genes were discovered in the early 1990s [1
]. In Sweden, oncogenetic clinics were established at university hospitals in the six national healthcare regions (North, Uppsala-Örebro, Stockholm, West, Southeast, and South) as BRCA
screening was implemented for women with hereditary breast and ovarian cancer. Subsequently, each oncogenetic clinic was comprised of a team of oncologists, clinical geneticists, clinical laboratory geneticists and oncology nurses that could provide genetic counseling and screening for patients and families with hereditary breast and ovarian cancer. The oncogenetic clinic at Sahlgrenska University Hospital in Gothenburg, Sweden was established in 1995 and currently serves Western Sweden with a population of 1.9 million inhabitants. Intriguingly, a novel recurrent founder variant in BRCA1
with origins in Western Sweden was discovered during the first years of BRCA
screening at the Clinical Genetics laboratory [3
]. During the first years of BRCA-testing, about 77% of women positive for any disease-causing variant in Western Sweden carried this variant [4
]. A regional follow-up register was initiated for women with BRCA
variants at the oncogenetic clinic to ensure that these women were given access to relevant clinical surveillance programs, and to be able to evaluate the long-term effects of measures, e.g. risk reducing surgery, performed within the cohort. Referral criteria for the assessment of hereditary breast and ovarian cancer has followed previous and current national healthcare programs (https://www.cancercentrum.se
), and are in general similar to international guidelines (https://www.nccn.org
In a prospective cohort study with data from three international consortia, the cumulative risk for breast and ovarian cancer at 80 years of age for women with BRCA
variants were estimated. The risks for BRCA1
-carriers were 72% (95% CI, 65–79%) and 44% (95% CI, 36–53%), and for BRCA2
-carriers 69% (95% CI, 61–77%) and 17% (95% CI, 11–25%), respectively [5
]. Risk reducing mastectomy (RRM) and risk reducing salpingo-oophorectomy (RRSO) substantially reduced these risks. A systematic review of risk reducing surgery in women with BRCA
showed a risk reduction of 90–100% of breast cancer after RRM and 72–88% for ovarian cancer [6
]. One study showed an annual risk of peritoneal cancer after RRSO of 0.20% for women with BRCA1
and 0.10% for women with BRCA2
, and in another study there was a 20-year cumulative risk of peritoneal cancer of 3.5% [7
Further, RRSO substantially reduced the increased mortality, both breast cancer–specific and ovarian cancer–specific mortality and all-cause mortality [6
]. To our knowledge, there are however no reports in the literature showing cancer incidence and mortality among women with BRCA
variants after risk reducing surgery in a defined geographic region in comparison with the general female population in the same region.
National guidelines regarding hereditary breast and ovarian cancer recommends annual diagnostic imaging with breast magnetic resonance imaging (MRI) and breast ultrasound from ages 25 to 55 and mammography from ages 25 to 74 years. RRM has routinely been offered as a choice for those women who wanted a higher degree of risk reduction instead of surveillance. Regarding the ovaries, annual gynecological ultrasound was offered from 30 years of age. But more importantly, RRSO was recommended from 35 years of age for women with BRCA1
variants and from 40 years of age for women with BRCA2
]. During the surveillance period of this study the standard surgical procedure for RRM was a complete removal of breast tissue, initially including removal of the nipple but later changing to nipple sparing technique. The standard surgical procedure for RRSO has been baging and removal of the ovaries and fallopian tubes with laparoscopic techniques.
The aim of the present study was to assess the impact of risk reducing surgery on cancer incidence and mortality during a 20-year period (1995–2016) in a geographically defined population-based cohort of women in Western Sweden with BRCA variants. We evaluated cancer incidence and mortality among cancer-free women who underwent RRM and RRSO in comparison with baseline risks without risk reducing surgery and in comparison with the general female population.
2. Materials and Methods
The study population consisted of women that had tested positive for disease causing variants in BRCA between 1995–2016 at the regional oncogenetic clinic in Western Sweden and were given information about participating in a follow-up register to evaluate the long-term effects of surveillance and risk reducing surgery. In total, 489 women with BRCA variants were included in the register. The register contained information on date of DNA analysis, the identified BRCA variant, date of any previous malignancy, date of annual breast MRIs, ultrasounds, mammograms, date of annual gyneacological ultrasounds, date of risk reducing mastectomy, date of risk reducing oophorectomy, date and type of any cancer diagnosis after registration, date of death, and cause of death. Approval of the register was obtained from the Swedish Central Ethical Review Board (registration number EPN S331-01 2001, EPN S331-01 2017), and all patients included in the register signed a consent form.
