Breast cancer is the commonest malignancy and leading cause of cancer‐related mortality in women. Breast‐conserving surgery (BCS) with radiotherapy or mastectomy are recommended treatments, with comparable oncological outcomes. Autologous abdominal‐based free flap and implant‐based procedures are the approaches used most frequently in immediate breast reconstruction (BRR). Autologous BRR has the inherent advantage of using the patient's own tissues, taken from a different part of the body where there is excess fat and skin, to restore breast volume and appearance after mastectomy. Various donor sites can be used, most commonly the abdomen.
Adjuvant locoregional postmastectomy radiotherapy (PMRT) of the chest wall, and potentially of the regional lymph nodes, has been indicated historically for locally advanced disease. These indications increased following the Early Breast Cancer Trialists' Collaborative Group meta‐analyses, which showed significantly improved disease‐free and overall survival after PMRT and regional node irradiation in women at intermediate risk (tumour size 50 mm or less and 1–3 positive lymph nodes). Newly proposed US guidelines emphasize the need to consider the lower recurrence rates associated with contemporary practice and the benefits of systemic therapy. Current recommendations for PMRT in the intermediate‐risk group remain controversial, pending the results of the SUPREMO (Selective Use of Postoperative Radiotherapy aftEr MastectOmy) trial, evaluating chest wall and/or axillary radiotherapy.
Adjuvant radiotherapy (PMRT) may have deleterious effects on breast cosmetic outcomes, quality of life (QOL) and surgical complications after immediate BRR. Previous studies evaluating the impact of PMRT on types of immediate BRR showed its potential feasibility in this setting, with lower morbidity rates compared with those of implant‐based procedures. Surprisingly, the rapid adoption of immediate implant‐based reconstruction in about 70 per cent of women, compared with 34 per cent of autologous procedures when PMRT is recommended, may be influenced by surgeon and patient preferences, regardless of current evidence.
Increasing recommendations for PMRT and immediate BRR have prompted a need to consider their optimal sequence. Previous systematic reviews have not provided clarity concerning the choice between immediate and delayed BRR. Despite this, immediate autologous BRR is commonly recommended in the setting of PMRT, given the potential long‐term benefits on patients' QOL and breast cosmetic satisfaction. Currently, immediate autologous BRR and PMRT recommendations are variable. A systematic review in 2011 showed methodological variations in the definitions of surgical complications, precluding interstudy comparisons.
Complications of autologous breast reconstruction with PMRT include: poor wound‐healing, flap‐related fat necrosis, fibrosis and contracture, which reduce breast volume. Surgical complications contribute variably to decreased patient satisfaction and impaired cosmetic outcomes. A standardized core set of outcomes for BRR has been proposed involving a range of complications, including flap‐related complications and the need for further unplanned surgery. The BRR core outcome set has yet to recommend a standardized measurement tool for evaluating surgical complications. Most surgeons use the Clavien–Dindo classification (CDC). Patient‐reported QOL outcomes using validated BRR questionnaires, such as the BREAST‐Q and the European Organisation for Research and Treatment of Cancer (EORTC) Quality‐of‐Life Questionnaire (QLQ)‐BRECON23, are recommended to evaluate comparative effectiveness.
This systematic review aimed to evaluate the quality and strengths of the current evidence regarding surgical complications in autologous abdominal flaps in the context of the receipt and timing of radiotherapy related to PMRT and, less commonly, neoadjuvant radiotherapy, generally administered before skin‐sparing mastectomy and immediate breast reconstruction, including assessment of QOL.
The protocol was registered and published on the Prospective Register of Systematic Reviews PROSPERO (CRD42017077945). The authors adhered to the PRISMA statement.
A comprehensive search of the MEDLINE (Ovid SP), Embase (Ovid SP), Google Scholar, Cochrane Controlled Register of Trials (CENTRAL), Science citation index databases and
Database‐related searches were entered into an EndNote™ X8 library (Clarivate Analytics, Philadelphia, Pennsylvania, USA). Study screening was performed independently in two stages by two investigators using prespecified screening criteria.
In stage 1, two authors independently screened titles and abstracts. Discrepancies were resolved by consensus with the senior author. Remaining doubts regarding an article resulted in a review of the complete publication.
