Intraoperative Predictors of Salvage in Infected Breast Implants: A Retrospective Study

Abstract

Background

Infectious complications associated with implant-based breast reconstruction (IBBR) can be devastating and may lead to reconstructive failure. Although there are known demographic risk factors for reconstructive failure, few studies have identified intraoperative findings that predict failure after attempted salvage.

Objectives

The objective of this study was to identify intraoperative findings in infected breasts that might be predictive of implant failure.

Methods

In total, 837 patients undergoing IBBR between January 2017 and July 2023 were included. Intraoperative records of patients who developed a major infection were reviewed. Reconstructive salvage denoted any intervention not resulting in explantation. Failure denoted explantation.

Results

Of 837 patients, 8% developed a major infection (n = 71). Within this group, 8% had successful treatment with intravenous antibiotics alone, 38% were salvaged after operative intervention, 28.2% failed without salvage attempt, and 25.4% underwent salvage attempt but ultimately failed. Overall, the rate of reconstructive failure was 53.5% and the rate of implant salvage was 46.5%. A total of 51% of patients returning to operating room were found to have unincorporated acellular dermal matrix (ADM). Seventy-eight percent of patients undergoing intervention had a positive culture, most commonly methicillin-susceptible Staphylococcus aureus, methicillin-resistant Staphylococcus aureus, Serratia, Enterobacter, Pseudomonas or Proteus. Unincorporated ADM was significantly associated with reconstructive failure (odds ratio 5.4, P = .003). Serratia infection, hematoma, and gram-negative infection were associated with implant failure, but these findings did not achieve statistical significance.

Conclusions

Presence of unincorporated ADM was associated with eventual implant failure. Surgeons should be aware that this finding may portend poor outcomes for patients. These results may be helpful to intraoperative decision-making.

Level of Evidence: 4 (Therapeutic)

Breast cancer is one of the most frequently diagnosed cancers in the US, with 287,850 new cases diagnosed in 2022. The American Cancer Society predicts that this number will continue to rise in the coming years.¹ Implant-based breast reconstruction, performed by placing either a tissue expander or an implant at the time of mastectomy, is the most common method of immediate breast reconstruction. Periprosthetic infection is a common complication of implant reconstruction, occurring in between 10% and 30% of cases.² Infection following implant-based reconstruction can result in hospital readmission, reoperation, and in some cases device removal leading to reconstructive failure. Patients who experience reconstructive failure have lower overall satisfaction scores compared to those who do not.³

When a patient presents with an infected implant, the surgeon may elect to attempt salvage through intravenous antibiotic administration or through washout of the implant pocket. Often the decision of whether to attempt salvage or move forward with explantation is made intraoperatively. Although several factors have been associated with higher rates of implant failure, such as smoking, radiation, and increasing BMI, little research has been done to elucidate intraoperative factors predictive of implant failure.^4-6 The purpose of this study was to investigate the relationship between patient characteristics, intraoperative findings, and implant salvage or failure.

METHODS

A prospectively maintained database consisting of 837 patients who underwent immediate breast reconstruction after mastectomy at several institutions between January of 2017 and July of 2023 was retrospectively analyzed. Patients were included if they underwent either unilateral or bilateral tissue expander and/or implant-based breast reconstruction following surgical treatment of breast cancer, and if they were 18 or older. Patients who underwent immediate autologous breast reconstruction after mastectomy, or who did not otherwise meet inclusion criteria, were excluded. Data collected included patient demographics such as preoperative BMI, race, age, smoking status, and previous diagnosis of diabetes; surgery-specific characteristics such as unilateral vs bilateral reconstruction, implant vs expander, volume of implant or expander, weight of resection, concomitant sentinel lymph node biopsy, type of mastectomy, and case length; and postoperative factors such as time to drain removal, concurrent oncologic treatment, and length of follow-up. Additionally, surgical complications such as mastectomy skin necrosis or development of seroma, hematoma, or infection were noted. In the prospective cohort, patients were followed for 90 days after their surgical date; in the retrospective cohort, noted complications were reviewed.

For the purposes of this study, “infection” was defined as any symptom or sign that resulted in administration of antibiotics. A minor infection was defined as an infection treated with outpatient antibiotics, whereas a major infection was defined as an infection requiring intravenous antibiotics with or without surgical intervention. Delayed healing denoted presence of epidermolysis or necrosis that necessitated the initiation of wound care such as application of Silvadene (Pfizer, New York, NY), bacitracin, or debridement. Implant failure was defined as removal of the original implant due to infection, with no replacement of the implant. Implant salvage was defined as return to operating room (OR) with washout but no removal, return to OR with washout and implant exchange, or patients admitted for inpatient intravenous (IV) antibiotics who did not require a return to the OR.

