Dysregulated inflammatory response to urogynecologic meshes in women with diabetes and its implications.

The BMI in women with diabetes was 11.6% higher than those without (32.6 ± 4.2 vs. 29.2 ± 5.8, p=0.005) while other parameters in the demographic and clinical characteristics ( Table 2 ) were not significantly different between groups (all p>0.05). In addition, no differences were observed in the mesh-associated characteristics between groups ( Table 3 , all p>0.05). For age- and BMI-matched cases undergoing tissue analysis, there were no differences in the clinical data between groups (all p>0.05). Considering that the distribution of complication types (exposure only, pain only, and exposure + pain) in the diabetic vs. non-diabetic groups was comparable, and that our prior investigation showed no significant differences of inflammatory mediators between tissues with mesh exposure and pain, 17 we collectively analyzed these samples to increase statistical power. Prior to mesh implantation, the HbA1c (%) was controlled at 5.7 (min 5.4, max 5.9) and the BG (mg/dL) at 95 (min 84, max 107) in patients with diabetes, similar to those in patients without diabetes ( Table 4 ). It should be noted that due to the unavailability of glucose management data before mesh implantation when the operations were performed in other hospitals, the pre-implantation HbA1c measurements were retrieved from the record of a limited number of patients (n=4 in diabetic group and 1 in non-diabetic group). Before and after mesh removal, 23% and 36% of the cases had HbA1c > 8%, and accordingly, 25% and 50% had poorly controlled BG, respectively ( Table 4 ). Overall, the longitudinal glucose managements appeared variable and suboptimal with 59% of the patients having loose (24%) or poor glucose control (35%), respectively. In the non-diabetic group, the levels of BG and HbA1c remained normal and consistent. We examined the gene expression of 32 cytokines/chemokines and 7 growth factors involved in inflammatory response ( Figure 1 ). The results showed that the majority of these factors were not differentially expressed except C-C motif chemokine ligand 13 ( Ccl13 ), C-C motif chemokine ligand 24 ( Ccl24 ), and interleukin 1 alpha ( Il1a ). Interestingly, the expression of proinflammatory chemokines Ccl13 and Ccl24 was upregulated by 1.5- and 1.8-fold (p=0.035, 0.027, respectively), while that of Il1a was downregulated by 2.2-fold (p=0.037) in the diabetic vs. non-diabetic groups, demonstrating a dysregulated inflammatory response to mesh in the former. We screened the gene expression profile of 29 enzymes involved in epigenetic modifications and 12 major regulatory factors implicated in the regulation of immune mediators ( Figure 2A and 2B ). The results showed that most of these regulatory factors were comparably expressed, except for a mild 1.2-fold upregulation of Setdb2 (p=0.035), a histone lysine methyltransferase, and 1.8-fold downregulation of Fosl1 (p=0.026), a transcription factor, in the diabetic vs. non-diabetic groups ( Figure 2C ). Interestingly, the expression of Setdb2 strongly correlated with Ccl13 in the diabetic (tau-b=0.633, p<0.001, δ=6) but not the non-diabetic group (tau-b=−0.247, p=0.08, δ=0) ( Figure 3A ), while the expression of Fosl1 strongly correlated with Il1a in both non-diabetic (tau-b=0.420, p=0.003, δ=5) and diabetic (tau-b=0.617, p<0.001, δ=6) groups, suggesting their involvements in the dysregulated expression of proinflammatory genes. To determine the immune cell populations in the mesh-tissue samples, we examined the expression profile of various immune cells markers, including leukocytes (protein tyrosine phosphatase receptor type C [ Ptprc ]), myeloid-lineage cells (integrin subunit alpha M [Itgam]), macrophages (CD68 molecule [ Cd68 ]), M1 macrophages (Fc gamma receptor Ia [ Fcgr1a ], CD86 molecule [ Cd86 ], CD80 molecule [ Cd80 ]), M2 macrophages (mannose receptor C-type 1 [ Mrc1 ], CD163 molecule [ Cd163 ], MER proto-oncogene tyrosine kinase [ Mertk ]) 31 , 32 , dendritic cells (CD1a molecule [ Cd1a ], CD1b molecule [ Cd1b ], CD1c molecule [ Cd1c ], CD1e molecule [ Cd1e ], fms related receptor tyrosine kinase 3 [ Flt3 ]), 33 – 36 mast cells (KIT proto-oncogene receptor tyrosine kinase [ Kit ], CD34 molecule [ Cd34 ]), neutrophils (Fc gamma receptor IIIb [ Fcgr3b ]), T cells (CD3 epsilon subunit of T-cell receptor complex [ Cd3e ]), helper T cells (CD4 molecule [ Cd4 ]), cytotoxic T cells (CD8 subunit alpha [ Cd8a ]), regulatory T cells (forkhead box P3 [ Foxp3 ]), and B cells (membrane spanning 4-domains A1 [ Ms4a1 ]). It is worth mentioning that protein products of Cd80 and Cd86 , which are usually used as markers for M1 macrophages, are also expressed in mature dendritic cells (DCs). 