Secondary Logo

Journal Logo

Review Articles

Molecular diagnostics for congenital heart disease: a narrative review of the current technologies and applications

Zu, Bailing; Zheng, Zhaojing; Fu, Qihua

Author Information
doi: 10.1097/JBR.0000000000000068
  • Open

Abstract

Introduction

Congenital heart disease (CHD) is a structural abnormality of the heart and/or great vessels that is present at birth. CHD is the most common birth defect with a reported prevalence of approximately 1% of all live births.[1] More than 130,000 new cases of CHD are diagnosed each year in China,[2] and CHD is recognized as a heavy socio-economic burden for the healthcare system. The pathogenic mechanism of CHD is complex and heterogeneous, involving genetic, epigenetic, and environmental factors.[3] However, accumulating evidence shows that the cause of CHD is primarily genetic. The updated scientific statement from the American Heart Association, published in 2018, states that 3% to 5% of CHD is caused by single-gene mutations, 8% to 10% is caused by chromosome anomaly/aneuploidy, and 3% to 25% syndromic CHD and 3% to 10% isolated CHD cases are caused by copy number variations (CNVs).[1] Therefore, the application of molecular genetic testing is of great significance for CHD healthcare.[4] Molecular genetic testing can facilitate earlier and more accurate diagnosis, as well as help with risk stratification of extracardiac abnormalities, and will aid in determining surgical intervention prognosis. In this review, we summarized the current molecular diagnosis techniques of CHD and focused on its practical applications for clinical CHD. This review also addressed the unmet issues and future directions in adapting novel genomic testing technologies to the molecular diagnosis of CHD in clinical settings.

Database search strategy

The authors performed an electronic search of the PubMed database for literatures describing molecular diagnostics of CHD from 1976 to 2020 using the following search terms: (1) congenital heart disease, (2) birth defect, combining with each of: (a) molecular diagnostics, (b) genetics, for example, “congenital heart disease molecular diagnostics”, viz. (1) + (a); “birth defect genetics”, viz. (2) + (b). Four queries were obtained. The results were further screened by title and abstract to select potentially useful studies. In addition, we further searched for the reference list of included studies to obtain more related studies. Articles that did not related with clinical molecular diagnosis of CHD were excluded.

Molecular diagnostic techniques

The rapid advancements in DNA sequencing technology, bioinformatics, and computing infrastructure for genomic data processing, have led to a shift in CHD genetic diagnosis from single loci testing to high throughput genome-wide testing. Traditional genetic analytical techniques, including karyotyping, Sanger sequencing, fluorescence in situ hybridization (FISH), and multiple ligation dependent probe amplification still play important, and sometimes indispensable, roles in clinical diagnostic workups of patients with CHD. However, high throughput genomic testing including chromosomal microarray (CMA), whole exome sequencing (WES), and whole genome sequencing (WGS) are increasingly used in clinical practice for patients with CHD. Here we summarize several testing techniques widely applied in the current clinical molecular diagnosis of CHD.

FISH is a macromolecule recognition technology that identifies sequences complementary to the probes.[5] FISH is achieved using nucleic acid labeled with fluorescent groups (the probe), which then binds to target DNA/RNA sequences in the specimen. Currently, commercially available FISH probes for CHD commonly focus on the 22q11 deletion and aneuploidy. For syndromic CHD, with typical clinical phenotypes and known genetic etiologies, FISH has demonstrated high diagnostic and clinical utility, as well as a cost-effect advantage over traditional karyotyping. Cuturilo et al[6] performed FISH on samples from 57 patients with suspected DiGeorge syndrome, and found that 24 of them were positive for the 22q11.2 microdeletion, with a diagnostic yield of 42.1%. Using FISH, Ramírez-Velazco et al[7] found that 22 of 39 Mexican patients with craniofacial dysmorphisms were positive for the 22q11.2 deletion, a diagnostic yield of 56%. Zheng et al[8] reported that 21-trisomy syndrome was confirmed in 14 of 16 children with suspected Down syndrome using a 21q22 FISH kit, a detection rate of 87.5%. According to the 2012 American Heart Association scientific statement,[9] FISH testing for the 22q11.2 microdeletion is suggested for all newborns and infants with conotruncal anomalies before surgical intervention. This analysis is also recommended for any child, adolescent, or adult with tetralogy of Fallot, ventricular septal defect, interrupted aortic arch, truncus arteriosus, or aortic arch anomaly, regardless of whether they have facial dysmorphisms. However, FISH only detects loci or genome regions known to cause a disease or phenotype.

