Risk of de novo severe carpal tunnel syndrome after bilateral oophorectomy: a population-based cohort study.

The M ayo Clinic Cohort Study of O ophorectomy and A ging-2 (MOA-2) included all of the premenopausal women who underwent bilateral oophorectomy for a nonmalignant indication in Olmsted County, Minnesota between January 1, 1988 and December 31, 2007. Each woman was randomly matched by age (+/−1 year) to a referent woman residing in the same county who had not undergone bilateral oophorectomy before the index date. Hysterectomy with at least one ovary conserved or unilateral oophorectomy were not exclusion criteria for referent women. Index date for each matched pair was the date of oophorectomy. Extensive details about the MOA-2 study were reported elsewhere. 7 , 12 - 14 Women were excluded from the study if they did not authorize the use of their medical records for research (2.5% of women in the overall population). 15 The project was approved by the institutional review boards for Olmsted Medical Center and for Mayo Clinic. 12 , 13 All data were obtained using the medical records-linkage system of the Rochester Epidemiology Project (REP). REP resources include complete records (inpatient and outpatient) from all major medical care providers in Olmsted County, Minnesota. Additional details about the REP were reported elsewhere. 15 - 18 Extensive demographic and clinical characteristics as well as surgical indication were collected as part of MOA-2 via abstraction of medical records. 12 , 13 The oophorectomy was classified as having a ‘benign indication’ when it was performed for presumed nonmalignant ovarian conditions such as adnexal masses, cysts or endometriosis, and as having ‘no ovarian condition’ when it was performed at the time of hysterectomy in the absence of any ovarian condition. The hysterectomy was performed most commonly for uterine fibroids, heavy menstrual bleeding, or chronic pelvic pain. International Classification of Diseases, Ninth Revision codes for CTS assigned to women over their entire lifetime were obtained from the REP diagnostic indexes through December 31, 2014 (ICD-9 code 354.0). The medical records for all women who screened positive because they received at least one diagnostic code for CTS, with the earliest code assigned on or after their index date, were manually reviewed by two physician abstractors (J.S. and V.J.M.M.S.). The physician abstractors confirmed the diagnosis of CTS and the date of diagnosis, and collected details on symptoms, diagnostic procedures, and treatments such as steroid injections or carpal tunnel release surgery. In addition, non-idiopathic causes for CTS such as cervical radiculopathy, trauma, peripheral neuropathy, or other causes were collected. By contrast, women who received at least one CTS code before the index date were considered to have prevalent CTS and their records were not reviewed to confirm the diagnosis. Therefore, women with CTS of any level of severity before the index date were excluded from the cohort analyses. Severe CTS was defined as women who were treated with steroid injection, underwent carpal tunnel release surgery, or had electrodiagnostic studies with moderate or more severe results (electromyography or nerve conduction study). 19 The onset of severe CTS was defined as the earliest date of injection, release surgery, or qualifying electrodiagnostic study. In secondary analyses, we also considered CTS of any level of severity (any CTS) and idiopathic CTS. The onset of any CTS or idiopathic CTS was defined as the earliest diagnosis date in the medical record. A preliminary study was done to validate the use of a single ICD-9 code to define prevalent CTS at the index date and to screen for CTS as an outcome during follow-up. Medical records were reviewed for a random sample of 50 women with no CTS diagnostic codes and 50 women with at least one CTS diagnostic code. The two physician abstractors (J.S. and V.J.M.M.S.) were kept unaware of the presence or absence of the diagnostic codes, and determined CTS status based solely on information abstracted from the medical records. Using the medical record abstraction as the standard for comparison, the sensitivity of codes was 93.6% (44 out of 47 women) and specificity was 88.7% (47 out of 53 women). The positive predictive value was 88.0% (44 out of 50 women) and the negative predictive value was 94.0% (47 out of 50 women). In addition, the medical records for 20 women were abstracted independently by both abstractors, and agreement on CTS status was observed for 18 women (90% agreement, kappa 0.76, 95% CI 0.46-1.00; 5 negative agreement, 13 positive agreement). Chronic conditions that were present at the index date and could act as confounders were identified using diagnostic codes suggested by the Department of Health and Human Services. 