Supplemental Digital Content is available in the text
Keywords: diabetic foot, ICAM1, polymorphisms
Abstract
The objective of the present study was to explore the association between intercellular adhesion molecule 1 (ICAM1) polymorphisms (rs5498 and rs3093030) and diabetic foot (DF) susceptibility in a Chinese Han population.
128 type 2 diabetes mellitus (T2DM) patients with DF, 147 T2DM patients without DF, and 155 healthy individuals were enrolled in this study. ICAM1 polymorphisms were genotyped by polymerase chain reaction-restriction fragment length polymorphism (PCR-RFLP). The genotypes and alleles of the polymorphisms were compared by χ2 test between the 2 groups. Association between ICAM1 polymorphisms and DF susceptibility was expressed through odds ratio (OR) with corresponding 95% confidence interval (95%CI). Effects of ICAM1 polymorphisms on DF clinical characteristics were analyzed by t test.
GG genotype of rs5498 polymorphism was distinctly correlated with decreased T2DM risk (OR = 0.369, 95%CI = 0.152–0.895) and reduced susceptibility to DF among healthy controls (OR = 0.316, 95%CI = 0.119–0.837). Similar results were discovered between rs5498 G allele and decreased risk of T2DM (OR = 0.676, 95%CI = 0.475–0.963) and DF (OR = 0.656, 95%CI = 0.453–0.950) among healthy controls. Individuals carrying rs3093030 T allele had low susceptibility to DF developed from T2DM (OR = 0.634, 95%CI = 0.412–0.974). DF patients carrying rs5498 AA genotype had significantly higher serum creatinine levels than GG genotype carriers (P = .003).
ICAM1 rs3093030 polymorphism may act as a protective factor against DF developed from T2DM, moreover, rs5498 may be involved in onset of T2DM.
Clinical trial number: ChiCTR-INR-18010231.
1. Introduction
Diabetes mellitus (DM) is a metabolic disorder and could be divided into type 1 DM (T1DM) and type 2 DM (T2DM) in clinic. Long-term DM usually leads to different complications.[1] Diabetic foot (DF) is a common complication which is caused by the combination of neuropathy and different levels of vascular lesions.[2] It shows high prevalence and also represents the main cause of disability.[3] Its occurrence significantly reduces the patients’ life quality and brings heavy economic burden to the patients and family members.[4] DF is usually caused by combined effects of infection and ulceration in lower limbs or vascular lesions in deep tissues.[5–7] Peripheral neuropathy in limb abates pain feeling, and leads to the loss of pain protection mechanism.[8] Therefore, even if foot lesions occur, patients could not find it in time. While infection will trigger nonspecific immune response, thus slowing down the ulceration.[9] Previous study indicated that various factors participate in the development of DF.[10] Genetic factors determine individual responses to DF risk factors.
Intercellular adhesion molecule 1 (ICAM1) is a cell surface glycoprotein and expressed in immune and endothelial cells.[11] ICAM1 could mediate interactions between cell–cell and cell–cellular matrix.[12] ICAM1 concentration is elevated in cardiovascular and inflammatory patients.[13] In peripheral nervous system, ICAM1 expression is induced by nerve injury.[14] ICAM1 is an important regulator for cardiovascular disorders and peripheral neuropathy in diabetic patients.[15]ICAM1 gene is located at 19p13.2 and its single nucleotide polymorphisms (SNPs) in exon regions may influence the expression or function of the protein. Minor allele frequencies of ICAM1 gene rs5498 (exon 6) and rs3093030 (non-coding exon c.-286C>T) SNPs were more than 0.1% in CHB (Chinese Han in Beijing) population. These 2 SNPs have been explored in various diseases.[16,17] Then we suggested that ICAM1 SNPs might be associated with DF onset.
However, no previous study is focused on such association. Therefore, we selected rs5498 (exon 6) and rs3093030 (non-coding exon c.-286C>T) polymorphisms to explore their association with DF susceptibility in a Chinese Han population (Fig. 1).
Figure 1.
The flow diagram of this case–control study.
