Over the past 30 years, the proportion of the Hong Kong population aged ≥65 years doubled from 6.6% in 1981 to 13.3% in 2011.1 The proportion of those aged ≥65 years among patients with tuberculosis (TB) tripled from 13%2 to 39%3 in the same period. Although the annual notification rate decreased from 149.1 to 65 per 100 000 population and the TB mortality rate decreased from 9.4 to 2.6 per 100 000 population from 1981 to 2011, the proportion of those aged ≥65 years increased from 53% to 82% among TB deaths.2 3 In older adults, TB is associated with other co-morbidities, hospitalisation, and delays in presentation and commencement of treatment.4 Missed opportunities for intervention might contribute to the higher mortality rates in older adults, and might also increase the risk of TB transmission. The present longitudinal study was conducted to assess the effects of age on the mortality rates of patients with TB and to elucidate the factors associated with missed TB diagnosis.

All consecutive cases of TB notified to the Department of Health in 2010 were retrospectively collected from the statutory TB notification registry. Hong Kong identity card numbers (or passport numbers for non-residents) were retrieved from the notification registry, together with date of notification, source of notification, and demographic and clinical information. Further clinical information and outcome data at 1 year after notification/initiation of treatment were retrieved from the TB programme record forms.3 These forms are filed by the TB and Chest Service for patients managed under its chest clinics and for patients managed by other health care providers. Treatment outcome was classified according to the World Health Organization (WHO) recommendations.5 Using the identity card number/passport number as the unique identifier, the 2010 TB cohort data were cross-matched with the statutory death registry from 1 January 2009 till 31 December 2012 for vital status, date and cause(s) of death. All cases with a date of TB notification after the date of death were recorded. A mortality survey was conducted on these recorded cases by retrieving relevant clinical information from records in public clinics and hospitals.

The demographics, co-morbidities, treatment outcomes, and mortality pattern of the cohort were analysed. Published data on patients of all ages with TB6 and on elderly patients with TB7 notified in 1996 were used for comparison. A date of TB notification after the date of death was considered as a surrogate marker of delayed diagnosis. Categorical variables were analysed by Pearson χ² test or Fisher’s exact test as appropriate; continuous variables were analysed by Mann-Whitney U test. Binary regression modelling was used to calculate the adjusted odds ratios (aORs) for risk factors for delayed diagnosis of TB after death using a backward conditional approach, with probability to remove being 0.10 and to retain being 0.05. A two-tailed P<0.05 was considered statistically significant. Statistical analyses were performed using SPSS for Windows, version 16.0 (SPSS Inc, Chicago [IL], US).

After exclusion of 336 cases subsequently denotified because of alternative diagnoses, a total of 5092 patients with TB were included in the 2010 TB cohort, at a notification rate of 72.5/100 000 person-years. Table 1 summarises their demographic data, clinical characteristics, and 1-year outcomes. Comparison with published data on the 1996 TB cohort6 7 was restricted to patients managed under the TB and Chest Service, for which the proportion of patients with TB aged ≥60 years increased from 34.5% in 1996 to 42.9% in 2010. There were more co-morbidities such as diabetes mellitus (16.0% vs 9.6%), lung cancer (2.1% vs 1.1%), and other cancers (5.0% vs 0.6%) in the 2010 TB cohort than in the 1996 TB cohort (χ² test, P<0.001). In 2010, the proportion of patients who died before completion of TB treatment was smaller for those managed under the TB and Chest Service (7.4%) than for the overall cohort (16.2%). However, the proportion of patients managed under the TB and Chest Service who died before completion of TB treatment nearly doubled between 1996 (3.9%) and 2010 (7.4%).

Among 5092 TB notifications, 1061 (20.9%) deaths occurred within 2 years of notification. Of the 1061 deaths, 211 (4.1%) occurred before the TB notification (ie, TB was notified after death; median delay in notification [interval between death and TB notification] 45 days, interquartile range 30-65 days). Of the deaths after notification, 683 (13.4%) died in the first year and 167 (3.3%) died in the second year. The reported causes of death were related to TB in only 191 (18.0%) of all deaths; 30 (14.2%) before TB notification, 158 (23.1%) in the first year, and three (1.8%) in the second year after notification.

Among the 211 deaths before TB notification, only 30 (14.2%) had TB as the main cause of death. There were 54 cases of ‘pneumonia unspecified’ and three cases of ‘sepsis unspecified’ reported as main cause of death. Of these cases, only five with potential causative organisms, such as Pseudomonas, Acinetobacter, or Escherichia coli, were identified. However, in the sputum that had been collected before death in these patients, Mycobacterium tuberculosis was subsequently isolated after prolonged culture, indicating that TB was the likely main cause of death in the remaining 52 deaths initially reported as ‘sepsis or pneumonia unspecified’. Including these revised results increases the 2010 TB-related mortality from 191 to 243, ie, an increase of 27% from the officially reported mortality figures of 2.6 to 3.4 per 100 000 person-years.8 The corresponding proportion of TB-related mortality increases to 38.8% (82/211) in cases with TB notified after death compared with 23.3% (158/683) who died in the first year after notification and 1.7% (3/167) who died in the second year. Therefore, a substantial proportion (15.5%) of TB-related deaths could potentially have been prevented by early diagnosis and treatment.

For the 211 deaths before TB notification, 25 cases of TB were notified from the public mortuary. Of the remaining 186 cases of TB that were notified from hospital, 173 hospital records were collected; 13 cases had missing data. None of the 198 patients with retrievable records were started on treatment. Of these 198 patients, 119 (60.1%) were aged ≥80 years at the time of death, and 93 (47%) had more than one admission to hospital before death. Prior to death, of these 198 patients, 83 (41.9%) were living in an old age home (OAH), 78 (39.4%) were bed-ridden, and 121 (61.1%) had an advance care directive such as ‘do not resuscitate’ or ‘do not investigate’ stated in the case notes. Table 2 summarises the univariate and multiple logistic regression analyses of these 198 early deaths, using deaths occurring within 1 year after notification as controls. Female sex, having a malignancy other than lung cancer, living in an OAH, drug abuser, sputum TB smear negative, sputum TB culture positive, and chest X-ray (CXR) not done or not available were independent risk factors for death before TB diagnosis.

Subgroup analysis was carried out for patients that most likely died of TB. The study group included the 30 patients who died of TB before diagnosis and the 52 patients whose deaths were initially reported as ‘sepsis or pneumonia unspecified’ but later sputum TB culture was positive. The control group was all patients who died of TB after notification (Table 3). Female sex, living in an OAH, sputum TB smear negative, sputum TB culture positive, and CXR not done or not available were independent risk factors for this group.

In the present study, the proportion of patients with TB aged ≥60 years increased by 25% from 1996 to 2010. However, over the same interval, the proportion of patients who had died before completion of treatment nearly doubled (Table 1). A substantial proportion (211 of 1061; 19.9%) of the TB-related deaths were notified after death. Over 60% of these cases were aged ≥80 years and none were started on treatment, suggesting a failure to detect TB rather than just a delay in notification. Over 60% of them had an advance care directive against resuscitation or investigation, likely indicating a concurrent terminal illness. Independent factors associated with TB notified after death were female sex, malignancies other than lung cancer, living in an OAH, drug abuse, sputum TB smear negative, sputum TB culture positive, and CXR not done. Although the recorded cause of death was TB in only 30 (14%) cases, in 52 (25%) cases the recorded cause of death was respiratory disease (predominantly pneumonia unspecified), particularly among those aged ≥80 years (19% vs 39%; P<0.005). In these cases, pulmonary TB is likely to have been the main or precipitating cause.

In the present study, the fatality rate in the first year of TB notification was 17.5% (4.1% died before TB notification and 13.4% died within 1 year after TB notification). This is much higher than rates reported earlier in Europe (7.8%9) and England and Wales (8.4%10), but similar to rates reported more recently in Taiwan (16.5%11). This is probably a reflection of differences among patient profiles in these regions, especially age and the associated co-morbidities. In the present study, 47% of patients with TB were aged ≥60 years (Table 1), whereas in the studies in Europe and the United Kingdom only 24.3%9 and 17.9%10 of the patients with TB were cohort aged ≥60 years.

Our finding that 4.1% of TB cases were notified after death is similar to rates reported in Taiwan in 2006 (4.0%12) and in the US in the 1980s (5.1%13 and 3.9%14). In all of these reports, advanced age was a consistent observation for this extreme form of delayed diagnosis. As expected from the relatively short turnover time for sputum TB smear tests and CXRs, sputum TB smear negative, and CXR unknown or not done were important risk factors for TB notified after death. The strong association between these cases and positive sputum TB culture might be explained by the fact that the sputum TB culture was the primary method of TB diagnosis, unless a diagnosis had already been made during autopsy.

Our findings that drug abusers have a higher chance of TB notification after death is in line with an earlier study that suggested such patients have difficulty completing medical evaluations.15 Drug abusers might be less aware of their TB symptoms because of the effects of the drugs taken, such as opiate suppression of the cough reflex.

Female sex was also an independent factor in the current study, similar to a previous study in Taiwan.11 This is expected, because there is a higher proportion of women among the geriatric population16 and among residents of OAH17 owing to their longer life expectancy and because conservative treatment is more frequently selected by these elderly female patients or their guardians. Patients with terminal conditions might have an advance care directive against resuscitation or investigation. An incorrect provisional diagnosis might also result from the readiness to accept a diagnosis of advanced disseminated malignancy in a patient with such an advance care directive. As lung cancer patients usually had CXR and sputum samples taken in their initial diagnostic investigation, coexisting TB could be discovered early. In addition, most lung cancer patients were diagnosed at an advanced stage and usually died within the first year after presentation.18

In our study, TB-related death occurred shortly before or after TB treatment was started, in line with findings from studies in Taiwan,19 the US,20 and Russia21 reporting a median time of 3 to 7 weeks from diagnosis or notification of TB to death. A study in Canada showed that a delay in TB treatment increased risk of death (aOR=3.3; 95% confidence interval=1.7-6.2) and intensive care unit admission (aOR=16.8; 95% confidence interval=2-144).22 Another study of hospitalised patients with TB also showed that late TB treatment guided by conventional TB culture was associated with a higher mortality than for treatment guided by polymerase chain reaction (PCR), liquid culture, positive histological findings or typical clinico-radiological manifestation.23 In settings with a high human immunodeficiency virus prevalence, the WHO advocates early empirical TB treatment based on clinical and radiological criteria in patients strongly suspected as having TB but with sputum TB smear negative, because this can improve survival.24 25

Although a timely diagnosis might not avert most non–TB-related deaths, early treatment could reduce the institutional transmission risk, because 42% of patients with TB were living in OAHs in the current study. The prevalence of active TB in OAHs has been estimated to be as high as 669 per 100 000 person-years in Hong Kong.26 The majority of patients in the present study did not have a positive sputum TB smear; however, a representative sputum sample might have been difficult to obtain from patients living in OAHs. That 73% of these patients had a positive sputum TB culture suggests that there was a sufficient degree of suspicion, either clinical or radiological, for initiation of bacteriological sampling. In total, 54 out of 211 patients who died before TB notification were recorded to have ‘pneumonia unspecified’ or ‘respiratory disease’ as the main cause of death. Past studies have shown that negative TB smear contributed to around 17% of TB transmission in San Francisco27 and Vancouver28 and even 30% in China.29 Thus, rapid diagnosis with effective isolation and early treatment can reduce transmission and even mortality. Sputum induction30 or gastric aspiration31 would improve specimen collection. However, in view of the infection risk, these bio-aerosol generating procedures would preferentially be performed in a negative pressure room with effective personal protective equipment as stipulated by the Institutional Infection Control Guidelines. Real-time PCR diagnostic tests such as Xpert® MTB/RIF assay32 may also be valuable, either as a primary diagnostic test or as an add-on test in patients previously found to be TB smear negative, to avoid the long turnover time for bacteriological cultures. In a study in Hong Kong,33 Xpert® MTB/ RIF assay was found to be a highly cost-effective strategy for TB diagnosis in terms of quality-adjusted life-years gained and lower first year mortality rate.

