Capsular contracture is the most common complication of breast augmentation and implant-based breast reconstruction in the United States.1–3 Visible or palpable deformity from contracture may occur in 2% to 15% of patients who undergo primary breast augmentation and 5% to 22% of patients after revision augmentation.4 Understanding the pathogenesis of capsular contracture is important, as it may lead to the identification of therapeutic targets that may prevent or mitigate disease progression. Although the development of capsular contracture is undoubtedly multifactorial, there have been reports that suggest implant texture is an important variable. Although textured implants have gained disfavor because of their association with breast implant–associated anaplastic large-cell lymphoma (BIA-ALCL),5 recent level I evidence reveals that smooth implants have a greater relative risk of capsular contracture than textured implants.6 The molecular environment during initial capsule formation likely influences the pathogenesis of contracture, and different patterns of gene expression based on capsule type may help explain the increased frequency of contracture in patients with smooth implants. Because textured and smooth implants are known to cause capsular contracture at different rates, we aimed to identify and characterize differences in gene expression in implant capsules that varied only in surface texture.
Although tissue from contracted human capsules has been used to study differences in gene expression profiles important to contracture pathogenesis, numerous variables including demographic differences, reason for initial implantation, underlying disease, lead surgeon, or insertion site used at the time of implantation may potentially confound these data.7 In this study, we placed miniature silicone implants in a rodent model to study and genetically characterize host response to implant texture, and correlate these findings with human capsular pathologic data. We aim to identify differences in patterns of gene expression between implant types that may shed insight into future therapeutic targets directed at minimizing or preventing capsular contracture.
PATIENTS AND METHODS
Age- and weight-matched female Fischer rats were used as the small-animal model of capsular contracture, as they are sufficiently large for implant placement. Textured or smooth 1 × 1-cm silicone implants (Allergan, Corp., Irvine, CA) were placed in the submammary gland position. After induction of anesthesia using an isoflurane cone mask, the animal’s chest was shaved and cleaned with alternating treatments of alcohol and povidone-iodine. Surgery was performed under aseptic techniques beginning with a 2-cm incision along the chest wall, followed by exposure of the submammary gland, implant placement, and surgical closure (Fig. 1). Postoperative analgesia was administered with a one-time dose of slow-release buprenorphine. Sutures were removed 2 weeks postoperatively to avoid excessive irritation to the surgical site. Animals were monitored every day postoperatively to ensure recovery and to note for signs of infection, inflammation, wound dehiscence, and contracture formation. The rats were euthanized at postoperative week 6 for implant and capsule extraction. Extracted capsules were snap-frozen in liquid nitrogen for RNA and protein isolation or sectioned after paraffin embedding for immunohistochemical analysis.
RNA Sequencing and Bioinformatics Pipelines
RNA was extracted using Qiagen miRNeasy kit and the RNA quality was checked on the Agilent Bioanalyzer 2100 with RNA integrity number greater than or equal to 9.0. RNA sequencing libraries were constructed using Kapa Biosystems mRNA library preparation kit (Roche, Basel, Switzerland) with 1 µg total RNA, and RNA sequencing libraries were sequenced at the University of Southern California Genomics Core on NextSeq 500 at 75 bp single-end read. To examine gene transcript expression in the extracted capsules and identify differentially expressed genes based on mean fold change, the BitSeq tool has been used to analyze RNA sequencing data sets.8 It uses a probability of positive log ratio (PPLR) approach, which uses a Bayesian hierarchical model to detect differentially expressed genes or other genomic features.9 Originally developed for microarray analysis, this tool has more recently been adjusted for RNA sequencing data.8 The PPLR expresses the probability or likelihood of the ratio being positive (i.e., second condition being up-regulated with respect to the first). A PPLR value close to 1 means that there is a very high probability of a given transcript being up-regulated in the second condition. When the PPLR is very low or close to 0, this implies a very low probability of up-regulation, and consequently the probability of down-regulation is high. There is no direct relationship between PPLR and P value, as they come from different perspectives of looking at the same problem, such that in the probabilistic approach, an uncertainty propagation between the subsequent stages of analysis is possible and desired. Here, we selected transcripts with either a PPLR greater than 0.975 and fold change greater than or equal to 2 or a PPLR less than 0.025 and fold change less than 0.5 for further profiling.