CMA provides comprehensive genome-wide screening for CNVs through using either comparative genomic hybridization or single nucleotide polymorphism microarray platforms,[10] and has a resolution of 50 to 100 kb for deletions and duplications. In 2004, using a comparative genomic hybridization array, the 5q11 deletion was first identified in a CHD patient by comparative genomic hybridization array,[11] Subsequently, CMA was widely applied to the molecular diagnosis of CHD in clinical laboratories.[12–16] In a retrospective study including 514 CHD patients (excluding cases with chromosomal aneuploidy), our group found that the diagnostic yield of CMA for syndromic CHD was 14.11% to 20.56% and was 4.32% to 9.26% for isolated CHD.[17] For CNV analysis in structural birth defects, the American College of Medical Genetics and Genomics recommended CMA as first-tier diagnostic test for congenital malformations with multiple anomalies not specific to a well-delineated genetic syndrome.[18] CMA testing is a powerful diagnostic tool for CHD and produces higher diagnostic yields than do karyotype and FISH analyses, but CMA cannot detect single-nucleotide variants, small deletions and insertions, and balanced chromosomal rearrangements.

Targeted sequencing, using next generation sequencing, was designed to detect variants in selected genes or genomic regions that are most likely involved in diseases or phenotypes of interests. Targeted sequencing is an excellent alternative to WES because of the relative simplicity of variant filtering and prioritization. Moreover, targeted sequencing has a shorter turnaround time, a low rate of incidental findings, and a much higher coverage level than does WES,[19] and is commonly used for the clinical molecular diagnosis of monogenic CHD. For patients with monogenic CHD, including Noonan or Marfan syndromes, the diagnostic yield of targeted sequencing is as high as 80% to 90%.[20] Lepri et al[21] reported a diagnostic yield of 46.3% in a cohort of 90 patients with Noonan syndrome-related diseases by targeted sequencing of RAS-mitogen-activated protein kinase signaling genes. Furthermore, targeted sequencing could be used as a highly cost-effective testing tool for clinical evaluation of familial CHD. Blue et al[22] performed targeted sequencing of 57 CHD-related genes in 16 families with strong CHD histories. Using this approach, they identified 5 disease-causing variants that segregated with the disease phenotype and achieved a molecular diagnosis of 31% of this cohort. Jia et al[23] reported a diagnostic yield of 46.2% after using targeted sequencing in a cohort of 36 patients from 13 families with autosomal dominant non-syndromic CHD. While targeted sequencing has many benefits and advantages, there are also disadvantages of using this technique for both clinical application and CHD research. First, the gene panel of targeted sequencing should be periodically updated to reflect the discovery of new causative variants for CHD or there is the risk of missing certain important genes for data analysis and diagnosis. Second, CNV prediction based on targeted sequencing data is less accurate than WGS and needs further validation. Finally, reanalysis of targeted sequencing data is unlikely to provide new information about the genetic causes of CHD in patients for whom targeted sequencing was ordered.

WES, which sequences all protein-coding regions, can detect the majority of reported CHD related genetic variants, including both common and rare variants.[24] WES is a powerful diagnostic tool for patients with CHD and non-specific clinical phenotypes. In the largest genetic investigation of CHD to date, Jin et al[25] analyzed recessive- and dominant-inherited variants in a single cohort of 2871 patients with CHD who were recruited from The Pediatric Cardiac Genomics Consortium and the Pediatric Heart Network programs. WES identified rare inherited detrimental variants and de novo detrimental variants in 1.8% and 8% of the 2871 CHD probands, respectively. Li et al[26] performed WES in 342 congenital cardiac left-sided lesion cases and identified pathogenic variants in 49 (14.3%) cases. Moreover, WES is also a powerful tool for the discovery of novel CHD genes. Fotiou et al[27] integrated the CNV data from 4634 non-syndromic CHD cases with the WES data from 829 patients with isolated tetralogy of Fallot. Using this approach, and by comparing vertebrate ohnologs, they identified 54 novel candidate protein-coding genes. However, WES cannot sensitively detect CHD-related structural variations,[28] such as inversions and translocations, because WES coverage is heterogeneous in exome regions less amenable to capture.