2. Materials and methods
2.1. Study population
This study adopted a case–control design, and was ratified by the Ethic Committee of Xuanwu Hospital of Capital Medical University (approval number: SN201704302). Study subjects were informed the study process and signed written informed consents. Study process conformed to the declaration of Helsinki. The controls were matched with cases in age and gender.
564 T2DM patients were diagnosed in Xuanwu Hospital of Capital Medical University from April, 2017 to October, 2018 based on previous guideline.[18] 128 T2DM patients with DF were selected as case group (DF group), while 147 T2DM patients without any complications were recruited as T2DM group. Besides, 155 healthy persons with normal glucose tolerance and receiving regular healthy checkup were randomly enrolled as control group. All participants suffered no severe systemic diseases, vascular diseases, neurogenic diseases, or tumors. Basic characteristics, including age, gender, body mass index (BMI), smoking, drinking, systolic blood pressure (SBP), diastolic blood pressure (DBP), fasting plasma glucose (FBG), serum creatinine (SCr), total serum cholesterol (TC), triglycerides (TG), high density lipoprotein (HDL), and low density lipoprotein (LDL) were collected by fully-fledged nurses through questionnaire.
2.2. Sample collection and genotyping method
Blood samples were collected from cubital vein for every subject in early morning and put into blood-collecting tubes with EDTA. Genomic DNA was extracted from whole blood using TIANamp Blood DNA Kit (TIANGEN Biotech, Beijing), and then stored at −20 °C.
Primer sequences for ICAM1 gene rs5498 and rs3093030 polymorphisms were designed by Primer Premier 5.0 and Primer-BLAST based on information from GenBank database. Forward primer for rs5498 SNP was 5′-TGTGCCAGGCTTGGGGGA-3′, while reverse primer was 5′-ACATTCACGGTCACCTTGCGG-3′. Primer sequences for rs3093030 were 5′-GGACCATAGGCTCACAACACCACA-3′ (forward) and 5′-GATTCCGCTCCCCAGGTCAAG-3′ (reverse). PCR application utilized a 25 μL system, including 2 μL template, 2.5 μL 10× buffer, 2.5 μL dNTPs, 0.8 μL each of forward and reverse primers, 0.5 μL DNA polymerase, and 15.9 μL ddH2O. PCR reaction process was as follows: initial degeneration at 94 °C for 10 min, 35 cycles of degeneration at 94 °C for 55 s, annealing at 60 °C for 50 s, extension at 72 °C for 60 s, and final extension at 72 °C for 10 min. PCR products were digested by restriction enzymes, BstUI for rs5498 (A: 169 bp, G: 151 + 18 bp) and MscI for rs3093030 (T: 373 bp, C: 173 + 200 bp), at 37 °C for 4 hours. Then the products were separated through 3% agarose gel electrophoresis.
2.3. Statistical analysis
Hardy–Weinberg equilibrium (HWE) examination detected the representativeness of participants in healthy control group. Quantitative variables were evaluated by t test. Genotype and allele distributions were compared between the case and control groups by χ2 test. Association strength between ICAM1 polymorphisms and DF risk was expressed by odds ratio (OR) with 95% confidence interval (95%CI). SPSS 18.0 was used to perform data syntheses. Two-sided P < .05 represented statistical significant level.
3. Results
3.1. Basic characteristics of study subjects
Mean age for healthy controls, T2DM patients, and DF patients were 58.13 ± 11.59, 57.35 ± 11.21, and 58.70 ± 11.04 years old, respectively. There were 71 males and 57 females in DF group, 85 males and 62 females in T2DM group, and 85 males and 70 females in control group. While, no significant difference was discovered among 3 groups (Table 1, P > .05). Besides, smoking and drinking status were not significantly associated with DF or T2DM development (P > .05). SBP, FBG, TC, SCr, and TG levels were significantly different among DF, T2DM patients, and healthy controls (Table 1, P < .05). However, DBP, HDL, and LDL levels had no significant difference between 3 groups (P > .05).
Table 1.
Characteristics of the subjects.
3.2. Association of ICAM1 polymorphisms with DF susceptibility
Genotype distributions of rs5498 and rs3093030 SNPs in healthy controls did not deviate from HWE (Table 2, P > .05). It suggested that the study subjects could represent general population.