Quantitative Polymerase Chain Reaction Validation
Quantitative reverse transcription polymerase chain reaction (qRT-PCR) was performed to confirm three selected differentially expressed genes, including matrix metalloproteinase-3 (MMP3), troponin T3 (TNNT3), and neuregulin-1 (NRG1) between textured and smooth silicone implant capsules. We obtained the following rat primers for the genes: National Center for Biotechnology Information reference sequence NM_133523.3 (MMP3), National Center for Biotechnology Information reference sequence NM_001270665.1 (TNNT3), and National Center for Biotechnology Information reference sequence NM_001271124.1 (NRG1). Gene expression was normalized to glyceraldehyde 3-phosphate dehydrogenase mRNA levels as an endogenous control in qRT-PCR. Using smooth implants as the control, statistical analysis of mean relative fold change in gene expression was conducted by means of unpaired t tests.
Immunohistochemical staining was performed according to the Abcam immunohistochemistry-paraffin protocol. After deparaffinization, heat-induced antigen retrieval with Immunohistochemical-Tek Epitope Retrieval Solution (IHCWorld, Woodstock, MD) was performed on selected transcripts. Endogenous peroxidase activity was blocked with 3% hydrogen peroxide in 10% normal serum, and nonspecific binding was blocked with 1% bovine serum albumin in phosphate-buffered saline with Tween. Tissue was incubated with the primary antibody overnight at 4°C. Antibody staining was visualized using horseradish peroxidase–conjugated secondary antibodies developed with the 3,3′-diaminobenzidine chromogenic reagent kit (Vector Laboratories, Burlingame, CA). MMP3, TNNT3, and NRG1 (Abcam, Cambridge, MA) were the primary antibodies used. All secondary antibodies were obtained from Vector Laboratories.
The slides were counterstained with hematoxylin and eosin. Capsule cellularity and collagen fibers were visualized and qualitatively assessed using bright-field microscopy to capture relative fiber density, and regions of interest were scanned using a Mirax slide scanner (Carl Zeiss, Munich, Germany).
To better understand the in vivo host response to smooth and textured silicone implants, we used a reproducible small-animal model with implantation of miniature silicone gel–filled breast devices (Allergan) that were identical except for their surface textures. Specifically, Fischer rats were divided into two groups: the first group had implantation of smooth silicone implants into the submammary gland position (n = 5), and the second group had implantation of textured silicone implants (n = 5). Six weeks after implantation, animals were euthanized, and implant capsules were harvested for histologic and molecular analysis.
As expected, hematoxylin and eosin staining revealed that capsules surrounding textured implants were ridged and less continuous compared to the smooth group. Collagen fibers were loosely arranged in bands parallel to the capsule-implant interface, and there was no difference in fiber density between the two implant types. Textured implant capsules had thicker capsular epithelium and more surrounding skeletal muscle fibers than smooth implant capsules. There was limited vascularity in the tissue surrounding both implants. Capsule cellularity did not vary with implant type and there was no evidence of synovial metaplasia (Fig. 2).
RNA from smooth and textured implant capsules was isolated to identify differential changes in gene expression. Because of quality control measures, three of five smooth implant capsules and all five textured implant capsules were included in the analysis. Through RNA sequencing, 18,555 transcripts were identified from the capsules analyzed. We used the PPLR algorithm to identify target transcripts expressed in extracted capsules. Using smooth implant capsule values as baseline, we selected transcripts with either PPLR greater than 0.975 and fold change greater than 2, or PPLR less than 0.025 and fold change less than 0.5 for further profiling and present this analysis as a volcano plot (Fig. 3). Table 1 shows the top 10 up-regulated and down-regulated transcripts identified from RNA sequencing data based on mean fold change.