20 , 21 The 18 chronic conditions included depression, anxiety, substance abuse disorders (drugs and alcohol), dementia, schizophrenia and psychosis, hyperlipidemia, hypertension, diabetes, cardiac arrhythmias, coronary artery disease, congestive heart failure, stroke, arthritis, cancer (all types), asthma, chronic obstructive pulmonary disease, osteoporosis, and chronic kidney disease. Each condition required the presence of two or more related diagnostic codes, separated by >30 days to reduce the risk of false-positive diagnoses. A finer dating of diagnostic codes was not available before 1994; therefore, a 1-year separation of codes was required before 1994. 7 Severe CTS was the primary outcome and any type of CTS or idiopathic CTS were secondary outcomes. Women who received a diagnostic code for CTS of any level of severity before the index date were excluded from all analyses. Therefore, we did not consider the possible effect of bilateral oophorectomy on the recurrence of CTS in women who had experienced CTS of any severity before the index date. Similarly, only the first episode of CTS during follow-up was considered. Follow-up was terminated at the first episode of CTS, regardless of severity, and recurrent episodes were not studied. Women were followed from their index date until the date of the first episode of CTS (of any severity), death, last visit with a REP care provider, or the end of the study (December 31, 2014). Cox proportional hazards models were used to estimate hazard ratios (HRs) and 95% confidence intervals (CI) using age as the time scale, with each woman entering the risk set at her respective age at index date. Cumulative incidence curves were estimated using the Kaplan-Meier method. Differences between the oophorectomy and referent cohorts were also measured using absolute risk increase (ARI) or reduction (ARR), calculated by subtracting the two absolute risks at 20 years after the index date. The Cox models and Kaplan-Meier curves were adjusted using inverse probability weights derived from a logistic regression model including characteristics at baseline. We included 18 chronic conditions (list provided earlier), years of education (≤12, 13-16, and >16), household income quartiles (<$42,000, $42,000-56,999, $57,000-71,999, ≥$72,000), race (white vs. nonwhite), body mass index (BMI; <30 vs. ≥30 kg/m 2 ), cigarette smoking (current or former vs. never), age (continuous), and calendar year (continuous). The inverse probability weights were calculated separately within each stratum to maximize the balance of the adjustment variables. After adjustment using inverse probability weights, the standardized differences for all of the conditions or characteristics considered were below the recommended threshold of 0.10 (i.e., negligible imbalance between the two cohorts; data not shown). The Cox models used robust sandwich covariance estimates to account for the use of estimated weights and for the inclusion of women in both cohorts (referent women who underwent subsequent bilateral oophorectomy). Analyses were performed for all women combined and stratified by age at oophorectomy (≤45 and 46-49 years), ET within each age group, BMI at index date (<30 vs. ≥30 kg/m 2 ), parity at index date (nulliparous vs. ≥1 birth), and surgical indication. Stratification by age, ET, and surgical indication were defined based on the status of the woman who underwent bilateral oophorectomy in each matched pair. By contrast, stratification by BMI and parity were defined based on the status for each woman separately (i.e., obese women who underwent oophorectomy were compared to obese referent women). We also performed two sets of sensitivity analyses to 1) censor referent women at the time of bilateral oophorectomy if it was performed after the index date, and 2) exclude women with any chronic condition at the index date. Tests of statistical significance were conducted at the two-tailed alpha level of 0.05. Analyses were performed using SAS version 9.4 (SAS Institute Inc., Cary, NC).