In stage 2, full‐text studies from stage 1 were screened independently for their eligibility by two reviewers. Discrepancies were resolved by consensus with a third reviewer. Authors of eligible studies were contacted (via e‐mail) to reconcile any methodological issues or to provide more detailed information on data for individual types of autologous flap.
All primary human studies evaluating surgical complications for autologous free flap (microvascular) abdominal BRR in breast cancer and types of radiotherapy (PMRT, neoadjuvant and no radiotherapy) were included. Outcomes also included patient‐reported QOL and cosmetic assessments. Radiotherapy groups were compared with a control or no radiotherapy group in comparative studies, compatible with immediate and delayed BRR. Commonly performed autologous abdominal flaps included: deep inferior epigastric perforator (DIEP), transverse rectus abdominis myocutaneous (TRAM) and the superficial inferior epigastric artery perforator (SIEA).
Inclusion criteria were: women aged at least 18 years with a diagnosis of invasive breast cancer (TNM categories: T0–3, N1–3, Mx, M0), undergoing immediate or delayed abdominal autologous BRR using free flaps (DIEP, TRAM or SIEA) who received adjuvant radiotherapy (PMRT), neoadjuvant radiotherapy or no radiotherapy.
Clinical studies that involved at least 50 patients were included (RCTs, prospective and retrospective comparative observational studies, and case series).
Review articles, conference abstracts, simulation studies and clinical studies in non‐human subjects were not included, along with studies involving patients who received segmental or partial mastectomy, technical descriptions of operative repair with no outcome measures, BRR unrelated to breast cancer, implant‐based reconstructions and other non‐abdominal autologous flaps.
Cochrane's ROBINS‐I (Risk Of Bias In Non‐randomised Studies – of Interventions) tool was used for comparative studies. This comprises seven domains from which the risk of bias may be ascertained to produce an overall risk‐of‐bias score. The Grading of Recommendations, Assessment, Development, and Evaluations (GRADE) tool was used to evaluate the methodological quality of individual studies.
Primary outcomes were surgical complications including: Clavien–Dindo classification (CDC) grades II and III26; partial flap loss; total flap loss; fat necrosis (CDC grades, when reported); number(s) of unplanned reoperations for surgical complications (excluding cosmetic revisions); and number(s) of total complications. A surgical complication was defined as an adverse, postoperative, surgery‐related event that required additional treatment. If CDC grades were not defined, the complications reported by the included studies were graded retrospectively according to the CDC by two independent authors; any discrepancy was discussed and agreed with the senior author.
Secondary outcomes were assessed using patient‐reported QOL‐validated questionnaires (COnsensus‐based Standards for the Selection of health Measurement INstruments (COSMIN), Breast Questionnaire (BREAST‐Q), the EORTC Quality‐of‐Life Questionnaire (QLQ) – Breast Cancer 23, the Quality‐of‐Life Cancer Generic Questionnaire (QLQ‐C30), the Numerical Pain Rating Scale (NPRS), the Patient‐Reported Outcomes Measurement Information System – Profile 29 (PROMIS‐29), the McGill Pain Questionnaire (MPQ), the Generalized Anxiety Disorder Scale (GAD‐7) and the Patient Health Questionnaire (PHQ‐9)), as well as assessment of cosmetic outcomes using independent panel or self assessments of medical photographs, and surface imaging using the Vectra® XT three‐dimensional system (Canfield Scientific, Parsippany, New Jersey, USA).
Two authors independently extracted data from full‐text articles using a standard data form. Any discrepancies were resolved by consensus with a third reviewer. Reporting authors of original articles were contacted on up to two occasions relating to missing data or where additional information was required.
Data extraction included: first author, year of publication, study design, study setting, number of centres, duration of follow‐up, study population and participant demographics (mean age, BMI, smoking, co‐morbidities).
Surgical complications were recorded using CDC: grades II–III26. Two authors reviewed eligible studies and classified each complication according to the CDC if unreported.
QOL and cosmetic outcomes were listed.