Table 1. -
Whole Transcriptome Expression Changes in Smooth versus Textured Implantsa
||Troponin T type 3 (skeletal, fast)
||LIM domain binding 3
||Crystallin, alpha A
||Myosin, heavy chain 7, cardiac muscle, beta
||Myosin, heavy chain 1, skeletal muscle, adult
||Potassium intermediate/small conductance calcium channel
||Chemokine (C-X-C motif) ligand 5
||Matrix metallopeptidase 3
||Tumor protein p63 regulated 1
||Sperm acrosome associated 1
||Olfactory receptor 772
||Olfactory receptor 1531
||Similar to ubiquitin 1 isoform 2
||Olfactory receptor 53
||Solute carrier family 5 (sodium/glucose cotransporter)
||Testis expressed 38
||Olfactory receptor 1286
aTop 10 up-regulated and down-regulated candidate genes with the largest fold change based on PPLR obtained through RNA sequencing.
bCandidate proteins were further characterized with immunohistochemical analysis.
To confirm the findings from RNA sequencing, the expression of several candidate genes was assessed based on the PPLR, fold change cutoffs, and investigator interest. qRT-PCR analysis was performed for MMP3, TNNT3, and NRG1. MMP3 and TNNT3 were both up-regulated in textured implant capsules compared to smooth implant capsules, with a mean relative fold change of 8.79 ± 6.79 (P = 0.0059) and 4.81 ± 3.30 (P = 0.0056), respectively. NRG1 was down-regulated in textured implant capsules, with a mean relative fold change of 0.40 ± 0.27 (P < 0.0001) compared to smooth implant capsules (Fig. 4).
To localize changes in candidate gene expression to the capsule anatomy of our experimental animals, we performed antibody-specific immunohistochemistry for these three gene products. Immunohistochemistry staining of MMP3 showed decreased protein expression in the smooth group that was weakly localized to the capsule-implant interface (Fig. 5). In the textured group, MMP3 signal was more abundant and localized to the capsule-implant interface. Expression of TNNT3 was also decreased in smooth implant capsules compared to textured implant capsules (Fig. 6). In the textured group, TNNT3 signal was strong in the capsule-implant interface and subcutaneous tissue. In contrast to MMP3 and TNNT3, staining for NRG1 was increased in the smooth group compared to the textured group, corroborating our RNA sequencing findings (Fig. 7). In the smooth group, NRG1 staining was strongly localized to the capsule-implant interface and dermal layer. In the textured group, NRG1 staining was distributed throughout the capsule section.
We further correlated these findings with immunohistochemistry data of human capsules derived from either normal breast implants or implants with capsular contracture. All human breast implants assessed for comparison were smooth implants. Immunohistochemistry staining of MMP3 showed decreased MMP3 expression in capsules with contracture compared to healthy capsules (Fig. 5). In healthy capsules, staining for MMP3 was evenly distributed throughout the section. Expression of TNNT3 was also decreased in capsules with contracture, whereas in healthy capsules, the intensity of the signal was strongest at the level of the capsule-implant interface (Fig. 6). In contrast to MMP3 and TNNT3, staining for NRG1 was increased in capsules with contracture compared to healthy capsules (Fig. 7). In contracted capsules, the capsule-implant interface was the region of greatest NRG1 expression.
