Body mass index and risk of inflammatory breast disease: a Mendelian randomization study

Wei, Changlong; Wang, Xiaofang; Zeng, Jinsheng; Zhang, Gongyin; Wei, Changlong; Wang, Xiaofang; Zeng, Jinsheng; Zhang, Gongyin

doi:10.20960/nh.04746

My SciELO

Custom services

Services on Demand

Journal

Article

Indicators

Cited by SciELO
Access statistics

Nutrición Hospitalaria

On-line version ISSN 1699-5198Print version ISSN 0212-1611

Nutr. Hosp. vol.41 n.1 Madrid Jan./Feb. 2024 Epub Mar 07, 2024

https://dx.doi.org/10.20960/nh.04746

ORIGINAL PAPERS

Body mass index and risk of inflammatory breast disease: a Mendelian randomization study

Índice de masa corporal y riesgo de mastitis: un estudio de aleatorización mendeliana

Changlong Wei, conception, initiation of the project and completing the data collection, analysis, and writing of the manuscript, reading and approving the final manuscript¹; Xiaofang Wang, assisted Changlong Wei with the data analysis and writing of the manuscript, complementary suggestions to the paper, supervision of the project, reading and approving the final manuscript¹; Jinsheng Zeng, constructive suggestions after reviewing the paper, supervision of the project, reading and approving the final manuscript¹; Gongyin Zhang, complementary suggestions to the paper, supervision of the project, reading and approving the final manuscript¹

¹The First Affiliated Hospital of Nanchang University. Nanchang, P. R. China

Abstract

Introduction:

in previous studies, obesity was identified as a risk factor for inflammatory breast disease, but its causality is uncertain. In the present study, we performed a two-sample Mendelian randomization (TSMR) analysis to investigate the causal relationship between obesity and inflammatory breast disease.

Methods:

we use body mass index (BMI) as a measure of obesity. Data for single nucleotide polymorphisms (SNPs) associated with BMI were obtained from UK Biobank. Data for single nucleotide polymorphisms (SNPs) associated with mastitis were obtained from FinnGen Biobank. We used several MR analysis methods, such as inverse-variance weighting (IVW), MR-Egger, weighted median, simple mode and weighted mode to make our results more convincing. We also performed MR-PRESSO test, MR-Egger test, heterogeneity test, pleiotropy test and leave-one-out analysis to make our analysis results more robust and credible. We used odds ratio (OR) to evaluate the causal relationship between BMI and mastitis.

Results:

based on the IVW random effects model, we found that a one-standard deviation (SD) increase in BMI increased the risk of mastitis by 62.1 % (OR = 1.621, 95 % CI: 1.262-2.083, p = 1.59E-4), which is almost consistent with the results of several other methods.

Conclusions:

in European individuals, an increase in the number of BMI increases the risk of inflammatory breast disease. People with high BMI need to control their weight to reduce the incidence of inflammatory breast disease.

Keywords: Mendelian randomization; GWAS; Body mass index; Mastitis

Resumen

Introducción:

en estudios previos, la obesidad se identificó como un factor de riesgo para la enfermedad inflamatoria de mama, pero su causalidad es incierta. En el presente estudio, se realizó un análisis de aleatorización mendeliana de dos muestras (TSMR) para investigar la relación causal entre la obesidad y la enfermedad inflamatoria de mama.

Métodos:

se empleó el índice de masa corporal (IMC) como medida de obesidad. Los datos de los polimorfismos de nucleótido único (SNP) asociados con el IMC se obtuvieron del Biobank de Reino Unido y los datos de los polimorfismos de nucleótido único (SNP) asociados con la mastitis se obtuvieron de FinnGen Biobank. Se utilizaron varios métodos de análisis de RM, como la ponderación inversa de la varianza (IVW), RM-Egger, mediana ponderada, modo simple y modo ponderado para que nuestros resultados fueran más convincentes. También se realizaron la prueba MR-PRESSO, la prueba MR-Egger, la prueba de heterogeneidad, el test de pleiotropía y la validación dejando uno fuera (en inglés, leave-one-out) para que los resultados de nuestro análisis fueran más sólidos y creíbles. Se utilizó la odds ratio (OR) para evaluar la relación causal entre el IMC y la mastitis.