The intraoperative records of patients who developed a major infection were reviewed and the descriptors of intraoperative findings reported by the surgeons were compiled. Descriptors included qualitative factors such as “purulent” or “murky” fluid, status of acellular dermal matrix (ADM) incorporation, and other intraoperative findings such as hypergranulation tissue or defined capsular rind.

Databases were compiled and basic statistics such as averages, standard deviations, and t tests were performed in Microsoft Excel (Microsoft Corporation, Redmond, WA). More sophisticated statistics such as multivariate logistic regression were performed in JMP (JMP Statistical Discovery, Cary, NC). P values < .05 were considered significant.

This study received approval from the Emory Institutional Review Board, approval number STUDY00002699.

RESULTS

A total of 837 patients were included in this comprehensive analysis. Of the 837 patients, age at time of surgery ranged from 16 to 86, with a mean of 52.1. The average patient follow-up was 216 days, with a range of 0 days to 1114 days. BMI ranged from 16.5 to 68.9 with an average BMI of 27.4. Five percent of our patient population had smoked within the previous 2 months (n = 42) and 4.4% were diabetic at the time of surgery (n = 37). Twenty percent of our cohort underwent preoperative or postoperative radiation (n = 168) and 32.1% of our cohort underwent neoadjuvant or adjuvant chemotherapy (n = 269) (Table 1).

Of our cohort, 36% (n = 301) underwent unilateral reconstruction of just the affected breast, and 64% (n = 536) opted for bilateral reconstruction, for a total of 1373 breasts undergoing reconstruction. Of these breasts, 55.4% were reconstructed with an implant (n = 761), whereas 44.6% were reconstructed with a tissue expander (n = 612). A total of 45.81% of breasts (n = 629) underwent skin-sparing mastectomy (SSM), 45.96% of breasts (n = 631) underwent nipple-sparing mastectomy (NSM), and 8.23% (n = 113) of breasts underwent mastectomy with other methods, such as skin-reducing mastectomy (SRM) (Table 2).

Of the 837 patients, 11.9% developed infection (n = 100) with approximately 8.5% of the total cohort developing a major infection (n = 71) and the rest developing a minor infection (n = 29). Of those patients who developed postoperative infection, the average postoperative day of infection was 31.2, with a standard deviation of 27.6 days (Figure 1). A total of 11.1% of the patients experienced delayed healing (n = 93), 7.5% of patients developed seroma (n = 67), and 3.7% of patients developed hematoma (n = 31) (Table 3).

Cultures obtained from patients who developed major infections grew a variety of organisms, most commonly methicillin-susceptible Staphylococcus aureus (MSSA; n = 19), Serratia (n = 7), Pseudomonas (n = 5), methicillin-resistant Staphylococcus aureus (MRSA; n = 4), Enterobacter (n = 4), and Proteus (n = 3) (Table 4).

Of the 71 patients who developed a major infection, 8.5% (n = 6) had successful treatment with IV antibiotics, 38% (n = 27) were salvaged after operative intervention, 28.2% (n = 20) underwent immediate operative implant removal, and 25.4% (n = 18) underwent salvage attempt but ultimately failed, resulting in explantation. Overall, the rate of reconstructive failure was 53.5% (n = 38) and the rate of implant salvage was 46.5% (n = 33) (Table 5).

Radiation before surgery, bilateral reconstruction, and development of seroma were significantly associated with the development of any infection, with odds ratios of 1.72, 2.81, and 5.18 respectively (P = .05, P = .003, P = .001). Bilateral reconstruction and development of seroma were significantly associated with the development of major infections, with odds ratios of 2.36 and 6.51 (P = .03, P = .01). Bilateral reconstruction and seroma were also significantly associated with implant failure due to infection, with odds ratios of 2.91 and 5.51 (P = .02, P = .001). Mastectomy skin flap necrosis was not significantly associated with the development of infection. Patient characteristics such as BMI, age, and smoking status were not found to be significantly associated with either development of major infection or eventual implant failure in our cohort. Hormone treatment, increased length of time between initial surgery and return to the operating room, and increased length of surgery were not found to be significantly associated with development of infection or implant failure.

Factors significantly associated with implant failure after development of major infection included reconstruction with a tissue expander (OR 3.56, P = .01). Factors which were associated with implant failure, but did not achieve statistical significance, were occurrences of hematoma (OR 3.5, P = .13), infection with Serratia (OR 5, P = .2), and infection with gram-negative organisms (OR 4.8, P = .2). White blood cell count on admission, length of intravenous antibiotic treatment before return to the operating room, mastectomy skin flap necrosis, and development of hematoma were not significantly associated with eventual implant failure.

