Pilot Study: Investigating the Local Breast Microbiome in Implant-based Breast Reconstruction Using 16S rRNA Sequencing

Abstract

Background:

Prior studies have used 16S rRNA sequencing to examine and define the local breast microbiome, but this has not been investigated with respect to breast reconstruction. Periexpander fluid can be readily collected in patients with a dual-port tissue expander, which could allow us to define the local breast microbiome at any given time point. This study aimed to determine the feasibility of obtaining microbiome data from periexpander fluid and explore its potential relevance for clinical implant infections.

Methods:

We designed a pilot study including patients undergoing mastectomy with 2-stage implant-based reconstruction using dual-port tissue expanders. The periexpander fluid was obtained by accessing the aspiration port during standard postoperative visits, and this fluid was stored in a 1:1 ratio with DNA/RNA shield at −20°C. The microbiome of each sample was defined using 16S rRNA microbiome sequencing.

Results:

Intraoperative and postoperative samples from 10 patients were sequenced to determine the feasibility of obtaining microbiome data from the periexpander aspirates. We were successful in obtaining microbiome data from all aspirates. Our results indicate that there are a large range of genera represented, but several genera appear to be more pervasive, including Pseudomonas, Corynebacterium, Phenylobacterium, Acinetobacter, and Staphylococcus.

Conclusions:

We found that it is feasible to perform microbiome sequencing of breast tissue and periexpander aspirates to define the local breast environment. Rather than focusing on eliminating bacteria, it is critical to learn more about how we can optimize the balance of microorganisms in the breast microbiome to minimize infection risk.

Takeaways

Question: Is it feasible to obtain microbiome samples from breast periexpander aspirates?

Findings: In our randomized controlled pilot study, we found that it is possible to obtain microbiome data from breast periexpander aspirates. Furthermore, we found certain bacteria to be most pervasive, including the Pseudomonas, Corynebacterium, Phenylobacterium, Acinetobacter, and Staphylococcus genera.

Meaning: Rather than focusing on eliminating bacteria with antibiotics, it may be more important to learn more about how we can optimize the balance of microorganisms in the breast microbiome.

INTRODUCTION

Approximately 50% of women diagnosed with breast cancer will undergo some form of breast reconstruction following mastectomy, with the most common method being implant-based reconstruction (>70%).^1,2 Postoperative infections are a significant cause of morbidity in this operation, with reported infection rates ranging from 5.8% to 28%.^3,4 Infections in the breast cancer population can have serious consequences, not only by delaying definitive reconstruction but also by delaying adjuvant therapy.⁵ Infections may lead to reconstructive failure requiring readmissions and reoperations. Furthermore, the broad-spectrum use of antibiotics to prevent and/or treat infections can lead to antibiotic resistance.⁶

Infections in postmastectomy reconstruction patients are prevalent and morbid but remain poorly understood. Although a wide range of bacteria are causative, in at least 25% of cases no organism is isolated in culture.⁷ There are certainly known risk factors for infection, including mastectomy flap necrosis, chemotherapy, radiation therapy, diabetes, and tobacco use; however, infections frequently occur in women, with no identifiable risk factors.⁸

Recent literature in other surgical fields suggests that a patient’s local microbiome can affect rates of infection and postoperative complications.^9,10 Prior studies specifically in breast surgery have used 16S rRNA sequencing to examine and define the local breast microbiome from breast tissue. They found that the breast microbiome and the gut microbiome in patients with breast cancer differ significantly in diversity measures and in the abundance of certain microbes when compared with healthy controls.^11–16

The breast microbiome has not been studied in the setting of mastectomy with implant-based reconstruction. Additionally, the advent of dual-port tissue expanders has made periexpander fluid aspiration a routine practice for many plastic surgeons. These aspirates could allow us to define the periexpander microbiome at any given point in time.

In this study, we aimed to determine the feasibility of obtaining microbiome sequencing data from breast tissue and periexpander fluid aspirates. We believe that a better understanding of the breast and periexpander microbiome will improve our understanding of postmastectomy implant complications and infections.

METHODS

We designed a trial including patients with breast cancer or genetic predisposition who were scheduled to undergo mastectomy with 2-stage implant-based reconstruction. All patients had a dual-port tissue expander placed in the first reconstructive stage. In all cases, either DermACELL or AlloDerm allograft was used without any unique solution or prewash. Before insertion of the tissue expander, the breast pocket was instilled with betadine solution, and this was allowed to sit for 3 minutes before being further irrigated out with normal saline. Patients in our study were randomized to 1 of 2 groups: 1 group received 24 hours of perioperative intravenous antibiotics and 7 days of postoperative oral antibiotics, whereas the other group received only 24 hours of perioperative intravenous antibiotics.

