Is a Seroma the “Kiss of Death” in Prepectoral Tissue Expander Reconstruction?

Abstract

Background:

A seroma following prepectoral tissue expander (TE) reconstruction often begets other complications, which may compromise the reconstruction. This study investigated the association between seroma and subsequent complications.

Methods:

All consecutive prepectoral TE reconstructions performed between March 2017 and December 2022 at a single center were reviewed. Demographics, operative characteristics, and complications data were extracted for all patients and analyzed.

Results:

Two hundred patients (318 breasts) underwent reconstruction and were, on average, 53 years of age, nonsmokers (98%), and nondiabetic (91%), with a body mass index of 26 kg/m². Seventy-six (24%) breasts were radiated, and 93 (47%) patients received chemotherapy. All 318 breasts underwent immediate reconstruction following prophylactic (34%) or therapeutic (66%) mastectomies. Seroma occurred in 50 (16%) breasts and was associated with higher body mass index (30 versus 27 kg/m², P < 0.05) and higher mastectomy weight (662 versus 515 g, P < 0.05). Half of all breasts with a seroma (24 of 50, 49%) went on to develop other complications. Infection and explantation commonly followed, occurring in 18 (36%) and 21 (42%) breasts with a prior seroma, respectively. In adjusted multivariable models, prior seroma was associated with 9 times higher odds of infection (odds ratio 9.2; 95% confidence interval, 4–21, P < 0.01) and 7 times higher odds of explantation (odds ratio 6.8, 95% confidence interval, 3–17, P < 0.01).

Conclusions:

Although causality cannot be determined, our data suggests that seroma may be the “kiss of death” in prepectoral TE reconstruction because half of all breasts with a seroma went on to develop other complications.

Takeaways

Question: How do seromas affect complications in prepectoral tissue expander–based breast reconstruction?

Findings: Fifty (16%) breasts had a seroma, 24 (49%) of which went on to develop other complications.

Meaning: A seroma is the “kiss of death,” increasing odds of infection by 9 times and explantation by 7 times.

INTRODUCTION

Annually, more than 150,000 women in the United States undergo postmastectomy breast reconstruction, with the majority being 2-stage tissue expander (TE)–based breast reconstruction.^1,2 Although the prepectoral plane for implant placement was first described in 1971, historically, the majority of breast reconstructions have utilized the submuscular plane.^3–5 However, improvements in TE design (tabs for securing TEs on the chest wall) and the introduction of soft tissue support (STS) such as synthetic meshes and acellularized dermal matrices (ADMs) have repopularized the prepectoral plane.^6,7 This plane offers several advantages over the subpectoral plane, namely decreased postoperative pain, shorter operative times, no animation deformity, and faster time to final TE fill volume and second stage of reconstruction.^8–10

The prepectoral plane has its own limitations, one of which is a higher incidence of seroma as compared with the subpectoral plane.^8,11,12 Although easier to place and position, prepectoral TEs lack the well-vascularized tissue from the pectoralis muscle, which may play an important role in reducing dead space and minimizing microshifting of the TE. Furthermore, a higher surface area of STS may be required to secure the pocket in prepectoral reconstruction, which may increase the risk of seroma, as partial or total failure of STS integration can lead to inflammation; increased exudate; and, eventually, a seroma.^12–17

The objective of this study was to identify if a relationship exists between seroma formation and the incidence of subsequent complications in prepectoral TE-based reconstruction. We hypothesized that seroma formation serves as a nidus for a cascade of other complications and is a negative predictor of reconstructive outcomes.

MATERIALS AND METHODS

All consecutive prepectoral TE reconstructions performed at a single institution between March 2017 and December 2022 were included. Mastectomies were conducted by 5 distinct breast surgeons, and reconstructions were performed by 5 distinct plastic surgeons (V.T., O.C., J.L., M.C., and N.S.K.). Inclusion in the study required patients to (1) be older than 18 years of age at the time of reconstruction and (2) undergo immediate TE reconstruction in the prepectoral plane.