Resultados:

basándonos en el modelo de efectos aleatorios IVW, se halló que un aumento de una desviación estándar (DE) en el IMC aumentaba el riesgo de mastitis en un 62,1 % (OR = 1,621, IC 95 %: 1,262-2,083, p = 1,59E-4), que es casi consistente con los resultados de otros diversos métodos.

Conclusiones:

en los individuos europeos, un aumento del número de IMC aumenta el riesgo de enfermedad inflamatoria mamaria. Las personas con un IMC elevado deben controlar su peso para reducir la incidencia de enfermedad inflamatoria de la mama.

Palabras clave: Aleatorización mendeliana; GWAS; Índice de masa corporal; Mastitis

INTRODUCTION

The global obesity epidemic continues to grow and now affects more than two billion people, or 30 % of the world’s population (¹). Obesity is ranked as the fifth leading cause of death worldwide and is associated with the development of many diseases (²). Body mass index (BMI) is commonly used to define overweight and obesity (³). It has been shown that obesity is associated with a higher risk of breast cancer, while relatively little research has been done on the relationship between obesity and inflammatory breast disease (⁴).

Inflammatory breast diseases are common benign breast diseases and are classified as lactational mastitis and non-lactational mastitis. Non-lactating mastitis is less common and includes idiopathic granulomatous mastitis, periductal mastitis, and tuberculous mastitis (⁵). Many previous observational studies have shown a strong association between obesity and mastitis (⁶,⁷). From the perspective of evidence-based medicine, the causal relationship between obesity and inflammatory breast disease is unclear. There are few clinical studies on obesity and mastitis and only a few observational studies, which are not convincing enough due to their shortcomings and many confounding factors.

Mendelian randomization (MR) studies have gotten a lot of attention since the recent addition of genome-wide association studies (GWAS) databases. Mendelian randomization (MR), conceptually similar to randomized controlled trials (RCTs), is based on the principle of random assignment of genetic variants at meiosis, which reduces the impact of confounding encountered in observational epidemiology (⁸-¹⁰). MR studies have been widely employed to determine the causal relationship between an exposure and an outcome.

In this study, we investigated the genetic causal relationship between BMI and inflammatory breast disease using MR analysis.

The overview of the research design is shown in figure 1.

Figure 1. The study assumptions of the two-sample Mendelian randomization analysis between body mass index (BMI) and mastitis. The assumptions included: a) single nucleotide polymorphisms (SNPs) selected as instrumental variables (IVs) should be closely related with exposure; b) selected SNPs must be independent of confounders; and c) IVs are associated to outcome only via exposure rather than through a direct association.

MATERIALS AND METHODS

DATA SOURCE

Data on BMI were obtained from the GWAS European Population Summary database supported by the UK Biobank (https://gwas.mrcieu.ac.uk/datasets/ukb-b-19953/). This database contains 461,460 European participants. Data on outcome variables for inflammatory breast disease were obtained from the FinnGen Biobank (https://r8.finngen.fi), which contains 1,435 cases of inflammatory breast disease and 176,011 controls of European ancestry (¹¹).

INSTRUMENT SELECTION CRITERIA

We extracted genetic instrumental variables (IVs) using the following criteria: a) p value (genome-wide significance threshold < 5e-8) (¹²); and b) in order to avoid the potential bias caused by strong linkage disequilibrium (LD) (¹³), a linkage disequilibrium correlation coefficient r² (r² < 0.001), and a number of bases between two SNPs (kb > 10,000) were set. Next, we combined the exposed and outcome data and removed the palindromic SNPs, leaving the remaining SNPs as the final instrumental variables used for analysis. We then used the F-statistic (F = beta²/se²) to assess the statistical power of the remaining SNPs and calculated the F-values of all SNPs and removed those with low statistical power (F-statistic < 10) (¹⁴). If the F value for the instrument exposure correlation was significantly greater than 10, the likelihood of weak instrumental variable bias was low (¹⁵).