When two or more studies reported outcome data, these were pooled using Review Manager 5.3 software (The Cochrane Collaboration, The Nordic Cochrane Centre, Copenhagen, Denmark). Odds ratios with 95 per cent confidence intervals were used to evaluate dichotomous outcomes (surgical complications). Standard mean differences (with 95 per cent c.i.) were used for continuous outcomes between treatment groups. Rates of each complication (fat necrosis, partial and total flap loss, infection and wound complications (dehiscence and delayed wound healing)) were compared for PMRT (versus no radiotherapy) and neoadjuvant radiotherapy (versus no radiotherapy). Data were also pooled to provide an overall summary measure of combined radiotherapy (adjuvant and neoadjuvant) compared with no radiotherapy.
Heterogeneity between studies was assessed in Review Manager 5.3 using the Higgins and Thompson I2 statistic. Levels of heterogeneity were defined as: low (I2 less than 50 per cent), moderate (I2 = 50–80 per cent) and high (I2 above 80 per cent). A random‐effects model was used for cohorts with heterogeneity (I2 above 50 per cent). As heterogeneity was generally moderate or high, and outcome measures differed between studies, these were combined using the DerSimonian and Laird random‐effects model. Results of meta‐analyses are shown as forest plots. A sensitivity analysis was performed where possible, to evaluate whether outcomes differed when restricting the analysis exclusively to high‐quality studies.
Clinically meaningful differences in QOL items/questions or domain scores may vary depending on response shift, that is a change in the meaning of QOL scores over time. This is relevant in longitudinal studies and may influence clinical significance, defined as greater than 5‐point score differences for EORTC QLQ‐C30 and QLQ‐BR23. Clinically meaningful differences are currently being evaluated using a number of methods such as qualitative interviews and using predefined clinical anchors. Clinically meaningful differences in QOL should be differentiated from statistical significance. BREAST‐Q findings have been compared with large population‐derived normative data, facilitating clinically meaningful interpretation of data.
A total of 697 studies were identified. Of these, 12 studies (including 1756 patients) evaluated adjuvant radiotherapy (350 patients), neoadjuvant radiotherapy (723) and no radiotherapy (683) (Fig. ). There were three prospective study designs and nine that were retrospective, but no RCTs. There were two multicentre (1 prospective and 1 retrospective) and ten single‐centre studies (2 prospective and 8 retrospective) (Table ). Study quality (GRADE) was low in eight studies and moderate in the other four, with an overall high risk of bias. A summary of baseline characteristics, including numbers of centres, country of origin, dates, patient numbers, breast cancer pathology and adjuvant medical treatments in comparative adjuvant and neoadjuvant radiotherapy groups, including non‐comparative studies, is provided in Table S1 (supporting information).
Study summaries: comparative adjuvant or neoadjuvant radiotherapy in autologous breast reconstruction, and non‐comparative studies (adjuvant radiotherapy or neoadjuvant radiotherapy only)
| Reference | Years | Country | No. of centres | Type of BRR flap | Overall follow‐up (months) | Group differences in baseline characteristics¶ | RT dose and regimen |
