Background Women testing positive for BRCA1/2 pathogenic variants have high lifetime risks of breast cancer (BC) and ovarian cancer. The effectiveness of risk reducing surgery (RRS) has been demonstrated in numerous previous studies. We evaluated long-term uptake, timing and effectiveness of risk reducing mastectomy (RRM) and bilateral salpingo-oophorectomy (RRSO) in healthy BRCA1/2 carriers.
Methods Women were prospectively followed up from positive genetic test (GT) result to censor date. χ² testing compared categorical variables; Cox regression model estimated HRs and 95% CI for BC/ovarian cancer cases associated with RRS, and impact on all-cause mortality; Kaplan-Meier curves estimated cumulative RRS uptake. The annual cancer incidence was estimated by women-years at risk.
Results In total, 887 women were included in this analysis. Mean follow-up was 6.26 years (range=0.01–24.3; total=4685.4 women-years). RRS was performed in 512 women, 73 before GT. Overall RRM uptake was 57.9% and RRSO uptake was 78.6%. The median time from GT to RRM was 18.4 months, and from GT to RRSO–10.0 months. Annual BC incidence in the study population was 1.28%. Relative BC risk reduction (RRM versus non-RRM) was 94%. Risk reduction of ovarian cancer (RRSO versus non-RRSO) was 100%.
Conclusion Over a 24-year period, we observed an increasing number of women opting for RRS. We showed that the timing of RRS remains suboptimal, especially in women undergoing RRSO. Both RRM and RRSO showed a significant effect on relevant cancer risk reduction. However, there was no statistically significant RRSO protective effect on BC.
- Genetic Predisposition to Disease
Data availability statement
All data relevant to the study are included in the article or uploaded as supplementary information.
Statistics from Altmetric.com
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.
It is well established that pathogenic variants (path_variants) in BRCA1 and BRCA2 (BRCA1/2) genes are strongly associated with a high lifetime risk of developing breast (BC) and/or ovarian cancer (OC). While the lifetime risk of BC and OC in the general population is 10%–12.5% and 1.5%–2%, respectively, various studies have shown that path_variants in the BRCA1/2 genes confer cumulative lifetime risk for developing BC of 35%–87% and 15%–60% for developing OC.1–3 Therefore, substantial effort has been deployed into developing specific surveillance and risk reducing strategies worldwide. Current risk management options for women with a BRCA1/2 path_variant include regular surveillance, chemoprevention and risk reducing surgery (RRS): risk reducing mastectomy (RRM), risk reducing salpingo-oophorectomy (RRSO) or both.4 While breast surveillance, including regular breast self-examination, clinical breast examination, mammography and breast MRI, is effective for early BC detection,5 it is not effective in cancer risk reduction. Moreover, there is no effective surveillance for early detection of OC that decreases mortality, as regular transvaginal ultrasound and CA125 estimation are ineffective.6 7 On the other hand, there is considerable evidence that RRS reduces BC8–10 and OC8 11–13 risks in BRCA1/2 path_variant carriers, although with negative consequences related to the surgery itself.14 Consequently, RRSO is widely recommended for BRCA1/2 path_variant carriers to reduce their risks of developing OC. In addition, RRM is discussed in depth and offered as an option for BRCA1/2 path_variant carriers, taking into account the efficacy of early detection of BC.
