Ovarian absence: a systematic literature review and case series report

A systematic search of the literature was conducted by a medical librarian in Cochrane Library, ClinicalTrials.gov, Google Scholar, Ovid Embase, Ovid MEDLINE, PubMed, Scopus, and Web of Science Core Collection databases to identify relevant articles published from the inception of each database to April 2022. The final searches were performed in all the databases on April 6, 2022. The search was peer-reviewed by a second medical librarian using PRESS (Peer Review of Electronic Search Strategies) [22]. Databases were searched using a combination of controlled vocabulary and free text terms for ovarian anomalies. The search was not limited by publication type or year. Details of the full search strategies are listed in Supplementary Table 2. CitationChaser was used to search the reference lists of included studies to find additional relevant studies not retrieved by the database search [23].

Case studies or series were included if the study reported ovarian absence with surgical confirmation. Studies were limited to cases of ovarian absence described in phenotypical females that had 46 XX karyotype or, when karyotype was not stated, absence of syndromic features suggestive of a genetic condition. Studies describing cases with Differences of Sexual Development (DSD) or those not available in English language were excluded.

Four independent reviewers (HAC, PV, MTG, AAG) assessed citations returned by the literature search in pairs, selected relevant abstracts for full text review, and identified full text articles eligible for inclusion in the final review. The lead investigator (AVM) resolved all identified conflicts at each stage of the review process and reviewed the final list of full-text articles to ensure each was relevant to this study’s objectives.

Three independent reviewers (HAC, PV, AAG) extracted variables from full texts. These included age of patients, presentation symptom(s), surgical procedure, OA laterality, genetic studies, imaging studies, anatomy of the remaining reproductive tract, renal anatomy, and study design. Data items are listed in Table 1.

Manuscripts were assessed for quality of evidence based on the Oxford Centre for Evidence-Based Medicine (OCEBM): Levels of Evidence. This scale grades manuscripts from 1 (highest) to 5 (lowest) according to the level of clinical evidence that can be derived from their study design. Randomized control trials are designated 1, cases-series are 4, while case reports and expert opinions are graded 5. Case reports were subsequently assessed using the JBI Critical Appraisal Checklist for Case Reports [24].

Statistical analysis was performed using SAS software (v9.4, SAS Institute. Cary, NC, USA) and Microsoft Excel (Microsoft Office, v16.16.27). We performed quantitative, qualitative, and formal narrative syntheses of data extracted from the case reports and literature reviews included in our systematic review of OA. We performed univariate analysis on the categorical variables and reported their frequencies and relative percentages. We conducted chi-square tests to evaluate the relationships between several categorical variables. During the data analysis process, we used the fixed-effect assumption. Publications included in this systematic review were primarily case reports and case series, and there was significant variability in reporting of case details.

A 15-year-old postmenarchal girl presented to Yale New Haven Children’s Hospital for a diagnostic laparoscopy due to chronic abdominal and pelvic pain. Menarche occurred at age 12. She reported regular menses with normal flow and acyclic pelvic and flank pain. She denied being sexually active. Past medical and surgical history were significant for ADHD, dyslexia, nephrolithiasis, constipation, tonsillectomy and adenoidectomy. There was no documented family history of endometriosis, but there was a suspicion for endometriosis necessitating oophorectomy in the maternal grandmother. Physical exam was notable for tenderness with deep palpation in the left upper and lower abdominal quadrants. She had age-appropriate secondary sexual characteristics with Tanner stage 5 breasts and Tanner stage 4 pubic hair. The external inspection of the genitalia revealed normal labia, clitoris, prepuce, external urethra meatus, and annular hymen. Renal ultrasound demonstrated left-sided nonobstructing kidney stones. The transabdominal pelvic ultrasound was unremarkable. Three prior ultrasounds reported normal-appearing ovaries bilaterally. The patient was started on continuous combined ethinyl estradiol-progesterone oral formulation to suppress menses due to suspicion of endometriosis. After four months, she continued to experience pelvic pain and irregular vaginal bleeding. Following a shared decision-making process, the patient and guardian elected to proceed with diagnostic laparoscopy. Surgery revealed a normal appearing uterus with bilateral round ligaments and a normal appearing right fallopian tube leading to an enlarged right ovary. Lesions suspicious for endometriosis were observed and biopsied in the pouch of Douglas, on the sigmoid colon, and the right uterosacral ligament. The left ovary, distal fallopian tube, and utero-ovarian ligament were absent. The left cornua of the uterus had a 3 cm tubular structure which was presumed to be the left fallopian tube remnant. (Fig. 1). The left ureter was clearly observed passing over the external iliac vessels, but no other adnexal structures, including the infundibulopelvic (IP) ligament, were observed. Multiple peritoneal biopsies revealed endometriosis and stage 3 was assigned per American Society for Reproductive Medicine classification [25]. A second gynecology attending physician was consulted intraoperatively and concurred with the incidental finding of the absent left ovary. The patient had an uncomplicated postoperative course. Endometriosis was managed by a 30 mcg-1.5 mg ethinyl estradiol-norethindrone pill in continuous regimen, resulting in amenorrhea and the resolution of both pelvic and flank pain. Postoperative MRI of the abdomen and pelvis did not identify ortho/heterotopic left ovary.