Table 2.
Genotype and allele distributions of ICAM1 polymorphisms.
AA, AG, and GG genotype frequencies of ICAM1 rs5498 polymorphism were 55.47%, 39.84%, 4.69% in DF patients, 55.10%, 39.46%, 5.44% in T2DM patients, and 45.81%, 41.93%, 12.26% in healthy controls. Frequencies of rs5498 A and G alleles were 75.39% and 24.61% in DF patients, 74.83% and 25.17% in T2DM patients, and 66.77% and 33.23% in healthy controls (Table 2). Allele distributions among 3 groups had statistically significant difference (P = .033).
Compared with AA genotype, rs5498 GG genotype significantly decreased susceptibility to T2DM (P = .023, OR = 0.369, 95%CI = 0.152–0.895) and DF (P = .016, OR = 0.316, 95%CI = 0.119–0.837) among healthy controls (Table 3). Meanwhile, individuals with rs5498 G allele faced reduced risk of T2DM (P = .030, OR = 0.676, 95%CI = 0.475–0.963) and DF (P = .025, OR = 0.656, 95%CI = 0.453–0.950) among controls, compared with A allele. However, genotypes and alleles of rs5498 had no significant association with susceptibility to DF developed from T2DM (P > .05).
Table 3.
Association of ICAM1 polymorphisms with DF susceptibility.
Rs3093030 CC, CT, and TT genotype frequencies were 69.53%, 28.91%, 1.56% in DF group, 59.18%, 35.37%, 5.44% in T2DM patients, and 57.42%, 36.77%, 5.81% in healthy controls. Genotype distributions were significantly different between 3 groups (Table 2, P = .013). However, there were no significant associations between rs3093030 genotypes and susceptibility to T2DM or DF (Table 3, P > .05). Meanwhile, C and T allele frequencies were 83.98% and 16.02% in DF group, 76.87% and 23.13% in T2DM group, and 75.81% and 24.19% in healthy controls. Significant differences were also discovered among these 3 groups (Table 2, P = .041). T allele of rs3093030 SNP was significantly correlated with decreased DF risk based on healthy controls (P = .016, OR = 0.598, 95%CI = 0.391–0.912) and T2DM cases (P = .037, OR = 0.634, 95%CI = 0.412–0.974) (Table 3).
3.3. Effects of ICAM1 SNP genotypes on DF characteristics
In this study, we also detected the effects of ICAM1 polymorphisms on the characteristics of DF patients. SCr level was significantly higher in rs5498 GG genotype carriers than in AA genotype carriers (Table 4, P = .003). However, no significant difference was discovered for rs5498 genotypes with SBP, DBP, disease duration, FBG, TC, TG, HDL, or LDL levels (Table 4, P > .05). Meanwhile, these basic characteristics were not significant difference between rs3093030 genotype carriers (Supplement Table 1).
Table 4.
Effects of rs5498 genotypes for DF characteristics.
4. Discussion
As a biomarker of cell–cell interaction, ICAM1 plays an important role in diabetic nephropathy.[19,20] ICAM1 level is elevated in inflammatory and malignant disorders.[21] Plasma ICAM1 levels were higher in diabetic nephropathy (DN) patients and T2DM patients than in healthy controls. Meanwhile, plasma ICAM1 level was significantly higher in healthy individuals with rs5498 AA genotype than persons with AG or GG genotypes.[22] Serum level of ICAM1 was influenced by rs3093030 SNP.[23] Therefore, we speculated that ICAM1 polymorphisms might contribute to individuals susceptibility to DF onset. However, no previous study focused on the association of ICAM1 polymorphisms with DF risk.