There were 1,653 women who underwent bilateral oophorectomy and 1,653 age-matched referent women who were assessed for CTS over their entire lifetime using diagnostic codes. Among these women, 254 women from the oophorectomy cohort (15.4%) and 175 women from the referent cohort (10.6%) had prevalent CTS at baseline and were excluded from all of the outcome analyses ( Figure 1 ). In a case-control analysis at baseline, the odds ratio for CTS of any type prior to the index date was 1.52 (95% CI 1.24-1.87; P <0.001). Case-control analyses conducted in the overall MOA-2 sample for baseline characteristics other than CTS have been reported elsewhere. 12 Table 1 shows the baseline characteristics of the women who were included in the cohort analyses (1,399 with bilateral oophorectomy, 1,478 referent). More referent women were of nonwhite race; however, the number of nonwhite women was small. Women who underwent bilateral oophorectomy were less educated, had lower income, higher BMI, were more likely to smoke, and had higher parity. Inverse probability weights were used to adjust all analyses for these baseline differences plus age, calendar year, and the presence of 18 chronic conditions. Table 1 also provides the indication for the bilateral oophorectomies. Approximately 41% of the woman did not have any ovarian condition mentioned in their medical record at the time of surgery, whereas the remaining 59% underwent bilateral oophorectomy for a benign ovarian condition. Only women who received at least one diagnostic code for CTS, with the earliest code assigned on or after the index date screened positive for the detection of CTS during follow-up and underwent medical record review. The results of the medical record review in the two cohorts are summarized in Figure 1 . The characteristics of women who developed de novo severe CTS, any CTS, or idiopathic CTS during follow-up are provided in the Supplemental Digital Content 1 . The median age at onset of severe CTS was 52 years (interquartile range [IQR] 46-56) in the oophorectomy cohort and 51 years (IQR 46-55) in the referent cohort, whereas the median time from index date to onset of severe CTS was 7.1 years (IQR 3.8-13.9) in the oophorectomy cohort and 7.0 years (IQR 3.0-12.4) in the referent group. For all three CTS definitions, the majority of women reported numbness and tingling, and over half of the women reported nocturnal symptoms and had positive Tinel or Phalen tests ( Supplemental Digital Content 1 ). Table 2 and Figures 2 and 3 show the results of primary analyses for severe CTS, overall and in strata. The overall risk of de novo severe CTS was increased in women who underwent bilateral oophorectomy (adjusted HR 1.65, 95% CI 1.20-2.25, ARI 4.0%). Increased risk was also noted in women age ≤45 years at oophorectomy and in women who were taking ET at their 46th birth date; however, the interactions by age group or by ET were not significant ( Table 2 , footnote b). No significantly increased risk of severe CTS was found for women age 46-49 years at the index date or for obese women of any age. Figure 3 shows significant increased risk of severe CTS in the oophorectomy cohort for non-obese women, for both nulliparous women and for women with at least one birth, and for women who underwent oophorectomy for a benign indication. However, interactions by BMI, parity, and indication were not significant ( Table 2 , footnote b). Supplemental Digital Content 2 shows the results for de novo CTS of any type (any level of severity). Although the magnitude of the HRs was not as high as the HRs from the primary analyses, the results were similar. Specifically, women who underwent bilateral oophorectomy had a significantly increased risk of any CTS in the overall analyses, among non-obese women, and in women with benign ovarian indications. However, only the interaction by indication was significant (interaction P =0.01; Supplemental Digital Content 2 , footnote b). Supplemental Digital Content 3 shows an overall higher risk of developing idiopathic CTS in women who underwent bilateral oophorectomy compared to referent women (adjusted HR 1.35, 95% 1.07-1.70, ARI 3.9%). A significantly increased risk of idiopathic CTS was also found in women who underwent oophorectomy at age 46-49 years, in non-obese women, and in women with a benign ovarian indication. However, only the interaction by indication was significant (interaction P =0.02; Supplemental Digital Content 3 , footnote b). In a first set of sensitivity analyses for de novo severe CTS, referent women who underwent bilateral oophorectomy after their index dates were censored at the date of subsequent surgery. The results were essentially unchanged from the primary analyses (results not shown). In a second set of sensitivity analyses, we assessed a sub-group of women who did not have any of the 18 chronic conditions at the index date for the risk of de novo severe CTS (607 women who underwent bilateral oophorectomy, 832 referent women). Again, the results were consistent with results from the primary analyses (results not shown).

