Fluid accumulation after mastectomy and/or axillary lymph node dissection (ALND) is among the most common issues surgeons have to face in the early postoperative management and a major factor contributing to the significant financial burden of breast care management.1,2,3,4,5 When the collection of serous fluid becomes clinically evident and symptomatic, it is defined as a seroma.6 Seroma is the most common complication after mastectomy, with an incidence rate of up to 60%, and a significant cause of perioperative morbidity and patient discomfort.3,4,5,6,7 Moreover, the presence of seroma increases wound complication rates and hospital stay, thus delaying recovery and the initiation of adjuvant therapy.8 Axillary dissection,9 obesity,10 neoadjuvant therapy,11 smoking,12 and electrocautery use13 have been proposed as risk factors for seroma formation and longer duration of drainage. Conventionally, mastectomy flaps are closed by approximating the edges, while fluid accumulation in the dead space, beneath the skin flaps, and in the axilla, is reduced through closed-suction drainage. Apart from suction drainage, other techniques have been described to reduce dead space, including quilting sutures (QS) of the skin flaps to the underlying pectoralis fascia, quilting of the axillary space, tissue adhesives,14,15 gentamicin-collagen sponges,16 etc. Using QS to aid in tissue approximation and decrease dead space is proposed as a simple and cheap intraoperative technique to reduce the seroma rate, thus improving wound outcomes and decreasing hospital stay and costs of care. We aimed to perform a systematic review, and analyse, in a meta-analytical model, the role of QS in improving wound outcomes and decrease the volume of drainage, duration of drain removal, and length of stay in hospital.

Materials and Methods

Literature Search and Study Selection

The study was registered with PROSPERO (International Prospective Register of Systematic Reviews), and had a study ID of CRD42021255268. A systematic search of the PubMed, EMBASE, and SCOPUS databases was performed for all comparative studies examining surgical outcomes in patients who underwent QS versus conventional closure (CC) after mastectomy with or without ALND. The following search algorithm was used: (quilting) OR (closure) OR (suture) AND (mastectomy) AND (seroma). Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines were used as the search protocol and the PRISMA checklist was followed to conduct the methodology17 (Fig. 1). Inclusion criteria were used according to the Problem, Intervention, Comparison and Outcome (PICO) formula. The latest search was performed on 3 April 2021. Two authors (SM and CC) assessed the titles and abstracts of studies found in the search and the full texts of potentially eligible trials were reviewed. Disagreements were resolved by consensus-based discussion. The Newcastle–Ottawa scale (Table 1) and the ROBINS-I tool (Fig. 2) were used to quantify the quality of eligible studies. The references of full texts reviewed were further screened for additional eligible studies. The corresponding author was contacted to clarify data extraction if additional information was necessary.

Fig. 1
figure 1

PRISMA diagram. PRISMA Preferred Reporting Items for Systematic Reviews and Meta-Analyses

Full size image
Table 1 Study characteristics
Full size table
Fig. 2
figure 2

ROBINS-I risk-of-bias assessment. Assessment of the risk of bias was performed by two authors (SM and CC). Each study was classified as low/moderate/serious risk for each of the seven domains. Disagreements were resolved via consensus.

Full size image

Eligibility Criteria

Studies written in English that included comparative surgical data between quilting versus no quilting sutures (CC with or without drainage) post mastectomy ± ALND were assessed for eligibility. The primary endpoints were seroma formation and total volume of drainage, while secondary endpoints included patient characteristics, wound complications, and early postoperative outcomes. Long-term and oncological outcomes were not recorded. Patients with immediate breast reconstruction were excluded as were male patients and studies including only ALND, without mastectomy. Studies without comparative data or with comparison between QS and tissue adhesives were also not included, along with studies in which QS were not the only intervention in the study group (SG).

Data Extraction and Outcomes

For each eligible study, the following data were recorded: author names, journal, year of publication, study type, total number of patients and number of patients included in each group, skin flaps quilting technique, axillary quilting, number of drains used in each group, mean age, mean body mass index (BMI), smoking status, and lymph node yield. The type of procedure was recorded and was defined as modified radical mastectomy (MRM), mastectomy only, ALND, sentinel lymph node biopsy (SLNB), and lumpectomy. For each study, the outcomes of interest were extracted and grouped into three main categories, which were further analyzed: (1) risk factors for seroma formation (BMI, neoadjuvant therapy, operative time, lymph node yield); (2) drainage outcomes (seroma, total volume of drained fluid, days to drain removal, length of stay); and (3) wound outcomes (surgical site infection, hematoma, flap necrosis, skin dimpling).