The second patient is a 30-year-old G3P1011 (obstetric history significant for one term delivery of a living child and one ectopic pregnancy) woman with a history of asthma and genital chlamydia infection. Her surgical history was significant for a supraumbilical laparotomy in infancy to retrieve a severed umbilical catheter remnant. She had no other abdominal surgery. She had a history of a right tubal ectopic pregnancy which was successfully treated with methotrexate. The patient presented to the hospital with severe abdominal pain and a positive home pregnancy test. On abdominal exam she had mild suprapubic tenderness with rebound. Her pelvic exam revealed normal appearing labia, clitoris, and external urethral meatus. Speculum exam demonstrated multiparous cervix, without discharge or blood. Bimanual examination revealed a small, anteverted uterus with no masses appreciated in the adnexa bilaterally, and mild tenderness in the posterior fornix. The serum beta-HCG level was 5,519 mIU/mL. The patient had two pelvic ultrasounds. The first revealed normal bilateral ovaries with normal blood flow and a complex cystic right adnexal structure possibly indicative of an ectopic pregnancy. The second ultrasound suggested torsion of the left fallopian tube, a cystic structure with a possible yolk sac in the right adnexa, and no distinct right ovary. Neither ultrasound indicated an intrauterine pregnancy. The patient elected to have diagnostic laparoscopy due to a suspected ectopic pregnancy and adnexal torsion. Operative findings were significant for the minimal adhesive disease and a right tubal ectopic pregnancy. The left ovary and adnexa appeared normal as did the bilateral round ligaments. An ectopic pregnancy was found extruding from the remnant of the right fallopian tube that had dived into the posterior leaf of the broad ligament. Notably, the right ovary, distal 2/3 of the right fallopian tube and the right IP ligament were missing. To verify the absence of the right adnexal structures, dissection of the right pelvic side wall was performed. This revealed normal appearing ureter and iliac vessels, but the right ovary and distal 2/3 of the fallopian tube were not visualized. Of note, a second gynecology attending physician was consulted intraoperatively to confirm the absence of the right ovary. The pathology of the extruded gestational tissue revealed decidualized stromal tissue with implantation site trophoblasts, consistent with an ectopic pregnancy. Postoperative recovery was uneventful.

Database searches resulted in 22,813 citations (Fig. 2). After removing duplicates, 12,120 citations underwent title and abstract screening. Of these, 513 citations met the criteria for full text review. Subsequently, 79 studies met the inclusion criteria for the study. An additional 10 studies were found through reference chasing and searching Google for a total of 89 included manuscripts (Supplementary Table 3). We excluded 434 studies as those presented data on ectopic ovaries, ineligible patient populations, non-English languages, had insufficient information to assess ovarian abnormality, were conference abstracts, were duplicates, had no original data, or presented duplicate study data (Supplementary Table 4).