Breast microbiome samples were obtained at the time of surgery using a breast tissue sample obtained directly from the periphery of the mastectomy specimen, and then at their 1-week postoperative visit by accessing the aspiration port of the tissue expander and sampling the periexpander fluid. To stabilize the DNA and RNA in the specimens, 2-mL CryoSafe sample tubes were all prefilled with 1 mL of DNA/RNA shield. Tissue samples of 0.5 cm³ were obtained and placed in these sample tubes at the time of a patient’s mastectomy. At each patient’s postoperative visit, 1 mL of periexpander fluid was added to the sample tubes for a ratio of 1:1 DNA/RNA shield to specimen. All specimens were stored at −20°C.

Within our pilot study cohort, we primarily aimed to test the feasibility of microbiome sequencing with 16S rRNA methods on tissue samples and periexpander aspirate samples in 10 consecutive patients. After determining feasibility, we analyzed each sample’s microbiome in terms of the top genera, their representative percentage, and the overall number of species present. Our secondary aim was to determine the impact of 24 hours of perioperative antibiotics versus 7 days of postoperative antibiotics on clinical infections and the associated breast microbiomes.

RESULTS

Five patients were randomized to receive both 24 hours of perioperative cefazolin and 7 days of postoperative oral cephalexin, and 5 to receive only 24 hours of perioperative cefazolin. A perioperative tissue sample and a postoperative aspirate sample from each of these 10 patients in our study were sequenced to ensure the feasibility of our planned sequencing techniques.

The top 10 genera represented in each sample and their percentage of representation in each sample are depicted in Supplemental Digital Content 1. (See table, Supplemental Digital Content 1, which displays the top 10 genera represented in each sample and their respective percentages in each sample in a cohort of patients receiving 7 d of antibiotics [A] and a cohort of patients receiving 24 h of antibiotics [B], https://links.lww.com/PRSGO/E157.) Table 1 demonstrates the genera that were in the top 10 genera of more than 1 sample and their average representative percentage. The top 5 most common genera identified in the 20 samples were Pseudomonas, Corynebacterium, Phenylobacterium, Acinetobacter, and Staphylococcus. Averaged across all samples, the top 5 genera representing the largest percentage of the microbiomes were Pseudomonas (32.77%), Staphylococcus (7.80%), Burkholderia (7.28%), Corynebacterium (7.09%), and Acinetobacter (4.19%). Between 10 and 37 genera were detected in each sample out of a total of 99 genera evaluated.

Table 1.

Genera Represented in the Top 10 of More Than 1 Sample and Their Average Representative Percentage

Genus	No. Samples Containing Genus Among Top 10 Represented	No. Samples Containing Genus Among Top 10 Represented, %	Average % per Sample Containing Genus	Average % Across All Samples
Pseudomonas	15	75	43.70	32.77
Corynebacterium	15	75	9.46	7.09
Phenylobacterium	12	60	6.79	4.07
Acinetobacter	11	55	7.61	4.19
Staphylococcus	11	55	14.19	7.80
Burkholderia	9	45	16.19	7.28
Bradyrhizobium	8	40	4.42	1.77
Sphingomonas	8	40	3.98	1.59
Dietzia	7	35	3.51	1.23
Vampirovibrio	6	30	4.00	1.20
Xanthomonas	5	25	6.30	1.57
Pelomonas	5	25	6.52	1.63
Bacillus	5	25	5.86	1.47
Enhydrobacter	4	20	3.06	0.61
Shigella	4	20	5.53	1.11
Ralstonia	3	15	1.94	0.29
Finegoldia	3	15	4.40	0.66
Methylobacterium	3	15	3.79	0.57
Streptococcus	3	15	7.36	1.10
Hydrogenispora	2	10	3.80	0.38
Moheibacter	2	10	7.25	0.73
Thermicanus	2	10	1.75	0.18
Brevundimonas	2	10	6.62	0.66
Hymenobacter	2	10	26.65	2.67
Comamonas	2	10	3.37	0.34

Looking at outcomes for these select patients, 1 patient (patient no. 5) in the 7-day antibiotic cohort (Supplemental Digital Content 1, https://links.lww.com/PRSGO/E157) developed late cellulitis after implant exchange, which was successfully treated with oral antibiotics (trimethoprim–sulfamethoxazole). We do not have culture data or 16s rRNA data from this timepoint for this patient, as the infection occurred when she had her permanent implant (with no seroma port or drain available for testing). She did not have a fever or leukocytosis. Interestingly, her microbiome data demonstrated a change in dominant genera from Burkholderia and Streptococcus to Sphingomonas and Pelomonas between the first and second samples. Another patient (patient no. 2) in the cohort who received 24 hours of antibiotics (Supplemental Digital Content 1, https://links.lww.com/PRSGO/E157) developed an infection that required explant of the tissue expander. The 16s rRNA sequencing showed a change of the Pseudomonas genus in this patient’s microbiome from 4.29% at the time of her initial operation to 71.14% at her first postoperative visit. This is consistent with intraoperative cultures from her explant, which grew Pseudomonas aeruginosa.

