Risk Factors Associated with Adnexal Torsion after Hysterectomy

Between June 2007 and September 2019, a total of 352 surgeries were performed for adnexal torsion. Of these, 46 occurred in patients who were status post hysterectomy. Interestingly, there were only 43 individual patients, however 3 of them suffered recurrent or contralateral adnexal torsion, for a total of 46 torsions. The estimated incidence of ovarian torsion after hysterectomy was 0.5% (46/8,538 ovarian-sparing hysterectomies). Table 1 depicts overall baseline demographic, clinical, and surgical characteristics of all hysterectomy patients by torsion status. Age at the time of hysterectomy, gynecologic history, surgical history, mode of previous hysterectomy, indication for previous hysterectomy, specialty of the surgeon who performed hysterectomy, any other ovarian procedure performed, and surgical approach showed an association with torsion in unadjusted models. Additionally, some variables had to be collapsed into other variables or excluded from analysis given low cell counts. Adjusted logistic regression analysis revealed that mode of previous hysterectomy (TLH/LAVH vs TAH/TVH), age at the time of hysterectomy (younger (17–40) vs older (41–51)), and gynecologic history (endometriosis vs none), were associated with adnexal torsion after hysterectomy ( Table 2 ). Patients who underwent laparoscopic or laparoscopic-assisted hysterectomy were 3.4 times more likely to develop adnexal torsion than those who underwent open or vaginal approach (OR 3.36 [95% CI: 0.86, 13.23]). Likewise, those patients who were 17–40 years old at the time of their hysterectomy were 3.4 times more likely to develop adnexal torsion after hysterectomy than patients who were 41–51 years old (OR 3.45 [95% CI: 1.33, 8.97]). Patients who had a gynecologic history of endometriosis were 4.1 times more likely to develop adnexal torsion after hysterectomy than patients with no history of endometriosis (OR 4.07 [95% CI: 1.04, 15.88]). Only the variables that were associated with torsion in unadjusted analyses and were able to be included without having convergence issues are presented in Table 2 , all other variables were not associated. Median time to torsion after hysterectomy was 40.4 months (95% CI: 19.5, 53.1), or 3.4 years. Kaplan Meier curves are shown in Figure 1 . In the overall group, by 1 year, 11 (24%) torsions had occurred; by 2 years, a total of 18 (40%); by 5 years, a total of 31 torsions (69%); and by 10 years, a total of 43 (96%). The longest recorded time to torsion was 147 months, or 12.3 years. This patient was 38 years old at the time of her hysterectomy. When broken down into age groups ( Figure 1 ), the risk of torsion persists for a longer duration in patients who had their hysterectomy at ages 17–40 years old than those who were between 41–51 years old (p=0.006). The rate to torsion of patients ages 17–40 occurs rapidly; for example, approximately 25% of the torsions had occurred by 18 days. The average age at time of torsion was 39.6 years (SD=6.9). All patients presented with lower abdominal pain, accompanied by a combination of nausea, emesis, and anorexia. Torsion occurred on the right side 70% of the time. Sixty-two percent of patients had a cyst or mass seen on ultrasound within the torsed ovary, and mean cyst diameter was 4.6 cm (SD=2.3 cm). 38% of patients did not have a cyst present. Cyst composition was most commonly complex or mixed appearing (69%), and only 21% were simple. Mean diameter of the entire affected ovary was 6.7 cm (SD=2.1 cm) but in contrast, the mean diameter of the contralateral non-torsed ovary was less than half that size at 3.2 cm (SD=0.09). Doppler flow to the affected ovary was absent or diminished in 54% of ultrasounds, but interestingly was still present with normal flow documented in 27%. The remaining imaging reports did not document presence or absence of flow. Thirty-one percent of ultrasounds showed increased amounts of free fluid in the cul-de-sac. Treatment consisted of unilateral oophorectomy in 91% of cases, likely due to a 74% reported intra-op appearance of a necrotic or ischemic ovary that could not be saved in the opinion of the surgeon. Final pathology confirmed this fact with 70% of specimens reported as necrotic or infarcted.