37 , 38 These markers were detected in both groups, demonstrating active immune responses at the mesh area with multi-linage immune cells presented. Notably, the markers for macrophages (predominantly M2 macrophages), neutrophils and helper T cells were prominently expressed in both groups, consistent with the scenario of chronic inflammation. While most of these cell markers were not differentially expressed, Mrc1 and Cd4 were increasingly expressed in the diabetic vs. nondiabetic groups (p=0.023, 0.047, respectively) ( Figure 3A ). To better understand the involvement of immune cells in the differentially expressed proinflammatory mediators, we examined their correlations. Interestingly, markers for M2 macrophages ( Mrc1 , Mertk ) showed robust positive correlations with Ccl13 (all δ = 5) in the diabetic group. This contrasts with the non-diabetic group, in which positive correlations were found between Ccl13 and markers of multiple types of cells except for Mrc1 and Mertk ( Figure 3B ). The Setdb2 expression, which was positively and strongly correlated with Ccl13 in the diabetic group, also explicitly exhibited strong positive correlations with M2 markers ( Mrc1 , Mertk , and Cd163 ) (all δ = 5 – 6), demonstrating the involvement of M2 cells in the upregulation of Ccl13 in the diabetic group. This was in contrast to the non-diabetic group, which only showed strong negative correlation between Setdb2 and Cd1b (δ = 5). In addition, the Il1a expression was strongly correlated to multiple cell type markers similarly in both groups (all δ = 5 – 6), including DCs ( Cd80 , Cd1a , Cd1b , Cd1e , Flt3 ), neutrophils ( Fcgr3b ), and regulatory T cells ( Foxp3 ). Notably, the Fosl1 expression, which showed a strong positive correlation with Il1a , was only positively correlated with DC markers ( Cd80 , Cd1b , Cd1e and Flt3 , δ = 4 – 5) in the diabetic but not non-diabetic group ( Figure 3B ), demonstrating the involvement of DCs in the downregulation of Il1a under diabetic conditions. We assessed the expression of SASP markers, including cyclin dependent kinase inhibitor 2A ( Cdkn2a ), cyclin dependent kinase inhibitor 2D ( Cdkn2d ), cyclin dependent kinase inhibitor 1A ( Cdkn1a ), tumor protein p53 ( Tp53 ), and NLR family pyrin domain containing 3 ( Nlrp3 ). No significant differences were found between groups (all p > 0.05). This result was consistent with the expression profile of secretory SASP markers, in which the expression of C-X-C motif chemokine ligand −1, −2, and −8 ( Cxcl1 , Cxcl2, Cxcl8 ), interleukin 1 beta and 6 ( Il1b , Il6 ), and vascular endothelial growth factor A ( Vegfa ), were not statistically different ( Figure 1 ). While IF did not reveal significant difference in the A/M between groups (p=0.176), the results from multiplex protein assays showed that CALR (Calreticulin, a prominent pro-phagocytic signal on the surface of apoptotic cells 39 – 41 ), and AXL (one of the Tyro3–Axl–Mer tyrosine kinase receptors involved in macrophage phagocytosis 42 , 43 ) were increased by 99% and 74% in the diabetic vs. nondiabetic group (p=0.031, 0.030, respectively) ( Figure 4 ). Other proteins in the panel (CD36, CD31, GAS6, LOX-1, MBL, MER, PAI-1, uPAR, RAGE, TYRO3) were not significantly different between the two groups (all p>0.05). The BG level (mg/dL) before mesh removal showed positive correlations with the expression of Setdb2 (tau-b=0.398, p=0.003, δ=4) moderately and AXL (tau-b=0.332, p=0.024, δ=3) mildly. No correlations between glucose/HbA1c measurements and other immune variables were observed (all p>0.05).