Higher mortality among patients with TB aged ≥80 years is a consistent finding among different TB programmes.34 The present study also found frequently missed diagnosis of TB and excessive mortality among patients aged ≥80 years who were frequently institutionalised and had multiple co-morbidities. A high index of suspicion and rapid diagnostic tools are necessary to reduce both mortality and transmission risk in a rapidly ageing population, in order to meet the WHO End TB 2035 target of a 95% reduction in TB mortality rate compared with the 2015 rate.35

This study shares an important limitation with other retrospective studies. The clinical data in this cohort were constructed from a database of the pre-assembled ‘TB programme record form’ which was not specifically designed for this study. Therefore, not all pertinent risk factors were identified and recorded. As this is a population-wide database, many health care professionals were involved and the measurement of risk factors and outcomes is less accurate and less consistent than a prospective study. Nonetheless, data from the TB programme record form have been used in previous studies on patients with TB6 and elderly patients with TB7 and were included for comparison in this study.

This study was a collaborative effort between the Hospital Authority and the Department of Health, and a database was compiled for all patients with TB treated in the public or the private sector. This study provides insight into the mortality of patients with TB and the risk factors associated with a delay in TB diagnosis. These factors include novel patient factors such as female sex, living in OAHs, advance care directives refusing further investigation or resuscitation, and drug abuse. Additional factors include lack of a representative sputum sample. which could be mitigated by sputum induction or gastric aspiration, and the relative insensitivity of sputum TB smear and long turnover time for conventional TB culture, which could be mitigated by using of real-time PCR tests. Information generated by this study will help frontline clinicians to be better aware of this important infectious disease among elderly people. Hopefully, more resources will be allocated to promote rapid diagnosis of TB for patients in high-risk scenarios in Hong Kong.

The authors would likely to thank the Nursing and General Grade staff in Department of Health and Hospital Authority for their assistance in collection and compilation of the demographical, clinical and laboratory data for this study.

Concept or design: ECC Leung, CC Leung, CK Chan, KC Chang.
Acquisition of data: TYW Mok, KS Chan, KS Lau, CH Chau, WKS Yee, WM Leung, KF Au.
Analysis or interpretation of data: WS Law, SN Lee, LB Tai.
Drafting of the article: ECC Leung, CC Leung, WM Leung, WS Law.
Critical revision for important intellectual content: All authors.

This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.

The authors have no conflicts of interest to disclose.

This study was approved by the Ethics Committee of the Department of Health and Ethics Committees of all hospital clusters from the Hospital Authority.

1. Demographic Statistics Section, Census and Statistics Department, Hong Kong SAR Government. Demographic Trends in Hong Kong 1981-2011. Available from: https://www.statistics.gov.hk/pub/B1120017032012XXXXB0100.pdf Accessed 13 Jul 2018. Crossref

2. Tuberculosis and Chest Service, Department of Health, Hong Kong SAR Government. Annual Report 1981. Hong Kong: Department of Health, Hong Kong SAR Government; 1981. Crossref

3. Tuberculosis and Chest Service, Department of Health, Hong Kong SAR Government. Annual Report 2011. Available from: http://www.info.gov.hk/tb_chest/doc/AnnualReport2011.pdf. Accessed 14 Jul 2018. Crossref

4. Leung CC, Yew WW, Chan CK, et al. Tuberculosis in older people: a retrospective and comparative study from Hong Kong. J Am Geriatr Soc 2002;50:1219-26. Crossref

5. World Health Organization. Definitions and reporting framework for tuberculosis—2013 revision. Geneva, Switzerland: World Health Organization; 2013. Crossref

6. Tam CM, Leung CC, Noertjojo K, Chan SL, Chan-Yeung M. Tuberculosis in Hong Kong-patient characteristics and treatment outcome. Hong Kong Med J 2003;9:83-90.

7. Chan-Yeung M, Noertjojo K, Tan J, Chan SL, Tam CM. Tuberculosis in the elderly in Hong Kong. Int J Tuberc Lung Dis 2002;6:771-9.

8. Centre for Health Protection, Department of Health, Hong Kong SAR Government. Notification & death rate of tuberculosis (all forms), 1947-2017. Available from: https://www.chp.gov.hk/en/statistics/data/10/26/43/88.html. Accessed 20 Jul 2018. Crossref

9. Lefebvre N, Falzon D. Risk factors for death among tuberculosis cases: analysis of European surveillance data. Eur Respir J 2008;31:1256-60. Crossref

10. Crofts JP, Pebody R, Grant A, Watson JM, Abubakar I. Estimating tuberculosis case mortality in England and Wales, 2001-2002. Int J Tuberc Lung Dis 2008;12:308-13.

11. Wu YC, Lo HY, Yang SL, Chu DC, Chou P. Comparing the factors correlated with tuberculosis-specific and non-tuberculosis-specific deaths in different age groups among tuberculosis-related deaths in Taiwan. PLoS ONE 2015;10:e0118929. Crossref

12. Wu YC, Lo HY, Yang SL, Chou P. Factors correlated with tuberculosis reported after death. Int J Tuberc Lung Dis 2014;18:1485-90. Crossref

13. Rieder HL, Kelly GD, Bloch AB, Cauthen GM, Snider DE Jr. Tuberculosis diagnosed at death in the United States. Chest 1991;100:678-81. Crossref

14. DeRiemer K, Rudoy I, Schecter GF, Hopewell PC, Daley CL. The epidemiology of tuberculosis diagnosed after death in San Francisco, 1986-1995. Int J Tuberc Lung Dis 1999;3:488-93.

15. Deiss RG, Rodwell TC, Garfein RS. Tuberculosis and illicit drug use: review and update. Clin Infect Dis 2009;48:72-82. Crossref

16. Census and Statistics Department, Hong Kong SAR Government. Population by Age Group and Sex. Hong Kong Population By-Census Main Report. 2015. Available from: http://www.censtatd.gov.hk/hong_kong_statistics/statistical_tables/index.jsp?charsetID=1<tableID=002. Accessed 16 Mar 2016.

17. Luk JK, Chan FH, Pau MM, Yu C. Outreach geriatric service to private old age homes in Hong Kong West Clusters. J HK Geriatr Soc 2002;11:5-11.

18. Cancer Research UK. Lung cancer survival statistics. Available from: http://www.cancerresearchuk.org/healthprofessional/cancer-statistics/statistics-by-cancer-type/lung-cancer/survival. Accessed 15 Mar 2016. Crossref

19. Lin CH, Lin CJ, Kuo YW, et al. Tuberculosis mortality: patient characteristics and causes. BMC Infect Dis 2014;14:5. Crossref

20. Oursler KK, Moore RD, Bishai WR, Harrington SM, Pope DS, Chaisson RE. Survival of patients with pulmonary tuberculosis: clinical and molecular epidemiologic factors. Clin Infect Dis 2002;34:752-9. Crossref

21. Mathew TA, Ovsyanikova TN, Shin SS, et al. Causes of death during tuberculosis treatment in Tomsk Oblast Russia. Int J Tuberc Lung Dis 2006;10:857-63.

22. Greenaway C, Menzies D, Fanning A, et al. Delay in diagnosis among hospitalized patients with active tuberculosis—predictors and outcomes. Am J Respir Crit Care Med 2002;165:927-33. Crossref

23. Lui G, Wong RY, Li F, et al. High mortality in adults hospitalized for active tuberculosis in a low HIV prevalence setting. PLoS One 2014;9:e92077. Crossref

24. Holtz TH, Kabera G, Mthiyane T, et al. Use of a WHO-recommended algorithm to reduce mortality in seriously ill patients with HIV infection and smear-negative pulmonary tuberculosis in South Africa: an observational cohort study. Lancet Infect Dis 2011;11:533-40. Crossref

25. Katagira W, Walter ND, Den Boon S, et al. Empiric TB treatment of severely ill patients with HIV and presumed pulmonary TB improves survival. J Acquir Immune Defic Syndr 2016;72:297-303. Crossref

26. Chan-Yeung M, Chan FH, Cheung AH, et al. Prevalence of tuberculous infection and active tuberculosis in old age homes in Hong Kong. J Am Geriatr Soc 2006;54:1334-40. Crossref

27. Behr MA, Warren SA, Salamon H, et al. Transmission of Mycobacterium tuberculosis from patients smear-negative for acid-fast bacilli. Lancet 1999;353:444-9. Crossref

28. Hernández-Garduño E, Cook V, Kunimoto D, Elwood RK, Black WA, FitzGerald JM. Transmission of tuberculosis from smear negative patients: a molecular epidemiology study. Thorax 2004;59:286-90. Crossref

29. Yang C, Shen X, Peng Y, et al. Transmission of Mycobacterium tuberculosis in China; a population-based molecular epidemiologic study. Clin Infect Dis 2015;61:219-27. Crossref

30. Chang KC, Leung CC, Yew WW, Tam CM. Supervised and induced sputum among patients with smear-negative pulmonary tuberculosis. Eur Respir J 2008;31:1085-90. Crossref

31. Brown M, Varia H, Bassett P, Davidson RN, Wall R, Pasvol G. Prospective study of sputum induction, gastric washing, and bronchoalveolar lavage for the diagnosis of pulmonary tuberculosis in patients who are unable to expectorate. Clin Infect Dis 2007;44:1415-20. Crossref

32. Steingart KR, Schiller I, Horne DJ, Pai M, Boehme CC, Dendukuri N. Xpert® MTB/RIF assay for pulmonary tuberculosis and rifampicin resistance in adults. Cochrane Database Syst Rev 2014;(1):CD009593. Crossref

33. You JH, Lui G, Kam KM, Lee NL. Cost-effectiveness analysis of the Xpert MTB/RIF assay for rapid diagnosis of suspected tuberculosis in an intermediate burden area. J Infect 2015;70:409-14. Crossref

34. Waitt CJ, Squire SB. A systematic review of risk factors for death in adults during and after tuberculosis treatment. Int J Tuberc Lung Dis 2011;15:871-85. Crossref

35. World Health Organization. Implementing the end TB strategy: the essentials. Available from: http://www.who.int/tb/publications/2015/The_Essentials_to_End_TB/en/.Accessed 3 May 2017. Crossref

Cardiac genetics is evolving rapidly and many new insights have recently been achieved. Genetic causes are found in various potentially lethal channelopathies and cardiomyopathies including long and short QT syndrome (LQTS and SQTS), Brugada syndrome, catecholaminergic polymorphic ventricular tachycardia (CPVT), hypertrophic cardiomyopathy (HCM), dilated cardiomyopathy (DCM), arrhythmogenic right ventricular dysplasia/cardiomyopathy (ARVD/C), Barth syndrome, and left ventricular non-compaction.1 Knowledge of genetics deepens the understanding of pathophysiology and remarkably changes the diagnosis, treatment, and genetic counselling for recurrence risk and family planning. This group is highly genetically heterogeneous (Table 12).

The genetic basis of inherited cardiac diseases in the Hong Kong Chinese population is largely unknown. The Princess Margaret Hospital provides a comprehensive cardiac genetic service. We conducted this study to review the clinical and genetic findings of 28 unrelated positive cases encountered between January 2006 and December 2015.