WGS is a sequencing technique that covers up to 98% of the entire human genome, providing more uniform coverage across coding and non-coding genomic regions than does WES. WGS enables the detection of all types of genetic variation throughout the genome, including single-nucleotide variants, small insertion/deletions, CNVs, and structural variations.[28] Therefore, the use of WGS has improved the CHD diagnostic yields. Using WGS, Alankarage et al[29] identified clinically relevant variants in 31% of 97 families with CHD probands. Using the same approach, Reuter et al[30] found causative variants in 14 (12.6%) of 111 families with cardiac lesions. WGS can uncover new insights into the genetic architecture and pathogenesis of CHD. McKean et al[31] employed WGS and RNAseq to assess allele-specific expression and biallelic loss-of-expression in 144 subjects with surgically repaired CHD an found that subjects with CHD had a significant burden of both loss-of-expression and allele-specific expression events. Using combined integrative analysis of trio WES data from 759 CHD families, single nucleotide polymorphism array data from 922 patients with CHD, trio WGS data from 33 CHD families, and RNAseq data from 55 patients with CHD, Liu et al[32] performed a comprehensive genetic study on the molecular pathogenesis of CHD and identified five novel CNV loci. In conclusion, when used for CHD genetic testing WGS significantly expands genome coverage and effectively promotes the discovery of new molecular mechanisms of CHD development.

Practical considerations

Molecular diagnosis has already been integrated into routine clinical healthcare practice for patients with CHD. It has become increasingly important for the clinical management CHD to facilitate subclassification, prognosis prediction, and the monitoring of response to therapy. For each patient with CHD, clinicians should choose the most appropriate molecular diagnostic test based on initial assessment,[33] including physical exam, ultrasonography evaluation, and previous laboratory testing results. A proposed strategy for CHD molecular diagnostic test is outlined in Figure 1. Which type of testing is appropriate mainly depends on clinical presentation and socioeconomic considerations. For example, FISH or karyotype analysis is suitable for genetic testing of patients of suspected Down syndrome, while Sanger sequencing is perfectly applied for those with suspected Marfan syndrome. However, it may prove challenging for clinicians to choose the most appropriate molecular diagnostic testing approach for each patient with CHD, because no typical clinical phenotype is observed for most patients with isolated or syndromic CHD. Moreover, polygenic, digenic, and oligogenic inheritance of CHD involving genetic modifiers increases the difficulty of choosing appropriate tests for patients.[34,35]

Figure 1
Figure 1:
Outline of proposed molecular diagnostic test for patients with congenital heart disease. CHD = congenital heart disease, CMA = chromosomal microarray, FISH = fluorescence in situ hybridization, WES = whole exome sequencing, WGS = whole genome sequencing.

In 2015, the American College of Medical Genetics and Genomics released standards and guidelines for the interpretation of sequence variants,[36] but the current variant categories do not imply 100% certainty. First, phenotypic expression is affected by many factors, including penetrance and expressivity,[37] genomic imprinting,[38] genetic modification,[34] and stochastic effects.[39] Thus, a genetic variant can result in a range of phenotypic expression levels and the chance of the disease developing may not be 100%. It is possible that CHD pathogenic variants are filtered out from the analysis when the same variants are observed in healthy individuals. The complex relationships between genetic variations and phenotypic outcome hamper etiology dissection, especially for oligogenic and polygenic inheritance of CHD involving genetic modifiers. In addition, a variety of commercially or publicly available in silico tools are used for screening damaging variants. However, the specificity and sensitivity of splice site prediction and missense variant prediction tools cannot reach 100%,[40,41] potentially leading to false positive or false negative results. Lastly, variants of uncertain significance, including a broad category of CNVs and single-nucleotide variants, can be detected ubiquitously using molecular diagnostic testing. As a result of the lack of scientific evidence, determining whether variants of uncertain significance are clinically correlated with CHD relies on the judgment of the clinicians.[42]

It is very important to establish standardized procedures and quality management systems to produce reliable testing results in laboratories providing genetic testing for CHD. Proficiency testing promotes next generation sequencing standardization and the quality assurance of genetic testing.[43] Genomic DNA from well-characterized cell lines, or other validated reference DNA samples, can be used as proficiency testing samples to evaluate upstream sequencing processes, while electronic data can be used as proficiency testing samples to evaluate the downstream bioinformatics abilities.[43,44] Performance indicators of proficiency testing include accuracy, robustness, precision, sensitivity, specificity, repeatability, and reproducibility across upstream sequencing processes and downstream pipeline data handling. For clinical runs, wherever possible, it is recommended to run negative and positive controls to monitor the overall workflow steps, which contain numerous variant types.