STATISTICAL APPROACHES AND SENSITIVITY ANALYSES

Data analysis in this study was performed using R (version 4.2.2) through TwoSampleMR (0.5.6) package and the MR-Pleiotropy Residual Sum and Outlier method (MR-PRESSO) (1.0) (¹⁶). First, to assess the level of significance of the causal relationship between BMI and mastitis, we used a two-sample MR analysis. In order to assess whether BMI has a causal effect on mastitis, we mostly used the inverse-variance weighted (IVW) method and the MR-Egger method, along with weighted median, simple mode, and weighted mode approaches (¹⁷-¹⁹). The IVW method is a crucial method for realizing unbiased estimates since it guarantees the necessary statistical power to establish a link between an exposure factor and an outcome. MR-Egger regression is similar to IVW, except that it includes an intercept term for the average pleiotropic effect. The intercept in the MR-Egger regression can be used to test the IV hypothesis (²⁰). Evaluation of potential heterogeneity and horizontal pleiotropy requires sensitivity analysis. Cochran’s Q test was performed to assess the heterogeneity of effect sizes for selected genetic IVs. The MR-PRESSO analysis was also applied to exclude outliers and moderate horizontal pleiotropy (²¹). The intercept derived from MR-Egger regression was employed to evaluate vertical pleiotropy (²²). We also performed a “leave-one-out” sensitivity test as part of our sensitivity analysis to check for significant variations in our models both before and after eliminating each IV (¹⁷,²³). The flow chart regarding the Mendelian randomization analysis of this study is shown in figure 2.

Figure 2. Flowchart of Mendelian randomization analysis of the effect of BMI on mastitis (BMI: body mass index; SNP: single nucleotide polymorphisms; MR: Mendelian randomization; IVW: inverse-variance weighted; MR-PRESSO: MR-Pleiotropy Residual Sum and Outlier method).

RESULTS

GENETICS VARIANTS SELECTION

We strictly followed the selection criteria for the genetic instrumental variables. The details of the SNPs finally selected for MR analysis are shown in Supplementary Table I.

Table I. Details of the GWAS included in the Mendelian randomization.

SNPs: single nucleotide polymorphisms.

THE CAUSAL RELATIONSHIP BETWEEN BMI AND MASTITIS

We evaluated the causal relationship of BMI on mastitis. The results are summarized in figure 3.

Figure 3. Forest plot of MR analysis results (BMI: body mass index; MR: Mendelian randomization; IVW: inverse-variance weighted; MR-PRESSO: MR-Pleiotropy Residual Sum and Outlier method).

Causal relationships assessed by IVW random effects models indicated that each standard deviation (SD) increased genetically determined BMI increased the risk of mastitis by 62.1 % (n = 413 SNPs, OR = 1.621, 95 % CI: 1.262-2.083, p = 1.59E-4),and this causality was further verified by MR Egger (OR = 1.541, 95 % CI: 0.789-3.007; p = 0.206), weighted median (OR = 1.359, 95 % CI: 0.885-2.086; p = 0.161), simple mode (OR = 1.857,95 % CI: 0.567-6.079; p = 0.307) and weighted mode (OR = 1.142, 95 % CI: 0.552-2.362; p = 0.720) methods. Figure 4 shows a scatter plot of the relationship between BMI and mastitis. Figure 5 shows a forest plot of the causal effect of single nucleotide polymorphisms (SNPs) associated with BMI on mastitis.

Figure 4. Scatter plot of MR analysis of the effect of BMI on mastitis (BMI: body mass index; SNP: single nucleotide polymorphisms; MR: Mendelian randomization).

Figure 5. Forest plots of the causal effects of BMI-associated SNPs on mastitis (BMI: body mass index; SNP: single nucleotide polymorphisms; MR: Mendelian randomization).

The p-value obtained by Cochran’s Q test was 0.017 < 0.05, so we used the random effects model as the gold standard. According to MR-PRESSO test, no outliers were found, which was consistent with the leave-one-out analysis (Fig. 6) and funnel plot (Fig. 7). Horizontal pleiotropy was absence in MR-Egger intercept analysis (p = 0.873).

Figure 6. “Leave-one-out” sensitivity analysis results in the two-sample MR analysis (BMI: body mass index; MR: Mendelian randomization).

Figure 7. Funnel plot of the causality of BMI to mastitis (BMI: body mass index; MR: Mendelian randomization).

DISCUSSION

In the present study, we applied MR methods to determine the causal relationship between BMI and mastitis, and we found that for every one standard deviation (SD) increase in BMI, the risk of mastitis increased by 62.1 %.