Diagnosis of the cardiac conditions was based on clinical assessments by a cardiologist and practice guidelines.3 4 5 The patients were referred by cardiologists from various public hospitals for genetic analysis. Only patients with positive genetic findings are reported in this study. There were seven patients with LQTS, two with Brugada syndrome, two with CPVT, nine with HCM, four with DCM, and four with ARVD/C. Local ethics board approval was obtained. Peripheral blood samples were collected from the proband after informed consent was obtained. Genomic DNA was extracted using a QIAamp Blood Kit (Qiagen, Hilden, Germany). The coding exons and the flanking introns (10 bp) of each gene were amplified by polymerase chain reaction. The primer sequences and protocol are available on request. Sanger sequencing was performed in the following order and stopped once a positive finding was detected: KCNQ1, KCNH2, KCNE1, KCNE2, and SCN5A for LQTS; SCN5A for Brugada syndrome; RYR2 for CPVT; MYH7 and MYBPC3 for HCM; LMNA for DCM; and PKP2 and DSP for ARVD/C. The order was based on prevalence according to the literature and local experience. All coding exons were amplified for each gene except selected exons 3, 8, 14, 45, 46, 47, 49, 88, 89, 90, 93, 96, 97, 100, 101, and 103 for RYR2.6 The GenBank accession numbers are shown in Table 2. The pathogenicity of novel missense variants was analysed by Alamut Visual (Interactive Biosoftware, Rouen, France) with Polymorphism Phenotyping v2 (PolyPhen-2), Sorting Intolerant from Tolerant (SIFT), MutationTaster, and Assessing Pathogenicity Probability in Arrhythmia by Integrating Statistical Evidence (APPRAISE, https://cardiodb.org/APPRAISE/) and that of novel splicing variants by Splice Site Finder-like, MaxEntScan, NNSPLIC, GeneSplicer, and Human Splicing Finder, wherever appropriate. Splicing variants were considered to be damaging if there was a >10% lower score when compared with the wild-type prediction. Allele frequencies among populations were referred to the Exome Aggregation Consortium (ExAC; http://exac.broadinstitute.org/).

During the 10-year study period more than 90 patients with channelopathies or cardiomyopathies were referred for genetic analysis. Among them, 28 unrelated patients had positive genetic results, comprising 17 males and 11 females. Their mean age at diagnosis was 39 years (range, 1-80 years). The major clinical presentations included syncope, palpitations, and abnormal electrocardiography (ECG) findings. Four patients were asymptomatic and were diagnosed following an incidental abnormal finding related to other medical issues. A family history was present in only 13 (46%) patients. All detected mutations were heterozygous, and 26 different heterozygous mutations were detected. These encompassed 11 missense, two nonsense, and five splicing mutations, as well as eight small insertions and deletions. There were six novel mutations—two in SCN5A (NM_198056.2:c.429del and c.2024-11T>A), two in MYBPC3 (NM_000256.3:c.906-22G>A and c.2105_2106del), and two in LMNA (NM_170707.3:c.73C>A and c.1209_1213dup) [Table 3]. All were considered pathogenic or likely pathogenic according to the Practice Guidelines for the Evaluation of Pathogenicity and the Reporting of Sequence Variants in Clinical Molecular Genetics by the Association for Clinical Genetic Science.7 Further clinical details and genotypes are shown in Table 3.

There were seven patients with LQTS, two with Brugada syndrome, two with CPVT, nine with HCM, four with DCM, and four with ARVD/C. Three patients with LQTS had mutations in KCNQ1 (cases 1-3) and four had mutations in KCNH2 (cases 4-7). Two patients (cases 8 and 9) with Brugada syndrome had mutations in SCN5A, including two novel mutations. Two patients (cases 10 and 11) with CPVT had mutations in RYR2. Four patients with HCM (cases 12-15) had MYH7 mutations and five (cases 16-20) had MYBPC3 mutations, including two novel mutations. Four patients with DCM (cases 21-24) had LMNA mutations, including two novel mutations. Finally, three patients with ARVD/C had PKP2 mutations (cases 25-27) and one had a DSP mutation (case 28).

This is the first report of a cardiac genetic case series among Hong Kong Chinese patients with channelopathies and cardiomyopathies. A total of 28 patients are reported, and 26 different mutations and six novel mutations have been identified. Wide genetic diversity is observed, with no common mutation found. Hereditary channelopathies and cardiomyopathies are mainly inherited in an autosomal dominant manner. Mutations can be either inherited or de novo. Risk to proband sibling(s) and first-degree relatives depends on the genetic status of the parents. Offspring of the proband have a 50% risk of inheriting the mutation. Siblings of the proband have the same risk if the mutation is transmitted from either parent. Patients carrying a mutation of these sudden arrhythmia death syndromes show incomplete penetrance. In general, a mutation carrier will show symptoms/signs in 80% of those with CPVT, 20% to 50% of ARVD/C patients, 18% to 63% of LQTS patients, 80% to 94% of SQTS patients, and 80% of patients with Brugada syndrome who have abnormal ECG findings when challenged with a sodium channel blocker.8 No exact figure is available for HCM. The data could be more specific if a particular mutation was considered alongside clinical findings and family history. Pre-symptomatic testing of at-risk family members cannot be used to predict age of onset, severity, type of symptoms, or rate of progression. Detailed clinical, ECG, and genetic characterisation of affected and unaffected family members is helpful.

Long QT syndrome is genetically heterogeneous, with at least 12 genes involved. Mutations in the four genes, KCNQ1, KCNH2, KCNE1, and KCNE2, are detected in 46%, 38%, 2%, and 1% of affected patients, respectively.8 A small proportion of patients (3%) have double heterozygous mutations in more than one disease loci.9 Specific arrhythmogenic triggers are associated with a particular subtype, such as exertion, swimming, and near-drowning for LQT1; auditory triggers and cardiac events occurring in the postpartum period for LQT2; and cardiac events during sleep or at rest for LQT3. Three patients had KCNQ1 mutations. Case 1 had recurrent syncope induced by exercise and swimming, but genetic testing confirmed LQTS type 1. Other patients had no specific provoking factor. LQTS type 2 caused by KCNH2 mutations accounts for about 38% of all LQTS.8 Four patients (cases 4-7) carried KCNH2 mutations and two (cases 4 and 6) presented with Torsades de pointes and one (case 7) had survived cardiac arrest requiring an implantable cardioverter defibrillator. Case 6 was the youngest patient, presenting at age 1 year. Genotype-guided treatment in LQTS is recommended and LQT1 responds best to beta-blockers.10 11

Brugada syndrome is characterised by cardiac conduction abnormalities (ST-segment abnormalities in leads V1-V3 on ECG and a high risk for ventricular arrhythmias) that can result in sudden death. The Shanghai Score System has been recently published for the diagnosis of Brugada syndrome.12 13 The prevalence of Brugada syndrome or its characteristic ECG pattern is reportedly higher among Asians, such as Japanese (0.14%-1.22%).14 15 16 17

Brugada syndrome is genetically heterogeneous and can be attributed to defects in at least 23 genes at the time of reporting.8 Mutations in SCN5A are detected in 11% to 14% of affected individuals in Japan and <10% in Taiwan where mutations in CACNA1C account for 1% to 7%.18 Approximately 65% to 70% of patients remain genetically undiagnosed. Expressivity is variable and penetrance is incomplete and low.

Conventionally, Brugada syndrome has been described as a monogenic disease that has autosomal dominant inheritance with incomplete penetrance; it is caused by rare genetic variants with a large effect size. Most individuals diagnosed with Brugada syndrome have an affected parent. The proportion of cases caused by a de-novo mutation is approximately 1%. Recent studies indicate that genetic inheritance is likely more complex, and models of an oligogenic disorder or susceptibility risk/genetic predisposition have been suggested.19 20 21 22

Among the two patients in this series, none had a positive family history. Symptoms were more non-specific, such as palpitation and syncope. It is noteworthy that convulsion can be a presentation of channelopathies (case 8). Clinical suspicion should be higher with more specific investigations, such as exercise-stress ECG and flecainide challenge tests, are required in order to reveal the real culprit. Sudden cardiac death can be the first presenting symptom in Brugada syndrome.

Two novel mutations are described in SCN5A: c.429del and c.2024-11T>A. The former is predicted to cause a frameshift and premature protein truncation. The latter is predicted to abolish the acceptor splice site and create a cryptic site upstream. At the time of reporting, both are absent from controls in the Exome Sequencing Project, 1000 Genomes Project, and ExAC. SCN5A mutations can cause either LQTS or Brugada syndrome.

Catecholaminergic polymorphic ventricular tachycardia can present with syncope and sudden death during physical exertion or emotion, due to catecholamine-induced bidirectional ventricular tachycardia, polymorphic ventricular tachycardia or ventricular fibrillation. The reported mean age of onset is between 7 and 12 years.8 Exercise stress testing or an adrenaline provocation test may induce ventricular arrhythmia and enable a clinical diagnosis. About half of these cases are related to a dominantly inherited RYR2 gene mutation, with a small proportion (1%-2%) related to recessively inherited CASQ2 gene mutations. RYR2 is a large gene with 105 exons. Tier testing has been proposed by Medeiros-Domingo et al.6 First-tier RYR2 genetic testing of the 16 selected exons allows identification of about 65% of CPVT cases. There were two paediatric CPVT patients (cases 10 and 11) in our series, with two known disease-causing mutations detected, namely NM_001035.2(RYR2):c.11836G>A (p.Gly3946Ser)23 24 25 26 and c.14848G>A (p.Glu4950Lys).23 24 Both mutations were detected in first-tier screening.

Hypertrophic cardiomyopathy is the most prevalent hereditary cardiac disease, causing about one third of sudden cardiac deaths in young athletes. Its prevalence in China is approximately 1 in 1250.27 The clinical manifestations are markedly variable, ranging from asymptomatic to sudden cardiac death. Genetic testing provides an accurate diagnosis in the probands and enables screening of asymptomatic family members. Although the genetic background of HCM is heterogeneous, involving at least 30 genes, MYH7 and MYBPC3 are the most common and each accounts for approximately 40%.8

Nine patients with HCM are reported here: four had known MYH7 mutations and five had MYBPC3 mutations, including two novel mutations. NM_000256.3:c.906-22G>A was detected in case 16 and was a novel variant. Neither population frequency nor known pathogenicity have been reported. In-silico analysis showed creation of a novel acceptor site and insertion of 20 nucleotides into exon 10. This conceivably would lead to a frameshift and premature protein termination. Exon 10 of MYBPC3 is a microexon in which the stability of its original splicing site is easily disrupted by intronic variants. A similar mutation has been reported as c.906-36G>A.28 Nonetheless, cDNA analysis was not performed. NM_000256.3(MYBPC3):c.1223+1G>A at the critical canonical +1 splice site is also novel. In addition, other known disease-causing splicing mutations affecting the same nucleotide have been reported.26 29 30 Case 19 had two variants detected in MYBPC3 (c.2215G>A and c.3624del). The small deletion c.3624del is a mutation known to cause HCM in the Chinese population31 and predicted to cause a frameshift and premature termination of the protein. The missense variant c.2215G>A is as yet unreported and is predicted by in-silico analyses to cause an amino acid change from glutamate to lysine at codon 739 and probably damage. At the time of reporting, the variant is absent from controls in the Exome Sequencing Project, 1000 Genomes Project, and ExAC databases. This variant is considered to have uncertain significance. The mother of the patient in case 19 was available for testing. She was 48 years old at the time of genetic testing, asymptomatic, and heterozygous for c.3624del only. Hence, the two variants c.2215G>A and c.3624del of MYBPC3 were in-trans in the patient and elder brother of the patient in case 19. Both had a more severe form of HCM, with a younger onset.

