Quality of life after risk-reducing surgery for breast and ovarian cancer prevention: a systematic review and meta-analysis

64 Objective: To assess the impact of risk-reducing surgery (RRS) for breast cancer (BC) and 65 ovarian cancer (OC) prevention on quality-of-life (QoL). We consider risk-reducing 66 mastectomy (RRM), risk-reducing salpingo-oophorectomy (RRSO), and risk-reducing early67 salpingectomy and delayed-oophorectomy (RRESDO). 68 Data sources: We followed a prospective protocol (PROSPERO: CRD42022319782) and 69 searched MEDLINE, EMBASE, PubMed, and Cochrane Library from inception to February 7


INTRODUCTION
gold-standard OC preventive strategy, reducing OC-risk by 80-97%. 29-31 RRSO has been 126 undertaken for BRCA1/BRCA2 carriers, or women with a strong FH of OC. Broadening access 127 has led to RRSO now being offered to women at >4-5% lifetime OC-risk, including newer 128 moderate-penetrance OC CSGs and women with a first-degree-relative with high-grade serous 129 OC. 19,32,33 130   131 Pre-menopausal oophorectomy leads to premature surgical menopause, impacting quality-of-132 life (QoL) outcomes like sexual function and vasomotor/menopausal symptoms. 34,35 It is 133 associated with long-term detrimental sequelae like coronary heart disease, osteoporosis, and 134 cognitive decline, although these may be ameliorated by hormone replacement therapy 135 (HRT). 36 Besides, a higher decision regret rate for pre-menopausal (compared to post-136 menopausal) RRSO has been reported. 37 The widespread acceptance of the fallopian tube as 137 the site of origin of most serous epithelial OC along-with the detrimental health sequelae of 138 early menopause has supported introduction of a novel two-step strategy of risk-reducing early-139 salpingectomy (RRES) and delayed-oophorectomy (DO) (RRESDO). [38][39][40] This allows pre-140 menopausal women wishing to decline/delay RRSO, a degree of OC risk-reduction, whilst 141 avoiding premature menopause. Given limited outcome data, it is not considered standard of 142 care 41 and currently offered in clinical trials within USA/Europe. 42-44 143 144 For women with increased BC/OC risk, the decision of whether and when to undergo RRS is 145 complex and changes over time. A number of factors may influence this such as, carrying a 146 PV, cancer risk perception, FH/personal history of cancer, menopause status, fertility wishes, 147 relationship status. 45 Whilst surgery significantly reduces BC or OC risk and improves cancer-148 related worry, 27 it encompasses surgical risks, particularly with complex breast reconstruction. 149 J o u r n a l P r e -p r o o f RRM may adversely impact the psychological/physical well-being of patients following 150 consequent morbidities and body image issues. 46 While HRT may ameliorate outcomes of 151 premature menopause, it remains contraindicated for many women with BC. RRES is of 152 unproven benefit, and unlike RRSO will not improve BC mortality in women with BC. 47 153 154 It is crucial for women and their clinicians to have robust data on relevant QoL outcomes to 155 guide informed decision-making and minimise decision regret. To our knowledge, no 156 systematic review has attempted to collectively summarise the impact of 157 RRM/RRSO/RRESDO on QoL outcomes including health-related QoL (HRQoL), sexual 158 function, menopause symptoms, body image, cancer-related distress or worry, anxiety or 159 depression. Therefore, robust evidence synthesis on generic and condition-specific QoL after 160 RRM, RRSO and RRESDO is required. 161 162 OBJECTIVES 163 The primary aim of this review is to assess the impact of RRS for BC and OC prevention on 164 QoL outcomes. We consider RRM, RRSO, and RRESDO. Secondary aims are to compare 165 long-term vs short-term QoL outcomes after RRS; the impact of menopausal status and/or use 166 of HRT following RRSO; and whether confirmed diagnosis of PV in BC or OC CSGs vs. FH-167 based diagnosis affects post-operative QoL outcomes. 168

