Female BRCA1 germline mutation carriers have a cumulative risk of 44% to develop ovarian cancer before the age of 80 years. For BRCA2 mutation carriers, this cumulative risk is 17% [1]. In the absence of an effective screening method for ovarian cancer, risk-reducing salpingo-oophorectomy (RRSO) is advised at an age when there is a high risk of developing ovarian cancer (before the age of 35-40 years for BRCA1 mutation carriers and at the age of 40-45 years for BRCA2 mutation carriers) provided that the woman no longer has a wish to have children [2,3]. During this procedure, both ovaries and fallopian tubes are removed. Since 2009, the uptake of RRSO in BRCA1/2 mutation carriers in the Netherlands has increased greatly from 81% to 98% currently [4-6]. While RRSO at the recommended age decreases the risk of ovarian cancer by around 96% [7,8], it also initiates premature surgical menopause owing to a sudden drop in the levels of ovarian hormones. Whether RRSO also reduces the risk of breast cancer is currently debated. Various early studies have reported a risk reduction of 50%, but this is a substantial overestimation owing to bias [8-11].

The relationship between early menopause (either surgically induced or due to premature ovarian insufficiency [POI]) and noncancer health outcomes, including cardiovascular disease (CVD), osteoporosis, cognition, and health-related quality of life (HRQoL), has been examined previously. However, most of the published studies had a limited follow-up (median <5 years), which is too short to establish an association with long-term health effects [12-15]. Additionally, hormone replacement therapy (HRT) use, cancer treatment, and lifestyle may alter the potential association of RRSO with noncancer health outcomes. BRCA1/2 mutation carriers may have been treated for breast cancer, which can be cardiotoxic and potentially have a negative impact on the brain [16-19]. Lastly, women with a high familial risk of breast/ovarian cancer may have a healthier lifestyle than age-matched women without a family history of breast cancer, rendering translation of results of previous research on early menopause to this specific population difficult [20].

Before the age of 50 years, female sex hormones, such as estrogen and progesterone, are likely to protect women against CVD as they have atheroprotective properties [21,22]. Previous studies have shown that women with POI are at higher risk of developing CVD and that earlier age at menopause is associated with a higher cardiovascular mortality rate [23-26]. Additionally, a higher risk of CVD has been observed after surgical menopause compared with natural menopause [27]. However, age at surgical menopause was often not specified or surgery was performed at later ages than the recommended young ages for RRSO in BRCA1/2 mutation carriers. Thus, information about the cardiovascular risk of early surgical menopause is still limited [27-29]. Furthermore, it is possible that early loss of ovarian hormones is not the cause of the higher cardiovascular disease risk in women with POI. The reverse causality hypothesis postulates that POI is the result of accelerated vascular aging, which would also explain a statistical (noncausal) association between earlier natural menopause and increased CVD risk [30]. If this is true, no elevated CVD risk would be expected in women with early surgical menopause due to RRSO.

Reproductive hormones play a role in maintaining bone health. Estrogen suppresses bone resorption and stimulates bone formation [31]. Postmenopausal women aged 50 to 55 years have a higher incidence of hip fractures than age-matched premenopausal women, and among postmenopausal women, the incidence of hip fractures rises with age [32]. However, data on the long-term effect of early (surgical or natural) menopause on bone health are limited owing to short follow-up periods.

The neuroprotective effects of estrogens, progestogens, and androgens have been described in several studies [33]. Rocca et al observed that women who underwent premenopausal oophorectomy (before 56 years of age) for benign reasons had a higher risk of cognitive impairment than age-matched women without oophorectomy [34]. There was a trend for more severe cognitive impairment with younger age at oophorectomy for both unilateral and bilateral oophorectomies. However, information on BRCA1/2 mutation carriership was not provided [34]. Additionally, Bove et al observed a faster decline in global cognition in women with an earlier age at surgical menopause [14]. Thus, while there are some data on the effect of surgical menopause on cognition, research on BRCA1/2 mutation carriers is limited.