In the current study, we found that in comparison with healthy controls, rs5498 GG genotype and G allele were significantly correlated with reduced susceptibility to T2DM and DF, but the presence of GG genotype and G allele had no obvious association with DF onset among T2DM cases. That meant, rs5498 might be only involved in T2DM occurrence. Rs5498 polymorphism is located at exon region of ICAM1 gene, and it is involved in transcription and translation of the gene. Thus, rs5498 SNP could influence individual susceptibility to T2DM. The SNP has been explored in several diabetic complications. A meta-analysis demonstrated that rs5498 A allele was positively correlated with susceptibility to diabetic microvascular complications (DMI).[16] Lv et al found that rs5498 SNP had no significant association with diabetic retinopathy (DR) risk in Chinese population.[24] Popović and colleagues suggested that rs5498 SNP was not significantly correlated with T2DM susceptibility, but its AA genotype was positively associated with annual plaques growth in T2DM patients.[25] However, a meta-analysis performed by Fan and Liu indicated that rs5498 G allele carriers had lower susceptibility to DR in Asians than those with A allele.[26] This SNP was also detected in other diseases. GG genotype of rs5498 was negatively correlated with susceptibility to myocardial infarction.[27] AA genotype of rs5498 SNP was more frequent in patients with acute pyelonephritis (APN).[28] However, another study found that rs5498 GG genotype might induce multiple sclerosis.[29] While rs5498 had no significant association with endometriosis susceptibility.[30]
For rs3093030, the presence of T allele indicated low risk of DF based on T2DM and healthy individuals. Rs3093030 polymorphism was located at the non-coding exon region of ICAM1 gene, and its variants might have the ability to regulate the expression of genetic information, thus influencing the function of ICAM1. The article published by Zhang et al found that rs3093030 CC genotype carriers had high susceptibility to cancers.[31] Their conclusion supported our results. In order to explore the effects of ICAM1 polymorphisms on DF susceptibility, we checked their association with basic characteristics of DF patients. Then we found that rs5498 GG genotype carriers had significantly higher SCr level than AA genotype carriers. While, rs3093030 SNP genotypes had no significant effects on other clinical characteristics.
Some limitations in the present study should be noted. First of all, the sample size was relatively small that reduced the accuracy of final results. Secondly, 1 nationality may not represent the whole Chinese Han population. Thirdly, a variety of factors are involved in the development of DF, but potential interactions between these factors were not explored in the current study. Finally, the exact mechanism of ICAM1 polymorphisms on DF development was not illustrated in current study. Therefore, further studies with enlarged sample size and ethnicity number should be conducted to verify our results and investigate the mechanism of ICAM1 polymorphisms functioning in DF etiology.
In conclusion, ICAM1 rs3093030 polymorphism may act as a protective factor against DF occurrence developed from T2DM. Furthermore, rs5498 polymorphism may be significantly associated with the decreased risk of T2DM development.
Author contributions
Conceptualization: Xue-Xia Cao.
Data curation: Xue-Xia Cao, Jin-Kui Yang.
Formal analysis: Xue-Xia Cao.
Funding acquisition: Xue-Xia Cao, Jin-Kui Yang, Li Wang.
Investigation: Xue-Xia Cao, Jin-Kui Yang.
Methodology: Xue-Xia Cao, Jin-Kui Yang.
Project administration: Xue-Xia Cao.
Resources: Xue-Xia Cao, Jin-Kui Yang.
Software: Xue-Xia Cao, Jin-Kui Yang.
Supervision: Xue-Xia Cao, Li Wang.
Validation: Xue-Xia Cao, Jin-Kui Yang.
Visualization: Xue-Xia Cao, Jin-Kui Yang, Li Wang.
Writing – original draft: Xue-Xia Cao, Jin-Kui Yang.
Writing – review & editing: Xue-Xia Cao, Li Wang.
Supplementary Material
Footnotes
Abbreviations: 95%CI = 95% confidence interval, APN = acute pyelonephritis, BMI = body mass index, CCC = coronary collateral circulation, DBP = diastolic blood pressure, DF = diabetic foot, DM = diabetes mellitus, DMI = diabetic microvascular complications, DN = diabetic nephropathy, DR = diabetic retinopathy, FBG = fasting plasma glucose, HDL = high density lipoprotein, HWE = Hardy–Weinberg equilibrium, ICAM1 = intercellular adhesion molecule 1, LDL = low density lipoprotein, OR = Odds ratio, PCR-RFLP = polymerase chain reaction-restriction fragment length polymorphism, SBP = systolic blood pressure, SCr = serum creatinine, SNPs = single nucleotide polymorphisms, T1DM = type 1 DM, T2DM = type 2 DM, TC = total serum cholesterol, TG = triglycerides.