The analysis comprised 89 studies with 113 cases of bilateral and unilateral ovarian absence, all of which were from English-language articles published from 1909 to 2022. All articles were case reports or case series, some of which were accompanied by literature reviews.

10 studies were OCEBM Level 4 (11%) and 79 studies were Level 5 (89%). Using the JBI Critical Appraisal Tool for Case Reports, the overall quality of 89 included records was generally good, with > 90% records scoring “yes” in 6 or more domains. The demographics domain was poorly reported across 59 papers, in most cases this was due to the lack of reporting on race and ethnicity. The detailed quality assessment for each study can be found in Supplementary Table 5 and Supplementary Fig. 1.

The average age of OA diagnosis was 22.47 years (Table 1). A torsion or vascular etiology was the suspected etiology of OA in 58 cases (52%), while an embryologic etiology was suspected in 24 cases (21%). Etiology was undetermined in 31 cases (27%). Ultrasound was the most common imaging technique performed (55%), and laparoscopy was the most common surgical modality (58%). Of the cases that included pre-operative ovarian imaging results, 16/57 (28%) did not mention ovarian abnormalities. Abdominal/pelvic pain (30%) and infertility/subfertility (19%) were the most common presenting symptoms, followed by evaluation of a mass (16%) and amenorrhea (12%). Ovarian absence was most commonly unilateral, with both sides approximately equally affected. Some degree of ipsilateral fallopian tube absence was present in 85% of cases. Approximately 17% of cases had concomitant uterine anomalies, while 22% had renal abnormalities. An additional 10% of cases were comprised of uterine leiomyoma (n = 3) or hypoplastic/rudimentary uterus (n = 8). The most common uterine anomaly recorded was unicornuate uterus, followed by non-surgical uterine absence. Of note, patients who presented with uterine anomalies were more likely to have a renal abnormality. (p < 0.005), and patients with absent fallopian tube were more likely to have renal agenesis (p < 0.005).

Ovarian torsion leading to autoamputation may be preceded by the formation of an ovarian cyst or other outgrowth that raises the risk of torsion. Torsion, in turn, would typically manifest with nausea, vomiting, and acute-onset abdominal pain. Evidence of scarring and adhesions would also be expected during surgery. While some UOA patients report a history of severe abdominal pain for which they did not seek treatment, others recall no such history [1, 18, 52‐54]. In a prior UOA article, Sirisena emphasized the idea that “asymptomatic torsion” is paradoxical – as it necessarily presents as acute abdominal pain [46]. The lack of a painful event in the patient’s history may be explained by torsion while in utero or during infancy when distress is difficult to attribute. Such torsion may lead to autoamputation of the ovary which would then undergo necrosis and calcification, resorption, or possible reimplantation. With reimplantation, one may expect to find histologically verified ovarian remnants in regions outside the normal path of descent. In this sense, the presentation of UOA torsion cases fall along a spectrum, from the initial autoamputation event to the complete degradation or resorption of the ovary, or persistence of a wandering abdominopelvic mass. Instances of rudimentary or hypoplastic ovaries may be a sign of partial autoamputation, falling within this spectrum but outside the scope of this review [2].

Multiple cases in our review support the existence of such a spectrum, consistent with the natural history we proposed above. At one institution, the rate of autoamputation following antenatal ovarian torsion was > 60% [55]. We identified cases of fetuses with ovarian cysts diagnosed in utero with subsequent autoamputation seen in postnatal surgery [56‐59]. These cases were found to have calcified or necrotic ovoid masses in the pelvis, most of which had indeterminant tissue origin on histology. We also noted that similar calcified pelvic masses were found incidentally in adult patients, suggesting that these masses may persist for decades following torsion [43, 60]. Several cases of mature cystic teratomas were observed in older adult patients, all of which were suspicious for ovarian torsion [34, 61]. It was undetermined whether these cases resulted from autoamputation of a torsed dermoid, or transformation of previously autoamputated and reimplanted ovarian tissue. Torsion of the remaining ovary in UOA patients, necessitating salpingoophorectomy, may induce primary ovarian insufficiency. Most reported instances of ovarian torsion occur on the right side – possibly because the ovarian ligament is typically longer, and the left side is relatively protected by the sigmoid colon [62]. However, we found no significant difference in UOA laterality overall or in our torsion etiology group.