We selected BMI as the exposed genetic phenotype and found a genetic causal relationship between BMI and mastitis. BMI is often used as a measure of obesity, and the relationship between obesity and mastitis can be reflected laterally through this study (²⁴). Many retrospective studies have been conducted to evaluate potential risk factors for mastitis, however, many of them suffer from recollection and selection bias (²⁵). It has been shown that obesity is a low-grade chronic inflammatory condition with changes in associated factors that can directly affect local estrogen receptor expression and the development of inflammation (²⁶,²⁷). Multiple studies have found that obese individuals are more likely to develop various types of infections than normal weight individuals (²⁸,²⁹). In a previous Mendelian randomization study, high BMI was found to be associated with an increased risk of any infection (³⁰), and infections may contribute to the development of some mastitis. Granulomatous lobular mastitis (GLM) is a benign chronic inflammatory breast disease of unknown etiology, and because GLM is a rare disease, prospective studies are difficult to perform. In a retrospective analysis, obesity or overweight was found to be associated with GLM recurrence (⁷). Periductal mastitis (PDM) is a long-term inflammatory disease, but little is known about the etiology of PDM. In a retrospective study, overweight/obesity was shown to be an independent risk factor for PDM (⁶). Non-bacterial mastitis (NBM) is a solid disease of the breast that accounts for 4-5 % of all benign breast diseases (³¹). In some studies, obesity has been suggested to be possibly related to NBM (³²).

There were several advantages for the current two-sample MR studies. First, because alleles follow the principle of random distribution when forming gametes during meiosis, the causal relationship between genotype and disease in MR studies will not be distorted by confounding factors, which is the main limitation of traditional observation studies (³³,³⁴). Second, it is much easier to assess the causal relationship between BMI and mastitis risk using the public GWAS database than using prospective cohort studies or randomized controlled trials. This study also required less time and cost than a general epidemiological study.

However, we acknowledge some limitations to our study. First, there might be an ethnic bias in our study due to all the selected GWAS database populations being of European ancestry. Second, there was some heterogeneity in our results in the heterogeneity test, which may have compromised the accuracy of our results, although the results of our other analytical methods were robust. Third, our study did not differentiate between the different types of mastitis, and the relationship between BMI values and each type of mastitis needs further study. Fourth, this study needs to be further validated by other types of studies (observational studies, cohort studies, etc.).

CONCLUSIONS

According to our study, there is a genetic causal relationship between an increased number of BMI and an increased risk of inflammatory breast disease. Patients with high BMI need to control their weight to reduce the incidence of mastitis, and clinicians may also need to be concerned about the potential risk of inflammatory breast disease in some patients with high BMI.

Acknowledgments:

we want to acknowledge the participants and investigators of the FinnGen consortium, IEU OpenGwas project and UK Biobank. Thank you to all investigators and instructions for making their genome-wide association studies (GWAS) data freely available.

REFERENCES

1. Caballero B. Humans against obesity: who will win? Adv Nutr 2019;10(suppl_1):S4-s9. DOI: 10.1093/advances/nmy055 [ Links ]

2. Safaei M, Sundararajan EA, Driss M, Boulila W, Shapi'i A. A systematic literature review on obesity: understanding the causes and consequences of obesity and reviewing various machine learning approaches used to predict obesity. Comput Biol Med 2021;136:104754. DOI: 10.1016/j.compbi-omed.2021.104754 [ Links ]

3. Chooi YC, Ding C, Magkos F. The epidemiology of obesity. Metabolism 2019;92:6-10. DOI: 10.1016/j.metabol.2018.09.005 [ Links ]

4. Picon-Ruiz M, Morata-Tarifa C, Valle-Goffin JJ, Friedman ER, Slingerland JM. Obesity and adverse breast cancer risk and outcome: mechanistic insights and strategies for intervention. CA Cancer J Clin 2017;67(5):378-97. DOI: 10.3322/caac.21405 [ Links ]

5. Scott DM. Inflammatory diseases of the breast. Best Pract Res Clin Obstet Gynaecol 2022;83:72-87. DOI: 10.1016/j.bpobgyn.2021.11.013 [ Links ]

6. Liu L, Zhou F, Wang P, Yu L, Ma Z, Li Y, et al. Periductal mastitis: an inflammatory disease related to bacterial infection and consequent immune responses? Mediators Inflamm 2017;2017:5309081. DOI: 10.1155/2017/5309081 [ Links ]