We demonstrated that diabetes was associated with dysregulated inflammatory response, extrapolating toward the mechanism underlying the increased risk of complications in women with diabetes who receive mesh implantation. The altered inflammatory response to mesh under diabetic conditions was characterized by divergently expressed proinflammatory mediators with concomitant upregulation of Ccl13 and Ccl24 and downregulation of Ila , increased expression of M2 macrophage marker ( Mrc1 ), and impaired apoptotic cell clearance. Importantly, innate immune cells, i.e., macrophages and DCs, which are the first line of defense in response to an assault and key to foreign body reactions, were closely related to the aberrantly expressed cytokine/chemokines and their upstream regulators, supporting our hypothesis. In addition, we found that suboptimal BG control following mesh implantation might be involved in the diabetes-associated immune aberrance, with BG before mesh removal positively correlated with the expression of Setdb2 and AXL level. Consistent with previous studies, 17 , 29 we found ongoing chronic inflammation in the mesh-tissue samples. While we were not able to compare the intensity of inflammation between nondiabetic and diabetic samples due to the presence of complications, our results support that proinflammatory mediators were dysregulated under diabetic conditions. As Ccl13 and Ccl24 are both chemoattractants recruiting various immune cells to the inflamed tissue, 44 , 45 their upregulation in the diabetic group indicates enhanced proinflammatory activities. Meanwhile, as Il1 cytokines are pivotal in host immunity against pathogens such as microbials and viruses, 46 , 47 the paradoxically downregulated expression of Il1a in the diabetic group indicates an impaired immunity against infection. Both aberrancies can interfere with normal mesh integration into tissue, leading to increased risk of mesh complications in women with diabetes. Macrophages play a central regulatory role in the process of wound healing and foreign body reactions. Their plasticity in phenotype switching, broadly dichotomized into pro-inflammatory M1 phenotype and pro-healing M2 phenotype, orchestrates the progression of host response. Diabetic wound studies showed that macrophage dysfunction (e.g. sustained inflammasome activity, impaired macrophage phenotypic transition and phagocytosis, aberrant response to implants, etc.) is associated with prolonged or non-healing of wounds in patients with diabetes. 48 – 51 Parallel to our previous data generated in a diabetic rat model 18 , the findings in this study showed that macrophages, particularly M2 phenotype, were profoundly involved in the diabetes-associated immune dysregulations, as indicated by the strong and positive correlations between Ccl13 and M2 markers ( Mrc1 and Mertk ), which were absent in the non-diabetic group. In addition, as macrophages are potent effector cells in efferocytosis to clear up apoptotic cells in inflamed tissue, 52 , 53 our finding of impaired apoptotic cell clearance (increased CALR and AXL) in the diabetic samples further supports that macrophage function was likely impaired under diabetic conditions. Glycemic normalization is a standard of care in peri-surgical management of patients with diabetes, with HbA1c below 8 or lower often preferred. 54 – 56 However, its long-term benefits on the outcome of a permanently implanted medical device such as urogynecologic mesh remains to be elucidated. 54 Our longitudinal observation of BG and HbA1c demonstrated that glucose management in patients with diabetes following mesh implantation was suboptimal in ~59% cases, which may contribute to the observed abnormal inflammatory response in the diabetic group. Thus, continued good control of blood glucose following mesh implantation might be beneficial in decreasing the risk of mesh complications in women with diabetes. Yet, hyperglycemic memory, 57 – 60 a phenomenon in which the adverse effect of hyperglycemia on cells and tissues persists for a long term after re-institution of normal glycemia, may also be a potential mechanism contributing to the diabetes-associated risks, requiring further studies. The upstream regulation responsible for the differentially expressed immune mediators under diabetic conditions remained unclear. We did not find significant differences in epigenetic modifications, SASP, and transcriptional factors at transcriptomic levels, except for a mild upregulation of Setdb2 expression and downregulation of Fosl1 expression in the diabetic vs. nondiabetic group. While there were robust correlations, i.e., between Setdb2 and Ccl13 , Fosl1 and Il1a , the implication of these changes requires further studies. As pro-healing M2 macrophage markers had strong correlations with both upstream ( Setdb2 ) and downstream ( Ccl13 ) changes, and DC cell markers with Fosl1 and Il1a , it is highly likely that