RRSO may have an important impact on HRQoL, including generic issues, such as physical, role, and emotional functioning, as well as cancer-related anxiety, sexual functioning, and menopause-specific symptoms [12,35,36]. However, it is not known whether these effects persist over time. A recent study with a mean follow-up of 7.9 years after RRSO found that moderate to severe menopausal complaints were still present in almost 70% of women who underwent premenopausal RRSO, without any improvement over time [37]. The magnitude of the effect of RRSO on sexual functioning and endocrine symptoms is modulated by HRT use [38,39]. However, HRT use is contraindicated for BRCA1/2 mutation carriers with a history of breast cancer.

The number of postmenopausal years is associated with an increased prevalence of vulvovaginal atrophy [40]. Atrophy of the urethral epithelium may lead to urinary incontinence. To our knowledge, no research has been conducted on the association between early surgical menopause and urinary incontinence, although this generally has a severe impact on sexual functioniƒng and quality of life [41,42].

Figure 1 provides a schematic overview of the effects of a decrease in estrogen levels due to menopause on various organ systems.

A recent Cochrane systematic review on the benefits and harms of RRSO in women with a BRCA1/2 mutation stated that no conclusions can be drawn regarding bone fracture incidence and HRQoL, and that further research is needed. CVD and cognition were not included as outcomes in this review, probably because of the paucity of studies [43]. Therefore, the aim of the HARMOny study (Health After eaRly Menopause due to Oophorectomy) is to investigate the long-term health effects of RRSO. We will conduct a cross-sectional study with prospective follow-up nested in a large Dutch cohort. We hypothesize that women with surgically induced premature menopause are at higher risk of subclinical CVD, compromised bone health and osteoporosis, reduced HRQoL, and impaired cognitive functioning, compared to age-matched women with natural menopause after the age of 50 years. Depending on the results, this may lead to active surveillance of the above health problems to improve long-term health and HRQoL in women who have undergone early RRSO.

The primary aim of our study is to assess the long-term health effects of RRSO on subclinical CVD, bone mineral density (BMD), and cognitive functioning in women at increased familial risk of breast or ovarian cancer mainly due to a BRCA1/2 mutation. Our secondary study aims are to assess overall HRQoL, endocrine symptoms, such as hot flashes and night sweats, sexual functioning, urogenital problems, and the prevalence of cardiovascular risk factors.

We will include 500 women who have undergone RRSO before the age of 45 years, with a follow-up period of at least 10 years (the early RRSO group). We will compare them with 250 women from the same cohort (frequency matched on current age) who have not undergone preventive RRSO or have undergone RRSO at the age of 55 years or above (the late/non-RRSO group). The exclusion criteria are a metastatic disease; a serious physical comorbidity or psychiatric disorder; a language barrier; a metal cardiac valve; a bare metal coronary stent, since the reflection of the stent would interfere with the computed tomography scan; and natural POI before the age of 40 years. A history of nonovarian cancer, including breast cancer, is not a reason to be excluded.

Eligible women who are alive will be invited for a clinic visit and questionnaire survey that will assess CVD status, BMD, cognitive functioning, and HRQoL. For eligible women who have died, we will collect information on the causes of death to examine selection bias.

Participants will be selected from the dynamic nationwide HEBON study cohort, a national collaboration on HEreditary Breast and Ovarian cancer in the Netherlands, with women tested for BRCA1/2 germline mutations included since 1997 by all eight Dutch University Medical Centers (UMCs) and the Netherlands Cancer Institute (NKI). The cohort currently consists of over 3600 proven female BRCA1/2 germline mutation carriers. Participants have given informed consent for linkage with the Netherlands Cancer Registry and the Dutch Nationwide Pathology Database (PALGA), which can provide information on all prophylactic mastectomies and salpingo-oophorectomies performed in the Netherlands since 1991. This provides us with nearly complete information on the occurrence of RRSO. Data on lifestyle, family history, reproductive factors, and hormone use have been collected from HEBON participants through questionnaires [44].

Figure 2 provides a schematic overview of the patient selection and eligibility criteria of our study population.