Overlap between embryological and vascular causes may exist. It has been suggested that vascular compromise to the primitive ovary or the Müllerian duct could occur during elongation and spinalization of the tube during fourth to fifth month of embryologic development, resulting in ovarian aplasia or hypoplasia [63]. The presence of a calcified remnant in the pelvic cavity may suggest torsion-related autoamputation, though this remains speculative [38]. Two cases of OA presented here occurred likely due to an undiagnosed torsion. Both had normal appearing ipsilateral ureters, suggesting against urogenital abnormalities that would imply an embryological cause. The patient in case 1 also received additional imaging that revealed bilateral kidneys. This finding is consistent with a prior review of ovarian autoamputation by Focseneanu et al. that reported a paucity of renal abnormalities across 94 cases [64]. The authors also noted that most reports of UOA featured partial/absent fallopian tubes, and that ovaries and tubes have distinct embryological origins. Because renal findings would be expected if tubal absence was congenital rather than torsional, and their series demonstrated no renal findings [64]. Focseneanu et al. concluded that congenital absence is not a plausible etiology of OA [64]. We would argue that congenital tubal and ovarian absence could be explained by embryological defect that produces concomitant uterine anomalies and ipsilateral renal anomalies as described below.

Cases in our study with suspected embryological origins were all notable for comorbid GU abnormalities– most commonly complete ipsilateral renal agenesis. Uterine abnormalities such as Müllerian agenesis, unicornuate uterus, and hypoplastic/rudimentary uterus were also frequently observed in this group. Bousfiha et al. reports on a 19-year-old female with bilateral ovarian dysgenesis, concomitant uterine aplasia (MRKH), and normal karyotype [31]. Similarly, Kumar et al. reported an 18-year-old female with primary amenorrhea, absent uterus, vagina and right ovary as well as single ectopic grossly hydronephrotic kidney [65]. These cases of combined ovarian absence and Müllerian anomalies suggest an insult that occurs early in embryologic development. Such an insult accounts for both fallopian tube and OA. This supports existence of a congenital etiology of UOA, while acknowledging the implausibility of OA in the absence of ipsilateral renal findings.

Animal studies showed that ablation of genes involved in genital ridge development can lead to gonadal agenesis. Lhx1-, Lhx9-, Emx2-, Nr5a1-, and Wt1-null mice lack gonads completely, but agenesis extends to the kidneys and sometimes other organs, including the adrenal glands, due to their common embryological origin [5‐9]. Ablation of additional factors has been shown to affect gonadal development. These genes include Foxl2, Wnt4, Tcf21 (Pod1), Six1/4, and members of the insulin/insulin-like growth factor family Insr, Igf1r and Insrr [66‐70]. However, such deletions result in gonadal hypoplasia, dysgenesis, or sex reversal rather than agenesis. These findings show how the phenotypic presentation is directly associated with the time and tissue of expression of the gene being mutated or deleted. It is sometimes difficult to transpose animal studies to human cases. For example, Wt1and Nr5a1mutations in humans are associated with gonadal dysgenesis and primary ovarian insufficiency respectively, but do not impair early gonadal formation [71, 72]. However, despite possible species-specific genetic differences, genetically-caused gonadal agenesis would be unlikely to occur in isolation. In addition, loss-of-function of genes that are necessary for genital ridge development and ovarian formation would be unlikely to cause unilateral agenesis. In such an event, mechanisms including mosaicism or hypomorphic mutations should be considered. However, without genomic investigation and functional validation, these hypotheses are difficult to demonstrate.