METHODS 170
We conducted the systematic review and meta-analysis using a prospectively registered 171 protocol (PROSPERO: CRD42022319782) and reported in line with PRISMA (Preferred 172 Reporting Items for Systematic Reviews and Meta-analyses). 48  We followed a population, intervention, comparison, outcome and study design (PICOS) 184 framework 53 to specify our inclusion criteria (Figure-1). Population: defined as women at 185 increased BC or OC risk, including diagnosis of PV in BC or OC CSGs or documented FH of 186 BC or OC, amounting to a >30-40% or >5% lifetime risk of BC or OC respectively. 19 187 Intervention: We focused on RRM for BC prevention, and RRSO or RRESDO for OC 188 prevention. Comparison: We compared QoL outcomes in women undergoing RRS vs those 189 who did not. We then compared QoL outcomes across different subgroups: (1) long-term vs. 190 short-term follow-up: for RRSO or RRESDO ≥1-year, and for RRM ≥2-years period was 191 defined as long-term follow-up; (2) women with PVs in BC/OC CSGs (e.g., BRCA1/BRCA2) 192 vs. those with FH-based risk; (3) post-menopausal vs. pre-menopausal RRSO; (4) pre-193 menopausal RRSO in HRT users vs. non-users. Outcome: We included studies reporting QoL 194 outcomes on HRQoL, sexual function, menopause symptoms, body image, cancer-related 195 distress or worry, anxiety or depression using validated questionnaires/tools. Study design : We 196 included any study design (prospective/retrospective cohort studies, randomised/non-197 randomised trials, or case-series), that follows our PICOS framework. from 0 (not reported), 1 (reported but inadequate), to 2 (reported and adequate). The maximum 216 global score is 16 for non-comparative (8 items) and 24 for comparative studies (12 items). A 217 score ≤12 for non-comparative and ≤20 for comparative studies was considered high-risk of 218 bias. 54 We also assessed the external validity of included studies (representativeness of findings) 219 based on whether the included population was definitely high-risk for BC or OC (PV in BC/OC 220 For quantitative synthesis, fixed-effects meta-analysis was used to calculate summary 233 estimates of QoL with 95%CI after RRS vs. no surgery where data allowed. We chose fixed-234 effects meta-analysis models, as the outcome measures comprised of the same validated 235 questionnaires considered consistent across studies. However, we also undertook sensitivity 236 analysis using random-effects meta-analysis. We undertook further pre-designed subgroup 237 analyses to assess any difference in QoL outcomes for (1) the first 2-years post-RRM vs. after; 238 (2) the first year post-RRSO/RRESDO vs. after; (3)  For MINORS score see Figure-3 and Appendix-3. The median MINORS score was 286 20(IQR:19-21) for 11 comparative and 12(IQR:12-13) for 23 non-comparative studies. Short 287 (<1-year post-RRSO or <2-years post-RRM) or no reported duration of follow-up, >5% of 288 participants lost to follow-up, and no sample size calculation were the main potential biases. 289 Thirteen studies (N=2801) were deemed low-risk of bias for methodological quality, whereas 290 21 studies (N=4046) were high-risk of bias. Regarding external validity, 9 studies (N=2255) 291 were deemed high-risk of bias and 25 studies (N=4509) were low-risk of bias. 292 293 Data synthesis 294 Amongst them, 29 studies provided data for meta-analysis. Based on the number of studies 296 using each questionnaire (Appendix-2), we undertook quantitative synthesis from studies 297 where means and standard deviation (SD) of questionnaire results was extractable. For HRQoL, 298 SF-36 data was meta-analysed. To maximize available data, we used SD estimates of SF-36 299 summary score from the country-specific general population 55 when studies lacked this 300 information. For sexual function, we meta-analysed SAQ results. BIS results for body image 301 were not meta-analysed due to data insufficiency. Results of FACT-ES and MRS were meta-302 analysed for menopause symptoms, while MENQOL results were not as only one study 303 provided SD. HADS results were meta-analysed for anxiety and depression, while IES and 304 STAI (cancer-related distress) lacked SD. Where data allowed, pre-specified subgroup 305 analyses were undertaken. The fixed-effects meta-analysis results are summarised in Table-   Decreased sexual pleasure, more sexual discomfort, and less frequent sex were reported after 385 RRSO in 13 studies. 34,35,[50][51][52]70,72,74,[76][77][78]81,82 This included both pre-menopausal and post-386 menopausal women. Four studies 50,52,81,82 showed that HRT may mitigate these risks for pre-387 menopausal women but not to pre-surgical levels. Fang 77 reported sexual discomfort improved 388 after 1-year follow-up compared to 6-months, while Mai 76 concluded sexual function declined 389 during 5-years follow-up. In contrast, three studies 75,80,83 found little difference in sexual 390 function post-RRSO vs. no surgery; and also reported little difference in sexual function 391 between pre-vs. post-menopausal RRSO. 83 392 393 Nine of ten studies using SAQ provided data for meta-analysis. However, four studies 72,76,81,82 394 used reversed score for the discomfort component of SAQ, and hence, could not be meta-395 analysed with the remaining studies. Our meta-analysis (Table-4) demonstrated a significant 396 decrease in the pleasure domain (-1.21(95%CI:-1.53,-0.89); I 2 =0%; N=3070), and an increase 397 in the discomfort domain (1.12(95%CI:0.93,1.31); I 2 =0%; N=1400) in women undergoing 398 RRSO vs. no surgery. There was a reduction in sexual pleasure (-0.70(95%CI:-1.33,-0.07); 399 I 2 =0%; N=313) across different timeframes after RRSO (>1-year vs. <1-year). In pre-400 menopausal RRSO, HRT (vs. no HRT) was associated with an increase in sexual pleasure (1.16 401 (95%CI:0.17,2.15); I 2 =0%; N=291) and a decrease in sexual discomfort (-1.20(95%CI:-1.75,-402 0.65); I 2 =0%; N=157). Little difference was reported across any other comparison. 403