Statistical Analysis

Random-effects models were used to measure all pooled outcomes as described by Der Simonian and Laird18 and the odds ratio (OR) was estimated with its variance and 95% confidence interval (CI). The random-effects analysis weighted the natural logarithm of each study’s OR by the inverse of its variance plus an estimate of the between-study variance in the presence of between-study heterogeneity. As described previously,19 heterogeneity between ORs for the same outcome between different studies was assessed using the I2 inconsistency test and Chi-square-based Cochran’s Q statistic test20 in which a p-value <0.05 is taken to indicate the presence of significant heterogeneity. Analyses were conducted using Review Manager 5.3.

Results

Eligible Studies

Twenty-one studies21,22,23,24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41 containing data comparing quilting versus no quilting sutures in patients undergoing mastectomy with or without ALND were included (Table 1). The initial search found 358 studies. After excluding duplicates and unrelated studies based on abstract triage, 47 full texts were assessed for eligibility, of which 21 matched the inclusion criteria and were systematically reviewed. Year of publication of the included studies ranged from 2002 to 2021. Most studies were randomized control trials (n = 10),23,24,25,27,31,34,36,38,39,41 followed by prospective cohort studies (n = 7)21,22,28,29,32,37,40 and retrospective studies (n = 4).26,30,33,35 The total number of included patients was 3473, split into two groups: study group (SG; n = 1736) and control group (CG; n = 1737). In the SG, the mastectomy flaps with (n = 9)21,23,31,34,35,36,37,38,40 or without the axilla (n = 10)22,24,25,26,27,28,30,32,33,41 were quilted using interrupted or continuous polyglactin sutures. In most studies, interrupted polyglactin 3/0 sutures were used between the skin flaps and the pectoralis major muscle. Mean age in the SG was 55.1 versus 54.9 in the CG. Mean BMI was 26.5 in both groups. By far, the most common procedure was modified mastectomy (SG, n = 844; CG, n = 873), followed by mastectomies with SLNB (M+SB) and simple mastectomies (M) (Table 1). Studies did not report if and what proportion of mastectomies were skin- or nipple-sparing.

Risk Factors for Seroma Formation

Body Mass Index

Fourteen studies21,23,24,25,26,27,31,32,33,34,35,36,39,40 describing 1838 patients included data on patients’ BMI in QS versus standard closure after breast cancer surgery. Both groups were similar in this regard, having comparable BMI values with a mean difference of 0.02, however with considerable interstudy heterogeneity (mean difference 0.02, 95% CI −0.60 to 0.64, p = 0.95, Chi-square = 55.82, I2 = 77%) (Fig. 3a).

Fig. 3
figure 3

Meta analysis of risk factors for seroma formation and increased postoperative drainage: a BMI; b neoadjuvant therapy; c operative time; and d lymph node yield. Each study is shown by the point estimate of the OR/mean difference (square proportional to the weight of each study) and 95% CI for the OR (extending lines); the combined ORs/mean difference and 95% CIs by random effects calculations are shown by diamonds. a SG versus CG and BMI (n = 1838, p = 0.95; test for heterogeneity Cochran Q: 55.82, df: 13, p < 0.00001, I2: 77%). b SG versus CG and neoadjuvant therapy (n = 1155, p = 0.43; test for heterogeneity Cochran Q: 6.51, df: 8, p = 0.59, I2: 0%). (c) SG versus CG and operative time (n = 782, p = 0.22; test for heterogeneity Cochran Q: 276.02, df: 6, p < 0.00001, I2: 98%). d SG versus CG and lymph node yield (n = 879, p = 0.13; test for heterogeneity Cochran Q: 106.73, df: 8, p < 0.00001, I2: 93%). BMI body mass index, OR odds ratio, CI confidence interval, SG study group, CG control group, df degrees of freedom

Full size image

Neoadjuvant Therapy

Nine studies21,24,26,27,29,33,35,36,37 including data on 1155 patients described how many had neoadjuvant therapy in the two groups. No statistical difference and heterogeneity were seen between the SG and the CG (OR 0.89, 95% CI 0.67–1.19, p = 0.43, Chi-square = 6.51, I2 = 0%) (Fig. 3b).