We conducted a nested case-control study of patients with and without adnexal torsion after hysterectomy from a large urban hospital system in Pittsburgh, Pennsylvania between July 2007 and September 2019. This time frame indicates the availability of electronic medical records at our institution. This study protocol was approved by the University of Pittsburgh Institutional Review Board. The exposure of interest in this study was hysterectomy (of any approach), and the outcome of interest was adnexal torsion. Inclusion criteria included women who underwent hysterectomy with preservation of at least 1 ovary within the designated time frame. Exclusion criteria included women who had a bilateral salpingo-oophorectomy at the time of hysterectomy, and who were post-menopausal (or over the age of 51, based on the average age of menopause), given the low incidence of torsion after menopause. Adnexal torsion was defined as greater than or equal to 180 degrees of twisting of the ovary and/or surrounding adnexal structures documented during surgical intervention. In order to identify cases, all charts of patients who underwent surgery for adnexal torsion were reviewed, specifically for the presence or absence of a uterus. Once all patients who were diagnosed with adnexal torsion after hysterectomy were identified, controls were randomly selected in a 1:1 ratio from patients who did not have adnexal torsion after their hysterectomy from the pool of all hysterectomies that were performed during the study period. Minimum group sample sizes of 44 were required in order to detect a Cohen’s h effect size (standardized difference between group proportions) of 0.6 with a 0.05 significance level and 80% power. The following demographic and operative variables were recorded: surgeon specialty or sub-specialty, indication for hysterectomy, gynecologic and surgical history, age, and BMI. Preoperative clinical presentation and imaging findings were also recorded. The following preoperative risk factors for torsion were analyzed: presence of ovarian cyst(s), size and characteristic of cyst(s), and a history of ovarian cysts or torsion. Relevant intraoperative variables included mode of hysterectomy, entry into and dissection of the retroperitoneal space, creation of a window in the posterior broad ligament to facilitate coagulation and division of the utero-ovarian ligament, removal of the fallopian tubes, and any other adnexal procedure that may have been performed. Our primary objective was to identify risk factors for adnexal torsion after hysterectomy, and our secondary objective was to estimate the incidence. The incidence of adnexal torsion was estimated by obtaining the total number of hysterectomy specimens without bilateral adnexa during the specified time period from pathology records. The exact number of ovarian-sparing hysterectomies could not be obtained due to inability to assess for specimens in which only 1 ovary was removed. Current Procedural Terminology (CPT) codes were not utilized due to the vague description of removal of adnexal structures (i.e., Laparoscopy with total hysterectomy; with removal of tube(s) and/or ovary(s)). Cases of adnexal torsion were compared to controls across demographic and clinical characteristics via unadjusted and adjusted logistic regression, with Wald-chi-square tests used for significance. The multivariable (adjusted) logistic regression model included age at the time of hysterectomy – younger (17–40 years) vs older (41–51years), gynecologic history, surgeon specialty who performed previous hysterectomy, and mode of previous hysterectomy. Due to the small groups within ‘mode of hysterectomy’, we collapsed to total laparoscopic hysterectomy (TLH) and laparoscopic-assisted vaginal hysterectomy (LAVH) (given similarity of approach to the adnexa) vs. total abdominal hysterectomy (TAH) and total vaginal hysterectomy (TVH). Odds ratios and 95% Wald confidence limits are reported. Additional variables could not be included in the final multivariable model due to small size convergence issues. The time to adnexal torsion after hysterectomy was also recorded in the cases. Post-hoc Kaplan-Meier curves were used to show the cumulative proportion of patients with torsions over time both overall and by age group (younger (17–40 years) vs older (41–51 years)) with log rank tests used to compare time to torsion across the two age groups. Statistical significance was defined as a two-sided p value less than 0.05. Data were analyzed using STATA version 15 (College Station, TX: StataCorp LLC).

