Outcomes From Opportunistic Salpingectomy for Ovarian Cancer Prevention

Key Points Question Is opportunistic salpingectomy associated with fewer than expected ovarian cancers? Findings This population-based cohort study included 25 889 individuals who underwent opportunistic salpingectomy and 32 080 individuals who underwent hysterectomy alone or tubal ligation. There were no serous ovarian cancers among individuals in the opportunistic salpingectomy group, which was significantly lower than the age-adjusted expected rate of 5.27 serous cancers. Meaning The opportunistic salpingectomy group had significantly fewer serous ovarian cancers than expected, suggesting that opportunistic salpingectomy is associated with reduced ovarian cancer risk.


Introduction
Approximately 70% of sporadic and nearly all ovarian cancers in BRCA variant carriers are high-grade serous carcinomas (HGSCs), 1 which is the most lethal of the 5 main histotypes and has a 5-year survival rate less than 50%. 2 Although the general population lifetime risk of ovarian cancer is 1.4%, 3 individuals with an inherited germline BRCA1 or BRCA2 variant have average cumulative risks of 40% to 75% and 8% to 34%, respectively. 4 In BRCA 1/2 variant carriers, bilateral salpingo-oophorectomy is recommended, which reduces the risk of ovarian or fallopian tube cancers by 80%. 5 Removal of the ovaries is not recommended for the general population because it is associated with increased total mortality, coronary heart disease, and osteoporosis. 6 Thus, a different preventive strategy is needed for individuals at average risk, who account for 80% of cases of HGSCs.
The recent understanding that HGSC often originates in the fallopian tube 7,8 has led to a primary prevention opportunity for the general population-namely, opportunistic salpingectomy (OS). OS collectively refers to the removal of the fallopian tubes at the time of hysterectomy or instead of tubal ligation, while leaving the ovaries intact. In 2010, the British Columbia (BC) ovarian cancer research team launched a province-wide strategy asking gynecologists to discuss OS with their patients as an ovarian cancer prevention strategy. The same recommendation has since been made in many countries, including Canada, the US, and the UK for individuals without identified genetic factors associated with increased risk of ovarian cancer. [9][10][11][12] Research has shown that OS is safe, both in terms of perioperative adverse events 13 and minor complications, 14 there are no indications of an earlier age of onset of menopause following OS, 15 and it is cost-effective. 16 Some retrospective data from individuals who underwent bilateral salpingectomy for conditions such as hydrosalpinx and pelvic inflammatory disease have suggested decreased risk of ovarian cancer among individuals without fallopian tubes. [17][18][19] However, we hypothesize that OS is associated with more protection than salpingectomies done for diseases that directly affect and distort the fallopian tube, because the intent of OS is complete removal of the fimbriated end of the fallopian tube, which may not occur when salpingectomy is done for other indications. Finally, these historical studies did not use the appropriate control groups, as OS is recommended only for individuals already undergoing gynecological surgery, and both hysterectomy and tubal ligation are associated with protection against ovarian cancer. 20 Here, we examine observed rates of ovarian cancer and compare these with expected rates (based on age-adjusted rates of ovarian cancer in the control group of individuals who underwent hysterectomy alone or tubal ligation).

Methods
This population-based, retrospective, cohort study examined data on all residents of the Canadian province of BC (population, 5 million). All individuals who underwent a hysterectomy or tubal sterilization in BC between 2008 and 2017 were included. Approvals were obtained from all relevant data stewards, and access to the Consolidation file, the BC Cancer Registry, the Discharge Abstract Database, and the BC Cancer Agency Screening Program was facilitated through Population Data BC.
More details, including citations to data sources, are presented in the eTable in the Supplement.

Ethics approval was obtained from the University of British Columbia's Behavioral Research Ethics
Board. Approval by the ethics board and the BC data stewards for use of deidentified administrative data files includes a waiver of informed consent from participants. This study follows the The grade data were incomplete, and all serous cancers were presented together rather than as low grade and high grade; however, 95% of serous cancers are HGSCs. To examine whether differences in the observed and expected ovarian cancers in the OS group might be explained by underlying differences in the likelihood of getting cancer associated with health status, lifestyle, genetics, and so forth, differences in observed and age-adjusted expected numbers of breast cancer (ICD-O code C50) and colorectal cancer (ICD-O code C18.X) but excluding cancer of the appendix (ICD-O code C18.1) were also examined.