How to cite this article: Cao XX, Yang JK, Wang L. Association between intercellular adhesion molecule 1 (ICAM1) polymorphisms and diabetic foot susceptibility: A case–control study. Medicine. 2020;99:11(e18052).
Funding support: None.
The authors have no conflicts of interest to disclose.
Supplemental Digital Content is available for this article.
References
- [1].Yamazaki D, Hitomi H, Nishiyama A. Hypertension with diabetes mellitus complications. Hypertens Res: Official Journal of the Japanese Society of Hypertension 2018. [DOI] [PubMed] [Google Scholar]
- [2].Noor S, Ahmad J, Parwez I, et al. Culture-based screening of aerobic microbiome in diabetic foot subjects and developing non-healing ulcers. Front Microbiol 2016;7:1792. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [3].Khan A, Junaid N. Prevalence of diabetic foot syndrome amongst population with type 2 diabetes in Pakistan in primary care settings. J Pak Med Assoc 2017;67:1818–24. [PubMed] [Google Scholar]
- [4].Raghav A, Khan ZA, Labala RK, et al. Financial burden of diabetic foot ulcers to world: a progressive topic to discuss always. Ther Adv Endocrinol Metab 2018;9:29–31. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [5].Chin YF, Yeh JT, Yu HY, et al. Knowledge of the warning signs of foot ulcer deterioration among patients with diabetes. J Nurs Res 2017. [DOI] [PubMed] [Google Scholar]
- [6].Suryaletha K, John J, Radhakrishnan MP, et al. Metataxonomic approach to decipher the polymicrobial burden in diabetic foot ulcer and its biofilm mode of infection. Int Wound J 2018. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [7].Babamahmoodi F, Shokohi T, Ahangarkani F, et al. Rare case of Aspergillus ochraceus Osteomyelitis of calcaneus bone in a patient with diabetic foot ulcers. Case Rep Med 2015;2015:509827. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [8].Toscano CM, Sugita TH, Rosa MQM, et al. Annual direct medical costs of diabetic foot disease in Brazil: a cost of illness study. Int J Environ Res Public Health 2018;15. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [9].Nussbaum SR, Carter MJ, Fife CE, et al. An economic evaluation of the impact, cost, and medicare policy implications of chronic nonhealing wounds. Value Health: The Journal of the International Society for Pharmacoeconomics and Outcomes Research 2018;21:27–32. [DOI] [PubMed] [Google Scholar]
- [10].Lefrancois T, Mehta K, Sullivan V, et al. Evidence based review of literature on detriments to healing of diabetic foot ulcers. Foot Ankle Surg: Official Journal of the European Society of Foot and Ankle Surgeons 2017;23:215–24. [DOI] [PubMed] [Google Scholar]
- [11].Liu CW, Lee TL, Chen YC, et al. PM2.5-induced oxidative stress increases intercellular adhesion molecule-1 expression in lung epithelial cells through the IL-6/AKT/STAT3/NF-kappaB-dependent pathway. Part Fibre Toxicol 2018;15:4. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [12].Zhang Q, Zhang J, Yan Y, et al. Proinflammatory cytokines correlate with early exercise attenuating anxiety-like behavior after cerebral ischemia. Brain Behav 2017;7:e00854. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [13].Nunez D, Comas L, Lanuza PM, et al. A functional analysis on the interspecies interaction between mouse LFA-1 and human intercellular adhesion molecule-1 at the cell level. Front Immunol 2017;8:1817. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [14].Yang J, Gu Y, Huang X, et al. Dynamic changes of ICAM-1 expression in peripheral nervous system following sciatic nerve injury. Neurol Res 2011;33:75–83. [DOI] [PubMed] [Google Scholar]