Over 300,000 breast implant procedures are performed in the United States each year.10 The most common complications of breast implant procedures are capsular contracture, capsular rupture, and revision operations.11 Less frequent but more serious complications include autoimmune syndrome induced by adjuvants and BIA-ALCL.12,13 Reducing complication rates of implant-based augmentation and reconstruction has the potential to result in a dramatic improvement in patient well-being and health care costs. In particular, increasing implant biocompatibility and modulating host inflammatory responses can reduce implant malfunction and postoperative complications of numerous implant-based procedures.14
Capsular contracture is the most common complication of implant-based breast surgery and most common reason for reoperation.4 Currently, the pathogenesis of capsular contracture is thought to involve a foreign body reaction with a component of bacterial colonization.14 High colonization rates are associated with states of chronic inflammation and are predominantly seen in moderate to severe contractures.15 Controlling bacterial colonization might therefore be an effective way to mitigate contracture severity. More recent data suggest that disruption of the capsule-implant interface from physical movement plays a role in contracture formation.16 This offers an explanation for the increased rate of contracture formation in smooth implants, which have more mobile surfaces than textured implants and are more likely to experience disruption of the capsule-implant interface.
The pathogenesis of capsule formation begins with the normal stages of wound repair. Platelets release transforming growth factor-β, platelet-derived growth factor, and interleukin-8 to recruit and activate macrophages, which in turn attach to the implant surface by means of integrin receptors and secrete fibroblast-recruiting agents. This leads to giant cell formation and granulation tissue deposition. Giant cells, formed from a failed attempt at foreign body phagocytosis, release reactive oxygen species and transforming growth factor-β to stimulate fibroblast-mediated collagen production and granulation tissue formation. The extracellular matrix (ECM) and vasculature in granulation tissue ultimately remodels to form a fibrous capsule, leading to capsular contracture.14 Matrix metalloproteinases (MMPs) secreted by macrophages are involved in contracture pathogenesis through the regulation of angiogenesis, collagen deposition, and ECM remodeling.14 Tissue inhibitors of metalloproteinases (TIMPs) are soluble extracellular proteins that directly antagonize MMPs. TIMPs are induced by growth factors and cytokines involved in chronic inflammation, including transforming growth factor-β, platelet-derived growth factor, and interleukin-8.17 The balance between MMPs and TIMPs influences collagen fiber deposition, ECM composition, and ultimately contracture formation.18 Numerous peer-reviewed studies have noted differences in collagen fiber deposition and myofibroblast orientation in smooth versus textured implant capsules. Long, parallel collagen fibers, necessary for myofibroblasts to exert enough force on the capsule to cause contracture, are less likely to form in textured implants, offering an explanation for the lower rates of contracture seen in textured implants.5,15,19,20
RNA expression patterns in capsules surrounding smooth and textured implants have not yet been characterized in the literature. Using a novel, clinically relevant breast implant animal model, we found that MMP3 and TNNT3 expression is up-regulated in textured implants, whereas NRG1 expression is up-regulated in smooth implants. We confirmed these findings quantitatively with qRT-PCR and qualitatively with immunohistochemistry. Importantly, we evaluated normal and contracted capsules from human pathologic smooth implant capsular samples to further characterize the relative expression patterns of MMP3, TNNT3, and NRG1 obtained from our small-animal model. A translational analysis such as this one offers invaluable insight into the pathogenesis of capsular contracture by describing the precontracture transcriptome in smooth versus textured implant capsules with pathologic correlation. Understanding this precontracture transcriptome can help elucidate the mechanism behind higher rates of capsular contracture in smooth implants and guide the development of future therapeutic targets against this disease.
This study reveals that NRG1 is up-regulated, both qualitatively and quantitatively, in capsules derived from smooth animal implants and smooth human implants with capsular contracture. The diffuse staining pattern observed throughout capsule sections derived from smooth implants indicates expression by multiple cell types. Elevated NRG1 and its receptors, HER2 and HER3, have previously been linked to hypertrophic scarring by means of increased connective tissue growth factor expression and ECM deposition.21 This suggests that elevated NRG1 expression in smooth implant capsules contributes to contracture pathogenesis by means of increased ECM deposition and scar tissue formation. Fibrous capsules differ microscopically from other fibrotic diseases, particularly in collagen orientation and cable formation, and NRG1 may play a role in modulating these differences.2 Future studies that follow NRG1 expression, collagen deposition, and cable orientation throughout the development of capsular contracture can help elucidate the pathogenesis of fibrosis in this disease.