COMMENT 457
Findings 458 Our systematic review summarizes published evidence and provides the first meta-analysis of 459 various QoL outcomes following RRS in women with increased BC/OC risk. Overall, HRQoL 460 was unlikely to be negatively affected after RRM or RRSO, although short-term physical 461 deficits were reported in a small number of studies for RRM and RRSO. For RRSO this was 462 supported by a meta-analysis including 1050 women (Table-4). Sexual function appeared 463 negatively affected (reduced sexual frequency, sensation and pleasure) in 4/13 studies post-464 RRM, although this could not be supported by a meta-analysis. However, our meta-analysis in 465 3070 women confirmed RRSO negatively impacted sexual function, particularly with respect 466 to sexual pleasure and sexual discomfort, which were worse in pre-menopausal women not on 467 HRT (Table-4). The evidence on body image after RRM was conflicting, with some studies 468 reporting long-term body image problems despite reconstruction. Body image is not a problem 469 reported post-RRSO, as there is no disfigurement. However, significant menopause symptoms 470 occur, especially in pre-menopausal women after RRSO. This was re-confirmed in our meta-471 analysis of RSSO vs no RRSO in 1745 women for FACT-ES score (  49,60,62,63,66,69,70 and potential deficits with reconstruction 56,57,64,65,67,68 . This is reflected in 498 the disutility of 0.88 which has been reported for RRM. 86 While a number of studies reported 499 reduced cancer-related distress after RRM, one study indicated perceived distress and body 500 image might be worse in BRCA1/2 carriers and women with a strong FH. 65 There is some 501 evidence of a negative impact of RRM with less frequent sex within 2-years post-surgery, 502 compared to after 2-years, although less sexual discomfort was also reported. The potential 503 effects of RRM on sexual function and/or body image should be discussed with women during 504 decision-making. Patient pathways in many/most centers include mandatory appointments with 505 a psychologist as part of the decision-making process. Nevertheless, RRM is cost-effective, 506 has high satisfaction of ~97% and minimal decision regret, 65 which along-with our systematic 507 review findings strongly supports RRM as an acceptable approach for BC prevention. 508 509 Current guidelines including NCCN, RCOG and UK Cancer Genetics Group recommend 510 RRSO as the standard of care for OC-risk reduction for women at increased risk of OC. 19,41,87 511 RRSO is the most clinically effective strategy for reducing OC-risk, it reduces OC mortality 512 and is cost-effective for BRCA1/2 carriers 88 and women >4-5% lifetime OC-risk 32,33 , saving a 513 mean 7-10 life years at this risk threshold. RRSO is normally performed via minimal-access 514 surgery and has a 3-5% complication rate. 89 In pre-menopausal women, RRSO increases the 515 long-term health risks of osteoporosis/osteopenia, heart disease and neurocognitive decline. 36 516 Our review and meta-analysis demonstrate that RRSO is unlikely to affect generic HRQoL, 517 and any short-term deficits usually seem to resolve in the long-term. Nevertheless, RRSO has 518 a negative impact on sexual function in pre-and post-menopausal women. Although sexual 519 function appeared worse in terms of effect size in post-menopausal compared to pre-520 menopausal women, there was a lack of baseline data prior to RRSO which precludes the ability 521 to determine the difference in effect of RRSO between the two groups. Additionally, most 522 studies (12/13) found that post-RRSO women reported de-novo or aggravation of menopause 523 symptoms both in pre-and post-menopausal women. Several studies 50,52,81,82 demonstrated 524 HRT may mitigate menopause symptoms and improve sexual function, and the latter was 525 confirmed in our meta-analysis (Table-4). However, HRT cannot fully resolve menopause 526 symptoms or sexual dysfunction, which remains worse compared to women not undergoing 527 surgery. Short-term HRT in these women appears safe and (if not contraindicated) is 528 recommended till age of natural menopause. 19,36 HRT management following premature 529 surgical menopause is thus critically important for symptom control, sexual function and 530 ameliorating long-term detrimental health consequences. HRT compliance and satisfaction 531 appear higher in women managed in specialist centres or high-risk familial cancer clinics. 36,90 532 RRSO also alleviates cancer-related distress, worry and has high acceptability and satisfaction 533 rates (>85%), 72 although the decision regret rate is much higher in pre-menopausal (~9%) than 534 post-menopausal (~1%) women. 36,37 Women undergoing RRSO should receive non-directive 535 counselling and support highlighting the pros and cons of surgery to facilitate informed 536 decision-making. Emerging data suggests women would like to be offered psychological 537 support and prefer to be managed in specialist clinics. 90 There is an emerging demand for joint 538 RRSO and RRM procedures undertaken concurrently, 37 but relevant QoL outcome data in this 539 context is lacking. 540