these two pivotal innate immune cells were primarily impacted by diabetes, contributing to the diabetes-associated immune dysregulations. Future studies may explore targeting these innate immune cell types to improve the mesh outcomes in women with diabetes. This study provides direct evidence of diabetes-associated immune dysregulation in the host response to urogynecologic meshes using excised mesh-vaginal tissue samples. In addition, it represents the pioneering effort to simultaneously investigate multiple accountable mechanisms, including epigenetics, cell senescence, and proinflammatory gene regulatory pathways, through advanced multiplex techniques. The relatively small sample size might have impacted our ability to identify more statistically significant results. The limited availability of HbA1c measurements might also have restricted our interpretation of its associations with the inflammatory endpoints. To corroborate our findings and, importantly, to determine the significance of long-term good glucose management following mesh implantation, a larger cohort study is warranted. In addition, since tissue biopsy from women without complications is unethical and difficult to obtain, we were not able to compare our findings to non-complicated controls. Future investigations will be conducted by using accrued cases of mesh removal for indications other than mesh-related complications (such as urinary retention) and by translating data from animal studies to further address this issue. Another limitation is that 97% of the subjects were Caucasians, which may affect the generalizability of our findings. Further effort will be placed on increasing racial and ethnic diversity of our biorepository and correlating our results with findings from other studies. Diabetes was associated with altered long-term inflammatory response in complicated mesh implantation, particularly involving innate immune cell dysfunction. Suboptimal glycemic control following mesh implantation may contribute to this immune dysregulation, underscoring the need for further mechanistic studies.

All women who underwent vaginal mesh removal for the primary indications of exposure or pain at Magee-Womens Hospital between February 2012 and August 2020 were offered study participation, approved by the Internal Review Board at the University of Pittsburgh. Exclusion criteria included 1) acute infection as determined by their surgeons; 2) history of chronic pelvic pain, endometriosis, irritable bowel, interstitial cystitis, fibromyalgia, or vulvodynia; 3) previous mesh revisit surgeries. Patients were consented for tissue sample collection at the time of mesh excision. Deidentified patient information including age, race, BMI, gravidity, parity, menopausal status, smoking status, menopausal hormone therapy (MHT), comorbidities as well as variables relevant to mesh implantation and removal were collected at the time of enrollment using standard questionnaires and forms. Complete vaginal mesh excisions were performed. 27 Mesh-tissue specimens obtained after excision were immediately transported to the laboratory in sterile saline on ice and stored at −80°C in a biorepository. Diabetic cases (n=25) were identified by medical records and the use of a blood glucose-lowering agent at the time of preoperative visit. For tissue analysis, non-diabetic tissue samples were age- (± 5 year) and BMI- (± 5%) matched to those with diabetes (n=24 in each group) using a customized MATLAB programming. Random blood glucose (BG) and hemoglobin A1c (HbA1c) measurements at the time periods prior to initial mesh implantation, before and after mesh removal, were retrieved. Poorly managed diabetes was defined as median BG level >=200 mg/dL. Based on HbA1c values, perioperative glucose status was classified as tight (8%). 28 In addition, longitudinal glucose management from prior to mesh implantation to after mesh removal was categorized as tight (HbA1c 8% at any one period). Total RNA was isolated from tissue using Trizol reagent (Invitrogen, Carlsbad, CA). 300ng RNA was analyzed via NanoString multiplex assays with customized Codesets for cytokines/chemokines, immune cell markers, cell senescence markers, transcriptional regulatory factors, and epigenetic enzymes per manufacturer’s instructions. Assay quality control and data normalization are performed using nSolver software (NanoString). ABL1, HMBS, and RPL13a were included as reference genes. Fold changes were represented by the ratio of geometric means of experimental (E)/control (C) for increase or minus C/E for decrease. Tissue cryosections at 7μm in thickness were processed and incubated with mouse anti-human CD68 antibody (Clone 298807, R&D Systems, Inc, Minneapolis, MN) followed by an Alex Fluor 596 conjugated Donkey anti-mouse antibody (Jackson ImmunoResearch laboratories, Inc, West Grove, PA). In situ TUNEL assay (Roche, Branchburg, NJ) was then performed as instructed by the manufacturer, with cells undergoing apoptosis labeled with fluorescein (FITC). The cell nuclei were labeled with 4′,6-diamidino-2-phenylindole (DAPI). 