Familial DCM is a group of genetically heterogeneous disorders. Laminopathy can manifest as several allelic disorders affecting muscle, nerve, adipose, and vascular tissues; one of them is cardiomyopathy, dilated 1A. We identified four patients with DCM, two of whom also had proximal muscle weakness. Two novel mutations in LMNA were detected (c.73C>A and c.1209_1213dup). NM_005572.3(LMNA):c.73C>A is a novel variant that is predicted to be deleterious by SIFT, probably causing damage according to PolyPhen-2 and disease-causing according to MutationTaster. Other missense mutations have been reported in the same amino acid codon.32 33 34 NM_005572.3(LMNA):c.1609-1G>A is predicted to significantly affect splicing by in-silico analysis. At the time of reporting, all variants are absent from controls in the Exome Sequencing Project, 1000 Genomes Project, and ExAC. In case 22 with NM_005572.3(LMNA):c.73C>A, one of the parents died of chronic heart failure in the fourth decade of life, and one sibling died of heart block and chronic heart failure with a diagnosis of muscular dystrophy at age 38 years. Nonetheless, there was no sample left for genotyping.

Arrhythmogenic right ventricular dysplasia/cardiomyopathy is associated with fibrofatty replacement of cardiomyocytes, ventricular tachyarrhythmias, and sudden cardiac death. Although the right ventricle is primarily affected in this condition, left-dominant arrhythmogenic cardiomyopathy has also been described, and mutations have been identified in DSP as well as in other genes.35 Four patients are reported here, with three having mutations in PKP2 and one in DSP. Interestingly, the patient in case 26 presented at age 80 years with episodic palpitations. His ECG results showed paroxysmal ventricular tachycardia. He had a deletion in PKP2, c.1125_1132del (p.Phe376Alafs*8), resulting in a truncated incomplete protein product. Age of onset in patients with PKP2 mutations is older than that of the patient with DSP mutation. The latter patient (case 28) died at age 23 years, with sudden collapse as the first presentation.

Primary arrhythmogenic disorders including LQTS/SQTS, CPVT, Brugada syndrome, and cardiomyopathies account for about one third of sudden cardiac deaths in the young.36 Identification of a pathogenic variant can solve the diagnostic mystery, provide relief to the family, and enable family screening and counselling for other at-risk family members. In some developed countries, molecular autopsy is an essential part of a formal forensic investigation in unexplained sudden death.37 We support the implementation of molecular autopsy in routine autopsy investigation of sudden cardiac death victims. Our group has conducted the first local prospective study to determine the prevalence and types of sudden arrhythmia death syndrome underlying sudden cardiac death among local young victims through clinical and molecular autopsy of sudden cardiac death victims and clinical and genetic evaluation of their first-degree relatives (http://www.sadshk.org/en/medical_research.php). Such data can serve as the groundwork for the feasibility of implementation of such investigations in Hong Kong.

Genetic tests for cardiac conditions can aid diagnosis and guide treatment. Nonetheless, there are limitations that complicate the translational use of genetic results in patient care, such as incomplete penetrance, variable expressivity, and findings of variants of uncertain significance. In addition, since the genetic heterogeneity is large among cardiomyopathies and channelopathies and more genes are yet to be discovered, a negative genetic finding does not necessarily exclude a genetic basis of disease in patients.

Major limitations of the current study include its small sample size, incomplete family data for co-segregation study, and lack of functional study of novel variants. We observed a lower rate of use of genetic tests in early years that might have been due to insufficient awareness among clinicians about the clinical usefulness of such tests for channelopathies and cardiomyopathies. Clinical indications published in an expert consensus statement on the state of genetic testing for channelopathies and cardiomyopathies from the Heart Rhythm Society and European Heart Rhythm Association provide a good reference to determine when a genetic test should be requested.5 In our hospital, referral information can be accessed on http://kwcpath.home/genetics/ and more information about genetic service provision in public hospitals is available in the Hong Kong Hospital Authority Genetic Test Formulary (http://gtf.home/). A comprehensive system of cardiac genetics service is required for an efficient referral system, resource funding, training, and appropriate long-term follow-up.

We present the phenotypic and genotypic characteristics of 28 unrelated Hong Kong Chinese patients diagnosed across a 10-year period. For each disease entity, it was beyond our reach in the past decade to exhaustively screen for all known genes. We therefore focus on the most common ones when investigating cardiac genetics. Even so, genetic analysis can provide an accurate diagnosis and is of utmost importance for the management of patients and their families. Non-penetrance, variable expressivity, phenotype-genotype correlation, susceptibility risk, and digenic inheritance have been reported. Genetic testing also allows for genetic counselling on the recurrence risk. Correct interpretation of genetic findings for careful genetic counselling requires professional expertise with relevant experience in both clinical medicine and molecular genetics. Next-generation sequencing will improve diagnostic performance in this genetically heterogeneous group of channelopathies and cardiomyopathies, and could become a mainstay diagnostic tool.

All authors have made substantial contributions to the concept or design of this study; acquisition of data; analysis or interpretation of data; drafting of the article; and critical revision for important intellectual content.

This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.

All authors have no conflicts of interest to disclose. All authors had full access to the data, contributed to the study, approved the final version for publication, and take responsibility for its accuracy and integrity.

Local ethical approval of this study was obtained (KW/EX/09-155).

1. Herman A, Bennett MT, Chakrabarti S, Krahn AD. Life threatening causes of syncope: channelopathies and cardiomyopathies. Auton Neurosci 2014;184:53-9. Crossref

2. Ackerman MJ, Marcou CA, Tester DJ. Personalized medicine: genetic diagnosis for inherited cardiomyopathies/channelopathies. Rev Esp Cardiol (Engl Ed) 2013;66:298-307. Crossref

3. Priori SG, Wilde AA, Horie M, et al. HRS/EHRA/APHRS expert consensus statement on the diagnosis and management of patients with inherited primary arrhythmia syndromes: document endorsed by HRS, EHRA, and APHRS in May 2013 and by ACCF, AHA, PACES, and AEPC in June 2013. Heart Rhythm 2013;10:1932-63. Crossref

4. Authors/Task Force members, Elliott PM, Anastasakis A, et al. 2014 ESC guidelines on diagnosis and management of hypertrophic cardiomyopathy: the task force for the diagnosis and management of hypertrophic cardiomyopathy of the European Society of Cardiology (ESC). Eur Heart J 2014;35:2733-79. Crossref

5. Ackerman MJ, Priori SG, Willems S, et al. HRS/EHRA expert consensus statement on the state of genetic testing for the channelopathies and cardiomyopathies: this document was developed as a partnership between the Heart Rhythm Society (HRS) and the European Heart Rhythm Association (EHRA). Europace 2011;13:1077-109. Crossref

6. Medeiros-Domingo A, Bhuiyan ZA, Tester DJ, et al. The RYR2-encoded ryanodine receptor/calcium release channel in patients diagnosed previously with either catecholaminergic polymorphic ventricular tachycardia or genotype negative, exercise-induced long QT syndrome: a comprehensive open reading frame mutational analysis. J Am Coll Cardiol 2009;54:2065-74. Crossref

7. Wallis Y, Payne S, McAnulty C, et al. Practice guidelines for the evaluation of pathogenicity and the reporting of sequence variants in clinical molecular genetics. Association for Clinical Genetic Science; 2013.

8. Adam MP, Ardinger HH, Pagon RA, Wallace SE, editors. GeneReviews [internet]. Seattle (WA): GeneReviews; 1993. Available from: https://www.ncbi.nlm.nih.gov/books/NBK1116/. Accessed 19 Feb 2018.

9. Bokil NJ, Baisden JM, Radford DJ, Summers KM. Molecular genetics of long QT syndrome. Mol Genet Metab 2010;101:1-8. Crossref

10. Giudicessi JR, Ackerman MJ. Genotype- and phenotype-guided management of congenital long QT syndrome. Curr Probl Cardiol 2013;38:417-55. Crossref

11. Priori SG, Napolitano C, Schwartz PJ, et al. Association of long QT syndrome loci and cardiac events among patients treated with beta-blockers. JAMA 2004;292:1341-4. Crossref

12. Antzelevitch C, Yan GX, Ackerman MJ, et al. J-Wave syndromes expert consensus conference report: emerging concepts and gaps in knowledge. Europace 2017;19:665-94.

13. Antzelevitch C, Yan GX, Ackerman MJ, et al. J-Wave syndromes expert consensus conference report: emerging concepts and gaps in knowledge. J Arrhythm 2016;32:315-39. Crossref

14. Furuhashi M, Uno K, Tsuchihashi K, et al. Prevalence of asymptomatic ST segment elevation in right precordial leads with right bundle branch block (Brugada-type ST shift) among the general Japanese population. Heart 2001;86:161-6. Crossref

15. Matsuo K, Akahoshi M, Nakashima E, et al. The prevalence, incidence and prognostic value of the Brugada-type electrocardiogram: a population-based study of four decades. J Am Coll Cardiol 2001;38:765-70. Crossref

16. Sakabe M, Fujiki A, Tani M, Nishida K, Mizumaki K, Inoue H. Proportion and prognosis of healthy people with coved or saddle-back type ST segment elevation in the right precordial leads during 10 years follow-up. Eur Heart J 2003;24:1488-93. Crossref

17. Hiraoka M. Brugada syndrome in Japan. Circ J 2007;71 Suppl A:A61-8.

18. Juang JM, Tsai CT, Lin LY, et al. Unique clinical characteristics and SCN5A mutations in patients with Brugada syndrome in Taiwan. J Formos Med Assoc 2015;114:620-6. Crossref

19. Bezzina CR, Barc J, Mizusawa Y, et al. Common variants at SCN5A-SCN10A and HEY2 are associated with Brugada syndrome, a rare disease with high risk of sudden cardiac death. Nat Genet 2013;45:1044-9. Crossref

20. Behr ER, Savio-Galimberti E, Barc J, et al. Role of common and rare variants in SCN10A: results from the Brugada syndrome QRS locus gene discovery collaborative study. Cardiovasc Res 2015;106:520-9. Crossref

21. Gourraud JB, Barc J, Thollet A, et al. The Brugada syndrome: a rare arrhythmia disorder with complex inheritance. Front Cardiovasc Med 2016;3:9. Crossref

22. Juang JJ, Horie M. Genetics of Brugada syndrome. J Arrhythm 2016;32:418-25. Crossref

23. Priori SG, Napolitano C, Memmi M, et al. Clinical and molecular characterization of patients with catecholaminergic polymorphic ventricular tachycardia. Circulation 2002;106:69-74. Crossref

24. Jabbari J, Jabbari R, Nielsen MW, et al. New exome data question the pathogenicity of genetic variants previously associated with catecholaminergic polymorphic ventricular tachycardia. Circ Cardiovasc Genet 2013;6:481-9. Crossref

25. Kawamura M, Ohno S, Naiki N, et al. Genetic background of catecholaminergic polymorphic ventricular tachycardia in Japan. Circ J 2013;77:1705-13. Crossref

26. Xiong HY, Alipanahi B, Lee LJ, et al. RNA splicing. The human splicing code reveals new insights into the genetic determinants of disease. Science 2015;347:1254806. Crossref

27. Zou Y, Song L, Wang Z, et al. Prevalence of idiopathic hypertrophic cardiomyopathy in China: a population-based echocardiographic analysis of 8080 adults. Am J Med 2004;116:14-8. Crossref