This study showed that overall, women who underwent bilateral oophorectomy prior to menopause were more likely to develop de novo severe CTS than referent women who did not undergo bilateral oophorectomy. The risk was suggestively greater in women with lower body mass index, nulliparity, and with a benign ovarian indication for oophorectomy (non-significant interactions). We did not observe a protective effect of estrogen therapy after the oophorectomy. Our primary analyses focused on severe CTS because we expected that the detection of more severe episodes of CTS would be more complete than for milder episodes in a records-linkage system. In addition, severe episodes of CTS are more clinically relevant because they cause a greater functional impairment in women. However, the findings were similar in secondary analyses considering the incidence of CTS of any severity or of idiopathic CTS. We may have identified a subgroup of women at higher risk of CTS. Our findings suggest that lean nulliparous women who undergo bilateral oophorectomy for a benign indication will have a higher risk of developing CTS compared to the referent population. If our results are confirmed in other studies with a larger sample size, they may suggest a profile of women who are more likely to encounter this particular sequela from oophorectomy, and may guide a more individualized prediction. Parity is a risk factor for CTS in the general population. 22 , 23 In our study, parity suggestively modified the strength of the associations. Nulliparous women who underwent bilateral oophorectomy had a 3-fold risk of severe CTS compared to nulliparous referent women. By contrast women with at least one birth who underwent oophorectomy had an only 1.5-fold risk of developing severe CTS compared to referent women with at least one birth. However, both associations were significant, and the test for interaction was not. Obesity is also a risk factor for idiopathic CTS in the general population. 22 , 24 , 25 However, bilateral oophorectomy did not increase the risk for de novo CTS in obese women in our study. Therefore, it is possible that the effect of obesity on CTS may be mediated in part by the higher risk of bilateral oophorectomy in obese women. This is in contrast to what we found for women who were not obese at the index date. Even though non-obese referent women had a relatively low risk for developing de novo severe CTS, bilateral oophorectomy resulted in a doubling of the risk of developing CTS. However, the test for interaction was not statistically significant. As for any observational study, the observed associations may be explained by some residual bias or some yet unknown confounding variables. 26 Therefore, we propose three possible explanations: 1) The associations are explained by a yet unknown risk factor shared by CTS and by some uterine or ovarian conditions prompting the bilateral oophorectomy. 2) The associations are explained by a lower pain threshold that predisposes women both to undergo a bilateral oophorectomy (with or without hysterectomy) and to be diagnosed with CTS. 3) Finally, the associations may be causal, and may be mediated by the loss of estrogen, the loss of other ovarian hormones, or by an alteration of the hypothalamic-hypophyseal system. 10 These three explanations are not necessarily exclusive alternatives, and more likely, the three explanations apply to specific subgroups of women. Figure 4 provides a simple schematic representation of the three explanations. A possible first explanation is confounding by indication. A subset of women may have shared a common risk factor that predisposed them to gynecologic conditions such as a uterine disease or a benign ovarian pathology that prompted oophorectomy and independently predisposed them to CTS. This common underlying risk factor may be a genetic factor, a developmental or early-life event, an environmental factor, or another yet unknown factor ( Figure 4 , panel A). This first explanation is supported by the different results of analyses stratified by ovarian indication. If the oophorectomy is causative per se , all women should have an increased risk of CTS. Possible susceptibility genes are genes that encode for the production and degradation of collagen, 27 - 29 genes responsible for the inflammatory response, and genes involved in the protection against oxidative stress. For example, a matrix metalloproteinase (MMP) up-regulation might contribute to CTS because it is involved with chronic inflammatory response processes and connective tissue remodeling throughout the body. 30 , 31 In addition, increased MMP-2 and MMP-9 activity may cause chronic low-grade inflammation leading to some ovarian conditions (e.g., ovarian cysts and especially polycystic ovarian syndrome). 