| Baumann et al.‡ | 2005–2009 | USA | 1 | msTRAM; DIEP; SIEA | 11* | n.a. | Total 60 Gy; missing details |
| Billig et al.§ | 2012–2017 | USA and Canada | 11 | TRAM; DIEP; SIEA | 24 | Adjuvant RT: more non‐Hispanic patients (P = 0·001), bilateral BRR (P = 0·002), DIEP/SIEA (P < 0·001), adjuvant chemotherapy (P < 0·001); less TRAM (P < 0·001)# | Total 50·4 Gy over 4 weeks, daily (28 fractions of 1·8 Gy) |
| Chatterjee et al.§ | 1995–2005 | UK | 1 | DIEP | 42 (12–120)† | Adjuvant RT: more IDC (P = 0·02), LVI (P = 0·044), positive axillary LN (P < 0·001) | Total 45 Gy over 4 weeks (20 fractions) |
| Cooke et al.§ | 2012–2015 | Canada | 1 | DIEP; SIEA | 12 | Adjuvant RT: higher TNM staging, positive LN, more chemotherapy (P values not provided) | Total 50/50·4 Gy over 4 weeks, daily (25 fractions of 2 Gy/28 fractions of 1·8 Gy) |
| Huang et al.‡ | 1997–2001 | Taiwan | 1 | TRAM | 40 (24–74)† | n.a. | Total 50 Gy; missing details |
| Levine et al.‡ | 1999–2011 | USA | 1 | msTRAM; DIEP; SIEA | 22·7* | n.a. | Missing details |
| Modarressi et al.‡ | 2007–2013 | Switzerland | 1 | DIEP | 1 | n.a. | Missing details |
| Mull et al.‡ | 2003–2014 | USA | 1 | msTRAM; TRAM; DIEP | 1 | Neoadjuvant RT: more chemotherapy (P < 0·01), higher TNM staging (P < 0·01); less hypertension/CAD (P = 0·03) | Missing details |
| O'Connell et al.‡ | 2009–2014 | UK | 1 | DIEP | 44·3 (i.q.r. 31·1–56·4)† | Adjuvant and neoadjuvant RT: more chemotherapy and endocrine therapy as less DCIS/less advanced invasive disease (P values not provided) | Total 40 Gy over 3 weeks (15 fractions) |
| Peeters et al.‡ | 1997–2003 | Belgium | 2 | DIEP | ≥ 12 | n.a. | Total 50 Gy; missing details |
| Rogers and Allen‡ | 1994–1999 | USA | 1 | DIEP | 18·7* | n.a. | Total 50·5 Gy over 6·5 weeks (missing details) |
| Temple et al.‡ | 1990–2001 | USA | 1 | TRAM | ≥ 12 | n.a. | Total 58 Gy; missing details |
| No. of patients | Follow‐up (months) | Total no. of complications | No. of reoperations for complications | |||||||
| Reference | GRADE | ROBINS‐I | Adjuvant RT | No adjuvant RT | Adjuvant RT | No adjuvant RT | Adjuvant RT | No adjuvant RT | Adjuvant RT | No adjuvant RT |
| Chatterjee et al. | Low | Serious | 22 | 46 | 54 | 36 | n.a. | n.a. | n.a. | n.a. |
| Cooke et al. | Moderate | Moderate | 64 | 61 | 12 | 12 | 20 | 16 | 6 | 1 |
| O'Connell et al. | Low | Serious | 28 | 80 | 27·5 | 48·7 | 11 | 20 | 4 | 8 |
| Peeters et al. | Low | Serious | 16 | 109 | ≥ 12 | ≥ 12 | n.a. | n.a. | n.a. | n.a. |
| Rogers and Allen | Low | Serious | 30 | 30 | 19·9 | 17·4 | 65 | 41 | 32 | 26 |
| Billig et al. | Moderate | Moderate | 108 | n.a. | 24 | n.a. | 81 | n.a. | 5 | n.a. |
| Huang et al. | Low | Serious | 82 | n.a. | 40 | n.a. | 131 | n.a. | 5 | n.a. |
| Adjuvant RT versus no adjuvant RT | |||||||
| Clavien‐Dindo complication grade† | |||||||
| Reference | Total flap loss | Partial flap loss* | Fat necrosis* | Wound dehiscence and delayed wound healing* | II | IIIa | IIIb |
| Chatterjee et al. | n.a. | n.a. | n.a. | n.a. | n.a. | n.a. | n.a. |
| Cooke et al. | 0 versus 0 | 9 versus 6 | 2 versus 1 | 3 versus 5 | 2 versus 4 | n.a. | 6 versus 1 |
| O'Connell et al. | 0 versus 0 | 0 versus 0 | 1 versus 2 | 4 versus 9 | 3 versus 3 | 3 versus 3 | 1 versus 5 |
| Peeters et al. | n.a. | n.a. | 6 versus 36 | n.a. | n.a. | n.a. | n.a. |
| Rogers and Allen | n.a. | n.a. | 7 versus 0‡ | 11 versus 8 | 5 versus 7 | 7 versus 0 | 25 versus 26 |