Patient clinical data in the register was curated during June–July 2017, compared to medical files, and was supplemented when required. The cohort consisted of the 489 women who were registered as positive for disease causing BRCA
variants during this 20-year period, either via BRCA
screening or tested for a known familial variant. The analytical cohort consisted of the 253 women who were cancer-free when registered (Figure 1
In total, 179 of the 489 women had been diagnosed with breast cancer only before the DNA analysis was performed, 35 had ovarian cancer only and 22 had both breast and ovarian cancer before DNA analysis, leaving 253 cancer-free women at the time of DNA analysis. Consequently, the 253 patients formed the analytic cohort, and consisted of 213 BRCA1- and 40 BRCA2-carriers.
The median age for DNA analysis in this cohort was 36.8 years of age (range, 20.3–78.9). The median age at the end of follow-up was 46.9 years of age (range, 22.2–87.3). All instances of breast and ovarian cancer diagnosed after the DNA analysis, i.e., during the surveillance period, were classified as diagnosed before, at, or after risk reducing surgery. A diagnosis at risk reducing surgery was defined as diagnosed from the final pathology report or within 90 days after surgery. These cases were excluded from the analysis of SIR for breast and ovarian cancer and for the analyses of SMR.
Four of the 253 women had already undergone a bilateral RRM before the time of DNA analysis, and therefore 249 women and 498 breasts were under surveillance after the time of DNA analysis. Fourteen of the 253 women had undergone a RRSO before the time of DNA analysis, and therefore 239 women with ovaries were under surveillance after the time of DNA analysis. When analyzing breast cancer incidence for women with breasts, breast years were calculated from the time of DNA analysis to the first of following events; diagnosis of breast cancer, RRM, death or end of surveillance period, i.e., 31 December 2016. Every breast was accounted for separately. For women who underwent RRM, breast years were calculated from the time of RRM to the first of following events; diagnosis of breast cancer, death or end of surveillance period, i.e., 31 December 2016. The SIR was defined as the observed number of breast cancers during these breast years divided by the expected number of cases, using incidence rates from the Swedish female population stratified for 5-year age groups (0–4, 5–9, … 80–84, 85–) and calendar year.
When analyzing ovarian cancer incidence for women with their ovaries intact, person years were calculated from the time of DNA analysis to the first event; ovarian cancer, RRSO, death or end of surveillance period. For women who underwent RRSO, person years were calculated from the time of RRSO to the first event; peritoneal cancer assessed to be of ovarian origin, death or end of surveillance period. SIR was defined as the observed number of peritoneal cancers of ovarian origin during these person years divided by the expected number of cases, using incidence rates from the Swedish female population stratified for 5-year age groups (0–4, 5–9, … 80–84, 85–) and calendar period. Incidence rates for breast and ovarian cancer produced by the NORDCAN project [11
] were used. The coding of cancer in our cohort was the same as used by the NORDCAN project and followed definitions according to International rules for multiple primary cancers [12
Events of peritoneal cancer of unknown origin were not included in the analyses of SIR and SMR but were described in the results section. The reason for this was that the incidence rate of peritoneal cancer in the background population is uncertain.
When analyzing the mortality rate for all women in the cohort, person years were calculated from the time of DNA analysis to death or end of surveillance period. When analyzing the mortality rate for women before any risk reducing surgery, person years were calculated from the time of DNA analysis to bilateral RRM or RRSO or death or end of surveillance period. When analyzing the mortality rate for women after bilateral RRM or after RRSO, person years were calculated from bilateral RRM or RRSO to death or end of surveillance period. Lastly, when analyzing the mortality rate for women after both bilateral RRM and RRSO, person years were calculated from the last risk reducing surgery event to death or end of surveillance period. The SMR was defined as the observed number of deaths during these person years divided by the expected number of deaths, using rates from the Swedish female population stratified for 5-year age groups (0–4, 5–9, …, 80–84, 85–) and calendar year.