Future directions

Population and disease databases are important tools for the molecular diagnosis of CHD. The 1000 Genomes Project contains human genetic variation with 2504 individuals from 26 populations and provides population specific allele frequencies.[45] However, this resource is not sufficient to identify rare and low frequency variants for a specific population because of the limited sample size. New cloud-based systems are emerging for larger genome and phenotype datasets, consolidating all the available in-house databases from different populations. Centralizing the storage and computation of large datasets will maximize the efficiency of genetic testing for CHD, and help to effectively convert data into individual medical diagnoses.

Advances in artificial intelligence (AI), especially deep learning algorithms and graphics processing units, have led to a rapid increase in applications for molecular diagnostics. The DeepVariant AI tool of Google Genomics can automatically call genetic variation by learning statistical relationships between images of read pileups around a putative variant and true genotype calls. DeepVariant demonstrates more than 50% fewer errors per genome (4652 errors) than commonly used bioinformatics methods, including GATK, FreeBayes, SAMtools, 16GT, and Strelka.[46] Based on convolutional neural networks to predict the pathogenicity of genetic variants, PrimateAI and SpliceAI significantly outperform similar prediction tools.[47,48] In the foreseeable future, AI applications in clinical genomics, including variant calling, variant classification, and phenotype-to-genotype correspondence, will significantly improve the molecular diagnosis of CHD.

The current molecular diagnosis result of CHD may not imply 100% accuracy, as described in Practical Consideration part. However, with the expansion of the population database, the optimization of AI analysis strategies, the accumulation of CHD phenotype-to-genotype data, and the declining price of sequencing, molecular diagnostic technologies will alter the management of CHD in the near future. With evidence generated from molecular diagnosis of CHD probands and family members, healthcare providers could: (1) estimate the CHD risk for offspring, siblings, or other close relatives; (2) assess the prognostic risk of extracardiac organ dysfunction, such as neurodevelopmental delays; and (3) provide more specific medical intervention prognosis including for transfusion and organ transplantation. The appropriate application of molecular diagnostics has greatly improved the clinical management of CHD and will significantly contribute to precision medicine for CHD in the future. Personalized, proactive, and precise CHD therapies are driven by the explosive developments in molecular diagnostic technologies, big data collection, and AI, which pose great challenges but present even greater opportunities.

Acknowledgments

None.

Author contributions

BZ and ZZ wrote the manuscript. QF designed and supervised the review. All authors edited the manuscript and approved the final manuscript.

Financial support

This work was supported by the National Natural Science Foundation of China, Nos. 81672090 and 81871717.

Conflicts of interest

The authors declare that they have no conflicts of interest.