Risk-reducing surgeries in BRCA1/2 carriers have been shown to be both clinically effective15 and cost effective.16 A meta-analysis of four prospective trials involving 2635 patients on the efficacy of RRM in BRCA1/2 path_variant carriers demonstrated a significant risk reduction in BC incidence (HR=0.06, 95% CI 0.01 to 0.41, p=0.005).17 The efficacy of RRSO was demonstrated by meta-analysis of three prospective trials involving 9192 BRCA1/2 path_variant carriers with a significant risk-reduction in OC/primary peritoneal cancer incidence after surgery (HR=0.19, 95% CI 0.13 to 0.27, p<0.00001).13
Filippo-Morton and colleagues found a steady and significant increase in the uptake of RRS since 2006, reported in the literature.18 The growing proportion of women undergoing RRM (18%–50%) and RRSO (27%–78%) is thought to be related to improved surgical and reconstructive options and techniques, better education of BRCA1/2 path_variant carriers and greater availability of genetic testing (GT) and counselling commensurate with cancer diagnosis.18 Moreover, women choosing one RRS are significantly more likely to then choose to undergo the other.19
A recent study of 87 cancer-free BRCA1/2 positive women found a 59% uptake of RRS with a median follow-up time to RRS of 4.8 months (median time to RRM=7.5 months; median time to RRSO=4.7 months; total follow-up time=30.4 months).18
Women are opting for early bilateral RRSO as a combined measure to reduce both BC and OC risk, since many studies containing retrospective data have demonstrated an association between RRSO and BC risk reduction.8 20–22 However, a prospective study by Heemskerk-Gerritsen et al found no evidence of a short-term protective effect on BC risk after RRSO in BRCA1/2 path_variant carriers.23
In the present study, we summarise the uptake, timing and effectiveness of RRM and bilateral RRSO in a large cohort of women with high lifetime risks of BC and OC who were unaffected by either cancer at positive GT for a BRCA1/2 path_variant in the northwest of England.
Materials and methods
Healthy women (unaffected with BC/OC) with a confirmed family history for a BRCA1/2 path_variant are offered a targeted test, testing only for their familial variant. In contrast, unaffected women with strong family histories of BC and/or OC are offered full BRCA1/2 path_variant screening if an affected family member is unavailable for testing and their a priori likelihood of a pathogenic BRCA1/2 variant is ≥10%.
Women with a lifetime BC risk >25%, including BRCA1/2 path_variant carriers, have been offered a discussion about bilateral RRM since 1994. BRCA1/2 path_variant carriers are also encouraged to undertake RRSO once their family is complete, ideally before 40 years of age for BRCA1 and 45 years for BRCA2.
We used a prospective cohort design to evaluate the long-term uptake of RRM and RRSO among BRCA1/2 path_variant carriers, tested at the Manchester Centre for Genomic Medicine during the period November 1994–March 2019. The individuals were identified from the prospectively maintained Manchester Genetic Medicine Database (North Manchester Research Ethics Committee (reference 08/H1006/77)).
A total of 887 women, born between 1930 and 2002 (median=1971), with a positive presymptomatic test for BRCA1/2 gene path_variants were included in our study. Only women without previous BC/OC diagnoses were included. GT was performed as per institutional guidelines and all patients received post-testing counselling to review medical management options, including RRS. We have included all RRM surgeries, regardless of type and technique used. Women participating in the MRI programme who developed BC and had contralateral RRM were described, but not included in our calculations. The characteristics of our study population are described in table 1.
We retrieved data on date of birth, parity, date of individual DNA test result, gene with a path_variant (BRCA1 or BRCA2), date of RRM/RRSO, date of BC and/or OC diagnosis and tumour characteristics, date and cause of death. Follow-up was started from GT or 25th birthday (whichever was later—we used 25 years as cancer incidence before this age is very low) and censored at date of BC or OC diagnosis, date of death (DOD), whichever was earliest for each cancer. Time-dependent analysis was used for calculations—women were treated as unexposed to surgery before RRM/RRSO and exposed after they had RRM/RRSO. Women who underwent RRS were censored for follow-up for the relevant cancer at date of surgery but were assessed prospectively for cancer incidence and death. All individuals without censorship were counted as unaffected on prevalence day 01/03/2019.
The statistical analyses were performed using the IBM SPSS Statistics 20.0 package. A χ² test was used to compare categorical variables. A Cox regression model was used to estimate the HR and 95% CI of BC and OC cases associated with RRS, and the impact of RRS on all-cause mortality. Kaplan-Meier curves were used to estimate cumulative percentage uptake of RRS. Patients were censored at time of RRM and RRSO. The uptake of RRS among different age groups was compared using log rank test and ORs with 95% CI. Women who underwent RRS prior to receiving BRCA1/2 test were excluded from Kaplan-Meier analysis. The annual BC incidence (%) was estimated by calculating the proportion of total women-years for the entire study population and the number of BC cases in the follow-up period. All p values were based on two-sided tests and were considered statistically significant if <0.05.