28. Frank-Hansen R, Page SP, Syrris P, McKenna WJ, Christiansen M, Andersen PS. Micro-exons of the cardiac myosin binding protein C gene: flanking introns contain a disproportionately large number of hypertrophic cardiomyopathy mutations. Eur J Hum Genet 2008;16:1062-9. Crossref

29. Millat G, Bouvagnet P, Chevalier P, et al. Prevalence and spectrum of mutations in a cohort of 192 unrelated patients with hypertrophic cardiomyopathy. Eur J Med Genet 2010;53:261-7. Crossref

30. Waldmuller S, Muller M, Rackebrandt K, et al. Array-based resequencing assay for mutations causing hypertrophic cardiomyopathy. Clin Chem 2008;54:682-7. Crossref

31. Liu X, Jiang T, Piao C, et al. Screening mutations of MYBPC3 in 114 unrelated patients with hypertrophic cardiomyopathy by targeted capture and next-generation sequencing. Sci Rep 2015;5:11411. Crossref

32. Narula N, Favalli V, Tarantino P, et al. Quantitative expression of the mutated lamin A/C gene in patients with cardiolaminopathy. J Am Coll Cardiol 2012;60:1916-20. Crossref

33. Vytopil M, Benedetti S, Ricci E, et al. Mutation analysis of the lamin A/C gene (LMNA) among patients with different cardiomuscular phenotypes. J Med Genet 2003;40:e132. Crossref

34. Yuan WL, Huang CY, Wang JF, et al. R25G mutation in exon 1 of LMNA gene is associated with dilated cardiomyopathy and limb-girdle muscular dystrophy 1B. Chin Med J (Engl) 2009;122:2840-5.

35. Sen-Chowdhry S, Syrris P, Prasad SK, et al. Left-dominant arrhythmogenic cardiomyopathy: an under-recognized clinical entity. J Am Coll Cardiol 2008;52:2175-87. Crossref

36. Semsarian C, Hamilton RM. Key role of the molecular autopsy in sudden unexpected death. Heart Rhythm 2012;9:145-50. Crossref

37. Torkamani A, Muse ED, Spencer EG, et al. Molecular autopsy for sudden unexpected death. JAMA 2016;316:1492-4. Crossref

The expansion of newborn screening (NBS) for various genetic disorders with a focus on inborn errors of metabolism (IEM) has become a mandatory part of health care policy worldwide. Multiplex testing by tandem mass spectrometry has extended the scope of NBS far beyond the traditional ‘one test for one disease’ paradigm, requiring only a tiny blood sample, obtained by a simple heel prick.1 2 As a result, many inherited diseases are now screened for to allow early diagnosis and intervention and thereby prevent permanent damage or potential deaths.

Inborn errors of metabolism are a group of rare metabolic diseases with heterogeneous clinical presentations and genetic aetiologies. They are individually rare but collectively common. In 2011, Lee et al3 reported a 5-year retrospective review on the laboratory diagnosis of amino acid disorders, organic acidurias, and fatty acid beta-oxidation defects in three regional hospitals. The overall local incidence of classical IEM was 1 in 4122 live births.3 No phenylketonuria was identified through the screening of 18 000 newborns in the early 1970s.4 Hyperphenylalaninaemia was the second most common amino acid disorder reported by Lee et al,3 with an incidence of 1 in 29 542 live births. Another study by Hui et al5 reported the overall incidence of common IEM as 1 in 5400. According to the Hong Kong Paediatric Metabolic Registry, there were two cohorts, the first one with 20 years from 1982 to 2002 with 89 IEM patients and the second one with 14 years from 1996 to 2010 with 120 IEM patients. The estimated incidence of IEM was 1 in 7580 (unpublished data); however, as that was a voluntary case-finding study from several hospitals, the incidence was likely to be an underestimate. These figures are similar to those reported worldwide, such as 1 in 5800 in mainland China,6 1 in 5882 in Taiwan,7 and 1 in 4000 in America.8

In 2000, a mandatory NBS programme for hyperphenylalaninaemia, congenital hypothyroidism, and congenital deafness was implemented in mainland China.9 In 2006, the American College of Medical Genetics recommended 29 metabolic diseases (IEM) for which screening should be mandated.10 Since then, the scope of this recommendation has been expanding (Recommended Uniform Screening Panel, the Secretary of the Department of Health and Human Services11).12 In Hong Kong, population screening for congenital hypothyroidism and glucose-6-phosphate dehydrogenase (G6PD) deficiency using umbilical cord blood has been mandatory since March 1984 under the Neonatal Screening Unit of the Clinical Genetic Service, Department of Health. This programme has resulted in a significant reduction in related morbidities and mortalities.

In 2008, a coroner inquest was called to investigate the sudden death of a 14-year-old boy with a postmortem genetic diagnosis of glutaric acidaemia type II.13 The Coroners’ Report demanded that “The Department of Health, the Hospital Authority, the Faculty of Medicine of various universities and others concerned should carry out a feasibility study to see whether universal check may be carried out on all newborn babies for congenital metabolism defect.”14

To be effective, an expanded NBS programme needs to be coupled with improved general awareness of IEM and NBS. Educational support and training are required for frontline clinicians engaged in the diagnosis and care of patients with IEM.15 Several studies have shown that health care professionals do not have satisfactory awareness and knowledge of IEM.15 16 17 18 Therefore, a better understanding of the awareness of IEM among health care professionals in Hong Kong is needed.

We have conducted the first feasibility pilot study on the expanded NBS service model in a hospital setting in Hong Kong and the first survey on the knowledge and opinions on NBS for IEM among health care professionals in Hong Kong.

This prospective pilot study was conducted in phases from 1 October 2012 to 31 August 2014, involving three public hospitals and The University of Hong Kong (HKU), with over 40 collaborators from departments of pathology, paediatrics, and obstetrics. Phases 1 and 2 involved a single-site study conducted at Princess Margaret Hospital from 1 October 2012 to 31 October 2013 and then at Tuen Mun Hospital from 1 November 2013 to 31 March 2014. Phase 3 was university (HKU)-based and the recruitment was open to the public from 3 March 2014 to 31 August 2014. Phase 4 was a two-site study at the Tuen Mun Hospital and Queen Mary Hospital from 4 April 2014. Phase 5 was carried out at all three hospitals from 2 July 2014 until 31 August 2014. The OPathPaed model for expanded NBS was used for evaluation.19 The OPathPaed model includes 10 steps: parental education, consent, sampling, sample dispatch, dried blood spot (DBS) preparation and testing, reporting, recall and counselling, confirmation test, treatment and monitoring, and cost-benefit analysis (Fig 1).

Educational talks were delivered by chemical pathologists during antenatal visits. With the help of the Save Babies Through Screening Foundation, we added Chinese subtitles to the video titled “Newborn Screening Saves Babies One Foot at a Time”. The video is available online (https://www.youtube.com/watch?v=dxFit_a601w). DVDs and a locally designed pamphlet with an email address and telephone number for enquiries were distributed to expectant mothers (Fig 2). In order to raise public awareness, several interviews with the media were arranged and reports were published in several newspapers20 21 22 and radio and television programmes.23 24

A consent form was designed for NBS for IEM (data not shown). Educational videos and pamphlets were used to inform the parents. Written informed consent was collected during a postnatal talk after the education session. The talk was conducted in group presentation for the mothers by chemical pathologists.

Paediatricians or pathologists organised training for phlebotomists on the heel prick technique, in compliance with the Clinical and Laboratory Standard Institute guidelines.25 An instruction sheet with photographs of valid and invalid DBS samples was provided as guidance for the phlebotomists (Fig 3). Samples were collected from neonates aged between 24 hours and 28 days.

Drying racks and special boxes designed for specimen transport before complete drying were delivered to the testing sites. Complete drying of blood spots was ensured for valid sample integrity. The blood spot cards were dried perpendicular to each other above and below the rack position to avoid contact contamination between blood spots of different patients.

Two commercial DBS assay kits: (1) MassChrom Amino Acids and Acylcarnitines from Dried Blood/Non-derivatised (Chromsystems Instruments & Chemicals GmBH, Gräfelfing, Germany); and (2) NeoBase Non-derivatized MSMS kit (with succinylacetone assay; PerkinElmer, Waltham [MA], US) were validated for use in the study. In addition to a manual puncher and an autopuncher for DBS preparation, a fully automated online extraction system (DBS-MS 500; CAMAG, Muttenz, Switzerland) was also evaluated. The precision and local reference intervals of the commercial assay kits are listed in Table 1. Our laboratory has participated in the Newborn Screening Quality Assurance Programme organised by the US Centers for Disease Control and Prevention (CDC) since 2011. The disease panel included in the study is shown in Table 2.8 10 11

Chemical pathologists were responsible for reporting of positive results to the paediatricians. The CDC cut-off for clinical decision (https://wwwn.cdc.gov/NSQAP/Restricted/CDCCutOffs.aspx) and the Region 4 Stork Collaborative Project (https://www.clir-r4s.org/) data interpretation tools were applied during interpretation of the results.

Newborn Screening ACT Sheets and Confirmatory Algorithms by the American College of Medical Genetics (https://www.ncbi.nlm.nih.gov/books/NBK55827/) were followed for patient recall. All abnormal results were examined by chemical pathologists. These chemical pathologists were also responsible for contacting the parents for post-test counselling and for arranging subsequent hospital referrals for care by paediatricians.

Confirmation of diagnosis was provided by regional laboratories through measurements of functional metabolites (mainly plasma amino acid levels, plasma acylcarnitine levels, and urine organic acid levels) and genetic diagnosis by DNA sequencing wherever appropriate.

Admission logistics and treatment protocols for neonatal units with on-call rosters were established by hospital paediatricians. The same regional laboratories mentioned in Step 8 continued to provide biochemical diagnostic services.

A cost-benefit analysis has been conducted and published previously.26 Hyperphenylalaninaemia due to 6-pyruvoyl-tetrahydropterin synthase deficiency was used as an example to evaluate the costs and benefits of implementing an expanded NBS programme in Hong Kong. Assuming an annual birth rate of 50 000 and hyperphenylalaninaemia incidence of 1 in 29 542 live births, the annual medical costs and adjusted loss of workforce would be HK$20 773 207. The implementation and operational costs of an expanded NBS programme are expected to be HK$10 473 848 annually. Thus, implementing the expanded NBS programme is expected to result in an annual saving of HK$9 632 750.26

A questionnaire was distributed by convenience sampling to 430 health care professionals who worked in hospitals and were not involved in the pilot study. These self-administered questionnaires were distributed to local health care professionals including medical doctors, nurses, and other allied health care professionals either in person with returning envelopes or via email to department heads for further distribution. The self-administered questionnaire in English was modified from a previously published questionnaire that was tested among parents.27 The self-administered questionnaire included 13 questions that covered the local practice of the existing NBS programme, as well as knowledge and opinions of an expanded NBS programme. No personal identifiers were included in the questionnaire and questions were mostly in a closed-ended format. Data analyses were performed using Excel 2000 (Microsoft Corp. Redmond [WA], US) and GraphPad QuickCalcs (http://graphpad.com/quickcalcs/ConfInterval1.cfm). Percentages for each question were calculated as the number of replies divided by the total number of respondents for that question. The questions and corresponding responses are shown in Table 3.