Demographic variables included age at surgery, body mass index (BMI), smoking status, medical comorbidities, adjuvant therapy, surgical and cancer history. Operative characteristics included the type of mastectomy, specimen weight, intraoperative TE fill, and the use of STS. Postoperative data included occurrence of seroma, timing of seroma formation, and the occurrence of systemic, major, and minor complications. Seroma was defined as a clinically identified postoperative fluid collection in the breast pocket on physical examination or imaging. Time to seroma formation was recorded as the postoperative day (POD) on which seroma was detected. Seromas were classified as major and minor complications. Major complications were defined as any complication that required readmission or reoperation, whereas minor complications included any complication requiring outpatient antibiotics, procedures, or wound care.

Descriptive statistics were reported as a mean or median and SD for continuous variables or frequency and percentage for categorical variables, respectively. The Pearson chi-square and Fisher exact tests were conducted for categorical variables and analysis of variance and Mann-Whitney tests for continuous variables. Factors found significant on univariate analysis were analyzed in a multivariate model. All statistical analyses were performed using R Statistical Software (version 1.1.447; R Foundation for Statistical Computing, Vienna, Austria) and a P value less than 0.05 was considered statistically significant.

RESULTS

Prepectoral TE-based reconstruction was performed in 200 patients (318 breasts). The mean age of patients was 53 ± 12 years (Table 1). Most patients were nonsmokers (99%) and nondiabetic (91%). Average BMI was 26 ± 8 kg/m². Ninety-three (47%) patients received chemotherapy (20% adjuvant, 17% neoadjuvant, and 10% both).

Table 1.

Baseline Characteristics

Patient Characteristics	Total Patients (N = 200)
Age at time of surgery, y, mean ± SD	53 ± 12
Smoking, n (%)
Current smoker	3 (2)
Former smoker	63 (32)
Never smoker	134 (67)
Diabetes, n (%)	18 (9)
BMI, kg/m2, median ± IQR	26 ± 8
Chemotherapy, n (%)	93 (47)
Adjuvant	39 (20)
Neoadjuvant	34 (17)
Both	20 (10)
Hormonal replacement therapy, n (%)	101 (52)
Breast characteristics	Total breasts (N = 318)
Radiation, n (%)	76 (24)
Adjuvant	56 (17)
Neoadjuvant	18 (6)
Both	2 (1)

Mastectomies were prophylactic in 34% of patients and therapeutic in 66% of cases (Table 2). Most mastectomies were skin sparing (59%), followed by nipple sparing (26%). Seventy-six (24%) breasts were radiated (17% adjuvant, 6% prior radiation, and 1% both). Median mastectomy weight was 546 (interquartile range [IQR]: 428) g.

Table 2.

Operative Characteristics by Breast

Operative Characteristics	Total Breasts (N = 318)
Mastectomy indication, n (%)
Prophylactic	107 (34)
Therapeutic	211 (66)
Mastectomy type, n (%)
Skin sparing	188 (59)
Nipple sparing	83 (26)
Simple	39 (12)
Other	8 (3)
Mastectomy weight, g, median ± IQR	546 ± 428
Intraoperative TE fill, mL, median ± IQR	175 ± 250
Final TE fill, mL, mean ± SD	417 ± 159
Use of STS, n (%)	61 (19)
AlloDerm	47 (77)
Durasorb	8 (13)
FlexHD	4 (7)
AlloDerm and Vicryl	2 (3)

Plastic surgeons varied in their surgical techniques, intraoperative TE fill, and use of STS. The median intraoperative TE fill volume was 175 (IQR: 250) mL, and the average final TE fill volume was 417 (±159) mL. STS was used in 61 (19%) breasts, and this was done in various iterations: 47 (77%) breasts used AlloDerm (Life Cell Corporation, Branchburg, NJ), 8 (13%) DuraSorb (SIA, Chicago, IL), 4 (7%) Flex HD (Ethicon Inc, Menlo Park, CA), and 2 (3%) AlloDerm and Vicryl mesh.