In total, 887 women underwent predictive BRCA1/2 testing and tested positive for the path_variant: 442 (49.8%) for BRCA1 and 445 (50.2%) for BRCA2 (table 1). The highest proportion of those testing positive was observed in 30–39-year-old women; this group was described as a reference group for further analysis. The mean age at individual DNA test result was 39.5 years (range=15.3–85.3 years). In total, 70 were tested positive before age 25 and only 4 over age 75 years. There were unusual circumstances that justified the unusual step of testing a minor with parental consent.
There have been 21 deaths (12 BRCA1 path_variant carriers and 9 BRCA2 path_variant carriers) during the study period. Three of them had undergone RRM, 4 RRSO, 3 both RRSs and 11 had no RRS. Cancer localisation profile: three had BC (one of whom had previously had RRSO), three OC (one of whom had previously had RRM) and six other types of cancer (oesophagus, pancreas, liver, rectum, renal and non-Hodgkin’s lymphoma). The mean age at death was 53.3 years (range=37.7–81.7 years).
There were 73/887 (8.2%) women who had RRS prior to GT, of which 22 (2.5%) had RRM (14 BRCA1 path_variant carriers and 8 BRCA2 path_variant carriers) and 56 (6.3%) underwent RRSO (35 BRCA1 path_variant carriers and 21 BRCA2 path_variant carriers) prior to GT. Five women underwent both surgeries prior to GT, all BRCA1 path_variant carriers.
Twenty-three originally healthy women (2.6%) (10 BRCA1 path_variant carriers and 13 BRCA2 path_variant carriers) underwent contralateral RRM after a prospective BC diagnosis.
At RRM, 5/306 (1.6%) women (3 BRCA1 and 2 BRCA2 path_variant carriers) were found to have an occult BC (online supplemental table 1) and 2/414 (0.5%) women (1 BRCA1 and 1 BRCA2 path_variant carriers) had an occult OC diagnosed at RRSO.
The mean period of time from positive predictive GT result or 25th birthday (whichever was later) to the censor date (DOD, BC, OC or last follow-up, whichever was earliest) was 6.26 years (range=0.01–24.3). This constituted 4685.4-women-years (when women are censored for both cancers; Note: in BC/OC incidence calculations, women are censored only for the relevant cancer). If time from RRS to GT was added in women who underwent RRS at an early age and prior to GT (two women had their RRSO surgery in 1970s), the total follow-up time would increase to 5339.9-women-years. This was not included in the follow-up calculations as study participants were followed from the date of GT or 25th birthday.
In total, 306/887 (34.5%) BRCA1/2 path_variant carriers underwent RRM (mean age at GT=37.9 years; mean age at RRM=39.2 years). BRCA1 path_variant carriers underwent RRM more frequently than BRCA2 path_variant carriers (165 and 141, respectively; p=0.04). RRM was most commonly performed in women aged 33–34 years. In total, 269/306 women (87.9%) underwent RRM before the age of 50 years, 29 of whom (9.5%) were <30 years of age.
The 5/306 women (1.6%) with occult cancer at RRM were not considered to have had RRM in the BC incidence calculations.
BRCA1/2 path_variant carriers who underwent RRM after GT result (284 women), had their surgery within 0.2–177.6 months (mean=28.4; median=18.4); only 20 of whom (7.0%) had their surgery within 6 months after GT.
Overall RRM uptake in patients followed up from GT to the censor date was 57.9% (online supplemental figure 1). The distribution of different age groups undertaking RRM is shown in figure 1. Women aged 50–59 had a statistically significantly lower RRM uptake rate 2, 5 and 10 years after GT, compared with the reference age group (30–39 years at GT) (table 2). RRM was associated with a statistically significantly decreased BC risk (HR=0.061; 95% CI 0.02 to 0.20, p<0.001) but not with overall mortality (HR=0.32; 95% CI 0.09 to 1.17, p=0.09).
In total, 414/887 (46.7%) BRCA1/2 path_variant carriers underwent RRSO (mean age at GT=43.8 years; mean age at RRSO=44.6 years; range=25.5–76.7).
