Article Text
PDF
Abstract
Objective Lynch syndrome is a hereditary cancer syndrome caused by mismatch repair gene mutations, and female carriers are at an increased risk of endometrial and ovarian cancer. The best approach to screening is not yet clear and practice varies across countries and centers. We aimed to provide evidence to inform the best approach to screening and risk reduction.
Methods A systematic search of the literature was conducted (Medline, Embase, PubMed). Studies evaluating the following were included: women with Lynch syndrome (by mismatch repair mutation or Amsterdam II criteria), screening methods for endometrial and/or ovarian cancer, intervention included endometrial biopsy, transvaginal ultrasound, or serum cancer antigen 125 (CA-125), outcomes evaluated were number of cancers and/or endometrial hyperplasia.
Results A total of 18 studies of Lynch syndrome carriers which screened for endometrial cancer using transvaginal ultrasound and/or hysteroscopy/endometrial biopsy revealed an incidence of 3.9% at the time of screening. Most (64.1%) endometrial cancers detected were from screening, with the balance detected in symptomatic women at the first screening visits, regular review, or between screening intervals. In mismatch repair carriers, the overall sensitivity of endometrial screening was 66.7%, and the number needed to screen ranged between 4 and 38 (median 7). The sensitivity of endometrial biopsy was 57.1% and the number needed to screen was 23–380 (median 78). The sensitivity of transvaginal ultrasound was 34.4% and the number needed to screen was 35–973 (median 170). Fourteen studies which screened for ovarian cancer using transvaginal ultrasound and/or CA-125 revealed an incidence of 1.3% at the time of screening and 42.9% of ovarian cancers were detected at asymptomatic screening. The sensitivity of ovarian screening was 54.6%, and the number needed to screen was 9–191 (median 23) in mismatch repair carriers. Thirteen studies reported 5.8% incident endometrial cancers and 0.5% ovarian cancers at time of risk reducing surgery.
Conclusions There is limited evidence to support screening for endometrial and ovarian cancer in Lynch syndrome and data on mortality reduction are not available. Further prospective, randomized trials comparing targeted screening methods are needed. Risk reducing surgery remains the most reliable way to reduce endometrial and ovarian cancer risk in Lynch syndrome.
- Lynch Syndrome II
- Endometrial Neoplasms
- Endometrial Hyperplasia
- Ovarian Cancer
- Hysterectomy
Data availability statement
Data are available upon reasonable request. Data used to calculate sensitivity, specificity, or number needed to screen are available on request.
This is an open access article distributed in accordance with the Creative Commons Attribution 4.0 Unported (CC BY 4.0) license, which permits others to copy, redistribute, remix, transform and build upon this work for any purpose, provided the original work is properly cited, a link to the licence is given, and indication of whether changes were made. See: https://creativecommons.org/licenses/by/4.0/.
Statistics from Altmetric.com
HIGHLIGHTS
Endometrial and ovarian cancer rates were highest in MLH1 and MSH2 carriers, respectively
Endometrial biopsy had a sensitivity of 57.1% and the number needed to screen was 23–380 (median 78)
Risk reducing surgery could be offered based on genetic pathogenic variant
Introduction
Lynch syndrome is an autosomal dominant hereditary cancer syndrome caused by mutations in mismatch repair genes (MLH1, MSH2, MSH6, PMS2). Epithelial cell adhesion molecule (EPCAM) can also predispose to MSH2 deficient cancers. Before genetic testing, patients were diagnosed with Lynch syndrome using Amsterdam II criteria based on family history.1
Mismatch repair carriers are at higher risk of multiple malignancies, including colorectal, endometrial, ovarian, gastric, small bowel, pancreatic, biliary, renal, bladder, prostate, skin, and brain cancers.2 Women have 13–47% lifetime risk of endometrial cancer, and 3–17% risk of ovarian cancer, depending on the mismatch repair gene.2 3 The management of gynecological cancer risk involves screening and/or risk reducing hysterectomy and salpingo-oophorectomy. The Manchester International Consensus Guidelines4 and Mallorca Group5 recommend risk reducing surgery in all mismatch repair carriers except for PMS2 carriers, without screening, while the US Multi-Society Task Force recommends screening for endometrial cancer by annual endometrial biopsy from 35 to 40 years.6 7
Only one previous systematic review has evaluated the benefits of gynecological cancer screening in Lynch syndrome. This included five studies and concluded there was insufficient evidence to support screening for either endometrial or ovarian cancer.8 Our review provides an updated analysis of the evidence for the effectiveness of endometrial and ovarian cancer screening and risk reducing hysterectomy and bilateral salpingo-oophorectomy in cancer prevention.