Average length of follow-up was 23 ± 17 months. As a general rule, surgeons removed drains when the output had been less than 30 mL for 3 consecutive days, and the median time of drain removal was 15 days. Surgeons commenced expansion in breasts without mastectomy flap necrosis or delayed healing approximately 2–4 weeks following TE placement. Expansion then proceeded incrementally to the patient’s desired size at variable amounts per session, depending on how the patient tolerated expansion. Seromas were aspirated when detected, and fluid was sent for culture when possible. Additionally, expansion was typically used as a means to mitigate seroma reformation. For example, if a seroma was noted on POD 14, the seroma was aspirated in office and the expander was filled simultaneously to prevent reaccumulation of seroma.

Seroma occurred in 43 (22%) patients and 50 (16%) breasts (Table 3). Median POD of seroma development was 26 days. Physical examination was used to detect seromas in 45 (90%) breasts, and the remaining 5 (10%) were detected on imaging upon presentation to the emergency department for an infectious workup. Patients who experienced seroma had a higher BMI (30 versus 27 kg/m², P < 0.05) when compared with patients without seroma. Breasts that ultimately developed a seroma were more likely to have undergone skin-sparing mastectomies (80% versus 55%, P < 0.05) and have higher mastectomy weight (662 versus 515 g, P < 0.05) when compared with breasts without seroma.

Table 3.

Characteristics of Breasts That Developed Seroma

	Incidence of Seroma
	No (N = 268)	Yes (N = 50)	P
Age, y, mean ± SD	51.4 ± 11.8	51.9 ± 13.1	0.778*
BMI, kg/m2, median (IQR)	26 (7)	29 (11)	0.002†
Current smoker, n (%)	4 (1.5)	0 (0.0)	1.000‡
Diabetes, n (%)	27 (10.1)	3 (6.1)	0.595‡
Chemotherapy, n (%)	123 (45.9)	27 (54.0)	0.292§
Radiation, n (%)	61 (22.8)	15 (30.0)	0.271§
Mastectomy type, n (%)			0.011‡
Skin sparing	148 (55.2)	40 (80.0)
Nipple sparing	77 (28.7)	6 (12.0)
Simple	35 (13.1)	4 (8.0)
Other	8 (3.0)	0 (0.0)
STS, n (%)	55 (20.5)	6 (12.0)	0.160§
Mastectomy weight, median (IQR)	516 (422)	662 (454)	0.007†
Intraoperative TE fill, median (IQR)	175 (225)	200 (275)	0.949†
Final TE fill, mean ± SD	409 ± 155	462 ± 180	0.077†

^*Student t test.

^†Kruskal-Wallis rank sum test.

^‡Fisher exact test for count data.

^§Pearson chi-square test.

In multivariable models controlling for age, BMI, diabetes, radiation, chemotherapy, mastectomy weight, intraoperative TE fill, final TE fill, and STS, only 2 variables were found to be predictive of seroma (Fig. 1). The use of STS decreased odds of seroma (odds ratio [OR], 0.3; 95% confidence interval [CI], 0.1–0.8; P < 0.05), and a 100-cc increase in intraoperative TE fill led to decreased odds of seroma (OR, 0.8; 95% CI, 0.6–0.9; P < 0.05).

Fig. 1.

Multivariable model for seroma. Incidence of seroma formation after controlling for patient demographics and operative characteristics. AIC, akaike information criterion.

Seroma was considered a minor complication in 29 (58%) of 50 breasts, that is, it resolved spontaneously or with interventions in the outpatient setting (Table 4). Thirteen (26%) of 50 breasts had a seroma requiring admission for intravenous antibiotics, drainage or sclerotherapy by interventional radiology (IR), and/or reoperation. The remaining 8 (16%) of 50 breasts with seroma were categorized as having both minor and major complications and received a combination of the aforementioned interventions. On average, major seromas presented later, on POD 54, whereas minor seromas presented earlier, on POD 22 (Fig. 2). Breasts with major seromas had a higher BMI (32 versus 27 kg/m²), more frequent history of chemotherapy (81% versus 35%), higher mastectomy weight (1074 versus 619 g), higher intraoperative TE fill (332 mL versus 98 mL), and higher final TE fill (578 versus 402 mL) when compared with minor seromas (Fig. 3; Table 4).