541
The detrimental long-term health sequelae, menopause symptoms and sexual dysfunction seen 542 post-RRSO and highlighted in our meta-analysis establishes the importance/need for using 543 HRT, extra efforts to improve symptom management, and study novel approaches like 544 RRESDO. RRESDO has high acceptability among women concerned about menopause/sexual 545 dysfunction, 37 but only two studies report preliminary results. 51,52 Preliminary data from the 546 TUBA study demonstrated improved sexual function and menopause symptoms compared to 547 RRSO with/without HRT. 52 However, the effect size of OC risk-reduction from early-548 salpingectomy and risk of interval cancers remains unknown. Additionally, the long-term 549 impact on menopause or endocrine function is not established. These issues need addressing 550 before recommending change in clinical practice guidelines and widespread 551 implementation. 87,91 RRESDO is not considered standard of care 41 and is currently offered in 552 the context of clinical trials within USA/Europe. 42-44 UK Cancer genetics Group and RCOG 553 recommend RRSO as the primary method of surgical prevention and that early-salpingectomy 554 is best offered in a research setting. 19,87 RRESDO requires comprehensive counselling, ideally 555 in specialist centres, along with thorough pathology evaluation incorporating the SEE-FIM 556 protocol 92 and pelvic peritoneal washings, with any serous tubal intraepithelial carcinoma 557 (STIC) lesions urgently referred for completion surgery and reviewed by a gynaecological 558 oncology MDT. 559 560 Our review summarises the QoL outcomes reported (HRQoL, sexual function, body image, 561 menopause symptoms, psychological well-being) and highlights the various commonly used 562 tools/questionnaires for each of them (Appendix-2). There is a clear need to establish a unified 563 approach and develop core outcome sets for reporting QoL outcomes after RRS to optimise 564 potential evidence synthesis. In addition, the questionnaires/methodologies used precludes the 565 ability to obtain utility scores of RRS from these studies, although the SF-36 used by some 566 could be converted to utility scores using algorithms. 93  To the best of our knowledge, this is the first comprehensive systematic review of all available 574 QoL outcomes after RRS in women at increased BC/OC risk. We followed high standard 575 prospective methodology as per PRISMA guidelines, and provided quantitative QoL outcome 576 data using meta-analysis to support our qualitative results. Sensitivity analysis with random-577 effects models showed similar results to fixed-effects models. Our results can guide future 578 prospective studies to address knowledge gaps and missing or conflicting evidence where 579 applicable. We clearly highlight the outcomes and reporting tools used in measuring QoL post-580 RRS, which can serve as a guide for future trials or evidence synthesis studies. 581