Unincorporated ADM was the major intraoperative finding significantly associated with eventual implant failure (OR 5.4, P = .003). Intraoperative findings that were not associated with eventual implant failure included character of the intraoperative fluid (murky, seromatous, or purulent), and quantity of fluid encountered intraoperatively (Table 6).

DISCUSSION

Reconstructive surgery of the chest after breast cancer is one of the most frequent operations that plastic surgeons may perform. Many patients with breast cancer elect to undergo breast reconstruction after mastectomy. These patients report a higher quality of life and improved body image compared to those who do not undergo reconstruction.^7-9 Patients may undergo different types of reconstruction depending on their comorbidities, cancer staging, and personal preferences. Implant-based breast reconstruction (IBBR) is the most common method of postmastectomy reconstruction, but rates of major complications may range anywhere from 2% to 94%.¹⁰ Because breast reconstruction is an important psychological and emotional component of breast cancer treatment, major complications can be devastating, particularly if they result in implant failure. Complete reconstructive failure can be devastating to patient satisfaction; therefore it is crucial to study and understand predictors of this outcome.³ Reconstructive failure is also responsible for increased costs to the healthcare system, and associated with increased length and quantity of hospital stays as well as greater return to the operating room.

Surgical site infection (SSI) is a major risk for implant failure. Reported rates of infection after IBBR in the literature range from 1% to 35%.^2,11 In our patient cohort there was an infection rate of 11.9%. Previous literature has found an association between patient factors such as BMI, smoking, diagnosis of diabetes mellitus type 2, or age, and the development of infection or implant failure.^4-6 We did not find any of these associations in our cohort. Radiation before or after surgery, seroma formation, and operative factors such as the use of ADM are positively associated with infection in the literature.^12-14 These findings were confirmed by our cohort, with radiation before surgery (OR 1.72, P = .05) and seroma formation (OR 6.51, P = .01) significantly associated with development of infection. Because ADM is utilized almost universally by our surgeons in IBBR, we were not able to assess ADM as an independent factor. Although hormone therapy and chemotherapy administration have been associated with SSI in the literature, we did not find this association in our cohort.^15,16 Cohen et al found the most commonly isolated organisms in infections after IBBR to be coagulase-negative staphylococcus, MSSA, MRSA, Pseudomonas, and Peptostreptococcus.¹⁷ In our cohort the most common bacteria were MSSA, Serratia, MRSA, Pseudomonas, and Enterobacter. We did find Serratia to be correlated with implant failure, although this correlation did not achieve statistical significance (P = .2). These results correlate with previously published data about the risk of gram-negative organisms in implant salvage.¹⁸

Previously described risk factors for implant failure after immediate breast reconstruction include radiation therapy either preoperatively or postoperatively, smoking, a history of diabetes, delayed wound healing, hematoma, increasing BMI, and previous infection.^19,20 A systematic review published in 2022 found that 37.6% of patients who developed severe complications after implant-based breast reconstruction required prosthesis explantation.²¹ In a retrospective review published by Asaad et al in 2022, approximately 5% of 6093 implant-based breast reconstructions resulted in explantation due to infection.²² Similarly, Reish et al in 2013 found the overall explantation rate after reconstruction to be 3.2%, of which 62.6% were secondary to infection.²³ This same paper found a higher white blood cell count at admission to the hospital as well as colonization with MRSA bacteria to be predictive of implant failure.²³ We did not find associations between either of these variables and implant failure in our cohort, however we did find an association between Serratia infection and implant failure (odds ratio of 5).

Intraoperative findings that were analyzed included description of intraoperative fluid and incorporation of acellular dermal matrix (ADM) in the healing process. Fluid was described qualitatively in terms such as “murky,” “purulent,” or “seromatous.” ADM was considered unincorporated based on surgeon observation and documentation. Character of fluid was not found to be associated with either implant salvage or failure. These descriptors are subjective and what one surgeon defines as murky may appear purulent or seromatous to another, making it difficult to objectively correlate quality of intraoperative fluid to outcomes. Unincorporated ADM was significantly associated with eventual implant failure (odds ratio 5.4, P = .003).

We identified unincorporated ADM as the main intraoperative factor that significantly predicted whether or not an implant would ultimately fail. Previous studies have associated usage of ADM in breast reconstruction with increased surgical site infection, but none have distinguished between ADM that was incorporated or unincorporated intraoperatively.^12,24,25 Previous studies have found that breasts reconstructed with ADM have higher rates of infection, reoperation, and explantation, however ADM has not been found to be an independent risk factor for infection.^24,26 We are not aware of any previous studies identifying unincorporated ADM as a factor associated with surgical site infection or implant failure.