Methods
Endometrial and Ovarian Cancer Screening
Medline (Ovid), Embase, and PubMed databases were searched in August 2020 using relevant medical subject headings and keywords (Online Supplemental Table 1). Reference lists were searched for relevant articles. The same search was updated in November 2021. Articles meeting all of the following criteria were included: women with Lynch syndrome (by mismatch repair mutation or Amsterdam II criteria), screening methods for endometrial and/or ovarian cancer, intervention included endometrial biopsy, transvaginal ultrasound, or serum cancer antigen 125 (CA-125), and those where outcomes were number of cancers and/or endometrial hyperplasia.
Supplemental material
Articles meeting any of the following criteria were excluded: personal history of endometrial or ovarian cancer, non-Lynch syndrome hereditary cancer syndromes, other screening methods, screening was not for gynecological cancers, outcomes were cost effectiveness of screening or patient perception of screening, those not published in English, or did not contain patient data.
Risk Reducing Surgery
Medline (Ovid), Embase, and PubMed databases were searched in August 2020 using relevant medical subject headings and keywords. Reference lists were searched for relevant articles. The same search was updated in November 2021. Similar inclusion and exclusion criteria to the above were applied detailed in Online Supplemental Table 1).
Results
Endometrial and Ovarian Cancer Screening
The search from August 2020 to November 2021 identified 338 studies meeting the inclusion criteria. After removing 95 duplicates and excluding 194 results through abstract screening, 43 full text articles were assessed for eligibility and six abstracts were included. Eleven full text articles were included, and four additional articles were added through reference searching. Hence 21 articles were included in the analysis (PRISMA9 diagram presented in Online Supplemental Figure 1).
Supplemental material
Of the 21 studies included, nine were retrospective and 12 were prospective (Online Supplemental Table 2). Eleven studies screened for both endometrial and ovarian cancers (Table 1). The age of patients screened ranged from 18 to 84 years. All studies used transvaginal ultrasound, and additional screening methods included CA-125 (12 studies), routine endometrial sampling (14 studies), and routine hysteroscopy (four studies) (Table 1).
Supplemental material
Screening programs for gynecological cancer and participant characteristics across studies which studied outcomes of gynecological cancer screening in female Lynch syndrome carriers
Rates of Endometrial Cancer Detected by Screening
Eighteen studies of endometrial cancer screening detected a total of 104 cancers among 2688 women (3.9%), diagnosed between at ages 36–72 years (Figure 1).10–27 A total of 1193 of 2688 (44.4%) patients had confirmed mismatch repair/EPCAM mutations and 1495 of 2688 (55.6%) were identified through Amsterdam criteria (Figure 1).
Detection of endometrial cancers (EC) through screening according to mismatch repair (MMR) carrier status. Screening methods included transvaginal ultrasound, endometrial biopsy, and hysteroscopy. There were 2688 individuals across 18 studies, 1189 of whom had a confirmed germline MMR mutation. Of MMR carriers, 6.5% were diagnosed with EC; screening detected 64.1% of these, while the remainder were diagnosed between screening intervals or presented with symptoms during a prevalent visit or regular review. 1495 participants did not have MMR carrier status available. Of these, 1.7% were diagnosed with EC; screening detected 53.8% of these, while the remainder were diagnosed at the first visit, between screening intervals, or due to symptoms.
A total of 78 of 1193 (6.5%) mismatch repair/EPCAM carriers were diagnosed with endometrial cancer (Figure 1), representing 75% of all endometrial cancers found. Fifty of 78 (64.1%) patients were detected through asymptomatic screening at their first or subsequent screening visit, while the remainder were diagnosed due to symptoms at or between screening intervals (Figure 1). Twenty-six of 1495 women (1.7%) with Lynch syndrome diagnosed through Amsterdam criteria (or where genetic testing information was not published) were diagnosed with endometrial cancers, representing 25% of all endometrial cancers. Fourteen of 26 (53.8%) of these were through screening, while the remainder were diagnosed due to symptoms, or between screening intervals (Figure 1).