582
We recognise a series of limitations. QoL is a heterogenous topic with several outcomes and 583 many reporting tools/questionnaires. This did not allow a good proportion of the data to be 584 used for meta-analysis for more robust results. An agreed standardised core outcome set for 585 RRS outcomes needs developing. We noted substantial heterogeneity (I 2 >75%) for only two 586 comparisons (Appendix-4), indicating that differences between study populations or 587 procedures might affect results. On several occasions aggregate data was not fully available to 588 include in the meta-analysis, despite contacting the authors. The majority of studies (21/34 589 studies) were assessed high-risk of bias for methodological quality, including short or 590 unspecified duration of follow-up, >5% participants lost to follow-up, and missing sample size 591 calculation. This was considered during qualitative synthesis of data to draw conclusions. Most 592 of our conclusions were compared and found to be in line with the high-quality studies. 593 Similarly studies that were deemed high-risk for external validity bias (9/34 studies) lacked 594

CONCLUSIONS AND IMPLICATIONS 598
RRS may be associated with QoL outcomes. RRM and RRSO are well tolerated procedures, 599 do not seem to impact generic HRQoL, and reduce cancer-related distress and worry. There is 600 strong evidence that RRSO detrimentally affects sexual function, leads to increased menopause 601 symptoms and HRT may mitigate those risks. Limited data suggests RRM may impact sexual 602 function, and studies stress the importance of discussing body image issues despite 603 reconstruction. Effects of RRM and RRSO on QoL should be part of counselling process, and 604 women and clinicians should be aware of the potential effects. RRESDO may be a promising 605 alternative to mitigate QoL-related risks compared to RRSO but ongoing/future trials need to 606 address evidence gaps such as cancer incidence, to properly inform clinical practice. 607 608 Contributors: All authors had full access to all the data in this study and take responsibility 609 for the integrity of the data and the accuracy of the data analysis.

939
J o u r n a l P r e -p r o o f

945
J o u r n a l P r e -p r o o f   J o u r n a l P r e -p r o o f