| Billig et al. | 0 versus n.a. | n.a. | 4 versus n.a. | 17 versus n.a. | 8 versus n.a. | n.a. | 5 versus n.a. |
| Huang et al. | 0 versus n.a. | n.a. | 7 versus n.a. | n.a. | 82 versus n.a. | 5 versus n.a. | n.a. |
| No. of patients | Follow‐up (months) | Total no. of complications | No. of reoperations for complications | |||||||
| Reference | GRADE | ROBINS‐I | Neoadjuvant RT | No neoadjuvant RT | Neoadjuvant RT | No neoadjuvant RT | Neoadjuvant RT | No neoadjuvant RT | Neoadjuvant RT | No neoadjuvant RT |
| Modarressi et al. | Low | Serious | 60 | 45 | 1 | 1 | 20 | 9 | n.a. | n.a. |
| Mull et al. | Low | Serious | 142 | 312 | 1 | 1 | 26 | 45 | 26 | 45 |
| O'Connell et al. | Low | Serious | 38 | 80 | 50·3* | 48·7* | 12 | 20 | 3 | 8 |
| Peeters et al. | Low | Serious | 77 | 109 | ≥ 12 | ≥ 12 | n.a. | n.a. | n.a. | n.a. |
| Baumann et al. | Moderate | Moderate | 189 | n.a. | 11† | n.a. | 88 | n.a. | 69 | n.a. |
| Billig et al. | Moderate | Moderate | 67 | n.a. | 24 | n.a. | 37 | n.a. | 1 | n.a. |
| Levine et al. | Moderate | Moderate | 50 | n.a. | 22·7† | n.a. | n.a. | n.a. | 3 | n.a. |
| Temple et al. | Low | Serious | 100 | n.a. | ≥ 12 | n.a. | 41 | n.a. | 18 | n.a. |
| Neoadjuvant RT versus no neoadjuvant RT | |||||||
| Clavien‐Dindo complication grade† | |||||||
| Reference | Total flap loss | Partial flap loss* | Fat necrosis* | Wound dehiscence and delayed wound healing* | II | IIIa | IIIb |
| Modarressi et al. | 2 versus 1 | 12 versus 2 | n.a. | n.a. | n.a. | n.a. | n.a. |
| Mull et al. | 5 versus 15 | 7 versus 5‡ | n.a. | n.a. | n.a. | n.a. | 26 versus 45 |
| O'Connell et al. | 0 versus 0 | 0 versus 0 | 2 versus 2 | 7 versus 9 | 2 versus 3 | 0 versus 3 | 3 versus 5 |
| Peeters et al. | n.a. | n.a. | 29 versus 36 | n.a. | n.a. | n.a. | n.a. |
| Baumann et al. | 5 versus n.a. | 14 versus n.a. | 15 versus n.a. | 22 versus n.a. | 4 versus n.a. | n.a. | 69 versus n.a. |
| Billig et al. | 0 versus n.a. | n.a. | 7 versus n.a. | 11 versus n.a. | 4 versus n.a. | n.a. | 1 versus n.a. |
| Levine et al. | n.a. | 1 versus n.a. | n.a. | n.a. | n.a. | n.a. | n.a. |
| Temple et al. | 2 versus n.a. | 7 versus n.a. | 16 versus n.a. | n.a. | n.a. | n.a. | 18 versus n.a. |
No study prospectively graded surgical complications according to an accepted classification such as CDC (fat necrosis, partial or total flap loss, infection and wound complications). One study graded partial flap loss using a novel flap necrosis classification system, adapted from Kwok et al.. Only 30 per cent of all surgical complications (30 of 99) reported across the 12 included studies were defined a priori.
Meta‐analyses comparing PMRT (350 patients; mean follow‐up 27·1 (range 12·0–54·0) months) and no radiotherapy (326 patients; mean follow‐up 25·2 (12·0–48·7) months) showed no interstudy differences in rates of: overall complications (233 patients; odds ratio (OR) 1·52 (95 per cent c.i. 0·84 to 2·75), Z = 1·40, P = 0·160) (Fig. a); CDC grade III surgical complications (233 patients; OR 2·35 (0·63 to 8·81), Z = 1·27, P = 0·200) (Fig. b); CDC grade II (293 patients; OR 0·94 (0·32 to 2·76), Z = 0·11, P = 0·910) (Fig. c); or fat necrosis (418 patients; OR 1·83 (0·67 to 5·00), Z = 1·18, P = 0·240) (Fig. d). There were no differences in rates of infection (293 patients; OR 0·94 (0·32 to 2·76), Z = 0·11, P = 0·910) (Fig. S1a, supporting information) or wound complications (293 patients; OR 1·16 (0·56 to 2·39), Z = 0·40, P = 0·690) (Fig. S1b, supporting information). There were no total flap losses.