DISCUSSION

Our results demonstrate that it is feasible to obtain 16S rRNA microbiome data from both the breast tissue and periexpander fluid. Furthermore, although a range of genera are represented in the breast microbiome, Pseudomonas, Corynebacterium, Phenylobacterium, Acinetobacter, and Staphylococcus may be the most pervasive. Staphylococcus and Pseudomonas are the most common Gram-positive and Gram-negative organisms, respectively, implicated in breast implant infections.⁷ Corynebacterium is a Gram-positive organism that has been reported to cause breast implant infections.⁷ Acinetobacter and Phenylobacterium are both Gram-negative organisms; only Acinetobacter has been reported in the literature to cause breast implant infections.⁷ Phenylobacterium is a Gram-negative rod that has been reported in the literature to cause cutaneous infectious granulomas.¹⁷

Our microbiome data paralleled our available clinical culture data when it was available. Patient 2 from the 24-hour antibiotic group had a dramatic shift in her breast microbiome to a dominant Pseudomonas genera before the development of her clinical infection. This was consistent with her operative cultures, which grew P. aeruginosa, indicating that our microbiome data are valid and possibly indicative of infectious potential before the development of clinical infection. However, in several other samples, we observed both consistently dominant genera and a shift in dominant genera without evidence of clinical infection, suggesting that the dominance of a single genus alone does not predict clinical infection (Supplemental Digital Content 1, https://links.lww.com/PRSGO/E157).

Many clinically significant implant infections are culture-negative, in part because patients are already on antibiotics when samples are obtained. However, it is clear that these infections represent active live bacteria, albeit at a much smaller load. We are aiming to better understand the implications and role of the microbiome in these more nuanced infections in particular. Although it is possible that dead bacterial DNA is captured in the microbiome 16s rRNA sequencing, it is widely agreed upon in the field of microbiome science that dead bacteria tend to be transient and do not have a significant impact.¹⁸ We believe the microbiome signatures reported here represent live bacterial data.

The significance of change in the type and number of species between samples after a shorter or longer antibiotics course is reported in raw data here, but does need to be further elucidated with a larger patient population. Furthermore, the relative diversity of species in a sample may prove to be a more important prognostic indicator for infection than the dominance of any particular genus.

One patient in each antibiotics randomization group developed a clinical infection. The use of postoperative antibiotics in this patient population is controversial; however, many plastic surgeons still opt to give postoperative prophylactic antibiotics despite the lack of evidence that this prevents infections.^19,20 Although this pilot study was not powered to detect a statistical significance or to determine equivalence in infection rates between the groups randomized to 24 hours of perioperative antibiotics versus 7 days of postoperative antibiotics, we hope to expand our randomized controlled trial to add to the body of literature on this topic.

This study was limited by its small sample size, the experimental nature of this technique, the lack of control specimens to account for noise in the results, and confounding variables in reconstruction, including both patient and surgical factors. Several factors that could influence various microbiomes in the body were not analyzed in this pilot study, such as adjuvant therapies, body mass index, and comorbidities. As we expand this study, we plan to include controls to clarify the analysis and include confounding variables in a regression analysis.

Clinically, microbiome data may be most useful and actionable in culture-negative infections as the ability to perform microbiome sequencing becomes more widespread. It may become a method to more quickly and efficiently gain bacterial information in the future as well.

CONCLUSIONS

It is feasible to perform microbiome sequencing of periexpander aspirates to define the local environment. From this pilot study alone, it is clear that bacteria are present in the breast periexpander environment and that the space will not be sterilized with antibiotics. Rather than focusing on eliminating bacteria, it is critical to learn more about how we can optimize the balance of microorganisms in the breast microbiome. The role of antibiotics in this endeavor remains unclear.

DISCLOSURES

Dr. Barnes is the recipient of the Scott Spear Award, a PSF grant, which funded this research. The other authors have no financial interest to declare in relation to the content of this article.