Table 4.

Characteristics of Breasts by Seroma Severity

	Major or Both Major and Minor (n = 21)	Minor Only (n = 29)	P
POD of seroma presentation, median (IQR)	POD 54 (71)	POD 22 (9)	<0.001*
Age, y, mean ± sSD	49 ± 15	54 ± 12	0.152*
BMI, kg/m2, median (IQR)	32 (9)	27 (7)	0.005†
Diabetes, n (%)	0 (0)	3 (10)	0.260‡
Chemotherapy, n (%)	17 (81)	10 (35)	0.001§
Radiation, n (%)	12 (57)	23 (79)	0.091§
Mastectomy type, n (%)			0.336‡
Skin sparing	15 (71)	25 (86)
Nipple sparing	3 (14)	3 (10)
Simple	3 (14)	1 (3)
Other	0 (0)	0 (0)
STS, n (%)	4 (19)	2 (7)	0.192§
Mastectomy weight, median (IQR)	1074 (620)	619 (322)	0.003†
Intraoperative TE fill, median (IQR)	332 (166)	98 (115)	<0.001†
Final TE fill, mean ± SD	578 ± 189	402 ± 145	0.002*

^*Student t test.

^†Kruskal-Wallis rank sum test.

^‡Fisher exact test for count data.

^§Pearson chi-square test.

Fig. 2.

Postoperative incidence by seroma severity. Minor seromas, defined as those that resolved spontaneously or with minor interventions in the outpatient setting, tended to occur earlier in a patient’s postoperative course as compared with major seromas, defined as those requiring readmission or reoperation. Median [Q1, Q3].

Fig. 3.

Demographic and operative characteristics related to seroma severity. Seroma severity based on patient demographics and operative characteristics.

Half of all breasts with a seroma (24 of 50, 49%) went on to develop other complications. As compared with breasts without seroma, breasts that developed seroma had higher incidence of subsequent postoperative infection (36% versus 7%, P < 0.001) and explantation (43% versus 9%, P < 0.001) compared with breasts without seroma.

Eighteen breasts with seroma went on to develop an infection. Detection of seroma occurred before the detection of infection in 8 (44%) breasts and at the same time as infection in 10 (55%) breasts. Of the 18 breasts with seroma and infection, 6 (33%) were successfully managed with outpatient antibiotics, but 9 (50%) of 18 breasts required explantation, 2 (11%) of 18 breasts required inpatient admission for drainage by IR or intravenous antibiotics, and 1 (5%) of 18 breasts was managed with TE exchange and conversion to subpectoral plane. In multivariable models controlling for age, BMI, diabetes, radiation, mastectomy weight, intraoperative fill, and STS, prior seroma was the sole predictor of infection, and breasts with prior seroma had a nine times higher odds of infection (OR, 9.2; 95% CI, 4.0–21.4; P < 0.01; Fig. 3) when compared with breasts without a seroma.

Of the 18 breasts with infected seromas, 9 (50%) had microorganisms detected on culture: 5 (56%) had Staphylococcus aureus, and the remaining 4 (44%) had other organisms such as Staphylococcus epidermis, Candida parapsilosi, Enterobacter cloacae, Enterococcus faecalis, and Pseudomonas aeruginosa. The remaining 5 of 18 breasts (28%) with infected seromas had no growth on culture, and 4 (22%) had no cultures sent at time of aspiration.