Operative Time

Seven studies21,23,25,32,33,34,39 including data on 782 patients measured the operative time between the SG and the CG. QS were not associated with a significant increase in operative time compared with CC, with a mean difference of 6.3 min in favor of the CG, however with considerable interstudy heterogeneity (mean difference 6.30, 95% CI −3.83 to 16.43, p = 0.22, Chi-square = 276.02, I2 = 98%) [Fig. 3c].

Lymph Node Yield

Nine studies21,23,25,32,33,36,37,38,39 including data on 879 patients described the number of excised lymph nodes showing the extent of dissection between the two groups. There was no significant difference between the SG and CG in lymph node yield, with a mean difference of 1.01 lymph nodes (mean difference 1.01, 95% CI 0.21–2.34, p = 0.13, Chi-square = 106.73, I2 = 93%) (Fig. 3d).

Outcomes of Quilting Sutures

Seroma Rate

Twenty studies21,22,23,24,25,26,27,28,30,31,32,33,34,35,36,37,38,39,40,41 including data on 3395 patients were analyzed based on seroma rates between the SG and the CG. The addition of QS significantly reduced the risk of seroma formation in the SG (OR 0.32, 95% CI 0.21–0.49, p < 0.00001, Chi-square = 94.22, I2 = 80%) (Fig. 4a).

Fig. 4
figure 4

Meta analysis of outcomes after quilting sutures in terms of a seroma rate; b total volume of drainage; c days to drain removal; and d length of stay. Each study is shown by the point estimate of the OR/mean difference (square proportional to the weight of each study) and 95% CI for the OR (extending lines); the combined ORs/mean difference and 95% CIs by random effects calculations are shown by diamonds. a SG versus CG and seroma rate (n = 3395, p < 0.00001; test for heterogeneity Cochran Q: 94.22, df: 19, p < 0.00001, I2: 80%). b SG versus CG and total volume of drainage (n = 2500, p < 0.00001; test for heterogeneity Cochran Q: 1909.81, df: 14, p < 0.00001, I2: 99%). c SG versus CG and days to drain removal (n = 1565, p = 0.001; test for heterogeneity Cochran Q: 790.17, df: 8, p < 0.00001, I2: 99%). d SG versus CG and length of stay (n = 1086, p = 0.007; test for heterogeneity Cochran Q: 68.06, df: 5, p < 0.00001, I2: 93%). OR odds ratio, CI confidence interval, SG study group, CG control group, df degrees of freedom

Full size image

Total Volume of Drainage

Fifteen studies21,22,23,24,25,26,27,29,31,34,35,36,37,38,39 including data on 2500 patients measured the total volume of drainage between the SG and the CG. The total volume of drainage was significantly less in the QS group, with a mean difference of 474.9 mL. Interstudy heterogeneity in terms of volume of drainage was high (mean difference 474.99, 95% CI 337.58–612.40, p < 0.00001, Chi-square = 1909.81, I2 = 99%) (Fig. 4b).

Days to Drain Removal

Nine studies21,22,23,25,31,34,37,38,39 including data on 1565 patients analyzed the number of days until drains were removed in QS versus CC. All studies used an output-based approach in deciding when to remove the drains. In all studies, drains were removed when output was between 20 and 50 mL/24 h. Some studies kept the drains until the output was similar (20–50 mL) for 2 or 3 consecutive days.22,23,25 Drains were removed earlier in the SG by a mean difference of 3.3 days (mean difference 3.32, 95% CI 1.29–5.34, p < 0.00001, Chi-square = 790.17, I2 = 99%) (Fig. 4c).

Length of Stay

Six studies25,26,32,33,40,41 including data on 1086 patients compared length of hospitalization between the two groups. Patients who had QS had a shorter length of stay compared with the CG, with a mean difference in hospitalization of 0.7 days (mean difference 0.71, 95% CI 0.19–1.22, p = 0.007, Chi-square = 68.06, I2 = 93%) [Fig. 4d].

Wound Complications

Surgical Site Infections

Fourteen studies21,23,24,25,26,27,30,31,35,37,38,39,40,41 including data on 2134 patients compared the rates of local wound infections between patients with QS and patients without. There was no statistical difference in terms of wound infections between the two groups (OR 0.70, 95% CI 0.48–1.02, p < 0.07, Chi-square = 17.11, I2 = 24%) (Fig. 5a).