32 - 34 Oxidative stress may trigger systemic inflammatory processes, which may in turn trigger down-stream cascades, ultimately resulting in malignancies. 35 , 36 In our second explanation, the associations between bilateral oophorectomy and CTS may be caused by a lower threshold to pain ( Figure 4 , panel B). Some women may be predisposed to experience pain more severely than others. Conditions that result in increased pain may be associated with collagen turnover. However, there is only limited evidence supporting this theory. 37 Other explanations include disorders affecting sympathomimetic amines 38 and estrogen deficiency disorders, 39 which may ultimately result in chronic pelvic pain, joint pain, headaches, fibromyalgia, and CTS. Therefore, a woman with a lower threshold to pain may be more likely to report to a care provider for chronic pelvic pain. If the pain does not respond to oral contraceptive therapy or to pain medications, the care provider and the woman may consider a hysterectomy with bilateral oophorectomy. 40 , 41 Independent of this chain of events, the same woman may be more likely to report to her care provider with wrist pain and be diagnosed with CTS. Therefore, a lower threshold to pain may create a spurious association between bilateral oophorectomy and CTS that may manifest both before and after the time of the oophorectomy (index date). In support of this interpretation, CTS was more common in the women who underwent bilateral oophorectomy compared to referent women, both before and after the index date. The third explanation fits best with our knowledge about the long-term sequelae of bilateral oophorectomy in other organs and systems. 7 , 14 The abrupt drop in the levels of all ovarian hormones at a younger age due to bilateral oophorectomy may have a direct deleterious effect on the carpal tunnel ( Figure 4 , panel C). In support of a direct protective effect of ovarian hormones on the carpal tunnel, ET with or without a combined progestin reduced the risk of CTS in postmenopausal women involved in the Women’s Health Initiative clinical trials. 3 Even though results specific for bilateral oophorectomy were not reported, a subgroup of women who underwent hysterectomy in the estrogen alone clinical trial also underwent bilateral oophorectomy. 3 However, we did not observe a significant interaction by ET in our study. Our finding may be due to limited statistical power. Alternatively, other ovarian hormones such as androgens may play an important protective role. Against a direct causal explanation are the findings of a similar median age of onset and median time between the index date and onset of severe CTS in the two cohorts. Our study has several strengths. First, details about the oophorectomy, baseline characteristics, and CTS were abstracted from medical records contained in a records-linkage system with no direct involvement of women in the study. This design should have minimized recall bias. Second, because the data collection was historical, women included in the study only needed to provide a general research authorization (per Minnesota legal requirements). This design should have minimized non-participation. 15 , 16 Third, the referent women were representative of the general population. We did not use women who underwent hysterectomy with ovarian conservation as a referent group because hysterectomy by itself may also be associated with CTS (unpublished findings from our group). Lastly, the study population included all women without restriction by race, ethnicity, socioeconomic status, health insurance coverage, or health care provider. 15 , 16 The study also has some limitations. First, we restricted the medical record abstraction to determine CTS status during follow-up to women who received at least one diagnostic code for CTS (screening for CTS). Although in our small validation study the diagnostic codes performed well as a screening tool, some misclassification may have occurred. However, this misclassification should be similar among the oophorectomy and referent cohorts (non-differential misclassification). Second, it was impossible to conceal the bilateral oophorectomy status for the two data abstractors who confirmed the diagnosis of CTS because they had access to the entire medical record. Third, our sample size was limited for some of the stratified analyses. Therefore, some of the interesting findings were suggestive but did not reach statistical significance. These differences across strata need to