Enlarge this image.Forest plots comparing adjuvant radiotherapy with no radiotherapya Overall complications, b Clavien–Dindo classification (CDC) grade III complications, c CDC grade II complications, d fat necrosis. A Mantel–Haenszel random‐effects model was used for meta‐analysis. Odds ratios are shown with 95 per cent confidence intervals. RT, radiotherapy.
Comparisons between neoadjuvant radiotherapy (723 patients; mean follow‐up 16·8 (range 1·0–50·3) months) and no radiotherapy (546 patients; mean follow‐up 15·7 (1·0–48·7) months) showed no differences in overall complications (677 patients; OR 1·45 (95 per cent c.i. 0·97 to 2·18), Z = 1·82, P = 0·070) (Fig. a) and CDC grade III surgical complications (572 patients; OR 1·24 (0·76 to 2·04), Z = 0·85, P = 0·390) (Fig. b). One comparative study reported similar CDC grade II complications between neoadjuvant and no radiotherapy (118 patients; OR 1·43 (0·23 to 8·91), Z = 0·38, P = 0·700). There were no differences in rates of fat necrosis (304 patients; OR 1·29 (0·72 to 2·30), Z = 0·85, P = 0·400) (Fig. c). Rates of partial flap loss were higher for neoadjuvant radiotherapy than for no radiotherapy (559 patients; OR 3·85 (1·51 to 9·76), Z = 2·83, P = 0·005) (Fig. S2a, supporting information), with no differences in rates of total flap loss (559 patients; OR 0·81 (0·31 to 2·09), Z = 0·44, P = 0·660) (Fig. S2b, supporting information).
Enlarge this image.Forest plot comparing neoadjuvant radiotherapy with no radiotherapya Overall complications, b Clavien–Dindo classification (CDC) grade III complications, c fat necrosis. A Mantel–Haenszel random‐effects model was used for meta‐analysis. Odds ratios are shown with 95 per cent confidence intervals. RT, radiotherapy.
Meta‐analyses of pooled PMRT and neoadjuvant radiotherapy compared with pooled no radiotherapy groups (mean follow‐up 18·3 (range 1·0–48·7) months) were performed as a potential hypothesis‐generating exercise. This showed significantly higher overall complications in the combined radiotherapy groups compared with no radiotherapy (830 patients; OR 1·46 (95 per cent c.i. 1·04 to 2·07), Z = 2·16, P = 0·030) (Fig. a). There were no interstudy differences in: CDC grade III complications (725 patients; OR 1·38 (0·83 to 2·32), Z = 1·24, P = 0·220) (Fig. b); CDC grade II complications (331 patients; OR 0·89 (0·37 to 2·10), Z = 0·28, P = 0·780) (Fig. S3a, supporting information); rates of fat necrosis (533 patients; OR 1·59 (0·96 to 2·64), Z = 1·79, P = 0·070) (Fig. c); or emergency reoperations for complications (725 patients; OR 1·38 (0·83 to 2·32), Z = 1·24, P = 0·220) (Fig. S3b, supporting information). Rates of partial flap loss were also higher in the combined versus no radiotherapy groups (684 patients; OR 2·59 (1·27 to 5·28), Z = 2·63, P = 0·009) (Fig. S3c, supporting information), with no differences in rates of total flap loss (559 patients; OR 0·81 (0·31 to 2·09), Z = 0·44, P = 0·660) (Fig. S3d, supporting information), infection (331 patients; OR 0·89 (0·37 to 2·10), Z = 0·28, P = 0·780) (Fig. S3e, supporting information) or wound complications (dehiscence/delayed wound healing) (331 patients; OR 1·29 (0·68 to 2·47), Z = 0·78, P = 0·430) (Fig. S3f, supporting 1information).
Enlarge this image.Forest plot comparing combined adjuvant and neoadjuvant radiotherapy with no radiotherapya Overall complications, b Clavien–Dindo classification (CDC) grade III complications, c fat necrosis. A Mantel–Haenszel random‐effects model was used for meta‐analysis. Odds ratios are shown with 95 per cent confidence intervals. RT, radiotherapy.
