularity ( Fig. 29-19 ). This has often been referred to as a “ground glass” appearance and is highly predictive of an endometrioma, with one retrospective series reporting this appearance in 95% of endometriomas. However, there is a spectrum of sonographic appearances, and in another series, 13 of 87 surgically proven endometriomas were felt to be atypical. In an investigation by Patel and coworkers echogenic mural foci, thought to represent cholesterol deposits, were observed in 36% of endometriomas compared with 6% of nonendometriomas, and septations were found in 45% of endometriomas. In that study, if an adnexal mass had diffuse low-level echoes, as well as hyperechoic mural foci and multilocularity, as well as no evidence of other neoplastic features, it was 32 times more likely to be an endometrioma than any other adnexal mass. Occasionally an endometrioma can mimic a simple ovarian cyst and appear completely anechoic. Other findings include fluid-fluid level, thickened wall, and mural or central calcifications. Avascular mural nodules representing an adherent or retracting blood clot can occasionally be observed, but a clot is less commonly observed in endometriomas than in hemorrhagic ovarian cysts. Rarely, flow within nodular areas has been described, possibly due to endometrial stromal tissue. However, tumor nodules due to malignant degeneration (usually into clear cell or endometroid carcinoma) of an endometrioma may mimic this appearance and demonstrate Doppler-detected blood flow. Chronic endometriomas may mimic solid masses due to the presence of older blood products and fibrosis. However, typically, there is only perilesional blood flow with no internal vascularity. Doppler waveform analysis is not helpful in diagnosis as low resistance waveforms resembling malignancy can be observed in the wall of an endometrioma.
Ultrasound examination findings of endometriomas may overlap considerably with other adnexal masses including hemorrhagic cysts, TOAs, dermoids, and cystic ovarian neoplasms ( Fig. 29-20 ). In a meta-analysis by Moore and colleagues, the sensitivity of transvaginal sonography for the diagnosis of an endometrioma ranged from 64% to 89% with specificity of 89% to 100%. The most common misdiagnoses were hemorrhagic cysts and dermoids. If a hemorrhagic cyst is a diagnostic consideration, short-interval follow-up ultrasound examination is useful because a hemorrhagic cyst should resolve whereas an endometrioma will persist. If ultrasound findings are not definitive and if ovarian neoplasm is of concern, MRI is the next recommended step. As a result of the repeated cycles of bleeding, endometriomas will contain old blood products with high iron and protein concentration creating characteristic findings at MRI with increased signal intensity on T1-weighted images and decreased signal on T2-weighted images, sometimes referred to as shading . Mural nodules representing an adherent clot will not enhance on postcontrast subtraction images. An enhancing nodule following the administration of gadolinium is indicative of either malignant degeneration or endometrial stromal tissue. Fat-sensitive images help to identify macroscopic fat in a dermoid (Fig. 29-21 ). MRI has a greater than 90% sensitivity and specificity for endometriomas.
Endometrial implants involve the fallopian tubes in approximately 6% of women with endometriosis, and adhesions involve the tubes in 26%. Endometriotic implants are most common on the serosal surface, with transmural and mucosal involvement occurring less frequently. Hematosalpinx may be an isolated finding in patients with endometriosis. At sonography, low-level echoes, corresponding to blood products, may be identified in the dilated fallopian tube ( Fig. 29-22 ). This appearance may simulate that of pyosalpinx, and clinical findings are important to differentiate between these entities. MRI may also be helpful by confirming the presence of blood products in the tube.
Endometriotic implants may occur in the anterior abdominal wall, which is the most frequent site of extrapelvic disease. Endometriosis in the abdominal wall is usually found next to a surgical scar or a needle or laparoscopic trocar tract. One of the most common associations is with prior cesarean delivery, reported in 0.03% to 1% of women. In a series of 455 cases of abdominal wall endometriosis, 57% were associated with cesarean delivery scar, and 11% were associated with hysterectomy. Spontaneously occurring implants may be seen adjacent to the umbilicus. Among women with scar endometriosis, only 14% to 26% have concomitant pelvic endometriosis. Implants may be confined to the superficial subcutaneous soft tissues but may also involve the rectus muscle. The most widely held theory of pathogenesis states that these implants arise from transportation and direct implantation of endometrial cells during procedures that open the uterus. Women may present with a tender abdominal wall mass adjacent to a scar or with an asymptomatic palpable lump. Pain and swelling may be cyclic, correlating with menstruation. Imaging findings vary depending upon phase of the menstrual cycle, chronicity, number of stromal and glandular elements, and degree of associated bleeding and inflammation. Sonography is usually the first-line imaging study performed, using a high-resolution linear transducer, which will identify a solid, heterogeneous hypoechoic mass with scattered internal echoes, although the appearance is variable ( Fig. 29-23 ). Areas of cystic change may be observed and the margins may be spiculated with infiltration of adjacent soft tissues. Most implants will demonstrate some vascularity at Doppler evaluation. Differential diagnostic considerations include abdominal wall masses such as desmoid tumor, metastasis, lymphoma, melanoma, hematoma, suture granuloma, or incisional hernia. Location of the mass next to a scar and history of cyclic pain related to menses suggest the diagnosis. MRI may also be a useful adjunct because it may identify the typical imaging characteristics of hemorrhage as well as avid gadolinium enhancement. The extent of disease may also be better depicted on MRI. Fine-needle aspiration/biopsy with cytologic analysis may help confirm a definitive diagnosis. Wide surgical excision with clear margins to prevent local recurrence is the treatment of choice.