be confirmed in a study with a larger sample size. Fourth, because the majority of women who underwent bilateral oophorectomy also had concurrent or previous hysterectomy, the women primarily used unopposed ET. Therefore, we were not able to assess the effect of progestins or combined hormone therapy. Additionally, we did not have enough power for analyses by specific type of ET, route of administration, or dosage. Fifth, we did not have information about occupation or family history of CTS. Finally, the women in our study all lived in Olmsted County, Minnesota, and the findings may differ in other populations with different demographic or socioeconomic characteristics. However, the demographic characteristics of this geographically-defined population are similar to the characteristics of the Upper Midwest and of a large segment of the entire United States population. 42

This study is one of the first to demonstrate an increased long-term risk of de novo severe CTS in women who underwent bilateral oophorectomy before menopause. Whether oophorectomy increases the risk of CTS because it causes abrupt estrogen deficiency, because of other endocrine disruption, or because of some confounding mechanism (shared risk factors or lower threshold to pain) remains unclear. In addition, the lack of beneficial effect of ET on the risk of CTS after bilateral oophorectomy should undergo further evaluation. These findings, in conjunction with the results of other studies addressing more severe disease outcomes such as cardiovascular disease or dementia, have important clinical implications and should prompt a reassessment of bilateral oophorectomy in premenopausal women who are not at high genetic risk of ovarian cancer. 43

Carpal tunnel syndrome (CTS) is the most common nerve entrapment in the upper extremity. Although the syndrome is predominantly of idiopathic nature, an association with sex hormones has been suggested given the higher incidence in women compared to men at all ages, and the peak incidence around the age of menopause in women (ages 50-59 years) but at later age in men (70-79 years). 1 In a case-control study, CTS was found associated with earlier age at menopause. 2 In addition, estrogen therapy (ET) with or without a combined progestin reduced the risk of CTS in postmenopausal women involved in the Women’s Health Initiative clinical trials. 3 Although the exact role of estrogen and progesterone in the etiology of CTS remains unclear, these hormones have been shown to affect the regulation of pro-inflammatory cytokines, which could link endocrine imbalance to CTS. 4 , 5 Interestingly, the number of estrogen receptors in biopsy samples of the transverse carpal ligament of CTS patients was highest in women 50-70 years old compared to women 70 years, and to men. 5 The impact of bilateral oophorectomy on hormone homeostasis is striking, and bilateral oophorectomy has been linked to increased morbidity and mortality, in particular to several neurological disorders. 6 - 9 Bilateral oophorectomy was found associated with musculoskeletal conditions such as osteoporosis and arthritis, but little is known about the impact of oophorectomy on the incidence of CTS. 7 Two earlier studies suggested an increased risk of CTS after premenopausal oophorectomy. However, these studies were limited by small sample sizes and did not address the role of possible confounders such as obesity. 10 , 11 After natural menopause, most circulating estrogen comes from ovarian androgens that undergo conversion to estrogen in the adipose tissue, and heavier women have higher estrogen levels. Therefore, studying women who underwent bilateral oophorectomy may clarify the general effects of sex hormones on CTS. We used an established cohort study to specifically address the following research questions: 1) Is there an increased risk of severe CTS after bilateral oophorectomy? 2) If so, does the risk differ in women with certain characteristics?

Supplemental Digital Content 1. Table showing the distribution of sociodemographic and clinical characteristics in women with de novo carpal tunnel syndrome (CTS), separately for severe CTS, idiopathic CTS, and any type of CTS. (pdf) Supplemental Digital Content 2. Table showing cumulative incidence of any type of carpal tunnel syndrome overall, and in strata by age at oophorectomy, estrogen therapy, body mass index, parity, and oophorectomy indication. (pdf) Supplemental Digital Content 3. Table showing cumulative incidence of idiopathic carpal tunnel syndrome overall, and in strata by age at oophorectomy, estrogen therapy, body mass index, parity, and oophorectomy indication. (pdf)