RRSO after GT result was performed in 357/887 BRCA1/2 path_variant carriers, who had their surgery within 0.4–237.6 months (mean=29.5; median=10.0); 105 (29.4%) of whom had their surgery within 6 months after GT. The overall RRSO uptake in patients followed up from GT to censor date was 78.6% (figure 2 and online supplemental figure 2). In 57 women, RRSO was performed before their GT results and were not included in the calculations; therefore, the actual percent of women undergoing RRSO is significantly higher. BRCA1 path_variant carriers underwent RRSO more frequently than BRCA2 path_variant carriers (226 vs 188, respectively; p=0.01). The majority of women (77%) underwent RRSO <50 years of age with surgery most commonly performed in women aged 39–40 years. Only 2.2% of women underwent RRSO at age <30 years.
Women having GT result at <30 years were less likely to undergo RRSO 2, 5 and 10 years after the GT, compared with the reference group. The reference age group was less likely to have RRSO within 2 years after GT, compared with other age groups, except the <30 group (table 3).
A small proportion (14/887 (1.6%) originally healthy women) underwent RRSO after BC diagnosis (14/60 (23.3%) patients with BC in this study). RRSO alone was not associated with a reduced risk of BC (HR=0.77; 95% CI 0.45 to 1.34, p=0.36), neither when stratified by gene (BRCA1 versus BRCA2) nor age: <50 years vs >50 years at RRSO (data not shown).
OC incidence was not statistically significantly decreased in those undergoing RRSO (HR=0.02; 95% CI 0.000 to 5.9, p=0.17), possibly due to an extremely small incidence rate (0 in RRSO group vs 8 OC in non-RRSO; 2/8 OC cases were occult, therefore calculated as non-RRSO). This is perhaps unsurprising as the presymptomatic testing population is very young and RRSO would have little impact on incidence <35 years in BRCA1 and <45 years in BRCA2. On the other hand, there was a 100% risk reduction of OC in RRSO group, compared with non-RRSO.
RRSO alone was not associated with a reduced risk of overall mortality in the total study population (HR=0.44; 95% CI 0.1 to 1.4, p=0.16).
RRM and RRSO
For every year increase in age (range=25–85 years), women were more likely to undergo RRSO (HR=1.04; 95% CI 1.03 to 1.05, p<0.0001) but slightly less likely to undergo RRM (HR=0.98; 95% CI 0.97 to 0.99, p<0.0001).
In total, 208/887 (23.5%) women (mean age at GT=39.9 years) underwent both RRSs. Of these, 106 (61 BRCA1 and 45 BRCA2 path_variant carriers) underwent RRSO before RRM, 98 underwent RRSO after RRM (54 BRCA1 and 44 BRCA2 path_variant carriers) and 4 underwent RRM and RRSO simultaneously (all BRCA1 path_variant carriers).
Over the study period, 60 new first primary BC cases were diagnosed in the total study population (6.8%). Mean age at BC was 46.4 years (range=29.4–79.3). Mean time from GT to BC diagnosis was 4.5 years (range=0.02–14.7). The annual incidence of BC among all women in our study population during the 4702.9 women-years was 1.28%. The detailed BC incidence by RRS and BRCA1/2 status is described in table 4 and online supplemental table 2.
BC and contralateral RRM in the MRI programme
There were 30 healthy BRCA1/2 carriers detected with BC on the MRI programme during the study period (7 had ductal carcinoma in situ, 22 had invasive ductal carcinoma and 1 had invasive lobular carcinoma). Contralateral RRM after the BC diagnosis was performed in 23/30 (76.7%) patients (mean age at GT=40.6 years; mean age at RRM=44.8 years). The mean follow-up time from BC to contralateral RRM was 8.3 months (range=0.1–73.7 months). Tumour size ranged 2–29 mm (median=9 mm, mean=10.1 mm). Following MRI screening one Stage 2+ patient died after 3 years, with a mean of 6.3 years follow-up post-MRI screening diagnosis in our total study population.
Breast cancers after RRM
There were three BC cases (two CIS) following RRM (two previously reported24): one on the chest wall 6.6 years after RRM in a BRCA1 path_variant carrier, the second behind a retained nipple in a BRCA2 path_variant carrier 9.9 years after RRM and the third on the skin flap in a BRCA2 path_variant carrier 7 years after RRM.