By 31 August 2014, 2440 neonates had been recruited. The DBSs were collected from neonates aged 24 to 48 hours (n=2064, 84.6%), 3 to 5 days (n=331, 13.6%), 5 to 7 days (n=9, 0.4%), and 7 to 28 days (n=36, 1.5%). The participation rate was 86.6% on the days when blood samples were collected. There were no recorded DBS sampling or dispatch failures. The method validation and results of the DBS amino acids and acylcarnitine assays have been published elsewhere28; further details are available from the corresponding author on request. Overall, no true-positive cases were found in this pilot study, likely because of the limited sample size. Six (0.25%) false-positive cases were detected in 2440 neonates; of these, two had mild elevations in long-chain acylcarnitine levels, two had high tyrosine levels, one had a high citrulline level, and one had a low free carnitine level. Subsequent laboratory findings were all normal. No false-negative cases were reported from the IEM clinics of the involved hospitals within 2 years after project completion. However, patients who emigrated or received treatment at private institutions could not be followed up.

A total of 430 questionnaires were distributed and 210 (48.8%) completed responses were received. Results are shown in Table 3. Of the respondents, 50.0% were nurses and 32.9% were doctors. The doctors worked mainly in departments of paediatrics (47.8%), pathology (21.7%), and obstetrics (17.4%). Most (89.6%) respondents were aware of the existing NBS programme for hypothyroidism and G6PD deficiency; however, 47.5% did not know about IEM and 73.6% had not heard of expanded NBS for IEM. Most (87.6%) respondents agreed that more education on IEM and NBS is needed.

This is the first prospective pilot study on NBS for IEM in Hong Kong, and it has successfully evaluated the feasibility of the OPathPaed model. This study is also the first to investigate the knowledge and opinions on NBS for IEM of local health care professionals.

To implement an expanded NBS programme for IEM successfully in Hong Kong, there are several important points that need to be addressed. First, awareness and knowledge of NBS for IEM among the general public and among health care professionals should be improved.27 Second, comprehensive data on the local disease spectrum and incidence should be made available; such data were not available until recently.3 5 Third, free flow of information and sharing of experiences among colleagues working in the acute care and public health sectors should be facilitated. Fourth, more emphasis should be given to regular updates on NBS health care policy, confirmatory investigation service support, and treatment protocols. Last, the use of umbilical cord blood samples in the existing programme is unsuitable for an expanded NBS programme for IEM because of unacceptably high false-negative rates.29 The metabolites associated with many amino acid disorders, organic acid disorders, and fatty acid oxidation disorders are not elevated in cord blood. In 2013, the hospital-based OPathPaed model was published for the implementation of an expanded NBS programme suitable for a local setting.19 The present study further confirms the feasibility of the OPathPaed model for use on a larger scale. The OPathPaed model integrates expert input from obstetricians, pathologists, and paediatricians. Because babies born in Hong Kong are normally delivered in hospitals, the OPathPaed model approach should be able to achieve full coverage.

The success of an expanded NBS programme for IEM would depend not only on the diagnostics but also on how well patients diagnosed with IEM could be managed. It is difficult to accumulate experience and the many metabolic diseases can easily cause confusion. In addition, sophisticated management requires individualised drug formulations, which may not be easily accessible or may involve off-label prescriptions. Overseas studies have identified significant knowledge gaps among clinicians involved in the follow-up care of newborns with IEM identified by NBS.15 16 17 18 Some were poorly prepared to follow up the initial diagnosis, provide appropriate counselling, or make appropriate clinical referrals.17 In our study, 73.6% of 210 health care professionals (who were not involved in the pilot study) were unaware of the expanded NBS programme, and 47.5% of respondents did not know what IEM were. The majority of respondents (87.6%) agreed that better education was needed and 91.3% supported expanding NBS for IEM immediately or within 3 years. According to a parental survey among 172 parents regarding NBS for IEM,27 over 89% had never heard of NBS for IEM or metabolic disorders. Although some IEM may be incurable, 97% of parents supported an expanded NBS programme and 82.8% of parents supported implementation of this expansion immediately or within 3 years.27

The present study also provides the first local evaluation of the fully automated DBS-MS 500 system. The DBS is directly eluted into the extraction chamber, with an online extraction system connecting with the tandem mass spectrometer. There is no need for DBS card punching. Together with the integrated optical card recognition and barcode reading module, this automation minimises the risk of sample misidentification during manual processing. The precision and accuracy demonstrated are comparable to those of conventional procedures. However, because the DBS-MS 500 system requires application of an internal standard solution before extraction, the financial cost per extraction would be higher than that for conventional methods. In addition, special DBS cards are required for the extraction chamber. Third-party DBS cards of a specific quality may not easily fit into the system. The throughput of up to 500 DBS cards per run is more than adequate for local needs, as there are about 50 000 live births annually in Hong Kong.

The limitations of the pilot study include small and non-representative sample size, a relatively short study period that may have been inadequate for follow-up to confirm true negatives, and the convenience sampling and low response rate of the health care professional survey.

The present pilot study investigated the feasibility of an expanded NBS for IEM in Hong Kong, and surveyed health care professionals for their knowledge and opinions on NBS for IEM. We successfully evaluated the OPathPaed model on a larger scale than has been attempted previously and demonstrated that health care professionals have a favourable opinion of implementing an expanded NBS programme in Hong Kong. It is timely that, as this pilot study was completed, the needs of parents and health care workers were addressed in the Hong Kong Special Administrative Region Chief Executive’s Policy Address of 2015, when a government-led initiative was announced to study the feasibility of NBS for IEM in the public health care system on a large scale.

All authors have made substantial contributions to the concept or design of this study; acquisition of data; analysis or interpretation of data; drafting of the article; and critical revision for important intellectual content.

We acknowledge all collaborators, doctors, nurses, medical technologists, phlebotomists, information technologists, and parents for their efforts and support. We thank the Save Babies Through Screening Foundation for allowing us to use their video for educational purpose. We thank CAMAG Germany for providing technical support during the evaluation of the DBS-MS 500. The CAMAG had no role in the study design, data collection, analysis, reporting, or manuscript preparation.

This work was funded by the SK Yee Medical Foundation. The funder had no role in study design, data collection, analysis, interpretation, or manuscript preparation.

All authors have no conflicts of interest to disclose. All authors had full access to the data, contributed to the study, approved the final version for publication, and take responsibility for its accuracy and integrity.

Local ethical approval was obtained from each of the regional hospitals involved in this study.

1. Millington DS, Kodo N, Norwood DL, Roe CR. Tandem mass spectrometry: a new method for acylcarnitine profiling with potential for neonatal screening for inborn errors of metabolism. J Inherit Metab Dis 1990;13:321-4. Crossref

2. Carpenter KH, Wiley V. Application of tandem mass spectrometry to biochemical genetics and newborn screening. Clin Chim Acta 2002;322:1-10. Crossref

3. Lee HC, Mak CM, Lam CW, et al. Analysis of inborn errors of metabolism: disease spectrum for expanded newborn screening in Hong Kong. Chin Med J (Engl) 2011;124:983-9.

4. Davies DP. Hong Kong Reflections: Health, Illness and Disability in Hong Kong Children. Hong Kong: The Chinese University Press; 1995.

5. Hui J, Tang NL, Li CK, et al. Inherited metabolic diseases in the Southern Chinese population: spectrum of diseases and estimated incidence from recurrent mutations. Pathology 2014;46:375-82. Crossref

6. Gu X, Wang Z, Ye J, Han L, Qiu W. Newborn screening in China: phenylketonuria, congenital hypothyroidism and expanded screening. Ann Acad Med Singapore 2008;37(12 Suppl):107-10.

7. Niu DM, Chien YH, Chiang CC, et al. Nationwide survey of extended newborn screening by tandem mass spectrometry in Taiwan. J Inherit Metab Dis 2010;33(Suppl 2):S295-305. Crossref

8. Chace DH, Kalas TA, Naylor EW. The application of tandem mass spectrometry to neonatal screening for inherited disorders of intermediary metabolism. Annu Rev Genomics Hum Genet 2002;3:17-45. Crossref

9. Zheng S, Song M, Wu L, et al. China: public health genomics. Public Health Genomics 2010;13:269-75. Crossref

10. American College of Medical Genetics Newborn Screening Expert Group. Newborn screening: toward a uniform screening panel and system—executive summary. Pediatrics 2006;117(5 Pt 2):S296-307. Crossref

11. Recommended Uniform Screening Panel, The Advisory Committee on Heritable Disorders in Newborns and Children, US Department of Health and Human Services. Available from: https://www.hrsa.gov/advisorycommittees/mchbadvisory/heritabledisorders/recommendedpanel/. Accessed 1 Aug 2017.

12. Therrell BL, Johnson A, Williams D. Status of newborn screening programs in the United States. Pediatrics 2006;117(5 Pt 2):S212-52. Crossref

13. Lee HC, Lai CK, Siu TS, et al. Role of postmortem genetic testing demonstrated in a case of glutaric aciduria type II. Diagn Mol Pathol 2010;19:184-6. Crossref

14. Coroners’ Report 2008. Available from: http://www.judiciary.hk/en/publications/coroner_report_july08.pdf. Accessed 1 Aug 2017.

15. Gennaccaro M, Waisbren SE, Marsden D. The knowledge gap in expanded newborn screening: survey results from paediatricians in Massachusetts. J Inherit Metab Dis 2005;28:819-24. Crossref

16. Wells AS, Northrup H, Crandell SS, et al. Expanded newborn screening in Texas: a survey and educational module addressing the knowledge of pediatric residents. Genet Med 2009;11:163-8. Crossref

17. Kemper AR, Uren RL, Moseley KL, Clark SJ. Primary care physicians’ attitudes regarding follow-up care for children with positive newborn screening results. Pediatrics 2006;118:1836-41. Crossref

18. Dunn L, Gordon K, Sein J, Ross K. Universal newborn screening: knowledge, attitudes, and satisfaction among public health professionals. South Med J 2012;105:218-22. Crossref

19. Mak CM, Lam C, Siu W, et al. OPathPaed service model for expanded newborn screening in Hong Kong SAR, China. Br J Biomed Sci 2013;70:84-8. Crossref

20. 一滴血驗出罕見遺傳病. Oriental Daily 2010 Sep 12. Available from: http://orientaldaily.on.cc/cnt/news/20100912/00176_002.html. Accessed 1 Aug 2017.

21. 二代新生嬰兒篩檢代謝疾病. am730 2013 May 6. Available from: http://archive.am730.com.hk/column-153216. Accessed 1 Aug 2017.

22. 篩查防智障代謝病, 社會可年省近千萬. Ming Pao 2014 Jun 9. Available from: https://news.mingpao.com/pns/篩查防智障代謝病%20社會可年省近千萬/web_tc/article/20140609/s00002/1402257010439. Accessed 1 Aug 2017.

23. 精靈一點 (RTHK radio programme, 2014 Apr 15). Available from: http://programme.rthk.hk/channel/radio/programme.php?name=radio1/adwiser&d=2014-04-15&p=1147&e=259149&m=episode. Accessed 1 Aug 2017.

24. 星期二檔案：這幾滴血 (TVB programme, 2014 Feb 25). Available from: http://programme.tvb.com/news/tuesdayreport/episode/20140225/#page-1. Accessed 1 Aug 2017.

25. NBS01-A6, Blood Collection on Filter Paper for Newborn Screening Programs; Approved Standard—Sixth Edition. Available from: https://clsi.org/standards/products/newborn-screening/documents/nbs01/. Accessed 1 Aug 2017.