References

1. Pierpont ME, Brueckner M, Chung WK, et al. Genetic basis for congenital heart disease: revisited: a scientific statement from the American Heart Association. Circulation 2018; 138:e653–e711.
2. Li H, Luo M, Zheng J, et al. An artificial neural network prediction model of congenital heart disease based on risk factors: A hospital-based case-control study. Medicine (Baltimore) 2017; 96:e6090.
3. Vecoli C, Pulignani S, Foffa I, et al. Congenital heart disease: the crossroads of genetics, epigenetics and environment. Curr Genomics 2014; 15:390–399.
4. Dorn C, Grunert M, Dopazo A. Rickert-Sperling S, Kelly R, Driscoll D, et al. Technologies to Study Genetics and Molecular Pathways. Springer, Congenital Heart Diseases: The Broken Heart. Vienna: 2016.
5. Savic S, Bubendorf L. Common fluorescence in situ hybridization applications in cytology. Arch Pathol Lab Med 2016; 140:1323–1330.
6. Cuturilo G, Drakulic D, Jovanovic I, et al. Improving the diagnosis of children with 22q11.2 deletion syndrome: a single-center experience from Serbia. Indian Pediatr 2016; 53:786–789.
7. Ramírez-Velazco A, Rivera H, Vásquez-Velázquez AI, et al. 22q11.2 deletion detected by in situ hybridization in Mexican patients with velocardiofacial syndrome-like features. Colomb Med (Cali) 2018; 49:219–222.
8. Zheng L, Liu DH, Hao SJ, et al. Value of oral mucosa cast-off cells as samples in fluorescent in situ hybridization for the diagnosis of Down's syndrome. Zhongguo Dang Dai Er Ke Za Zhi 2012; 14:202–204.
9. Marino BS, Lipkin PH, Newburger JW, et al. Neurodevelopmental outcomes in children with congenital heart disease: evaluation and management: a scientific statement from the American Heart Association. Circulation 2012; 126:1143–1172.
10. Wiszniewska J, Bi W, Shaw C, et al. Combined array CGH plus SNP genome analyses in a single assay for optimized clinical testing. Eur J Hum Genet 2014; 22:79–87.
11. Prescott K, Woodfine K, Stubbs P, et al. A novel 5q11.2 deletion detected by microarray comparative genomic hybridisation in a child referred as a case of suspected 22q11 deletion syndrome. Hum Genet 2005; 116:83–90.
12. Schmid M, Stary S, Blaicher W, et al. Prenatal genetic diagnosis using microarray analysis in fetuses with congenital heart defects. Prenat Diagn 2012; 32:376–382.
13. Goldmuntz E, Paluru P, Glessner J, et al. Microdeletions and microduplications in patients with congenital heart disease and multiple congenital anomalies. Congenit Heart Dis 2011; 6:592–602.
14. Payne AR, Chang SW, Koenig SN, et al. Submicroscopic chromosomal copy number variations identified in children with hypoplastic left heart syndrome. Pediatr Cardiol 2012; 33:757–763.
15. Thienpont B, Mertens L, de Ravel T, et al. Submicroscopic chromosomal imbalances detected by array-CGH are a frequent cause of congenital heart defects in selected patients. Eur Heart J 2007; 28:2778–2784.
16. Mademont-Soler I, Morales C, Soler A, et al. Prenatal diagnosis of chromosomal abnormalities in fetuses with abnormal cardiac ultrasound findings: evaluation of chromosomal microarray-based analysis. Ultrasound Obstet Gynecol 2013; 41:375–382.
17. Geng J, Picker J, Zheng Z, et al. Chromosome microarray testing for patients with congenital heart defects reveals novel disease causing loci and high diagnostic yield. BMC Genomics 2014; 15:1127.
18. Manning M, Hudgins L. Professional Practice and Guidelines Committee. Array-based technology and recommendations for utilization in medical genetics practice for detection of chromosomal abnormalities. Genet Med 2010; 12:742–745.
19. Ballester LY, Luthra R, Kanagal-Shamanna R, et al. Advances in clinical next-generation sequencing: target enrichment and sequencing technologies. Expert Rev Mol Diagn 2016; 16:357–372.
20. Zaidi S, Brueckner M. Genetics and genomics of congenital heart disease. Circ Res 2017; 120:923–940.
21. Lepri FR, Scavelli R, Digilio MC, et al. Diagnosis of Noonan syndrome and related disorders using target next generation sequencing. BMC Med Genet 2014; 15:14.
22. Blue GM, Kirk EP, Giannoulatou E, et al. Targeted next-generation sequencing identifies pathogenic variants in familial congenital heart disease. J Am Coll Cardiol 2014; 64:2498–2506.
23. Jia Y, Louw JJ, Breckpot J, et al. The diagnostic value of next generation sequencing in familial nonsyndromic congenital heart defects. Am J Med Genet A 2015; 167A:1822–1829.
24. Bamshad MJ, Ng SB, Bigham AW, et al. Exome sequencing as a tool for Mendelian disease gene discovery. Nat Rev Genet 2011; 12:745–755.