Forty of 57 (70.2%) cases of endometrial hyperplasias were found in mismatch repair carriers, 50% of which were in MLH1 carriers (Online Supplemental Table 3). The number needed to screen, defined as the number of people needed to be screened for a diagnosis of cancer or hyperplasia, ranged between 4 and 135 (median 13) (Online Supplemental Table 4). This reduced to between 4 and 38 (median 7) when only mismatch repair carriers were included. In the mismatch repair carrier population, the sensitivity of screening to detect endometrial cancer (excluding hyperplasia) was 66.7%.
Supplemental material
Supplemental material
Combining studies from Table 2 where sufficient data were provided to inform the number of cancers or hyperplasia detected by each screening method, endometrial biopsy found 20 of 64 endometrial cancers detected via screening in total (65% stage I, 15% stage II, 5% stage III, remainder unreported) and 29 hyperplasias of 36 detected via screening in total. The sensitivity and specificity of endometrial biopsy in detecting cancer (excluding hyperplasia) were 57.1% and 97.7%, respectively. Number needed to screen, defined as the number of endometrial biopsies required to detect cancer or hyperplasia, ranged between 12 and 380 (median 19) (Online Supplemental Table 4), and between 23 and 380 (median 78) in detecting cancer only. Transvaginal ultrasound detected 11 endometrial cancers (81.1% stage I, remainder unreported) and seven cases of hyperplasia in total. Sensitivity and specificity in detecting endometrial cancer was 34.4% and 87.1%, respectively. The number needed to screen to detect either endometrial cancer or hyperplasia by transvaginal ultrasound ranged between 35 and 973 (median 89); this range remained the same to detect cancer only, however, the median increased to 170 (Online Supplemental Table 4). In studies of mismatch repair carriers only, two12 24 studies provided sufficient data to inform sensitivity. The sensitivity of endometrial biopsy and transvaginal ultrasound were 79.3% and 53.8%, respectively. In three studies13 19 22 which specified cancers detected by hysteroscopy, no additional cancers were detected when hysteroscopy was performed with endometrial biopsy.
Incidence of endometrial and ovarian cancer in female Lynch syndrome carriers detected through screening, or through symptoms during interval visits or regular review
Rates of Ovarian Cancer Detected by Screening
Fourteen studies detected 29 cancers among 2224 women (1.3%), diagnosed between 35 and 83 years old (Online Supplemental Figure 3).11 12 14 15 20 21 23–30 Twenty-eight of 1458 (1.9%) mismatch repair/EPCAM carriers were diagnosed with ovarian cancer, representing 96.6% of all ovarian cancers found. Twelve of 28 (42.9%) were in asymptomatic women detected through screening at their first or subsequent screening visit (Online Supplemental Figure 3). Of these, 11 of 12 (91.7%) were detected through transvaginal ultrasound and 3 of 12 (25%) were detected through CA-125. A total of 58.3% of ovarian cancers detected in gene carriers through screening were stage I, 25% were stage II, 8.3% were stage III, and 8.3% did not have staging reported. Eight of 28 were interval ovarian cancers which presented between regular screening visits; 50.0% were stage I, 12.5% stage II (12.5%), 25.0% stage III, and 12.5% had no reported staging. Interval ovarian cancers are defined as ovarian cancers which present clinically between regular screening visits. They are either cancers missed by screening tests at the previous screening visit, or cancers which rapidly developed between screening intervals. Two incident ovarian cancers were detected at risk reducing salpingo-oophorectomy. Eight of 28 (28.6%) patients with ovarian cancers were symptomatic: 50% were stage I, 25% stage II, 12.5% stage IV, and the remainder had no staging reported (Table 2).
Supplemental material
One (0.1%) stage III ovarian cancer was diagnosed in 766 Amsterdam criteria patients through CA-125, with no ultrasound abnormalities. The calculated sensitivity of ovarian cancer screening was 54.6% and the specificity of transvaginal ultrasound and CA-125 was 96% and 99%, respectively. The number needed to screen was 9–218 (median 72) (Online Supplemental Table 5). This reduced to 9–191 (median 23) across five studies where only mismatch repair carriers were included (Online Supplemental Table 5). There were insufficient data to calculate number needed to screen for individual ovarian cancer screening methods (transvaginal ultrasound, CA-125).