26. Lee HH, Mak CM, Poon GW, Wong KY, Lam CW. Cost-benefit analysis of hyperphenylalaninemia due to 6-pyruvoyl-tetrahydropterin synthase (PTPS) deficiency: for consideration of expanded newborn screening in Hong Kong. J Med Screen 2014;21:61-70. Crossref

27. Mak CM, Lam CW, Law CY, et al. Parental attitudes on expanded newborn screening in Hong Kong. Public Health 2012;126:954-9. Crossref

28. Mak M. Chemical pathology analysis of inborn errors of metabolism for expanded newborn screening in Hong Kong [thesis]. The University of Hong Kong; 2012. Available from: http://hub.hku.hk/handle/10722/180075. Accessed 1 Aug 2017.

29. Walter JH, Patterson A, Till J, Besley GT, Fleming G, Henderson MJ. Bloodspot acylcarnitine and amino acid analysis in cord blood samples: efficacy and reference data from a large cohort study. J Inherit Metab Dis 2009;32:95-101. Crossref

Connective tissue disease is a group of disorders characterised by the presence of antinuclear antibodies (ANAs) and clinical autoimmune phenomena. The investigations that are performed depend on both purpose and performance characteristics. For example, to rule out a diagnosis, a test with high sensitivity is needed, such as testing for the absence of ANAs to rule out systemic lupus erythematosus (SLE). In contrast, to establish a diagnosis, a test with high specificity is more desirable, such as testing for antibodies to double-stranded DNA (dsDNA) or anti-Sm antigens in SLE. Therefore, after an initial positive ANA test result, subsequent tests for specific antibodies, such as those against dsDNA and certain extractable nuclear antigens (ENAs), are necessary.

Conventionally, ANAs are detected by indirect immunofluorescence (IIF). This method is sensitive and essentially detects all antibodies against cellular constituents, with antibody profile having varying clinical significance. However, it is labour-intensive, and technical interpretation of the results can be subjective. The enzyme-linked immunosorbent assay (ELISA), which can be automated and high-throughput–enabled, is gaining popularity over IIF. When ELISA is used to screen for ANAs, the source of antigens has major implications on the sensitivity and specificity of the assay. Although the ELISA technique has improved with time, concerns over false-negative ANA cases persist. Therefore, the American College of Rheumatology (ACR) still recommends IIF as the gold standard in ANA testing.1

To detect antibodies against ENAs, gel precipitation assays have been used for more than five decades, and countercurrent immunoelectrophoresis (CIEP) has been accepted as the reference method for anti-ENA antibody testing. Positive results from CIEP are highly specific. The majority of the literature on autoantibodies and disease association has been established with this technique.2 However, other methods such as ELISA, immunoblot, and line blot are gradually replacing CIEP. In recent years, multiplex assays have been introduced. The BioPlex 2200 ANA Screen assay (Bio-Rad Laboratories, Hercules [CA], US) is an automated multiplex immunoassay using flow cytometry to detect a panel of autoantibodies, including ANAs and antibodies against ENAs. There are a few published studies showing reasonable agreement between this system and ELISA.3 4 5 6 7

Conventional assays are being replaced with newer automated high-throughput assays. However, the performance of the newer techniques may not be equivalent to that of conventional assays. This difference will have important implications to clinicians, who may base their clinical judgement on their knowledge of how conventional assays perform.8 9 10 11 12 13 14 15 16 In this study, we evaluate the performance of BioPlex 2200 using serum samples from a local cohort of SLE patients, and compare it with the performance of three established techniques (CIEP, ELISA, line blot) in terms of anti-ENA antibody detection. The sensitivity of BioPlex 2200 ANA Screen assay was also compared with IIF.

This cross-sectional study was conducted at the Queen Mary Hospital, Hong Kong, a tertiary university teaching hospital. Patients were recruited from the hospital’s lupus clinic from 1 December 2013 to 31 December 2013. All patients attending the clinic underwent routine serology screening during their visit. Of 160 consecutive patients, 140 with adequate serum stored in the clinical immunology laboratory were recruited. All patients had an established diagnosis of SLE, according to the ACR classification criteria.17 Patients who were <18 years or >80 years and pregnant patients were excluded from the study. Data of serum samples in 41 healthy controls, who were mainly laboratory staff and had given verbal consent for the blood donation were also included; their age ranged from 18 to 54 years. All stocked serum was stored at -70°C.

Electronic and written medical records of the recruited patients were reviewed, and relevant clinical and laboratory data were collected. Global disease activity was assessed according to the SLE disease activity index,18 19 and cumulative organ damage was assessed in terms of the Systemic Lupus International Collaborating Clinics/ACR Damage Index score.20

The BioPlex 2200 ANA Screen system was used to detect 13 types of autoantibodies simultaneously in one test—namely, those against dsDNA, chromatin, centromere B, Scl-70, RNP (RNP-A, RNP-68), Sm, RNP/Sm, Ro (SSA-52, SSA-60), SSB/La, Jo-1, and ribosomal P protein. For BioPlex results, anti-RNP was reported separately from anti-RNP/Sm; the kit’s RNP antigen is a recombinant antigen (RNP-A and RNP-68) whereas RNP/Sm is an affinity-purified antigen, which is similar to the antigen used for the RNP test in ELISA and line blot in this study.

The presence of anti-dsDNA antibody was classified as negative if levels were ≤4 IU/mL, indeterminate if 5 to 9 IU/mL, and positive if ≥10 IU/mL, as recommended by the manufacturer. For the other autoantibodies, the results were expressed as an antibody index (AI). An AI of 1.0 was the cut-off concentration that corresponded to approximately the 99th percentile of values obtained from a nondiseased population in the manufacturer’s study. Results of ≥1.0 were reported as positive (range, 0.2-8.0 AI). A test result was considered positive for ANAs if one or more of the antibody tests in the panel was positive.

The IIF assay was adopted as the reference method for ANA detection. All serum samples were diluted in 1:80 in phosphate-buffered saline and tested on slides pretreated with substrate from a human epithelial cell line (Kallestad HEp-2 Cell Line Substrate Slides; Bio-Rad Laboratories, Hercules [CA], US) according to the manufacturer’s instructions. The slides were read using the same microscope and setting as routine clinical samples by a single observer. Slides that were negative for ANA by IIF were reviewed by an independent second observer to confirm negativity. In cases of discrepancy, a third adjudicator was sought.

The performance of BioPlex in the detection of anti-ENA antibodies was compared with that of the following assays.

The CIEP assay used in this study was optimised in-house. Rabbit thymus extract (ImmunoVision Inc, United States) was used for typing of anti-Sm, anti-RNP, and anti-La, whereas human spleen extract (ImmunoVision Inc) was used as a source of Ro antigen.

The EUROASSAY test kit (EUROIMMUN, Lübeck, Germany) was used as the line blot immunoassay in this study. The kit qualitatively assessed the presence of human immunoglobulin G (IgG) autoantibodies against six different antigens: RNP, Sm, SS-A, SS-B, SCl-70, and Jo-1. On the basis of signal intensity, the results were categorised as negative, borderline, and positive.

The QUANTA Lite ENA Profile EIA kit (INOVA Diagnostics, San Diego [CA], US) was used for ELISA in this study. The kit qualitatively screened for the presence of IgG autoantibodies against specific ENAs—namely, SSA (60 and 52 kDa), SSB, Sm, RNP/Sm, Scl-70, and Jo-1. The results were calculated using the following formula (where OD = optical density at 450 nm):

Results of <8 U/mL were classified as negative, 8 to 12 U/mL as equivocal, and >12 U/mL as positive.

The diagnostic performance of BioPlex versus that of IIF was compared for the detection of ANAs in the SLE cohort and controls. Assay sensitivity and specificity were calculated and compared using a paired McNemar’s test. Cohen’s kappa coefficient and percentage of observed agreement were also calculated for the two methods.

For individual anti-ENA antibodies (against RNP, Sm, Ro, La, Scl-70, and Jo-1), agreement analysis was calculated for the four laboratory methods. Fleiss’ kappa coefficient with its 95% confidence interval and the percentage of observed agreement were calculated to assess overall agreement among the four methods. Pairwise agreement analysis for the four methods was also performed by calculating Cohen’s kappa coefficient and percentage of observed agreement. Weak and borderline results in the ELISA and line blots were treated as negative in the analysis.

The diagnostic value of anti-ENA antibody detection to predict various disease manifestations was examined, along with comparisons between the different methods. In particular, we studied diagnostic performance for the association of anti-Sm antibodies with nephritis, anti-RNP antibodies with Raynaud’s phenomenon, and anti-Ro/La antibodies with photosensitivity, discoid rash, Sicca symptoms, leukopenia, and lymphopenia.21 22 23 24 25 26 Sensitivity, specificity, positive and negative predictive values, and diagnostic accuracies were calculated.

The SPSS (Windows version 20.0; IBM Corp, Armonk [NY], US) and Microsoft Excel 2010 for Windows were used for statistical analysis and calculation of confidence intervals, respectively. P values of <0.05 were regarded as statistical significance. The STARD 2015 guidelines were used during the writing of this article.27

All 140 SLE patients were local Chinese patients, with a female predominance (n=128, 91.4%). The mean age was 46.8 (range, 24-69) years and the median disease duration was 17 years (Table 1). The majority of our cohort had at least one anti-ENA antibody present (n=114, 81.4%) as detected by the BioPlex method (Table 2). Anti-Ro antibody was the most commonly detected antibody, ranging from 50.7% to 62.9% of the cohort depending on the assay method (Table 2). Methods other than CIEP had a positivity rate of 1.4% to 5.0% for anti–Scl-70 antibody and 0.7% to 2.1% for anti–Jo-1 antibody.

The sensitivity of the BioPlex 2200 ANA Screen assay in the SLE cohort was 91.4%, which was comparable to that of IIF (90.7%; Table 3). The specificity of BioPlex among healthy controls was high, reaching 95.1%, compared with 85.4% for IIF, although the difference was not statistically significant. The agreement between BioPlex and IIF was moderate (κ=0.657). Eight patients tested positive by IIF, but negative by BioPlex. The IIF patterns of these cases were either weak homogeneous or weak fine-speckled. Nine patients were negative by IIF, but positive by BioPlex; these included two with a low titre of anti-dsDNA antibodies, one with anti-Sm antibodies, one with anti-RNP antibodies, and five with anti-Ro antibodies. Four patients tested ANA-negative by both methods; all four had a long-standing history of SLE (12-40 years). All had had severe disease manifestations, including cerebral lupus and lupus nephritis, and all had been taking powerful immunosuppressants for years, although the disease had become stable and inactive in recent years. Interestingly, they had been ANA-positive in the past. Possible changes in their serology after a long period of heavy immunosuppression for disease control may have accounted for the observed results.

In terms of agreement between different methods, BioPlex achieved the best agreement with ELISA, of >90% for the detection of most of the antibodies tested by ELISA (Table 4). The agreement between BioPlex and ELISA was 95.6% for anti-RNP/Sm antibodies (κ=0.89), 93.9% for anti-Sm (κ=0.79), 97.2% for anti-Ro (κ=0.94), and 95.6% for anti-La (κ=0.87). In contrast, the agreement between BioPlex and CIEP was not as strong. Agreement was 84.5% for detection of anti-RNP antibodies (κ=0.57), 85.6% for anti-Sm (κ=0.33), 93.9% for anti-Ro (κ=0.88), and 89.0% for anti-La (κ=0.6). Overall, the CIEP tended to agree better with the line blot assay than with ELISA or BioPlex.

Overall, BioPlex and ELISA had a higher sensitivity in detecting autoantibodies in the SLE cohort than the other two methods (Table 2). There were a few positive cases of anti–Scl-70 and anti–Jo-1 antibody detection in the cohort by all assays except CIEP, although the clinical significance of these antibodies in patients with SLE is uncertain.