Over the study period, 8 new first primary OC cases were diagnosed among 887 (0.9%) women in our cohort. Mean age at OC diagnosis was 48.2 years (range=37.7–62.7). Mean time from GT to OC diagnosis was 2.32 years (range=0.4–9.3). The annual incidence of OC in our study population during 5084.1-women-years follow-up was 0.16%. The incidence of OC among women who had RRSO was 0%, compared with 0.31% in women without RRSO. The detailed OC incidence by RRS and BRCA1/2 status is described in online supplemental table 3.
This study assessed the uptake, timing and relative risk reduction from RRM and RRSO in 887 path_variant carriers unaffected at GT. Most BC and OC associated with germline BRCA1/2 variants are diagnosed at younger ages with most ovarian being high-grade and advanced-stage serous carcinomas.25 The risk and age at onset of BC or/and OC depends on the gene and path_variant involved, with BRCA1 carriers affected more often and at younger ages than BRCA2 carriers, which is in keeping with our results.
The decision to undergo RRS is undoubtedly difficult, especially in cases with a complex family history of cancer or unsatisfactory outcomes from RRS in the family. Moreover, physical, mental and emotional consequences, which may result from RRS, should be taken into consideration. This makes presurgery genetic counselling, including psychological assessment for RRM, an essential part of the decision-making process. A woman who opts to undergo RRS must be fully prepared for all potential consequences. Howard et al reviewed 43 studies on factors influencing the decision-making on risk-reducing strategies for women at high-risk for BC/OC. Key factors included medical and physical, psychological and social context.26 Tong and colleagues found that younger age, more years of education, higher cancer-related distress and higher perceived risk of BC were independently associated with RRM intentions, whereas older age, perceived path_variant risk and perceived risk for OC were independently associated with RRSO intentions.27 Galmor et al reported that being married and having a first-degree relative with BC were positively associated with RRM, while having a previous benign breast biopsy was negatively associated with RRM in cancer-free Israeli BRCA1/2 path_variant carriers.28
Evans et al showed that RRS in unaffected women at high risk of BC and OC are risk, time and age dependent.29 Thus, women continue to opt for RRM and RRSO many months and even years after their positive BRCA1/2 path_variant gene test result.12 In this cohort, the mean time from GT to RRM was 28.4 months (range=0.2–177.6) and 29.5 months (range=0.4–237.6) to RRSO, respectively. We did not evaluate reasons of uptake and timing of RRS; however, the influence on decision making has been reported in detail previously12 18 30 and reasons in our cohort were similar. In particular, there is an appropriate delay in RRSO in the <30 years age group and much of the delay in the 30–39 age group may be explained by waiting until being over 35 years old or completion of childbirth.
Skytte et al reported a higher rate of uptake of RRSO within the first 6 months after GT at 27% compared with 11.8% (105/887 women) in our population. We also observed a much lower rate of uptake of RRM with only 20 women (2.25%) within the first 6 months after receiving the GT result, compared with 11%, reported by Skytte et al.31 This lower short-term uptake of both RRM and RRSO is likely to be related with healthcare management and the patient preparation process for major surgery. This is particularly true of RRM which requires a psychological assessment and at least two surgical appointments.32–34 In terms of health economics, undergoing RRSO should be recommended to BRCA1/2 path_variant carriers due to its favourable comparative cost-effectiveness,35 with RRM being discussed as an option.
In our cohort of 887 healthy BRCA1/2 path_variant carriers, the overall uptake of RRM and RRSO was 57.9% and 78.6%, respectively. It is important to note that in our study only women who had their RRS after receiving GT results were included in uptake calculations and therefore the actual uptake of both RRSs in our Centre is higher. Skytte et al reported similar data from 306 healthy BRCA1/2 path_variant carriers with 10-year RRM uptake of 50% and RRSO uptake of 75% by time to event analysis.31 Earlier studies report lower than 50% uptake of either RRS.29 30 36 37 These findings are consistent with recent literature, reporting growing numbers of RRS uptake in unaffected high-risk BRCA1/2 path_variant carrier population.18 Galmor et al reported that only 9.6% of cancer-free BRCA1/2 path_variant carriers underwent RRM in their institution over a median follow-up time of 4.4 years. This is significantly lower than we report and could relate to cultural differences.28 In reality, as demonstrated in our study, around 50% of RRM are undertaken in the first 2 years and later uptake may be driven by false positive screens, new BC diagnoses in the family or BC related deaths.29 Indeed, there are women who choose to undertake surgery more than 15 years post-GT. As with prior studies that have examined RRM and RRSO in a single institution,19 38 39 we found a higher uptake of RRSO than RRM. Beattie and colleagues suggested cultural factors, different counselling practices, body image effects of RRS and the greater accuracy of screening tests for BC compared with those for OC as the main determinants for decision making regarding uptake and timing of RRS.19
Although the uptake of RRSO is high, the surgeries are performed later than recommended.36 RRSO should be performed shortly after GT if women are aged >40 years, or around 35 years of age (later in BRCA2), to get the optimal effect on cancer risk reduction, while reported uptake is still on average up to 10 years later.31 We found two occult OC cases on RRSO in women aged 35–39 years that potentially could have been avoided if RRSO was performed earlier, as both women had delayed surgery post-GT. The decision to delay surgery is mostly related to later childbearing and concerns about early menopause with its consequences40 despite the possibility of hormonal replacement therapy.