Supplemental material
Risk Reducing Surgery
The initial search identified 438 results and the updated search in November 2021 identified 171 results, resulting in 609 total results. After removing 135 duplicates and excluding 424 reports through abstract screening, 48 full text articles were assessed for eligibility. Eleven articles were included, and two additional articles were added through reference searching. Hence 13 articles were included in the analysis (PRISMA9 diagram presented in Online Supplemental Figure 2).
Supplemental material
Of the 13 studies included, three were prospective cohort studies31–33 and 10 were retrospective histopathological analyses24 27 34–41 (Online Supplemental Table 6). Eight studies reported preoperative screening results. The total sample size across all included studies was 450, of whom 433 of 450 (96%) were germline mismatch repair/EPCAM carriers. Participants were aged 20–77 years at risk reducing salpingo-oophorectomy (Table 3). Only one study37 included symptomatic women (n=7) (1.5%) (Table 3).
Supplemental material
Incidence of endometrial cancer or hyperplasia and ovarian cancer in female mismatch repair carriers undergoing risk reducing surgery
Twenty-six of 450 women were diagnosed with endometrial cancer at the time of risk reducing hysterectomy (5.8%), aged 38–59 years (Figure 2). Twenty-seven of 450 (6.0%) patients had endometrial hyperplasia, at ages 35–53 years, 23 of 27 (85.2%) of which were atypical (Figure 2). Two of 413 (0.5%) ovarian cancers were detected at risk reducing salpingo-oophorectomy, both in MSH2 carriers. One was 44 years old and the other patient’s age was not provided.
Rate of endometrial cancer (EC) or endometrial hyperplasia (EH) in prophylactic specimens from risk reducing surgery in Lynch syndrome carriers according to mismatch repair (MMR) carrier status. Of 450 participants, 433 had germline mutations, of whom 5.7% had EC at the time of risk reducing surgery; 6.0% had EH, 84.6% of which were atypical. Germline mutations and cancer stages are listed where specified.
Discussion
Summary of Main Results
Endometrial Cancer Screening
In the 18 studies, the incidence of endometrial cancer was 3.9% over a median screening duration of 4.5 years, compared with 13–47% lifetime incidence reported in the Prospective Lynch Syndrome Database.3 In addition to cancer, both atypical and non-atypical hyperplasia were included as pre-malignant lesions. Atypical hyperplasia has a 27.5% cumulative risk of progression to cancer in the general population, compared with 4.6% for non-atypical hyperplasia.42 Despite the lower cumulative risk, we included non-atypical hyperplasia as it occurs with mismatch repair protein deficiency,43 and represents an opportunity for risk reduction.
More cancers were detected by screening in mismatch repair carriers compared with those who were clinically diagnosed. This may be due to a lower prevalence, more targeted screening or frequent medical reviews in carriers, or a higher index of suspicion in use of screening. Also, mismatch repair carriers were older on average than those who were not genetically tested (51 years vs 44.5 years). No endometrial cancers were diagnosed in PMS2 carriers, consistent with findings from the Prospective Lynch Syndrome Database (12.8%), suggesting that screening is not indicated in this population.
Interval cancers, defined as cancers which present between regular screening visits, are either cancers missed by a screening method at a previous screening visit or cancers which rapidly develop between screening visits and hence were not detected previously. Interval cancers were detected in 14.3% of studies of annual screening20 24 with transvaginal ultrasound and biopsy, and most (80%) were stage I. No interval cancers occurred where screening included hysteroscopy.13 18 19 21 22 25 However, these studies had low cancer detection rates overall. The specificity of hysteroscopy alone as a screening tool cannot be determined because it was always combined with other methods.
The false negative rate for endometrial cancer screening is uncertain because there is no gold standard. However, interval cancers may indicate a false negative rate. Interval cancers could overestimate false negatives, as they will include rapidly growing de novo cancers. However, most studies reporting interval cancers used annual screening, making this less likely. Cancers presenting after missed screening visits were excluded in this sensitivity analysis. We found a 57.1% sensitivity rate for endometrial biopsy with most cancers detected early (65% stage I). However, this rate was no better than staging information available on interval cancers (72.2%), suggesting that asymptomatic screening does not detect endometrial cancers at an earlier stage than symptomatic presentations. However, not all studies reported staging. The low transvaginal ultrasound sensitivity could be influenced by a higher proportion of premenopausal women included.