Fig. 5
figure 5

Meta-analysis of complications after quilting sutures in terms of a surgical site infection; b hematoma; c flap necrosis; and d skin dimpling. Each study is shown by the point estimate of the OR/mean difference (square proportional to the weight of each study) and 95% CI for the OR (extending lines); the combined ORs/mean difference and 95% CIs by random effects calculations are shown by diamonds. a SG versus CG and surgical site infections (n = 2134, p = 0.07; test for heterogeneity Cochran Q: 17.11, df: 13, p = 0.19, I2: 24%). b SG versus CG and hematoma formation (n = 978, p = 0.38; test for heterogeneity Cochran Q: 5.11, df: 6, p = 0.53, I2: 0%). (c) SG versus CG and flap necrosis (n = 685, p = 0.15; test for heterogeneity Cochran Q: 1.75, df: 6, p = 0.94, I2: 0%). d SG versus CG and skin dimpling (n = 649, p = 0.61; test for heterogeneity Cochran Q: 1.12, df: 2, p = 0.57, I2 0%). OR odds ratio, CI confidence interval, SG study group, CG control group, df degrees of freedom

Full size image

Hematoma Formation

Seven studies21,23,25,26,30,32,34 including data on 978 patients compared the rates of hematoma formation between the two groups. QS were not associated with an increased risk of hematoma formation (OR 1.30, 95% CI 0.72–2.34, p = 0.38, Chi-square = 5.11, I2 = 0%) (Fig. 5b).

Flap Necrosis

Seven studies21,23,25,31,34,37,39 including data on 685 patients analyzed the incidence of mastectomy flap necrosis between patients with QS and patients without. There was no significant difference between the two groups (OR 0.58, 95% CI 0.27–1.23, p = 0.15 Chi-square 1.75, I2 = 0%) (Fig. 5c).

Skin Dimpling

Four studies23,24,25,27 including data on 649 patients analyzed whether patients with QS have higher rates of skin dimpling on the skin flaps. QS were not associated with an increased risk of skin dimpling (OR 1.13, 95% CI 0.71–1.82, p = 0.61, Chi-square = 1.12, I2 = 0%) [Fig. 5d].

Discussion

Our analysis showed that the addition of QS resulted in fewer seromas, less drainage, and shorter length of stay without increasing wound complications. In terms of risk factors for seroma formation, both groups were comparable. Fixation of the mastectomy flaps can be safely used to reduce postoperative drainage. While individual studies showed good outcomes even when drainage is omitted, it is too early to say whether QS are the missing piece towards drain-free mastectomies, as mentioned in De Rooij’s recent systematic review.42 This is the most updated meta-analysis on the role of QS after mastectomy and the only one to meta-analyse the two groups in terms of risk factors for seroma formation, outcomes, and associated wound complications. We chose to include all comparative studies, not only RCTs, as the data provided were easily matched with the data in the RCTs and the cohorts were comparable without overpowering the rest. In the last 5 years, 11 studies have been published on this subject,21,22,23,24,25,26,27,28,29,30,31 of which five were RCTs.23,24,25,27,31 This is a strong suggestion that QS should be adopted more in standard surgical practice.

Increased drainage and seroma formation are regarded as the most common complications after breast surgery, thus increasing costs etc.1,2,3 Closed suction drainage is the single widespread technique for reducing seroma formation and while it clearly has a crucial role, research efforts have been made to reduce drainage even more. Quilting the mastectomy flaps and the axilla is a simple yet promising technique based on the building evidence validating its beneficial role. In 2016, Chen et al.43 performed a meta-analysis that included RCTs and which compared quilting versus non-quilting after axillary dissection, excluding case-control studies, and concluded by supporting the use of QS. In 2018, van Bastelaar et al.2 performed a systematic review comparing tissue glue and QS in reducing seroma formation. While both showed better outcomes than CC, QS were not superior to tissue glue. This was contradicted by the RCT published by de Rooij et al. in 202142 that showed better outcomes in the sutures group compared with tissue glue.