In contrast to NRG1, TNNT3 was up-regulated in textured animal capsules and healthy human capsules. TNNT3 is an isotype of troponin T found in fast-twitch skeletal muscle. Autosomal dominant mutations in TNNT3 have been linked to distal arthrogryposis type 2B, a disorder characterized by congenital contractures of the distal limb joints in the setting of a hypercontractile state.22,23 Although the relationship between TNNT3 and capsular contracture has not been well established, histologic analyses found increased skeletal muscle fibers present around capsules derived from textured implants compared to smooth implants. Future studies are needed to quantify skeletal muscle formation around implant capsules and ascertain the role of skeletal muscle formation in contracture pathogenesis.
Similar to TNNT3, MMP3 was up-regulated in textured animal capsules and healthy human capsules, with staining predominantly localized to the capsule-implant interface. This is consistent with previous studies showing that MMPs—specifically, MMP2—stain positive at the interface because of increased localization of macrophages, which are largely responsible for MMP release into the ECM.18 MMPs are expressed by macrophages during acute and chronic inflammation and function by breaking down ECM. In contrast, TIMPs inhibit MMPs and prevent ECM degradation. The TIMP-to-MMP ratio is increased in human capsular contractures found around both textured and smooth implants and is proportional to contracture severity based on Baker grade.6 This suggests that TIMPs contribute to excess fibrosis in contractures by inhibiting MMP-mediated ECM breakdown. Our RNA sequencing analysis of precontracted capsules reveals that TIMP expression is not significantly changed in textured implant capsules compared to smooth implant capsules, with the fold change of TIMP1, TIMP2, TIMP3, and TIMP4 all failing to reach the previously established greater than or equal to 2 or less than −2 cutoff (respectively, the fold changes were −0.796617, −0.49299, −0.619805, and −0.27549). Conversely, MMP3 was significantly up-regulated in textured implant capsules. The lower TIMP-to-MMP ratio seen in our textured implant capsules may represent a state of protection against contracture pathogenesis, such that MMP up-regulation by macrophages in response to differences in surface texture diminishes fibrosis by means of increased ECM breakdown. Higher levels of MMPs have previously been observed in textured implants at different points in time within the capsule lifetime.6 This further suggests that capsular contracture in textured implants may take longer to develop because of continuously higher expression of MMPs relative to TIMPs. Only when TIMPs are sufficiently up-regulated does the inhibition of ECM breakdown occur, followed by the development of fibrosis. These findings are contradicted by a recent study of human contractures suggesting that the TIMP-to-MMP ratio decreases with increasing contracture severity.24 However, it has not been clearly determined whether changes in the TIMP-to-MMP ratio are responsible for contracture development or are instead the result of contracture. Studies examining capsular gene expression in smooth and textured implants at different time points will be necessary to elucidate the temporal relationship between MMPs, TIMPs, and contracture severity.
Recent meta-analyses reveal that smooth implants have a significantly greater rate of capsular contracture development than textured implants, with a relative risk of 3.10 to 4.20. Implant texture is cited as the most important determinant of contracture development when compared to incision type, pocket location, and saline versus silicone implant type.1,5 This strongly suggests that differential gene expression patterns of capsules from differently textured implant surfaces may be used to predict the risk of contracture formation. From our study, possible gene targets for future analysis are NRG1, TNNT3, and MMP3, in addition to TIMPs previously identified in the literature. The study of contracture formation can be continued in animal models using siRNA to down-regulate these gene transcripts, possibly linking NRG1 down-regulation with slower contracture progression, and TNNT3 or MMP3 down-regulation with hastening of fibrosis. If future studies have promising findings, these proteins may become efficacious therapeutic targets.