Clinical outcomes within studies of PMRT versus no radiotherapy were homogeneous (I2 values below 50 per cent). All remaining meta‐analyses of outcomes were similar (neoadjuvant radiotherapy versus no radiotherapy, pooled PMRT and neoadjuvant radiotherapy versus no radiotherapy).
There was limited reporting of patient‐reported QOL; outcomes were detailed in only two prospective studies and one retrospective study, with small patient numbers and short follow‐ups for the PMRT groups. A priori hypothesis‐driven selection of QOL domains was absent from methods, with no reporting of missing data or how this problem was tackled.
Three studies used the BREAST‐Q and one used the breast cancer‐specific questionnaire (EORTC QLQ‐BR23). One small study reported significantly better ‘satisfaction with breast’ (P = 0·008) after a median follow‐up of 27·5 months for PMRT compared with 48·7 months for no radiotherapy (Table S2, supporting information). The moderate‐quality comparative prospective study found a significant adverse impact of PMRT on breast symptoms at 1 year (P < 0·001) compared with no radiotherapy (Table S2, supporting information).
The third study evaluated serial QOL outcomes, concluding a significant impact of PMRT on QOL domains (BREAST‐Q) at 1 and 2 years, despite the absence of a control group (no radiotherapy). Moreover, clinical significance was defined as P = 0·05, which may not account for multiple variables (Table S2, supporting information). Highly significant abdominal adverse effects in a small patient group (108 patients) may be unrelated to PMRT, but rather an indication of donor site morbidity. Interestingly, when evaluating the impact of neoadjuvant radiotherapy in a small non‐comparative study, significant time‐related improvements in most QOL domains were observed, except lower physical well‐being relating to the abdomen at 1 year (Table S3, supporting information).
Three studies evaluated PMRT and the effects on aesthetic outcomes (187 patients). There was no standardized evaluation of cosmetic outcomes, precluding meta‐analyses. Studies lacked robust methodology.
The mixture of underpowered observational studies included in this review were, in large part, lacking contemporaneous data to reflect current practice. Most were retrospective single‐centre cohorts, demonstrating poor levels of clinical evidence (levels 3 and 4) with insufficient follow‐up11.
A previous study of over 40 000 women undergoing BRR in 134 studies found that only 20 per cent reported a priori surgical complications, as well as inconsistent interstudy definitions. The present review found similar interstudy discrepancies, without uniform adoption of the CDC. The present authors graded all reported surgical complications using the CDC. All surgical interventions were graded as CDC IIIa or IIIb, and surgical reoperations were differentiated according to whether they were for complications or cosmetic revisions. Some complications were not amenable to retrospective grading in three studies. In one, it was not possible to determine whether fat necrosis required surgical revision for each radiotherapy group (adjuvant or neoadjuvant), compared with no radiotherapy. A second omitted individual abdominal complications relative to timings of radiotherapy, and the third omitted overall numbers of complications. Reviewed studies also failed to define postoperative wound infections according to Centers for Disease Control and Prevention criteria.
The IDEAL (Idea, Development, Exploration, Assessment, Long‐term study) Collaboration describes key methodological criteria for robust prospective cohort studies: studies should be powered on the effect size of primary outcomes evaluating interventions of interest. The Mastectomy and Breast Reconstruction Outcomes Collaborative (MROC) is a multicentre prospective cohort study that provides IDEAL level 2b evidence for clinical safety and satisfactory QOL outcomes in the evaluation of surgical complications in immediate autologous reconstructions with PMRT versus no radiotherapy (delayed BRR) in 11 US centres. The MROC cohort data were excluded from this systematic review based on its reporting of group‐related summative data for all types of autologous reconstruction, as opposed to individual abdominal donor sites.
The MROC has reported all surgical complications at 2 years and demonstrated that PMRT (versus no radiotherapy) was significantly associated with a greater risk of developing any complication (OR 1·50 (95 per cent c.i. 1·20 to 1·86); P < 0·001), reoperative complications (OR 1·52 (1·17 to 1·97); P < 0·002) and wound infection (OR 2·77 (1·78 to 4·31); P < 0·001). Autologous BRR was done more commonly in irradiated than non‐irradiated patients (38 versus 25 per cent respectively; P < 0·001), with similarly low rates (1–2·4 per cent) of reconstruction failure at 2 years.