Our study found that in premenopausal patients who undergo hysterectomy, a laparoscopic approach, younger age at time of hysterectomy, and a history of endometriosis independently contribute to the risk of developing subsequent adnexal torsion. Surgical approach to the retroperitoneal space and broad ligament did not contribute. Our study also showed that the estimated incidence of adnexal torsion after any type of hysterectomy is consistent with what has been published in previous literature. This study confirmed the theories of previous authors which state that laparoscopic approach to hysterectomy may be associated with subsequent adnexal torsion [ 7 – 9 ]; however, the exact mechanism is still unclear. Opening the retroperitoneal space may disrupt the peritoneal attachment to the pelvic sidewall, and creating a window in the posterior leaf of the broad ligament may skeletonize the IP. Both of these surgical steps could contribute to a more isolated pedicle on which the adnexa may twist; however, neither step was statistically significant in the statistical analysis. Therefore, the hypothesis of fewer postoperative adhesions and less ovarian trauma is more likely to be responsible. The fact that laparoscopic approaches to common procedures, for example myomectomy, lead to less adhesive disease than when performed open is well documented in the literature [ 10 ], and our data also supports this theory. Younger age likely contributes to the increased risk of subsequent torsion due to the propensity of reproductive age women to create ovarian cysts that serve as a nidus for torsion. The younger the patient is at the time of her hysterectomy, the more cumulative opportunity she has to develop adnexal pathology that may lead to torsion. It is unclear why patients with a history of endometriosis had a higher risk of subsequent ovarian torsion. Typically, one would think of endometriosis causing adhesions, especially endometriomas adhering to the pelvic sidewall, leading to less risk of torsion due to immobility of cysts. It may be that advanced dissection by specialized surgeons of the retroperitoneal spaces and aggressive skeletonization of the IP away from its peritoneal attachment that may occur in cases of endometriosis could contribute, however these procedures were performed too infrequently to assess if they could be associated with subsequent torsion. Our data may instead reflect the fact that early stage endometriosis does not cause as much adhesive disease as the more rare late stage, advanced disease described above. Alternatively, it may be that some other aspect of the surgical approach is different and may be contributing in a way that we are not yet able to identify, and is an area for further research. In recent years, there has been a shift towards performing opportunistic salpingectomy for ovarian cancer prevention. It was theorized that the loss of yet another stabilizing structure of the adnexal pedicle may lead to an increased risk of torsion [ 8 ], however our study showed that performing a bilateral salpingectomy does not increase the risk of subsequent torsion. This is likely because without the attachment to the cornua of the uterus, the fallopian tube serves no role in stabilizing the remaining adnexal pedicle. The recommendation for opportunistic salpingectomy was published by the Society of Gynecologic Oncology in 2013, and by the American College of Obstetricians and Gynecologists in 2015, so this study captured data from both before and after the recommendation and showed no difference. We also found that a very high percentage of patients with adnexal torsion after hysterectomy underwent oophorectomy instead of detorsion. In patients without a uterus, surgeons may be less inclined to perform ovarian-conserving surgery, given no potential for future fertility in these patients. The high rate of oophorectomy may have also been due to the high rate of ovarian necrosis and irreversible ischemic damage that was found on surgical entry to the pelvis, and was confirmed on pathology review. A factor that is not proven but may theoretically lead to necrosis is the ability of the pedicle to twist several times on itself given the lack of attachment to the utero-ovarian ligament, thus leading to more complete torsion and faster loss of flow to the adnexa. In addition, a delay in torsion diagnosis is well described in the literature [ 11 – 12 ], and this delay might only be prolonged in patients who are status post hysterectomy leading to necrosis and significant periovarian adhesions as well. Strengths of the study include the fact that this sample size is the largest published in the literature on this topic. The medical system at which this study was performed has a high-volume surgical practice which allowed