Finally, there is a growing body of evidence examining the occurrence of BIA-ALCL, a malignancy of T lymphocytes exclusively associated with textured breast implants. More than 500 cases of BIA-ALCL have occurred worldwide, none in the setting of smooth implant insertion.13 Suspected BIA-ALCL presents as a delayed seroma not otherwise explained by infection or trauma, occurring at least 1 year after implant insertion. Localized and advanced BIA-ALCL treated with complete surgical resection has a favorable prognosis.25 Although the pathogenesis of BIA-ALCL remains poorly understood, it is hypothesized that an interaction between the breast microbiome, textured implants, genetic susceptibility, and time plays a role.13 Future studies on the expression profiles of capsules derived from smooth and textured silicone implants might not only help us better understand the pathogenesis of capsular contracture, but might also shed some light on the genes responsible for the predisposition of textured implants to tumorigenesis.
There are some important limitations to consider with this work. Although our qRT-PCR RNA analyses used rat primers for MMP3, TNNT3, and NRG1, it is unclear whether our RNA sequencing libraries reflected the characteristic, posttranslational eukaryotic modifications identified with MMPs.26 This distinction will be important for future genetic analyses of capsular RNA isolates to differentiate host from bacterial MMP expression and, in particular, bacterial colonization that predisposes to biofilm formation. In addition, given individual surgeon preferences and patient anatomical specifications, human capsular analyses were performed for implants placed in both prepectoral and subpectoral reconstructive planes. To examine the relationship between reconstructive plane and consequent capsular contracture and gene expression, future work can appropriately subdivide analyses. To broaden the scope of genes considered as contributory to capsular contracture pathogenesis, we recognize selecting transcripts based on different PPLR for profiling. We did, however, note the top 10 up-regulated and down-regulated transcripts identified from our RNA sequencing analysis based on mean fold change. Lastly, given our finding that textured implants will likely experience contracture at a later time point than smooth implants, it would also be worthwhile to analyze the precontracture transcriptome during the weeks postoperatively after implant placement to verify the durability of gene expression in this period.
1. Wong CH, Samuel M, Tan BK, Song C. Capsular contracture in subglandular breast augmentation with textured versus smooth breast implants: a systematic review. Plast Reconstr Surg. 2006;118:1224–1236.
2. Moyer KE, Ehrlich HP. Capsular contracture after breast reconstruction: collagen fiber orientation and organization. Plast Reconstr Surg. 2013;131:680–685.
3. Spear SL, Baker JL Jr. Classification of capsular contracture after prosthetic breast reconstruction. Plast Reconstr Surg. 1995;96:1119–1123; discussion 1124.
4. Wan D, Rohrich RJ. Revisiting the management of capsular contracture in breast augmentation: a systematic review. Plast Reconstr Surg. 2016;137:826–841.
5. Collett DJ, Rakhorst H, Lennox P, Magnusson M, Cooter R, Deva AK. Current risk estimate of breast implant-associated anaplastic large cell lymphoma in textured breast implants. Plast Reconstr Surg. 2019;143(A Review of Breast Implant-Associated Anaplastic Large Cell Lymphoma):30S–40S.
6. Liu X, Zhou L, Pan F, Gao Y, Yuan X, Fan D. Comparison of the postoperative incidence rate of capsular contracture among different breast implants: a cumulative meta-analysis. PLoS One. 2015;10:e0116071.
7. Ulrich D, Ulrich F, Pallua N, Eisenmann-Klein M. Effect of tissue inhibitors of metalloproteinases and matrix metalloproteinases on capsular formation around smooth and textured silicone gel implants. Aesthetic Plast Surg. 2009;33:555–562.
8. Glaus P, Honkela A, Rattray M. Identifying differentially expressed transcripts from RNA-seq data with biological variation. Bioinformatics. 2012;28:1721–1728.
9. Liu X, Milo M, Lawrence ND, Rattray M. Probe-level measurement error improves accuracy in detecting differential gene expression. Bioinformatics. 2006;22:2107–2113.