Although more invasive, the specificity of endometrial biopsy is substantially higher than transvaginal ultrasound, especially in premenopausal women.16 False positives are more likely in premenopausal women which may lead to anxiety and over treatment. Only a small number of studies have reported sufficient data to inform sensitivity and specificity. Whether screening impacts on endometrial cancer mortality in Lynch syndrome is unknown. The 5 year survival rates from endometrial cancer in the Prospective Lynch Syndrome Database compared with sporadic endometrial cancers in the Australian population were 89% and 85%, respectively.2 44 Since screening is not routinely recommended in the general population, this suggests that endometrial cancer screening does not improve survival in Lynch syndrome. The number needed to screen to detect endometrial cancer varied widely. This might reflect variation in screening intervals, age at first screen (19–82 years),23 variable attendance at screening, and the percentage of mismatch repair carriers in each study. The number needed to screen with endometrial biopsy was less than with transvaginal ultrasounds, however, this could only be calculated for six studies, and sample sizes were small.
Ovarian Cancer Screening
The overall incidence of ovarian cancer was low (1.3%) and was higher among mismatch repair carriers (Online Supplemental Figure 3). As expected, all ovarian cancers detected were in MSH2 carriers, consistent with ovarian cancer risk of 17%.3 Screening led to diagnosis in half of the studies (Online Supplemental Figure 3). In the only study with a control group,20 an equal number of participants in both the screening and control groups were diagnosed with ovarian cancer, suggesting no benefit for screening. The number needed to screen for ovarian cancer varied greatly, even in mismatch repair carriers (Online Supplemental Table 5).
The false negative rate for screening is unknown, but one-third of ovarian cancers in mismatch repair carriers were interval cancers (Online Supplemental Figure 3). In some of these cases, the cancers were diagnosed within a year after a negative screening test which would lead to false reassurance for patients, assuming they were missed at the previous screening test.20 Transvaginal ultrasound and CA-125 had high specificity rates based on three11 14 21 studies, one of which had no false positives or ovarian cancers, questioning the selection of patients for screening. False positives are common with CA-12514,21, which may increase the rates of unnecessary oophorectomy.28
Risk Reducing Surgery
Risk reducing surgery for Lynch syndrome includes hysterectomy with bilateral salpingo-oophorectomy. A multicenter study of 315 women (aged 20–63 years)31 found no endometrial or ovarian cancers after risk reducing surgery, compared with 33% endometrial and 5% ovarian cancers in controls.31 This study has formed the basis of guidelines recommending hysterectomy and bilateral salpingo-oophorectomy from 35 to 40 years of age or completion of childbearing.4 6 7
Eighteen of 26 (69.2%) endometrial cancers identified at hysterectomy and salpingo-oophorectomy were early stage (stage I), with only two stage II cancers (Figure 2). The earliest endometrial cancer was 38 years old,31 consistent with recommended age for risk reducing surgery.4 From the Prospective Lynch Syndrome Database, we would expect 59–212 endometrial cancers to be diagnosed by 70 years in 450 patients, based on a 13–47% lifetime risk, depending on pathogenic variant.3 A number needed to treat between 2 and 8 is calculated. This means that two surgeries are needed to diagnose one endometrial cancer in MSH2 carriers, compared with eight surgeries in PMS2 carriers.
In some studies, preoperative diagnoses prior to risk reducing surgery were obtained using transvaginal ultrasound or biopsy, and then compared with histopathological findings post-hysterectomy. Only 7 of 9 studies used biopsy and one study used ultrasound, limiting generalizability. Three biopsies did not correctly diagnose endometrial cancer, and one misdiagnosed hyperplasia as cancer (Table 3). These findings are consistent with those from other studies of endometrial biopsy and transvaginal ultrasound.