25. Jin SC, Homsy J, Zaidi S, et al. Contribution of rare inherited and de novo variants in 2,871 congenital heart disease probands. Nat Genet 2017; 49:1593–1601.
26. Li AH, Hanchard NA, Furthner D, et al. Whole exome sequencing in 342 congenital cardiac left sided lesion cases reveals extensive genetic heterogeneity and complex inheritance patterns. Genome Med 2017; 9:95.
27. Fotiou E, Williams S, Martin-Geary A, et al. Integration of large-scale genomic data sources with evolutionary history reveals novel genetic loci for congenital heart disease. Circ Genom Precis Med 2019; 12:442–451.
28. Meienberg J, Bruggmann R, Oexle K, et al. Clinical sequencing: is WGS the better WES? Hum Genet 2016; 135:359–362.
29. Alankarage D, Ip E, Szot JO, et al. Identification of clinically actionable variants from genome sequencing of families with congenital heart disease. Genet Med 2019; 21:1111–1120.
30. Reuter MS, Chaturvedi RR, Liston E, et al. The Cardiac Genome Clinic: implementing genome sequencing in pediatric heart disease. Genet Med 2020; 22:1015–1024.
31. McKean DM, Homsy J, Wakimoto H, et al. Loss of RNA expression and allele-specific expression associated with congenital heart disease. Nat Commun 2016; 7:12824.
32. Liu Y, Chang X, Glessner J, et al. Association of rare recurrent copy number variants with congenital heart defects based on next-generation sequencing data from family trios. Front Genet 2019; 10:819.
33. Xue Y, Ankala A, Wilcox WR, et al. Solving the molecular diagnostic testing conundrum for Mendelian disorders in the era of next-generation sequencing: single-gene, gene panel, or exome/genome sequencing. Genet Med 2015; 17:444–451.
34. Gifford CA, Ranade SS, Samarakoon R, et al. Oligogenic inheritance of a human heart disease involving a genetic modifier. Science 2019; 364:865–870.
35. Blue GM, Kirk EP, Giannoulatou E, et al. Advances in the genetics of congenital heart disease: a Clinician's guide. J Am Coll Cardiol 2017; 69:859–870.
36. Richards S, Aziz N, Bale S, et al. Standards and guidelines for the interpretation of sequence variants: a joint consensus recommendation of the American College of Medical Genetics and Genomics and the Association for Molecular Pathology. Genet Med 2015; 17:405–424.
37. Wright CF, West B, Tuke M, et al. Assessing the pathogenicity, penetrance, and expressivity of putative disease-causing variants in a population setting. Am J Hum Genet 2019; 104:275–286.
38. Tucci V, Isles AR, Kelsey G, et al. Genomic imprinting and physiological processes in mammals. Cell 2019; 176:952–965.
39. Lana-Elola E, Watson-Scales S, Slender A, et al. Genetic dissection of Down syndrome-associated congenital heart defects using a new mouse mapping panel. Elife 2016; 5:e11614.
40. Jian X, Boerwinkle E, Liu X. In silico prediction of splice-altering single nucleotide variants in the human genome. Nucleic Acids Res 2014; 42:13534–13544.
41. Zhou J, Theesfeld CL, Yao K, et al. Deep learning sequence-based ab initio prediction of variant effects on expression and disease risk. Nat Genet 2018; 50:1171–1179.
42. Char DS, Lee SS, Magnus D, et al. Anticipating uncertainty and irrevocable decisions: provider perspectives on implementing whole-genome sequencing in critically ill children with heart disease. Genet Med 2018; 20:1455–1461.
43. Endrullat C, Glökler J, Franke P, et al. Standardization and quality management in next-generation sequencing. Appl Transl Genom 2016; 10:2–9.
44. Duncavage EJ, Abel HJ, Pfeifer JD. In silico proficiency testing for clinical next-generation sequencing. J Mol Diagn 2017; 19:35–42.
45. 1000 Genomes Project Consortium, Auton A, Brooks LD, et al. A global reference for human genetic variation. Nature 2015; 526:68–74.
46. Poplin R, Chang PC, Alexander D, et al. A universal SNP and small-indel variant caller using deep neural networks. Nat Biotechnol 2018; 36:983–987.
47. Sundaram L, Gao H, Padigepati SR, et al. Predicting the clinical impact of human mutation with deep neural networks. Nat Genet 2018; 50:1161–1170.
48. Rowlands CF, Baralle D, Ellingford JM. Machine learning approaches for the prioritization of genomic variants impacting Pre-mRNA splicing. Cells 2019; 8:1513.
Keywords:

congenital heart diseases; diagnostic yield; molecular diagnosis; whole exome sequencing; whole genome sequencing

Copyright © 2020 The Chinese Medical Association, Published by Wolters Kluwer Health, Inc. under the CCBY-NC-ND license.