It is unclear why unincorporated ADM potentially indicates a more severe infection. The ADM may not incorporate as quickly due to poor mastectomy flap vascularity, which leads to eventual development of a more severe infection. Unincorporated ADM also may be an indicator of the presence of biofilm on ADM or the device, which has been shown to be more difficult to eradicate than a planktonic infection. These findings of unincorporated ADM may also represent the downstream sequelae of severe inflammation, inflammatory cascade, or multiple infectious insults. Delayed wound healing could lead to bacterial access to the implant cavity, leading to inflammatory fluid and infected seroma, which finally results in unincorporated ADM. This unincorporated ADM could then provide a setup for biofilm formation and eventual development of a more severe infection.

Additionally, we found seroma to be strongly associated with major infection and implant failure. The 2 may be connected: for example, perhaps ADM does not incorporate because seroma prevents adherence to the mastectomy skin flap, or perhaps unincorporated ADM is in fact seromagenic. A systematic review of the literature in 2021 by Caputo et al found that ADM was not causative of seroma formation, however previous more broadly encompassing reviews had demonstrated a relationship between ADM in breast reconstruction and increased seroma development.^27,28 Future studies should aim to elucidate the mechanism around ADM incorporation, seroma formation, and wound healing. Many institutions have begun to place synthetic or biosynthetic meshes in their breast reconstruction practice, because they may be more cost effective than acellular dermal matrices.²⁹ Current research suggests that these newer meshes have similar complication profiles to ADM.^30,31 Future studies should aim to investigate intraoperative characteristics that may be predictive of implant salvage or failure when employing these ADM alternatives.

One important distinction between our study and the current literature is that we distinguished patients who underwent attempted salvage (an operative intervention aimed at salvaging the implant that ultimately failed) from those who had immediate explantation without attempted salvage. Surgeon risk aversion may affect how quickly a surgeon decides to remove the implant rather than attempt salvage. Some removed implants may have been ultimately salvaged, whereas some salvage attempts are doomed to fail. This group of “attempted salvage” patients adds to the nuance and depth of our study. Within this group, unincorporated ADM, bilateral reconstruction as opposed to unilateral reconstruction, and increased BMI were associated with implant failure after attempted salvage, although these p-values did not achieve statistical significance. It is important to note that patient-specific factors such as patient cancer stage and possible need to progress to adjuvant therapy may also affect surgical decision-making. For example, continued infection could delay radiation therapy. In these patients, regardless of whether the implant eventually may have been salvaged, it is in the patient's best interest to remove the nidus of infection and resolve the infection as quickly as possible. This may partially explain why reconstruction with a tissue expander was associated with implant failure in our cohort, because a tissue expander is often placed when there is anticipation of continued radiotherapy after initial reconstructive surgery.

Limitations of this study include nonstandardized intraoperative note-taking, as well as a 90-day follow-up period. Not all surgeons detailed intraoperative findings to the same extent in their surgical notes. Additionally, a new infection prevention protocol was introduced approximately halfway through the duration of data collection, which could provide a confounding variable. There was unfortunately little documentation of the subjective patient or clinical findings that most affected surgeons' decisions to attempt salvage or proceed with explantation. Furthermore, a follow-up period of 90 days may not be sufficient to catch quiescent bacterial infections. Our groups of patients with implant failure and attempted salvage are still relatively small and are limited to 1 practicing group. The small size of the group who developed infection and even smaller size of the individual treatment groups could impact the ability to discern factors that are statistically significant in predicting implant failure. Future studies should include investigation of intraoperative observations and their associations with implant failure with standardized templates to document intraoperative findings in a larger patient cohort. These studies should furthermore include consideration of quantifying the amount of unincorporated ADM in the implant pocket, to assess whether this affects outcomes. Future studies should also aim to elucidate whether unincorporated ADM alone is a factor that causes the surgeon to consider explantation in breasts that otherwise may have been salvaged, or whether it truly represents an altered healing process and more severe infection that is less likely to survive salvage attempts.

CONCLUSIONS

The purpose of this study was to investigate factors specifically predictive of infection after IBBR, and ultimately, intraoperative findings that were predictive of implant failure. We identified unincorporated ADM as an intraoperative factor that was significantly associated with eventual implant failure. Plastic surgeons may be able to utilize the incorporation status of ADM to aid intraoperative decision-making in the future.

Disclosures

The authors declared no potential conflicts of interest with respect to the research, authorship, and publication of this article.

Funding

The authors received no financial support for the research, authorship, and publication of this article.

REFERENCES

Author notes