Ovarian cancer is much less common than endometrial cancer in Lynch syndrome (0.5% vs 5.8%), consistent with the Prospective Lynch Syndrome Database (3–17%).3 Based on these rates, 12–70 ovarian cancers would be expected to occur by 70 years in 413 women depending on pathogenic variant, with an estimated number needed to treat of 6–34.3 Six surgeries would need to occur in MSH2 carriers to detect one ovarian cancer, compared with 34 surgeries needed in PMS2 carriers. This aligns with recent recommendations for risk reducing salpingo-oophorectomy for MSH2 and MLH1 carriers and not for MSH6 and PMS2 carriers.3 In the Prospective Lynch Syndrome Database, ovarian cancer had a high 5 year and 10 year survival rate of 84%3 compared with 46% in the general Australian population.44 This could be due to the highly screened targeted population and younger age group included in the former, or the specific phenotype of Lynch syndrome cancers.
Results in the Context of Published Literature
We found limited evidence to support ovarian cancer screening in Lynch syndrome. Similarly, the general population showed no reduction in deaths due to ovarian cancer in screened women compared with those who were not screened.45 Although ovarian cancer screening showed a stage shift in those at elevated risk,30 these were mostly in BRCA1/2 carriers who have a higher penetrance for ovarian cancer compared with mismatch repair carriers.
In Lynch syndrome, risk reducing hysterectomy and salpingo-oophorectomy is cost effective at age 4046 and prevents 13–45% of endometrial cancers if performed at age 40 years, and 4–18% if performed at age 60.3 Despite this, uptake by 50 years of age is low, at 9–28% for hysterectomy and 13–26% for salpingo-oophorectomy,47 and centers across the world vary in their approach to recommending risk reducing surgery.48 Female Lynch syndrome carriers regard cancer surveillance to be very important, and would opt for more regular investigations if offered.49 For those who have not completed childbearing, contact with a gynecologist as part of their multidisciplinary care could offer reassurance even without intervention. Providing information about symptoms such as abnormal bleeding and managing non-genetic risk factors for endometrial and ovarian cancer allows them to proactively manage their cancer risk. It also offers the opportunity to reinforce the use of aspirin for chemoprevention of Lynch syndrome cancers, including endometrial and ovarian cancer.50 Potential less invasive methods also present hope for the future in endometrial cancer screening. Cervico-vaginal cytology detected almost half of endometrial cancer cases in a meta-analysis,51 and vaginal and urine cytology showed high combined sensitivity.52 Urinary biomarkers show promise in detecting malignancy.51 53
Strengths and Weaknesses
This is the first systematic review to report number needed to screen in endometrial and ovarian cancers associated with Lynch syndrome. Limitations include that approximately half the studies were retrospective, and the median sample size was only 29 women. Most had no control groups. Inclusion criteria for participants were heterogeneous with respect to family history, inclusion of mismatch repair carriers, age of onset of screening, period of follow-up, and sample size. Reduction in mortality due to screening could not be determined.
Implications for Practice and Future Research
Adequately powered randomized controlled trials are needed to evaluate the sensitivity and specificity of each potential screening method and their effect on morbidity and mortality. The relative risks and benefits of surgery versus screening are also uncertain.
Conclusion
Transvaginal ultrasound for the detection of endometrial or ovarian cancers does not appear to reduce morbidity in Lynch syndrome. Endometrial biopsy is more sensitive and specific with a lower number needed to screen. However, endometrial biopsy is invasive. Risk reducing hysterectomy and bilateral salpingo-oophorectomy is the mainstay of prevention in Lynch syndrome but recommendations vary depending on the mismatch repair gene.
Data availability statement
Data are available upon reasonable request. Data used to calculate sensitivity, specificity, or number needed to screen are available on request.
Ethics statements
Patient consent for publication
Ethics approval
Not applicable.
Acknowledgments
The authors thank Dr David H Wrede for his advice and review of the work.
References
Supplementary materials
Supplementary Data
This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.
Footnotes
Contributors NL, guarantor, completed the literature search and drafted the manuscript. Titles and abstracts of articles identified through the search strategy were evaluated against inclusion and exclusion criteria by the primary authors NL and FAM. Full text articles were further excluded by NL against the same criteria. CK, FAM, MH, and GPY reviewed and edited the manuscript.
Funding The authors have not declared a specific grant for this research from any funding agency in the public, commercial, or not-for-profit sectors.
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.