Of the 21 breasts that developed seroma and subsequently had TE explanted, 16 (76%) of 21 were true TE removals due to sequalae related to the seroma. In 3 (10%) of 21 breasts with a seroma, TEs were explanted due to patient decision—in all cases, women decided not to proceed with breast reconstruction; one of these 3 patients later came back for delayed autologous reconstruction. In 2 (10%) of 21 breasts, TEs were exchanged. In 1 aforementioned case, the TE was successfully converted to subpectoral plane, whereas in another, the TE was exchanged but later removed due to patient decision not to complete reconstruction. In using multivariate regression to predict explantation, prior seroma led to a 7-fold increase in odds of explantation (OR, 6.8; 95% CI, 2.9–16.8; P < 0.01; Fig. 4). Radiation led to a 3-fold increase in odds of explantation (OR, 3.1; 95% CI, 1.4–7.0; P < 0.01), and a 10-cc increase in intraoperative fill was associated with a mild increase in odds of explantation (OR, 1.04; 95% CI, 1.01–1.06; P < 0.01) (Fig. 5).

Fig. 4.

Multivariable model for infection. Incidence of infection after controlling for patient demographics and operative characteristics. AIC, akaike information criterion; XRT, radiation.

Fig. 5.

Multivariable model for explantation. Incidence of explantation after controlling for patient demographics and operative characteristics. AIC, akaike information criterion; XRT, radiation.

After excluding cases where patients elected to have TE removed after experiencing a seroma, 17 breasts remained where seroma led to explantation. Of these, 8 (47%) of 17 breasts successfully completed reconstruction at the time of chart review using a deep inferior epigastric artery flap or latissimus dorsi flap and 9 (53%) of 17 awaited reconstruction pending resolution of other health issues.

DISCUSSION

The incidence of seroma in prepectoral TE reconstructions ranges from 2% to 15% in the literature.^{6,8,11,18–20} In this cohort of 200 patients and 318 breasts with an average follow-up of approximately 2 years, seroma occurred in 22% of patients and 16% of breasts. We hypothesized that seroma formation often begets subsequent complications, and this was confirmed by our results, which showed a temporal association between seroma occurrence and the incidence of other complications.

In line with the concept of seroma developing where there is more dead space, we found that breasts with postoperative seromas had significantly higher mastectomy weight on univariate statistics. This finding has been reproduced in the literature, which shows mastectomy specimen weights greater than 500 g having higher odds of seroma and skin necrosis.^21,22 High mastectomy specimen weights are often highly correlated with BMI, which was also predictive of seroma incidence in this cohort on univariate analysis.

Interestingly, neither BMI nor mastectomy weight were predictive of seroma occurrence on multivariate regression analysis. A 100-cc increase in intraoperative TE fill was found to lower the odds of seroma occurrence by a factor of 0.8. This intuitively makes sense, as reducing dead space in the mastectomy pocket will limit the potential space for seroma formation. However, the literature suggests that surpassing a threshold fill volume of 300 mL may be associated with higher odds of complications such as skin necrosis, reoperation, and explantation.^21,23 Despite this association with other postoperative complications, no dedicated studies of seroma have explicitly found an explicit link between it and intraoperative TE fill.^12,16 In our study, an association between TE fill and seroma was evident when seromas were delineated by severity: seromas characterized as major complications had higher intraoperative and final TE fills than those seromas characterized as minor. Minimizing dead space and jumpstarting the expansion process by filling a TE intraoperatively is directly at odds with avoiding tension on fragile mastectomy flaps in immediate prepectoral breast reconstruction. Managing these competing forces and finding the optimal TE fill is the focus of other studies by our group.²⁴

The use of certain types of STS is considered seromagenic by many due to its associated inflammation and increased exudate.^12–16 In our study, where nearly a fifth of reconstructions used STS, we found that STS use was associated with lower odds of seroma when accounting for other variables in multivariable regression. This finding is likely contingent on the type of STS predominantly used at our institution, with AlloDerm being the STS of choice in 77% of breasts.²⁵ Prior studies have shown certain types of STS, such as FlexHD, to be more seromagenic than others, and this was used in only four breasts in our cohort.¹⁶ In a meta-analysis of 1881 prepectoral breast reconstructions, Wagner et al²⁶ also concluded that STS use was associated with lower rates of seroma: 4%, versus 7% in breasts that did not utilize STS.