Participants will be asked to complete an online questionnaire and an online cognitive assessment, and they will be asked to visit the outpatient clinic for a cardiovascular and bone health assessment. We will assess several surrogate endpoints of cardiovascular health, that is, coronary artery calcium (CAC) score (CAC scoring), pulse wave velocity (PWV), advanced glycation end products (AGEs), anthropometric measurements, and blood sampling. We will determine nonfasting blood levels of lipids, glucose, HbA_1c, high-sensitivity cardiac troponin, and high‐sensitivity C‐reactive protein. BMD is assessed by a dual-energy X-ray absorptiometry (DXA) scan and laboratory blood sampling, and cognitive functioning is assessed by the Amsterdam cognition scan (ACS).

The online questionnaire is divided into the following several topics: general cardiovascular history, family history of CVD, traditional CVD risk factors (ie, diabetes, hypertension, and cholesterol), lifestyle (ie, smoking, alcohol use, and physical activity), female-specific risk factors (hypertensive pregnancy complications, gestational diabetes, contraceptives use, and HRT use, ie, type, duration, and timing), history of fractures, calcium intake, vitamin D supplement use, glucocorticoid use, family history of osteoporosis/fractures, history of inflammatory diseases (rheumatoid disease and thyroid disorders), and perceived cognitive problems. We will assess generic HRQoL with the SF-36 health survey [45,46] and the body image items of the European Organization for Research and Treatment of Cancer (EORTC) QLQ-BR23 [47,48]. Cancer worries will be measured with the eight items adapted from Lerman et al [49-51]. To assess menopausal symptoms, we will employ the 18-item Functional Assessment of Cancer Therapy-Endocrine Subscale (FACT-ES) and the Hot Flush Rating Scale (HFRS) [52,53]. To assess urogenital problems, the Urogenital Distress Inventory (UDI-6) and Incontinence Impact Questionnaire (IIQ-7) [54] will be used. The Sexual Activity Questionnaire (SAQ) [55] will be used to assess sexual functioning. For patients with a breast cancer history, either before or after RRSO, we will collect information on radiotherapy (yes/no and radiation fields), chemotherapy and immunotherapy regimens, and endocrine treatment from medical records.

Measurements of the CAC score as calculated by coronary calcium scoring will be performed according to standardized local protocols at the various UMCs using the Agatston score (AS) and the Multi-Ethnic Study of Atherosclerosis (MESA) coronary heart disease risk score calculator [56,57]. A disadvantage of CAC scoring might be the use of ionizing radiation. However, the radiation dose in the CAC protocol is as low as 1 mSv, which is not expected to be harmful [58]. All patients in our study are aged 55 years or above; thus, the radiation-induced risk of breast cancer from CAC scoring will be negligible with the current protocol [59,60].

The PWV will be assessed with an ambulatory arteriograph (TensioMed Kft) by the research physician. The arteriograph uses an oscillometric occlusive method, with an upper arm cuff to measure the time interval between the peak of the first systolic wave and the peak of the reflected systolic wave (return time). The device has been validated in numerous studies [61-65].

With aging, the end products of nonenzymatic modification of proteins, lipids, and nucleic acids (AGEs) accumulate endogenously in the serum and tissues. Levels of AGEs in the skin are measured noninvasively by standardized autofluorescence and can predict future cardiovascular events [66,67]. Skin autofluorescence is a quick and noninvasive surrogate marker of tissue accumulation of AGEs [68-70]. AGEs will be measured using the AGE Reader (DiagnOptics Technologies BV; software V 2.3.0.7) on the forearm.

Bone health will be assessed by a DXA scan, a vertebral fracture assessment, and blood levels of bone turnover markers (BTMs) for osteoclast and osteoblast activity (beta-carboxy-terminal collagen crosslinks [β-CTX] and N-terminal procollagen type 1 [P1NP]). For logistic reasons, it is not possible to get fasting blood samples for every woman. Since β-CTX values are only valid in fasting blood samples, we will assess this BTM only in those women for whom we have fasting blood samples. P1NP will be measured in all samples.