Whenever used, QS showed reduced postoperative drainage, however few surgeons still use them and this is likely explained by the initial longer operative time. Furthermore, there is a fear that anchoring the flaps to the muscle will lead to increased pain on shoulder movement and may lead to dimpling and bumping of the skin surface. In our study, we could not compare the postoperative pain associated with QS as insufficient studies provided data on the visual analog scale of pain. However, the operative time and dimpling were similar in both groups. Once mastered, QS should not affect the aesthetics of wound closure as they are widely used in anchoring various flaps in breast reconstructive procedures without reported issues.44,45,46

High BMI, increased operative time, neoadjuvant therapy, use of cautery, extent of lymph node dissection, type and number of drains used, tumor size, and presence of positive axillary lymph nodes have all been studied and have arguably been proven as risk factors for increased postoperative drainage and seroma formation.9,10,11,12,13 In our analysis, we compared the two groups in terms of BMI, operative time, neoadjuvant therapy, and lymph node yield. Number of lymph nodes was used as an indirect marker of extent of lymph node dissection, which in turn increases the risk of prolonged drainage. These four were constant variables analyzed in the included manuscripts. No significant difference was found between the SG and the CG, reducing the risk of selection bias.

There are a couple of important differences in how authors reported each variable that needs to be considered, although the overall results are not affected. While quilting was performed with the same goal of decreasing the dead space, there was interstudy variability between quilting techniques. In most cases, interrupted polyglactin 3/0 was used on the upper and lower skin flaps. Some used a single or double layer of continuous polyglactin sutures in both flaps.23,31,33,34,35 Furthermore, some authors, e.g. Cong et al.,23 quilted the axilla also.

Not all studies defined seroma in the same way. In most scenarios, seromas were defined according to the Common Terminology Criteria for Adverse Events (CTCAE) Classification: grade 0, no seroma; grade 1, asymptomatic—clinically identified, but no intervention needed; grade 2, symptomatic, medical intervention required; grade 3, severe symptoms, radiological or surgical intervention needed. Other studies defined seroma only when clear fluid was aspirated from a palpable collection,27,28,32,35,41 while others defined it as any palpable fluid collection after drain removal.22,36,39 There were differences in how the drains were placed in both the SG and CG. Fifteen studies used the same number of drains in both groups (one or two drains),21,22,23,24,27,29,30,32,34,35,39 while six studies25,26,28,33,40,41 used fewer drains in the SG, thus proving QS can be used to replace at least the prepectoral drain. Seroma rate and volume and duration of drainage were better in the SG, both from a statistical and practical point of view. In terms of bedside management, a volume difference of 475 mL is important because it subsequently led to removing the drains sooner, by a mean of 3.3 days. Three days is a significant interval that should have a beneficial impact on the postoperative recovery of patients. In our analysis, quicker drain removal was not followed by a clinically relevant shorter hospital stay (mean difference 0.7 days) although it was statistically significant.

Using multiple QS between the dermis and the pectoralis muscle will naturally lead one to believe this will increase the risk of bleeding from the muscular bed and asymmetric dimpling of the skin. For this reason, we sought to analyze wound complications. Most studies had enough data, apart from skin dimpling, which was compared in only four manuscripts.23,24,25,27 This was the smallest dataset in our study and it may need further research. In three studies,23,24,27 skin dimpling alongside wound cosmesis was assessed by an independent, blinded surgeon during outpatient follow-up. The investigator assessed for skin dimpling at every follow-up visit. In one study,23 skin dimpling was only quantified at 3 months. Even though not as important as in reconstructions, the wound cosmesis of patients who undergo non-reconstructive mastectomy is still an important factor to consider. In such scenarios, dimpling caused either by misplaced sutures or skin excess leads to wound deformity and impacts the patient-reported cosmesis. In our analysis, dimpling was not increased by QS when sutures were placed by an experienced surgeon. de Rooij et al.24 also analyzed shoulder discomfort, as QS would, in theory, increase pain on movement due to the sutures anchored to the pectoralis muscle. They did not report any correlation between QS and shoulder discomfort.

While all studies described in detail the quilting technique in non-skin-sparing mastectomies, none showed how QS should be placed in skin- or nipple-sparing mastectomies without immediate breast reconstruction. In such cases, it is more difficult to attach the flaps to the muscular bed, and, even when done, the redundant skin will still lead to skin dimpling and wound asymmetry. In modified radical mastectomies, QS have a clear beneficial role, but further research should clarify how QS should be used in oncoplastic mastectomies, which are now standard procedures.

Conclusion

The building body of evidence strongly supports the use of QS after non-skin-sparing mastectomies to reduce seroma rate, volume and duration of drainage, and length of stay. This meta-analysis reiterated the beneficial role of QS in improving postoperative outcomes without increasing wound complication rates.