Statistical Analysis
For privacy reasons, data steward agreements require that we not publish cell sizes between 1 and 5.
The number of observed epithelial and serous ovarian cancers in the OS group are presented according to privacy requirements. We also present the number of observed breast and colorectal cancers. These observed numbers were then compared with the number expected on the basis of age-adjusted (in 5-year age groups) rates in the control group multiplied by the person-time contribution in the OS group. Given the low number of ovarian cancers in both groups, statistical models were not run. Instead the distribution of the potential confounders across the OS and control groups and their standardized differences were presented. A difference between covariates was considered meaningful if the standardized difference was greater than 0.1. 23 The age-adjusted rates of serous ovarian cancers were also used to project the number of expected serous ovarian cancers in the OS group 5 and 10 years beyond our study period if serous ovarian cancers were to occur at the same rate as in the control group. To project, we used data from

Discussion
The realization that the fallopian tube fimbriae is the tissue of origin for most HGSCs opened the door for OS as a primary ovarian cancer prevention strategy. 8 Prevention of ovarian cancer seems more critical today than ever as the largest screening trial found that although a stage shift was achieved, there was no mortality benefit. 24 In BC in 2010, a population-wide primary prevention campaign was initiated to remove the fallopian tubes of individuals at general population risk for ovarian cancer when they were undergoing hysterectomy for benign indications or seeking tubal ligation. The acceptability, safety, and cost-effectiveness of this OS campaign has already been established. [13][14][15][16] In this cohort study, we now present data strongly suggesting that OS is effective as an ovarian cancer primary prevention strategy at the population level. There was not a single serous ovarian cancer in the OS group, which was significantly fewer than the slightly more than 5 that were expected. We have further shown that the OS group had the same risk of breast and colorectal cancers compared with the control group, indicating that the lack of ovarian cancers in the OS groups is unlikely to be associated with selection bias. The rates of common risk and protective factors for the OS group place them at slightly higher risk of ovarian cancer (eg, lower parity, lower gravidity, and higher age), indicating that our results are unlikely to be explained by confounding.
There were 15 serous cancers observed in the control group, and our calculations show that as we continue follow-up, there will be 45.1 serous ovarian cancers in this group by 2027. It is difficult to determine the preventable fraction given that we did not observe any serous cancers in our OS group. The least conservative interpretation would be that OS prevents all serous cancers, but more  Year 2017 Projection to 2022 Projection to 2027 Error bars denote 95% CIs. Our findings are consistent with previous epidemiological research from the US, Denmark, and Sweden, [17][18][19] which studied the association between excisional tubal surgery or salpingectomy and the risk of ovarian cancer among individuals with a medical indication for these procedures. The observed relative risks showed a 42% to 65% reduction in risk of ovarian cancer for individuals who underwent a major surgical procedure involving the fallopian tube. [17][18][19] Furthermore, tubal ligation has long been recognized as being inversely associated with ovarian cancer risk, although the magnitude of the association is lower than those for the more extensive fallopian tube procedures mentioned already. 35

Limitations
Our work has some important limitations, including that these are observational data and not derived from a randomized clinical trial; thus, selection factors could introduce bias. However, the Table   illustrates that there are few differences between the OS group and control group with respect to the most well-known risk and protective factors for ovarian cancer (eg, parity, oral contraceptive pill use, BRCA variant status, and endometriosis), and the differences that do exist would bias toward increased risk for ovarian cancer in the OS group. Although it remains possible that there are important unmeasured differences between the groups, such as lifestyle factors that are associated with cancer, the findings of no difference between the observed and expected numbers of breast and colorectal cancers in the OS group suggest that selection bias is unlikely to explain these results.
The study was limited by the small number of cancers and relatively short follow-up time in our groups. Although uptake of OS has been substantial in BC, the province has a relatively small population (approximately 5 million), and the surgical procedures at which OS is performed occur at young mean ages. Thus, our numbers of ovarian cancers were small, making it impossible to run Cox proportional hazards models controlling for potential confounders. In addition, although the preliminary data suggest that there are no indicators of an earlier age of onset of menopause, it is time to conduct a long-term follow-up study on the age of onset of menopause as self-reported by those who undergo OS or a control surgery. Given the reduction in observed ovarian cancer compared with expected, which strongly supports reduced risk of ovarian cancer following OS, it is important to ensure OS does not alter the age of onset of menopause.