The stset, stsplit and strate macros in STATA 15.1 were used for the incidence analyses [13
This population-based follow-up study was restricted to a defined geographic region, Western Sweden, with a cohort of women with BRCA
variants, primarily registered at the time of genetic testing and thereafter prospectively followed regarding cancer incidence, risk reducing surgery and death between 1995–2016. In concordance with previous studies, we observed a reduction in breast and ovarian cancer incidence after RRM and RRSO as well as mortality [6
]. However, the overall mortality rates for women with BRCA
variants were still significantly increased compared to age matched-women in the general population, even after risk reducing surgery, SMR 4.32 (95% CI, 1.62–11.5). As far as we know, a persistent increase in SMR even after risk reducing surgery has not been previously shown in a study with a population-based cohort from a geographically defined region. These findings emphasize the need for further research to identify additional complementary measures than risk reducing surgery for high-risk women with BRCA
For example, a more precise and individualized risk evaluation combining genetic screening for polygenic risk score in combination with hormonal factors and family pedigree could influence the woman to take preventive measures; engaging in surveillance programs or having risk reducing surgery, in time [14
]. Further, ovarian cancer may arise in the fallopian tubes [15
] which raises the question if sequential removal of tubes and ovaries would decrease their risk of ovarian cancer. PARP-inhibition and antihormonal drugs may with the right timing help in preventing cancer incidence and mortality [16
]. RRSO for premenopausal women results in an immediate onset of menopause, which can bring the physical and emotional symptoms of natural menopause, and there is also an elevated risk of osteoporosis and probably also cardiovascular disease [17
]. Hormone replacement therapy can be offered to women after RRSO to partly counteract the symptoms. Still, the negative aspects of early-induced menopause, including the impact on quality of life, are not fully known [18
] and further preventive studies among these patients is warranted. Furthermore, early detection of cancer as well as early detection of a relapse of treated cancer through biomarkers such as circulating tumor cells, cell free tumor DNA and micro-RNA in blood may improve the possibilities for earlier intervention and treatment [19
In our cohort, we observed two cases of breast cancer among 80 women who had a bilateral RRM. Furthermore, we observed three cases of peritoneal cancers of ovarian origin after RRSO among 136 women. This is in line with studies showing no or small remaining risks for cancer [6
] after risk reducing surgery. Patients as young as 30–34 and 35–39 years of age at the time of diagnosis were identified with breast and ovarian cancer, respectively. Current clinical guidelines recommend starting surveillance of the breasts at age 25 and for the ovaries at age 30, which was in concordance with the findings in this study. There is no age recommendation for RRM, but a firm recommendation to consider RRSO from 35 years of age for BRCA1
and 40 years of age for BRCA2
. These recommendations were thus adequate in relation to the ages of cancer diagnoses in our cohort.
In this study cohort of 253 women, 213 had BRCA1
variants and only 40 had BRCA2
variants, which affects the generalizability as BRCA1
is associated with a higher risk of both breast and ovarian cancer. Further, there is in the literature evidence for variant genotype–phenotype correlations in BRCA
]. Four out of 10 women with BRCA1
variants carried the founder variant c.3171ins5 with an origin in Western Sweden, but if there is a correlation between this very variant and the breast and ovarian cancer ratios, is not known. BRCA
variants may be associated with other cancers such as pancreas and colon cancer, in particular BRCA2
variants are associated to a relatively low increase in risk of pancreas cancer [22
]. Those two cancer diagnoses were absent in our analytical cohort.
All women in the region of Western Sweden with a BRCA variant were asked to participate in a prospective follow-up register study. There were only a few lacking consents and the genetic counselors did not perceive the lacking consents to be related to whether the patient had cancer or risk reducing surgery. The high degree of accepting registration and surveillance in our cohort may depend on the fact that healthcare in Sweden is mainly financed by taxes, access to healthcare is subsidized or free for the patient, and the healthcare region is responsible for funding and providing healthcare services, including this type of surveillance program, risk reducing surgery and cancer treatment.
A strength of this study was the prospectively-followed primarily registered cohort of 489 women with BRCA variants in a defined geographic region during the period 1995–2016. Follow up was complete and the medical exams during surveillance as well as risk reducing surgery were well described in the register. Limitations of the study was the limited number of person years of follow-up and limited time of follow-up due to the majority of the women entering the study during the latter part of the inclusion period, at which time genetic testing gradually became more common. This resulted in wide confidence intervals for the estimates.