The incidence of subsequent complications was staggeringly high in breasts that had seroma compared with those that did not have seroma, with infection at 36% versus 7% and explantation at 43%versus 9%. Moreover, this association between seroma and other postoperative complications held even after adjusting for demographic and operative characteristics in multivariable models. This suggests that the development of seroma can be considered a harbinger for future postoperative complications in prepectoral TE reconstruction.

Numerous studies have also acknowledged and established the link between seroma and downstream complications in prosthetic breast reconstruction, but ours is the first to do so in the population of patients undergoing prepectoral TE reconstruction.^14,17 In a series of 1605 consecutive prosthetic breast reconstructions, Jordan et al¹⁴ reported a 4-fold higher odds of infection and 6.7-fold higher odds of explantation. Prior data from our institution also note a 6.4-fold increase in the risk of infection in breasts with seroma in a study of 595 breasts undergoing submuscular implant-based breast reconstruction.¹⁷ Parks et al²⁷ studied 511 breasts undergoing immediate submuscular TE-based breast reconstruction and found that breasts reconstructed with ADM were twice as likely to develop seroma. Breasts that developed a seroma had a 4.6-times higher odds of TE loss, regardless of whether ADM was used.²⁷ These findings are mirrored and magnified in our cohort of prepectoral TE patients, where breasts with seroma had a 9-fold increase in odds of infection and 7-fold increase in odds of explantation.

Another aim of this study was to characterize seroma, as not all of them are created equal. Seroma fluid in the postmastectomy breast is thought to be comprised of a combination of lymph and inflammatory exudate.^28,29 How and when a simple seroma becomes an infected seroma and leads to explantation is a question we sought to answer. In our cohort, a major seroma was defined as one that required admission for intravenous antibiotics, drainage or sclerotherapy by IR, and/or reoperation for exchange or explantation. These seromas were more likely to occur in a delayed fashion and were associated with chemotherapy, higher BMI, mastectomy weight, higher intraoperative and final TE fill. Thus, surgeons should be on high alert and aggressively manage late seromas presenting in breasts with such high-risk features. Furthermore, if there is preoperative suspicion that a patient is at high risk for a seroma due to need for adjuvant radiation or chemotherapy, risk of seroma can be mitigated by minimizing intraoperative TE fill and, to the extent that it is possible, final TE fill.

LIMITATIONS

This study is a retrospective cohort study, which is unable to identify causation between variables. Furthermore, these patients have been operated on by a heterogeneous group of breast and plastic surgeons, each with slight variation in their operative technique. This variance in surgical technique among different surgeons is certainly a limitation, but inclusion of many surgeons allowed for a greater number of patients to be studied. Furthermore, each surgeon was evolved in their use of the prepectoral plane over time. We attempted to mitigate these variances by documenting as many intraoperative and postoperative variables as possible and controlling for them in our analyses. Finally, we included different mastectomy techniques in our analysis for the sake of mimicking modern practice, but the variability may bias our data by affecting the incidence of seroma and the ultimate reconstructive outcome.

CONCLUSIONS

Although causality cannot be determined, the data support the hypothesis of seroma being the “kiss of death” in patients undergoing prepectoral TE reconstruction, as more than half of patients who developed a seroma went on to develop other related complications. The development of seroma was associated with 9 times higher odds of infection and 7 times higher odds of explantation. Thus, seroma should be considered a harbinger for further downstream complications.

Not all seromas are created equal. Seromas that required readmission, reoperation, or IR presented in a delayed fashion and were associated with chemotherapy, higher BMI, mastectomy weight, intraoperative, and final TE fill. Aggressive management of seromas should be pursued in high-risk patients who develop a complicated seroma, to mitigate the risk of subsequent postoperative complications and potential for reconstruction loss.

DISCLOSURE

The authors have no financial interest to declare in relation to the content of this article.

PATIENT CONSENT

The authors assert that informed consent was obtained for all procedures performed in this study. Per institutional policy, informed consent for research is not required for retrospective chart review of patient records with minimal risk of harm to the patient.