Lastly, women will receive an invitation to fill out the online ACS, which has recently been developed and validated [71]. The ACS is an easy to use tool to obtain online measures of various cognitive abilities [72]. The online test battery is based on seven traditional neuropsychological tests and covers the following domains: verbal memory, attention, executive functioning, information processing speed, and motor functioning.

Table 1 provides an overview of the tests that will be performed [58,61-63,66,67,69,71-77]. Table 2 provides an overview of the questionnaires used [45-50,52-55].

We will examine the effect of age at RRSO on the risk of breast cancer, contralateral breast cancer, and ovarian cancer, as well as associations between RRSO and the prognosis of subsequent breast cancer and ovarian cancer. Additionally, after completion of the cross-sectional study, we will perform longitudinal follow-up every 4 to 5 years to evaluate incidence rates of outcomes of major interest, such as ischemic heart disease, and evaluate changes in outcomes (CAC, PWV, BMD, and cognitive functioning as assessed by the ACS) over time between the early RRSO group and the late/non-RRSO group. Moreover, with the anticipated prospective follow-up of our study population, we will look at mortality of cardiovascular disease.

All data, including blood and DNA samples, will be stored for 30 years to enable prospective longitudinal follow-up of the study population on the long-term health effects of RRSO. The stored blood samples will allow for future international studies, for example, those on the modifying effects of genetic factors, such as single nucleotide polymorphisms, on the outcomes of interest, and the novel cardiovascular markers or BTMs for these outcomes. All blood samples will be frozen and stored at −80°C at the NKI.

We will analyze differences in the cardiovascular risk profile, BMD, and neurocognitive functioning between the early RRSO group and the late/non-RRSO group using multivariable linear regression analyses and logistic regression analyses with clinically relevant cutoff points for the outcomes of interest. The analyses will be adjusted for confounders, including lifestyle, reproductive factors, and medication use, where applicable. For the analysis of patient-reported outcomes, such as HRQoL, sexual functioning, and menopausal symptoms, we will use mixed effect models.

We expect approximately 40% of the women to have a breast cancer history and will perform the analyses separately in women with and those without a breast cancer history, as breast cancer treatment may affect outcomes. Additional subgroup analyses will be performed according to HRT use (yes/no), BRCA1/2 mutation status, age at RRSO (<40 years/≥40 years), and time since RRSO (10-15 years/>15 years).

We based our sample size calculations for cardiovascular status on subclinical atherosclerosis by assessing the differences in the CAC score between women in the early RRSO group and women in the late/non-RRSO group. We chose a clinically relevant cutoff for the CAC score of 100 Agatston units (AU). According to the MESA trial, 10% of subjects between 55 and 65 years of age exceed this value [78]. For sample size estimation, we hypothesized that 20% of the early RRSO group exceeds this value of 100 AU. In logistic regression with an elevated CAC score (>100) as the outcome and with 250 women in the late/non-RRSO group and 500 women in the early RRSO group, we have 80% power to detect an odds ratio of 1.9 at an α value of 5%.

With 750 women in our study, we will be able to detect a difference of 0.07 g/cm² in BMD between the early RRSO and late/non-RRSO groups, as observed between women with premature and normal menopause and assuming an SD of 0.15 [79]. Statistical power for analysis of cognitive functioning is based on a previous neuropsychological study on the effects of oophorectomy on cognitive impairment and dementia [34]. This study reports relevant differences in 427 women with bilateral oophorectomy before the age of 49 years, compared to age-matched women without oophorectomy, with a median follow‐up of 25 years. As our neurocognitive tests are more sensitive than the tests screening for dementia, we expect to be able to detect smaller differences in cognitive functioning and distinguish between different cognitive domains, and with 750 participants, we have 90% power to detect an effect size of 0.25 with a P value set at .05 (two tailed). To detect a relevant difference between groups for the various HRQoL measures, we will use a half SD (an effect size of 0.5) to define clinically relevant group differences (over time). This requires about 67 patients per group. The available sample will be more than adequate to conduct subgroup analyses as well.