- [15].Roustit M, Loader J, Deusenbery C, et al. Endothelial dysfunction as a link between cardiovascular risk factors and peripheral neuropathy in diabetes. J Clin Endocrinol Metab 2016;101:3401–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [16].Su X, Chen X, Liu L, et al. Intracellular adhesion molecule-1 K469E gene polymorphism and risk of diabetic microvascular complications: a meta-analysis. PloS One 2013;8:e69940. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [17].Chou CH, Ueng KC, Liu YF, et al. Impact of intercellular adhesion molecule-1 genetic polymorphisms on coronary artery disease susceptibility in Taiwanese subjects. Int J Med Sci 2015;12:510–6. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [18].American Diabetes Association Standards of medical care in diabetes – 2011. Diabetes Care 2011;34: Suppl 1: S11–61. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [19].Elhadd TA, Khan F, Kirk G, et al. Influence of puberty on endothelial dysfunction and oxidative stress in young patients with type 1 diabetes. Diabetes Care 1998;21:1990–6. [DOI] [PubMed] [Google Scholar]
- [20].Lee S, Kim W, Moon SO, et al. Renoprotective effect of COMP-angiopoietin-1 in db/db mice with type 2 diabetes. Nephrol Dial Transplant: Official Publication of the European Dialysis and Transplant Association – European Renal Association 2007;22:396–408. [DOI] [PubMed] [Google Scholar]
- [21].Lin CW, Chuang CY, Tang CH, et al. Combined effects of icam-1 single-nucleotide polymorphisms and environmental carcinogens on oral cancer susceptibility and clinicopathologic development. PloS One 2013;8:e72940. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [22].Abu Seman N, Anderstam B, Wan Mohamud WN, et al. Genetic, epigenetic and protein analyses of intercellular adhesion molecule 1 in Malaysian subjects with type 2 diabetes and diabetic nephropathy. J Diabetes Complicat 2015;29:1234–9. [DOI] [PubMed] [Google Scholar]
- [23].Barbalic M, Dupuis J, Dehghan A, et al. Large-scale genomic studies reveal central role of ABO in sP-selectin and sICAM-1 levels. Hum Mol Genet 2010;19:1863–72. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [24].Lv Z, Li Y, Wu Y, et al. Association of ICAM-1 and HMGA1 gene variants with retinopathy in type 2 diabetes mellitus among Chinese individuals. Curr Eye Res 2016;41:1118–22. [DOI] [PubMed] [Google Scholar]
- [25].Popović D, Starcevic JN, Letonja MS, et al. Polymorphism rs5498 of the ICAM-1 gene affects the progression of carotid atherosclerosis in patients with type 2 diabetes mellitus. Lipids Health Dis 2016;15:79. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [26].Fan WY, Liu NP. Meta-analysis of association between K469E polymorphism of the ICAM-1 gene and retinopathy in type 2 diabetes. Int J Ophthalmol 2015;8:603–7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [27].Hu P, Dai T, Yu W, et al. Intercellular adhesion molecule 1 rs5498 polymorphism is associated with the risk of myocardial infarction. Oncotarget 2017;8:52594–603. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [28].Hussein A, Saad K, Askar E, et al. Functional variants in intercellular adhesion molecule-1 and toll-like receptor-4 genes are more frequent in children with febrile urinary tract infection with renal parenchymal involvement. Acta Paediatr 2018;107:339–46. [DOI] [PubMed] [Google Scholar]
- [29].Shawkatova I, Javor J, Parnicka Z, et al. Analysis of ICAM1 gene polymorphism in Slovak multiple sclerosis patients. Folia Microbiol 2017;62:287–93. [DOI] [PubMed] [Google Scholar]
- [30].Bessa NZ, Francisco DO, Andres MP, et al. Polymorphisms of ICAM-1 and IL-6 genes related to endometriosis in a sample of Brazilian women. J Assist Reprod Gen 2016;33:1487–92. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [31].Zhang X, Huang J, Bai J, et al. Association of polymorphisms in intercellular adhesion molecule 1 (ICAM-1) gene with cancer susceptibility: a meta-analysis of 14 case–control studies. Med Sci Monit: International Medical Journal of Experimental and Clinical Research 2016;22:569–79. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.