10. American Society of Plastic Surgeons. 2017 plastic surgery statistics report. Available at: https://www.plasticsurgery.org/documents/News/Statistics/2017/plastic-surgery-statistics-full-report-2017.pdf
. Accessed May 1, 2019.
11. Spear SL, Parikh PM, Goldstein JA. History of breast implants and the Food and Drug Administration. Clin Plast Surg. 2009;36:15–21, v.
12. Nunes E Silva D, Gründler C, Spengler MDGMT, et al. Autoimmune syndrome induced by adjuvants (ASIA) after silicone breast augmentation surgery. Plast Reconstr Surg Glob Open. 2017;5:e1487.
13. Rastogi P, Deva AK, Prince HM. Breast implant-associated anaplastic large cell lymphoma. Curr Hematol Malig Rep. 2018;13:516–524.
14. Major MR, Wong VW, Nelson ER, Longaker MT, Gurtner GC. The foreign body response: at the interface of surgery and bioengineering. Plast Reconstr Surg. 2015;135:1489–1498.
15. Bui JM, Perry T, Ren CD, Nofrey B, Teitelbaum S, Van Epps DE. Histological characterization of human breast implant capsules. Aesthetic Plast Surg. 2015;39:306–315.
16. Irkoren S, Ozkan HS, Ceylan E, Sivrioglu N, Tataroglu C, Durum Y. The effect of Botox on the implant stabilization and capsular formation: an experimental study. Ann Plast Surg. 2015;75:91–97.
17. Moore L, Fan D, Basu R, Kandalam V, Kassiri Z. Tissue inhibitor of metalloproteinases (TIMPs) in heart failure. Heart Fail Rev. 2012;17:693–706.
18. Ulrich D, Lichtenegger F, Eblenkamp M, Repper D, Pallua N. Matrix metalloproteinases, tissue inhibitors of metalloproteinases, aminoterminal propeptide of procollagen type III, and hyaluronan in sera and tissue of patients with capsular contracture after augmentation with Trilucent breast implants. Plast Reconstr Surg. 2004;114:229–236.
19. Rubino C, Mazzarello V, Farace F, et al. Ultrastructural anatomy of contracted capsules around textured implants in augmented breasts. Ann Plast Surg. 2001;46:95–102.
20. Danino AM, Basmacioglu P, Saito S, et al. Comparison of the capsular response to the Biocell RTV and Mentor 1600 Siltex breast implant surface texturing: a scanning electron microscopic study. Plast Reconstr Surg. 2001;108:2047–2052.
21. Kim JS, Choi IG, Lee BC, et al. Neuregulin induces CTGF expression in hypertrophic scarring fibroblasts. Mol Cell Biochem. 2012;365:181–189.
22. Li X, Jiang M, Han W, et al. A novel TNNI2 mutation causes Freeman-Sheldon syndrome in a Chinese family with an affected adult with only facial contractures. Gene. 2013;527:630–635.
23. Zhao N, Jiang M, Han W, et al. A novel mutation in TNNT3 associated with Sheldon-Hall syndrome in a Chinese family with vertical talus. Eur J Med Genet. 2011;54:351–353.
24. Kyle DJ, Harvey AG, Shih B, Tan KT, Chaudhry IH, Bayat A. Identification of molecular phenotypic descriptors of breast capsular contracture formation using informatics analysis of the whole genome transcriptome. Wound Repair Regen. 2013;21:762–769.
25. Clemens MW, Brody GS, Mahabir RC, Miranda RN. How to diagnose and treat breast implant-associated anaplastic large cell lymphoma. Plast Reconstr Surg. 2018;141:586e–599e.
26. Gonçalves AN, Meschiari CA, Stetler-Stevenson WG, et al. Expression of soluble and functional full-length human matrix metalloproteinase-2 in Escherichia coli
. J Biotechnol. 2012;157:20–24.