7. Huang Y, Wu H. A retrospective analysis of recurrence risk factors for granulomatous lobular mastitis in 130 patients: more attention should be paid to prolactin level. Ann Palliat Med 2021;10(3):2824-31. DOI: 10.21037/apm-20-1972 [ Links ]

8. Nitsch D, Molokhia M, Smeeth L, DeStavola BL, Whittaker JC, Leon DA. Limits to causal inference based on Mendelian randomization: a comparison with randomized controlled trials. Am J Epidemiol 2006;163(5):397-403. DOI: 10.1093/aje/kwj062 [ Links ]

9. Lawlor DA, Harbord RM, Sterne JA, Timpson N, Smith GD. Mendelian randomization: using genes as instruments for making causal inferences in epidemiology. Stat Med 2008;27(8):1133-63. DOI: 10.1002/sim.3034 [ Links ]

10. Davey Smith G, Ebrahim S. What can Mendelian randomisation tell us about modifiable behavioural and environmental exposures? BMJ 2005;330(7499):1076-9. DOI: 10.1136/bmj.330.7499.1076 [ Links ]

11. Kurki MI, Karjalainen J, Palta P, Sipila TP, Kristiansson K, Donner KM, et al. FinnGen provides genetic insights from a well-phenotyped isolated population. Nature 2023;613(7944):508-18. DOI: 10.1038/s41586-022-05473-8 [ Links ]

12. Hemani G, Bowden J, Davey Smith G. Evaluating the potential role of pleiotropy in Mendelian randomization studies. Hum Mol Genet 2018;27(R2):R195-r208. DOI: 10.1093/hmg/ddy163 [ Links ]

13. Myers TA, Chanock SJ, Machiela MJ. LDlinkR: an R package for rapidly calculating linkage disequilibrium statistics in diverse populations. Front Genet 2020;11:157. DOI: 10.3389/fgene.2020.00157 [ Links ]

14. Chen L, Yang H, Li H, He C, Yang L, Lv G. Insights into modifiable risk factors of cholelithiasis: a Mendelian randomization study. Hepatology 2022;75(4):785-96. DOI: 10.1002/hep.32183 [ Links ]

15. Davies NM, Holmes MV, Davey Smith G. Reading Mendelian randomisation studies: a guide, glossary, and checklist for clinicians. BMJ 2018;362:k601. DOI: 10.1136/bmj.k601 [ Links ]

16. Rasooly D, Patel CJ. Conducting a reproducible Mendelian randomization analysis using the R analytic statistical environment. Curr Protoc Hum Genet 2019;101(1):e82. DOI: 10.1002/cphg.82 [ Links ]

17. Hemani G, Zheng J, Elsworth B, Wade KH, Haberland V, Baird D, et al. The MR-base platform supports systematic causal inference across the human phenome. Elife 2018;7. DOI: 10.7554/eLife.34408 [ Links ]

18. Bowden J, Davey Smith G, Burgess S. Mendelian randomization with invalid instruments: effect estimation and bias detection through Egger regression. Int J Epidemiol 2015;44(2):512-25. DOI: 10.1093/ije/dyv080 [ Links ]

19. Bowden J, Davey Smith G, Haycock PC, Burgess S. Consistent estimation in mendelian randomization with some invalid instruments using a weighted Mmdian estimator. Genet Epidemiol. 2016;40(4):304-14. DOI: 10.1002/gepi.21965. [ Links ]

20. Burgess S, Thompson SG. Interpreting findings from Mendelian randomization using the MR-Egger method. Eur J Epidemiol 2017;32(5):377-89. DOI: 10.1007/s10654-017-0255-x [ Links ]

21. Verbanck M, Chen CY, Neale B, Do R. Detection of widespread horizontal pleiotropy in causal relationships inferred from Mendelian randomization between complex traits and diseases. Nat Genet 2018;50(5):693-8. DOI: 10.1038/s41588-018-0099-7 [ Links ]

22. Burgess S, Bowden J, Fall T, Ingelsson E, Thompson SG. Sensitivity analyses for robust causal inference from Mendelian randomization analyses with multiple genetic variants. Epidemiology 2017;28(1):30-42. DOI: 10.1097/EDE.0000000000000559 [ Links ]

23. Yin KJ, Huang JX, Wang P, Yang XK, Tao SS, Li HM, et al. No genetic causal association between periodontitis and arthritis: a bidirectional two-sample Mendelian randomization analysis. Front Immunol 2022;13:808832. DOI: 10.3389/fimmu.2022.808832 [ Links ]