us to obtain meaningful numbers. In addition, including a control cohort allowed us to perform a more robust statistical analysis with adjusted logistic regression and obtain more meaningful results. Limitations of the study include the estimation of the number of ovarian-sparing hysterectomies performed in the study period, though more reliable than if CPT codes had been used. Additionally, there is a possibility that the incidence of torsion may actually be higher than reported here if patients were diagnosed and treated at an outside institution of which we have no record. Furthermore, only surgically proven cases of adnexal torsion were included in this study, so we may have also missed those patients with less “typical” presentations who did not undergo surgery. There is also the ongoing risk of torsion that may occur after the study period has ended, especially in those who had their hysterectomy more recently, and as such, those events may have been missed as well. Lastly, statistical analysis was somewhat restricted due to the study design of not using a larger control cohort and the small event numbers of certain variables, which restricted the multivariable analysis. While the study was powered to show moderate associations which were still clinically significant, a larger study is needed to definitively answer the question of statistical significance of some variables. A comparison of patients diagnosed with adnexal torsion with and without a previous hysterectomy was not performed in this study but could serve an area for further research. Additionally, while the reported incidence of adnexal torsion in this study (0.5%) is lower than the reported incidence of adnexal torsion in any patient (with or without a uterus) 2.7–3% [ 1 – 3 ], that number is actually very outdated and was based off of data published in the 1980s which has been propagated over time. Research should be performed in the modern era of updated ultrasound technology and surgical diagnostic approaches to update the true incidence of ovarian torsion. In conclusion, this study provides data to better counsel patients who plan to undergo hysterectomy on their future risks. It is very important to note that the findings of this study should not change our decision to perform a hysterectomy when indicated, and do not support performing an alternative approach to hysterectomy over laparoscopic, but rather bring to light the fact that torsion after hysterectomy is still a possibility. Importantly, ovarian torsion should not be excluded from the differential in a patient who presents with acute-onset abdominal pain at any time after hysterectomy, even years later. In fact, in young patients who are status post laparoscopic hysterectomy, it should perhaps heighten our suspicion of torsion and help us reach an accurate diagnosis faster to prevent loss of an ovary due to irreversible damage.

Adnexal torsion is a relatively common surgical emergency with a reported prevalence of 2.7–3% [ 1 – 3 ] and well understood risk factors. These include an increased weight or diameter of the ovary, a known ovarian cyst, and a benign nature of the cyst [ 4 – 5 ]. It often presents in women in the reproductive age group with acute-onset lower abdominal pain, accompanied by nausea and emesis [ 6 ]. We know far less, however, about adnexal torsion that occurs after hysterectomy. The true incidence and associated risk factors are unclear, given that the literature consists of only case reports and a handful of small retrospective studies [ 7 ]. The largest study to date published by Ravid, et al. in 2019 consists of only 8 patients, and cites the risk of torsion after any type hysterectomy at 0.16%, but cites a significantly higher risk after laparoscopic hysterectomy at 1% [ 8 ]. In addition to the risk factors for adnexal torsion in general, it seems that a laparoscopic approach to hysterectomy may be an independent risk factor for subsequent adnexal torsion. Mashcish, et al. theorized that this may be because the laparoscopic technique allows for greater mobility of the ovary and infundibulopelvic ligament (IP), less ovarian trauma, and less adhesion formation than after laparotomy; this is known as “Wattiez Triple Factor Theory” [ 9 ]. However, without an adequate number of patients we are unable to verify these theories or identify surgical and clinical characteristics that are associated with torsion after hysterectomy. The aim of this study is to compare patients who had adnexal torsion after hysterectomy to a controls who did not have torsion after hysterectomy in order to identify risk factors. We also aim to estimate the incidence of adnexal torsion after hysterectomy in order to provide clinicians with better data to counsel their patients on the risks of hysterectomy.