In the past, ultrasound examination was considered most useful for characterization of an adnexal mass as an endometrioma, whereas extraovarian disease and deep pelvic endometriosis was better evaluated with MRI, offering important mapping of extent of disease prior to surgical intervention. However, more recently, an increasing role for sonography in identification of nonovarian disease has been advocated. Gentle probing with the transvaginal ultrasound transducer may identify extraovarian disease, particularly in the deep pelvis. Recent studies have shown that sonography may be useful for detecting implants in the bladder, rectovaginal septum, rectum, and sigmoid colon. Bazot and associates evaluated a group of women with deep endometriosis and found that sonography had a sensitivity and specificity for detection of disease of 78.5% and 95.2%, respectively, with the highest sensitivity for detecting intestinal and bladder disease. Endometriosis of the uterine serosa, uterosacral ligaments, and retrocervical and retrovaginal spaces may be diagnosed as hypoechoic solid lesions with indistinct margins due to smooth muscle proliferation and prominent fibrotic component ( Fig. 29-24 ). Occasionally, these lesions may also contain cystic spaces. Pelvic adhesions can be evaluated by moving the transducer back and forth to assess whether the uterus, adnexa, and bowel slide freely. Bowel implants may be more conspicuous after bowel preparation and appear as hypoechoic nodules with variable extension into the bowel wall. A recent meta-analysis showed that transvaginal sonography with or without bowel preparation is an accurate test for noninvasive presurgical detection of deep infiltrating endometriosis of the rectosigmoid with pooled estimates of sensitivity and specificity of 91% and 98%, respectively, and positive and negative predictive value (NPV) of 98% and 95%. Another study reported that transvaginal sonography yields few false positive results but that negative findings are less reliable and that the accuracy of sonography is affected by the location and number of endometriotic lesions.
Although endometriosis most often causes chronic pelvic symptoms, complications may lead to acute pelvic pain. Occasionally an endometrioma may rupture and have a clinical presentation similar to a ruptured hemorrhagic ovarian cyst. On sonography, complex free fluid compatible with hemoperitoneum will be noted in addition to a hemorrhagic-appearing adnexal mass ( Fig. 29-25 ). This occurs more frequently during pregnancy as a result of rapid growth from hormonal stimulation and may become a surgical emergency if there is extensive bleeding. A known history of endometriosis is helpful in establishing the diagnosis. Endometriosis may also be associated with massive hemorrhagic ascites, usually in the setting of pelvic adhesions and ovarian endometriomas. Exudative ascites may result from peritoneal inflammation due to hemoperitoneum following rupture of an endometrioma. An endometrioma may also become superinfected, clinically mimicking PID. This most commonly occurs after surgical drainage or aspiration. Direct spread from adjacent inflammation or hematogenous spread in a bacteremic patient may also cause superinfection. The sonographic appearance of an infected endometrioma may be identical to an uninfected endometrioma or may appear more complex ( Fig. 29-26 ). Clinical signs of infection are thus crucial in establishing the diagnosis.
During pregnancy, decidualization of an endometrioma may occur as a result of hypertrophy of the ectopic stromal cells, primarily from the effects of progesterone. A decidualized endometrioma increases in size and becomes more complex in appearance, with solid mural nodules or papillary excrescences that may be vascular on Doppler imaging. These changes may mimic a malignant ovarian neoplasm. Prior documentation of an endometrioma can be helpful in suggesting this diagnosis, allowing for sonographic surveillance rather than surgical intervention. MRI may also be useful by demonstrating that the mural nodules have signal intensity and texture similar to the decidualized uterine endometrium.
Malignant transformation is a well-described but rare complication of endometriomas, with an estimated incidence of approximately 1%. Women with a history of endometriosis are 4.2 times more likely to develop ovarian cancer than other women, the most common histologic findings being clear cell carcinoma and endometrioid carcinoma arising from glandular elements. Less commonly, endometrial stromal sarcoma can arise from stromal elements. A specific genetic mutation ( ARID1A ) has been implicated in this process. Because of this malignant potential, the Society of Radiologists in Ultrasound Consensus Conference Statement on management of asymptomatic ovarian cysts advised that adnexal masses with classic features of endometriomas be followed at least yearly with serial ultrasound examinations to search for worrisome features, including development of solid vascular components, usually mural based, and rapid interval growth ( Fig. 29-27 ). If such findings are observed, surgical exploration should be recommended. MRI may be helpful to confirm the diagnosis of malignant transformation. The frequency of follow-up will vary depending on the patient’s age and symptoms. With advancing age, patients may be followed more closely owing to increased incidence of malignant transformation in older women.