24. Lin X, Li H. Obesity: epidemiology, pathophysiology, and therapeutics. Front Endocrinol (Lausanne) 2021;12:706978. DOI: 10.3389/fendo.2021.706978 [ Links ]

25. Wilson E, Woodd SL, Benova L. Incidence of and risk factors for lactational mastitis: a systematic review. J Hum Lact 2020;36(4):673-86. DOI: 10.1177/0890334420907898 [ Links ]

26. Brown KA. Impact of obesity on mammary gland inflammation and local estrogen production. J Mammary Gland Biol Neoplasia 2014;19(2):183-9. DOI: 10.1007/s10911-014-9321-0 [ Links ]

27. Wensveen FM, Valentić S, Šestan M, Wensveen TT, Polić B. The “Big Bang” in obese fat: events initiating obesity-induced adipose tissue inflammation. Eur J Immunol 2015;45(9):2446-56. DOI: 10.1002/eji.201545502 [ Links ]

28. Dobner J, Kaser S. Body mass index and the risk of infection - From underweight to obesity. Clin Microbiol Infect 2018;24(1):24-8. DOI: 10.1016/j.cmi.2017.02.013 [ Links ]

29. Falagas ME, Kompoti M. Obesity and infection. Lancet Infect Dis 2006;6(7):438-46. DOI: 10.1016/S1473-3099(06)70523-0 [ Links ]

30. Winter-Jensen M, Afzal S, Jess T, Nordestgaard B, Allin KH. Body mass index and risk of infections: a Mendelian randomization study of 101,447 individuals. Eur J Epidemiol 2020;35(4):347-54. DOI: 10.1007/s10654-020-00630-7 [ Links ]

31. Wang S, Ma HF, Wang XF, Xu B. Diagnosis of plasma cell mastitis with multi-slice spiral CT. Zhong Xi Yi Jie He Xue Bao 2005;3(3):199-202. DOI: 10.3736/jcim20050310 [ Links ]

32. Li XQ, Sun HG, Wang XH, Zhang HJ, Zhang XS, Yu Y, et al. Activation of C3 and C5 may be involved in the inflammatory progression of PCM and GM. Inflammation 2022;45(2):739-52. DOI: 10.1007/s10753-021-01580-2 [ Links ]

33. Emdin CA, Khera AV, Kathiresan S. Mendelian randomization. JAMA 2017;318(19):1925-6. DOI: 10.34.1001/jama.2017.17219 [ Links ]

34. Richmond RC, Davey Smith G. Mendelian randomization: concepts and scope. Cold Spring Harb Perspect Med 2022;12(1). DOI: 10.1101/cshperspect.a040501 [ Links ]

Data availability statement:

the data supporting the results of this study are available in the IEU GWAS database (https://gwas.mrcieu.ac.uk), the UK Biobank and the FinnGen consortium.

Artificial intelligence:the authors declare not to have used artificial intelligence (AI) or any AI-assisted technologies in the elaboration of the article.

Statement of ethic:this article does not contain any studies conducted by any of the authors on human or animal subjects.

Wei C, Wang X, Zeng J, Zhang G. Body mass index and risk of inflammatory breast disease: a Mendelian randomization study. Nutr Hosp 2024;41(1):96-111

Supplementary Table I. Basic information about the SNPs used in this study.

SNPs: single nucleotide polymorphisms; EA: effect allele; OA: other allele; SE: standard error.

Received: April 25, 2023; Accepted: July 07, 2023

Correspondence: Jinsheng Zeng. The First Affiliated Hospital of Nanchang University. 330006 Nanchang, P. R. China e-mail: 594691901@qq.com

^{Conflict of interest:}

the authors declare no conflict of interest.

^{Author contributions:}

Changlong Wei conceived and initiated the project and completed the data collection, analysis, and writing of the manuscript. Xiaofang Wang assisted Changlong Wei with the data analysis and writing of the manuscript. Jinsheng Zeng made constructive suggestions after reviewing the paper, and Xiaofang Wang and Gongyin Zhang made complementary suggestions to the paper. Jinsheng Zeng, Xiaofang Wang and Gongyin Zhang supervised this project. All authors read and approved the final manuscript.

This is an open-access article distributed under the terms of the Creative Commons Attribution License