Eligible studies in the present systematic review were significantly underpowered in comparison with the MROC study, which evaluated irradiated autologous BRR at 1 year (236 patients) and 2 years (199), and non‐irradiated procedures at 1 year (1625) and 2 years (332). The MROC data showed no differences between radiotherapy and no radiotherapy groups in the rates of total complications (25·6 versus 28·3 per cent respectively), major complications (17·6 versus 22·9 per cent) or flap failure (1·0 versus 2·4 per cent) at 2 years after immediate autologous reconstruction. Studies in the present review showed significantly lower rates of major complications after radiotherapy compared with the MROC results, suggesting suboptimal overall reporting of surgical complications in the reviewed studies.
The retrospective grading of surgical complications in the two moderate‐quality studies reported showed a rate of major complications (CDC grade IIIb) of 9 per cent (6 of 64) at 1 year, and 4·6 per cent (5 of 108) at 2 years. These rates are also likely to reflect under‐reporting compared with the MROC rates of 14·8 per cent (35 of 236) at 1 year and 17·6 per cent (35 of 199) at 2 years. Despite its strengths, the MROC cohort is based on the review of complications from electronic patient records, potentially also underestimating true complication rates.
One way to measure what matters to patients is to use patient‐reported outcome measures (PROMs) to assess the effects of disease or treatment on symptoms, functioning and health‐related QOL. In this systematic review, PROMs were poorly reported and underpowered for overall small effect sizes of individual QOL domains. Preliminary conclusions regarding statistical significance were not substantiated by adequate patient numbers, lack of a comparator group or prospectively defined time points for questionnaire collection. Standardized and objective evaluations of cosmetic outcome have also remained elusive with emerging adoption of newer technologies such as the Vectra® XT. Robust study designs evaluating these innovations should be accompanied by surgery‐ and disease‐specific questionnaires.
Clear recommendations for the optimal timing of radiotherapy in relation to autologous BRR will remain elusive until information from high‐quality systematic reviews forms part of shared preoperative decision‐making.
Adequately powered prospective studies and ongoing audits, to allow comparisons of postoperative radiotherapy with neoadjuvant radiotherapy, are warranted. Current evidence for irradiating autologous abdominal flaps remains weak, involving only two moderate‐quality studies of the 12 included in this report. Future cohort studies should be designed and powered to take advantage of newly evolving study designs, such as multiple‐cohort RCTs or trials within cohorts. These designs permit collection of big data within registry or cohort platforms, and allow multiple synchronous randomized trials to be conducted in a cost‐effective manner.
The authors thank R. Davidson, who assisted in the formatting of tables and figures for this publication, and K. Cocks (Chartered Medical Statistician in QOL and clinical trials, Select Statistics, UK) for her advice on ‘clinically meaningful differences’ in QOL assessment.
A.K. is a Kellogg Scholar at the University of Oxford and receives funding equating to the scholarship amount. A.L.P. is the co‐developer of the BREAST‐Q and receives royalties when the BREAST‐Q is used in industry‐sponsored clinical trials. Z.E.W. is the co‐developer of the EORTC QLQ‐BRECON23.
Disclosure: The authors declare no other conflict of interest.
© 2020. This work is published under http://creativecommons.org/licenses/by/4.0/ (the “License”). Notwithstanding the ProQuest Terms and Conditions, you may use this content in accordance with the terms of the License.
Details
; Charles, W N 2 ; Prokopenko, M 3 ; Beswick, A 4 ; Pusic, A L 5 ; Mosahebi, A 3 ; Dodwell, D J 6 ; Winters, Z E 7
1 Kellogg College, Nuffield Department of Surgery, University of Oxford, Oxford, UK; Department of Surgery and Cancer, Imperial College London, London, UK
2 Department of Surgery and Cancer, Imperial College London, London, UK
3 Department of Plastic Surgery, Royal Free Hospital, London, UK
4 School of Clinical Sciences, University of Bristol, Bristol, UK
5 Patient‐Reported Outcomes, Value and Experience Centre, Brigham and Women's Hospital, Harvard Medical School, Boston, Massachusetts, USA
6 Nuffield Department of Population Health, University of Oxford, Oxford, UK
7 Surgical Intervention Trials Unit, Division of Surgery and Interventional Science, University College London, London, UK