Our results suggest that imaging alone is insufficient to diagnose UOA. Of the case reports that noted results of ovarian imaging prior to confirmatory surgery, 28% (16/57) of imaging results were inconsistent with operative findings. These findings echo Case 1, where three ultrasounds reported normal bilateral ovaries prior to surgery. As UOA is relatively uncommon, some degree of expectancy bias may contribute to these inconstancies. Imaging is also limited by the skill of the technician obtaining the study, as well as the individual who interprets it. Ultrasound was the most common imaging modality reported, and the modality most prone to error. More sensitive modalities, such as MRI, were often conducted post-operatively to ensure that the ovary was not heterotopic/ectopic or missed during surgery. Arguably the most sensitive test for UOA was laparoscopy, as it incidentally revealed every instance of UOA that was missed on imaging. While laparoscopy is generally a safe operative procedure, it still carries risks inherent to surgical intervention, and is likely unnecessary in the absence of other indications. For patients pursuing infertility evaluation, laparoscopy may be useful. It gives patients definite knowledge of their anatomy and can be performed in conjunction with procedures aimed at the evaluation of tubal patency.

Questions of how UOA may affect fertility are of great importance to patients with the condition. As approximately 20% of our cases presented with fertility concerns, it may be tempting to infer a causal relationship. While fertility was a major cause of presentation, approximately 35% of our cases with ages ≥ 18 years had a history of pregnancy prior to their UOA diagnosis. Perhaps the most extreme example was an incidental UOA finding in a G11P11 woman with a history of 11 uncomplicated deliveries [73]. Another incidental UOA finding occurred during a cesarean delivery [74]. Interestingly, fertility was the focus of a prior review of 60 cases [75]. The study found UOA was not associated with adverse reproductive outcomes in the subset of women without other conditions that could decrease fertility (e.g. endometriosis, leiomyoma, uterine malformation). That is, comorbidities were likely responsible for sub/infertility rather than UOA itself. UOA patients with contralateral tube obstruction have had successful pregnancies following salpingostomy [76], or mucous plug removal during HSG [77]. Two other instances of contralateral tube obstruction were also noted [78, 79].

Another way to conceptualize fertility in UOA involves examining women with a surgically absent ovary. A metanalysis of 21 studies examining ovarian reserve status-post unilateral oophorectomy found decreased ovarian pool quantity, but not quality [80]. Despite a decreased response to induction during in vitro fertilization, the likelihood of a clinical pregnancy in women with one ovary was statistically comparable to women with two ovaries. Taken together, these findings suggest that infertility should not be presumed in UOA patients. In UOA patients without additional conditions that affect fertility the great majority were able to conceive, though some required assistive reproductive techniques. This group of patients roughly corresponds to our cases with suspected non-embryological UOA etiologies – as structural uterine anomalies were much less common in this group.

In addition to documenting reported cases of UOA in-depth, our systematic review revealed a discrepancy in the literature regarding the semantics of “absence” and “agenesis” to describe patients with a solitary ovary. We also highlight the association between OA and Müllerian and renal anomalies, instances of which may denote true agenesis. We used a strict inclusion/exclusion criterion to ensure that our reported cases reflected surgically confirmed cases of UOA. By excluding articles without confirmation, we attempted to eliminate the possibility of false positive reports – in line with our finding that imaging may be insufficient to diagnose UOA.

Our study has several limitations that must be considered in light of our study design. First, we chose to exclude conference abstracts and non-English language publications. Conference abstracts, generally speaking, are less rigorous than peer-reviewed articles [89, 90]. Non-English language articles were excluded due to insufficient translation resources. Our study included data from case reports and series, which are study designs that inherently have a high potential for bias, despite overall good critical appraisal results. We were unable to secure a full text copy of one manuscript that may have included relevant cases (see Supplementary Table 2). Some older case reports predated the advent of karyotyping, raising the possibility that some patients with DSD are reported in our results. We attempted to minimize this possibility by excluding all cases with suspicion for DSD. Lastly, aside from clinical intuition, there is no reliable way of completely ensuring that our division of cases reports by suspected etiology was verifiably correct. This was somewhat limited by the information that case report authors chose to include. We believe our determinations of etiology are relevant to the discussion of absence versus agenesis, and still can inform clinical judgement. Taken together, these limitations suggest that the incidence of UOA is higher than our results would suggest.