Garcia et al reported that with increasing age, women are more likely to undergo RRSO (OR=1.04, 95% CI 1.01 to 1.07) and less likely to undergo RRM (OR=0.94, 95% CI 0.91 to 0.97) in keeping with our results.29
We did not include women with VUSs (variants of uncertain significance). Welsh et al reported RRS rates among unaffected women with BRCA1/2 VUS as high as 39%.41
RRM in high-risk BRCA1/2 path_variant carriers is clearly effective and this is supported by our data. Li et al analysed six non-overlapping studies with 2555 participants and showed that BC after RRM corresponded with a RR of 0.11 (95% CI 0.04 to 0.3).42
BC risk reduction after RRSO remains uncertain and most probably this is due to bias in previous studies. Stjepanovic and colleagues showed that premenopausal RRSO significantly decreased BC risk in local BRCA1 path_variant carriers, while both BRCA1/2 carriers benefitted in their systematic review of published work.43 Eisen et al previously reported that the BC risk reduction with RRSO was greater in BRCA1/2 path_variant carriers who underwent surgery <50 years compared with >50 years at surgery.44 However, Heemskerk-Gerritsen et al reported no effect on BC risk and Kotosopoulos et al found no effect of RRSO on premenopausal BC risk in BRCA1 gene path_variant carriers.23 45 Terry et al found no association of RRSO on BC risk when RRSO was used as a time-dependent variable.46 Similarly, our data do not support a reduction in BC risk for healthy BRCA1/2 path_variant carriers undergoing RRSO.
Schrag et al reported that benefit for BRCA1/2 path_variant carriers, received from RRS, declined with age with little gain obtained from RRS after the age of 60 years.47 There are only a few reports on long-term follow-up of BRCA1/2 carriers. It is therefore important to continue following-up our patients to determine the potential incidence and mortality from later cancer occurrences. Some cancer risk persists after RRS. Consequently, additional preventive and screening strategies are required to improve outcomes in high-risk BRCA1/2 carriers.11
In conclusion, this is one of the largest prospective studies evaluating healthy BRCA1/2 path_variant carriers. Despite the strong evidence for RRS effectiveness and rising uptake of RRM and RRSO compared with previous studies, women at high risk still need more encouragement to opt for RRS, especially RRSO. The timing of RRS is still suboptimal. Although RRM reduces BC incidence, we did not find a protective effect of RRSO on BC risk, similar to that observed by other groups. Further studies with long-term follow-up are needed.
Data availability statement
All data relevant to the study are included in the article or uploaded as supplementary information.
Patient consent for publication
North Manchester Research Ethics Committee (reference 08/H1006/77).
Twitter @ER_Woodward, @DrEmmaCrosbie
Contributors The research, literature review and manuscript text was conducted by RM, with a supervision and significant input from DGE. The other authors contributed with clinical work, patients inclusion, manuscript review and revision review.
Funding EJC is a National Institute for Health Research (NIHR) Clinician Scientist (NIHR-CS-012–009) and DGE is an NIHR Senior Investigator (NF-SI-0513–10076). DGE, EJC, EFH and ERW are supported by the all Manchester NIHR Biomedical Research Centre (IS-BRC-1215–20007).
Competing interests None declared.
Provenance and peer review Not commissioned; externally peer reviewed.
Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.