2% BSA omitting primary antibody was used as a negative control for the IF and an enzyme solution was used as a negative control for the TUNEL assay. Positive control for cell apoptosis was set up by using the DNase-induced apoptosis of vaginal fibroblast in culture. Full size images were taken under a Nikon confocal microscope at 20x magnification. The cells undergoing apoptosis were distinguished by nuclei double labeled with FITC and DAPI. NIS-Elements AR3.2 (Nikon) software was used to quantify the numbers of CD68+ cells and apoptotic cells. Cell counting was performed at mesh-tissue interfaces, specified as areas within 350μm from the mesh fiber surface. Ratio of CD68+/apoptotic cells (A/M) was calculated to indicate macrophage function in efferocytosis. Protein was extracted from ground frozen tissue and concentration was determined. 29 Luminex assays for 12-plex human apoptotic cell clearance panel (ThermoFisher Scientific, Waltham, MA) were performed following the instructions from the manufacturer. Results were presented as pg/μg total protein. All samples were run in duplicates. A power analysis, performed by leveraging leukocyte infiltration data in a prior diabetic wound healing study using human tissue biopsies, indicates that 9 samples per group would yield statistically significant results with an effect size of 1.72 and 90% power (p<0.05). 30 For categorical data, Chi-square tests were performed for the comparisons between groups. For non-categorical data, Shapiro-Wilk tests were used to determine data distribution patterns. Student-t or Mann-Whitney tests were used for group comparisons accordingly. Quantitative measurements were expressed as mean ± standard deviation or median (minimum, maximum). For differential gene expression analysis, p<0.05 with false discovery rate <0.1 was accepted as significant. Correlations between variables were tested with Kendall’s coefficient test (tau-b) with p<0.05 suggesting significance. Correlation strength (δ) was determined by both tau-b and p values ( Table 1 ) with δ = 5 – 6 considered as a strong correlation. IBM SPSS 25 (IBM, Armonk, NY) was used for the statistics.

The prevalence of type 2 diabetes mellitus is increasingly high in aging women, with ~87% of cases diagnosed in women aged 45 years or older. 1 In this aging group, pelvic floor disorders such as stress urinary incontinence (SUI) and/or pelvic organ prolapse (POP) are prevalent conditions with over 25% women affected. 2 – 4 Polypropylene urogynecologic meshes have been used as implantable devices to augment the surgical repairs of these disorders. It is estimated that among the 350,000 women who receive urogynecologic meshes annually in the US, ~14% have diabetes. 1 , 5 In this cohort, unfortunately, the risk of mesh-related complications, most commonly mesh exposure through vaginal epithelium, has been shown to increase about 2- to 11-fold. 6 – 11 In addition, mesh-associated pelvic pain has been reported to increase in women with diabetes and may not resolve with mesh removal. 12 , 13 Yet, the mechanism underlying the diabetes-associated risk remains unclear. Since the number of women undergoing SUI and POP surgeries is predicted to increase 50% by 2050, 14 insight into this issue is critical for developing preventive strategies to improve the outcomes of mesh and other permanent implantable medical devices in women with diabetes. The implantation of urogynecologic mesh typically triggers a default foreign body reaction, which constitutes progressive multilineage cell-mediated inflammatory and fibrotic responses. 15 , 16 We previously showed that macrophage-dominated innate immune response to mesh is key to the mesh outcomes. 17 , 18 However, diabetes has been shown to compromise the function of macrophages, e.g., through alternating epigenetic modulations 19 and inducing pre-matured senescence-associated secretory phenotype (SASP), 20 – 22 which leads to sustained inflammation in wounds and other diabetes-associated chronic diseases. 23 – 26 Thus, we hypothesized that the increased risk of mesh complications in women with diabetes is closely related to immune cell dysfunction in the foreign body response to mesh, with macrophages primarily affected. In this study, we aimed to define the diabetes-associated aberrances in the host inflammatory response to mesh, with an in-depth mechanism-oriented investigation at cellular and molecular levels. We took advantage of a large biorepository containing mesh-vaginal tissue samples excised from women with mesh complications. As secondary outcomes, we also investigated the status of blood glucose management at peri-mesh implantation/removal and correlated the results to the inflammatory response to better understand their associations.