In the healthy control group, CIEP had the highest specificity; none of the healthy subjects had anti-ENA autoantibodies when tested by CIEP (Table 2). With BioPlex, however, 2.4% (1/41) of the controls for each were positive for anti-La and anti–Scl-70 antibodies. For the line blot, if a weak borderline band were considered positive, then 4.9% (2/41) of the subjects were positive for anti-La antibodies (data not shown). If a weak borderline band were considered negative, then none of the healthy controls tested positive. For ELISA, if borderline results were counted negative then 2.4% (1/41) of the healthy controls still tested positive for anti-Sm antibodies.

Among the panel of anti-ENA autoantibodies tested, anti-Sm antibody had the best predictive value for the presence of lupus nephritis. However, the predictive value was method-dependent (Table 5). Anti-Sm antibody detection by CIEP had the best positive predictive value for lupus nephritis, reaching 87.5%. The specificity of anti-Sm antibody detection by CIEP for lupus nephritis was high, reaching 98.6%, although the sensitivity was only 10.4%. Anti-Sm antibody detection by BioPlex in nephritis had a higher sensitivity of 26.9%, however, the specificity and positive predictive value were lower than those achieved by CIEP (78.1% and 52.9%, respectively).

Anti-RNP antibody detection by CIEP had a specificity of 84.1% for Raynaud’s phenomenon, whereas the specificity by other methods was lower (69.2% for ELISA, 78.5% for line blot, and 65.4% for BioPlex). As the prevalence of Raynaud’s phenomenon in the cohort was not high, the positive predictive value was at best 37.0% only, by CIEP.

BioPlex generally had a higher sensitivity than the other methods, with the trade-off of lower specificity. However, CIEP generally performed better than BioPlex in disease-antibody associations (Table 6). The superiority of CIEP over BioPlex was most obvious in the diagnostic accuracy of linking anti-RNP antibody detection to Raynaud’s phenomenon (71.4% for CIEP vs 62.9% for BioPlex; P<0.001). Detection of antibodies to RNP (recombinant) and RNP/Sm by BioPlex did not differ significantly in diagnostic accuracy for association with Raynaud’s phenomenon.

In recent years, the multiplex method has been introduced in ANA testing. However, on the basis of the existing literature, this method is considered suboptimal in sensitivity compared with IIF, and its false-negative rate is similar to that of ELISA, ranging from 0.2% to 41.5% in the different populations studied.4 7 28 29 30 When Tozzoli et al31 compared the detection of ANAs between IIF using a 1:80 cut-off and BioPlex 2200 ANA Screen in a cohort of 95 SLE patients, they found that IIF had superior sensitivity over BioPlex (85/95 [89.5%] positive vs 77/95 [81.1%] positive, respectively). Generally, multiplex methods are considered to be simple to operate, have potential for automated and high-throughput processing, and can detect multiple specific antibodies simultaneously. Nonetheless, the main limitation is that such methods do not detect all the autoantibodies that can be detected by IIF. Hence, multiplex systems are considered to be insufficient in sensitivity and negative predictive value, and IIF remains the reference method of ANA testing.32 33

For our cohort, the BioPlex system demonstrated good sensitivity (91.4%), comparable to that of IIF (90.7%), with an agreement of 86.2% (κ=0.657). The specificity of BioPlex was slightly higher than that of IIF (95.1% vs 85.4%) but the difference did not reach statistical significance, perhaps because of the relatively small control group. Notably, a large proportion (6 of 9 cases) of the BioPlex-positive IIF-negative cases were actually positive for anti-Ro antibodies. Although IIF is the preferred method for ANA screening, as recommended by the ACR and the European Autoimmunity Standardisation Initiative, inconsistency among IIF assays exists.33 Slides from different vendors vary in sensitivity, especially for anti-SSA/Ro antibody detection.34 Moreover, the reading and interpretation of slides are reader- and skill-dependent.

Overall, BioPlex as used in our study showed a higher performance when compared with other studies. This difference could be due to different cohort characteristics, disease activities, and ethnicities. For example, the vast majority of the literature reports on studies of Caucasian populations, and studies of Chinese populations are scarce. In addition, we recruited only patients with SLE, but not other autoimmune diseases, thereby precluding direct comparisons. We also had a relatively small number of SLE cases; hence, some ANA staining patterns (eg, nuclear dots, proliferating cell nuclear antigen, nuclear lamina) were not encountered in the cohort, which limits the evaluation. Given the available literature and international recommendations, IIF remains the preferred method for ANA test until more supportive data for BioPlex are available.

In our study, CIEP performed best in terms of specificity, with none of the healthy controls testing positive for anti-ENA antibodies. In contrast, the specificity of the other platforms, especially ELISA and BioPlex, was less optimal, and positivity for antibodies to Sm (ELISA), Scl-70 (BioPlex), and La (BioPlex) was recorded. In addition, anti–Scl-70 and anti–Jo-1 antibodies were detected in assays other than CIEP in the SLE group. If appropriate disease controls, such as vasculitis, rheumatoid arthritis, and chronic infections are included, the performance of these assays would be better characterised.

There were several important limitations in our study. First, this was a cross-sectional study, and the clinical features and manifestations were retrospectively reviewed. The reviewer of the medical records was not blinded to the results of assays, which may have led to potential bias in record review and data extraction. Second, the performance of BioPlex was not evaluated in other rheumatic or autoimmune disease groups, which limits the generalisability of the results in other settings. Third, the number of participants included, especially that of healthy controls, was relatively small; disease controls were not included; and the controls were not age- and sex- matched with cases. These limitations may have led to bias in the evaluation of anti-ENA antibody assays. Finally, autoantibodies may precede clinical manifestations for years. A prospective study with parallel assessment of cases referred to the laboratory for ANA and anti-ENA antibody assessment by different techniques, as well as follow-up of the clinical manifestations and diagnosis, may provide a better assessment of the BioPlex system.

The BioPlex 2200 ANA Screen demonstrated comparable sensitivity to IIF in a local SLE cohort. The detection of specific antibodies, including those against ENAs, by the BioPlex system was more sensitive than that by CIEP, although with less specificity. Overall performance of BioPlex resembled that of the conventional ELISA technique, but with higher speed and turnaround time. Hence, BioPlex can be considered as a high-throughput ELISA-like assay for the detection of anti-ENA antibodies in SLE.

All authors have made substantial contributions to the concept or design of this study; acquisition of data; analysis or interpretation of data; drafting of the article; and critical revision for important intellectual content.

This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.

All authors have no conflicts of interest to disclose. All authors had full access to the data, contributed to the study, approved the final version for publication, and take responsibility for its accuracy and integrity.

This study was approved by The University of Hong Kong/Hospital Authority Hong Kong West Cluster Institutional Review Board (Ref No. UW14-442). The requirement for patient consent was waived by the ethics board.

1. Meroni PL, Schur PH. ANA screening: an old test with new recommendations. Ann Rheum Dis 2010;69:1420-2. Crossref

2. Clotet B, Guardia J, Pigrau C, et al. Incidence and clinical significance of anti-ENA antibodies in systemic lupus erythematosus. Estimation by counterimmunoelectrophoresis. Scand J Rheumatol 1984;13:15-20. Crossref

3. Hanly JG, Su L, Farewell V, Fritzler MJ. Comparison between multiplex assays for autoantibody detection in systemic lupus erythematosus. J Immunol Methods 2010;358:75-80. Crossref

4. Hanly JG, Thompson K, McCurdy G, Fougere L, Theriault C, Wilton K. Measurement of autoantibodies using multiplex methodology in patients with systemic lupus erythematosus. J Immunol Methods 2010;352:147-52. Crossref

5. Kim Y, Park Y, Lee EY, Kim HS. Comparison of automated multiplexed bead-based ANA screening assay with ELISA for detecting five common anti-extractable nuclear antigens and anti-dsDNA in systemic rheumatic diseases. Clin Chim Acta 2012;413:308-11. Crossref

6. Shovman O, Gilburd B, Barzilai O, et al. Evaluation of the BioPlex 2200 ANA screen: analysis of 510 healthy subjects: incidence of natural/predictive autoantibodies. Ann N Y Acad Sci 2005;1050:380-8. Crossref

7. Desplat-Jego S, Bardin N, Larida B, Sanmarco M. Evaluation of the BioPlex 2200 ANA screen for the detection of antinuclear antibodies and comparison with conventional methods. Ann N Y Acad Sci 2007;1109:245-55. Crossref

8. Orton SM, Peace-Brewer A, Schmitz JL, Freeman K, Miller WC, Folds JD. Practical evaluation of methods for detection and specificity of autoantibodies to extractable nuclear antigens. Clin Diagn Lab Immunol 2004;11:297-301. Crossref

9. Lock RJ, Unsworth DJ. Antibodies to extractable nuclear antigens. Has technological drift affected clinical interpretation? J Clin Pathol 2001;54:187-90. Crossref

10. Phan TG, Wong RC, Adelstein S. Autoantibodies to extractable nuclear antigens: making detection and interpretation more meaningful. Clin Diagn Lab Immunol 2002;9:1-7. Crossref

11. Kumar Y, Bhatia A, Minz RW. Antinuclear antibodies and their detection methods in diagnosis of connective tissue diseases: a journey revisited. Diagn Pathol 2009;4:1. Crossref

12. Emlen W, O’Neill L. Clinical significance of antinuclear antibodies: comparison of detection with immunofluorescence and enzyme-linked immunosorbent assays. Arthritis Rheum 1997;40:1612-8. Crossref

13. González C, Martin T, Arroyo T, García-Isidoro M, Navajo JA, González-Buitrago JM. Comparison and variation of different methodologies for the detection of autoantibodies to nuclear antigens (ANA). J Clin Lab Anal 1997;11:388-92. Crossref

14. Bruner BF, Guthridge JM, Lu R, et al. Comparison of autoantibody specificities between traditional and bead-based assays in a large, diverse collection of patients with systemic lupus erythematosus and family members. Arthritis Rheum 2012;64:3677-86. Crossref

15. Egner W. The use of laboratory tests in the diagnosis of SLE. J Clin Pathol 2000;53:424-32. Crossref

16. Wiik AS, Gordon TP, Kavanaugh AF, et al. Cutting edge diagnostics in rheumatology: the role of patients, clinicians, and laboratory scientists in optimizing the use of autoimmune serology. Arthritis Rheum 2004;51:291-8. Crossref

17. Tan EM, Cohen AS, Fries JF, et al. The 1982 revised criteria for the classification of systemic lupus erythematosus. Arthritis Rheum 1982;25:1271-7. Crossref

18. Petri M, Hellmann D, Hochberg M. Validity and reliability of lupus activity measures in the routine clinic setting. J Rheumatol 1992;19:53-9.

19. Urowitz MB, Gladman DD. Measures of disease activity and damage in SLE. Baillieres Clin Rheumatol 1998;12:405-13. Crossref

20. Gladman D, Ginzler E, Goldsmith C, et al. The development and initial validation of the systemic lupus international collaborating clinics/American College of Rheumatology damage index for systemic lupus erythematosus. Arthritis Rheum 1996;39:363-9. Crossref

21. Tan EM, Fritzler MJ, McDougal JS, et al. Reference sera for antinuclear antibodies. I. Antibodies to native DNA, Sm, nuclear RNP, and SS-B/La. Arthritis Rheum 1982;25:1003-5. Crossref

22. Isenberg DA, Maddison PJ. Detection of antibodies to double stranded DNA and extractable nuclear antigen. J Clin Pathol 1987;40:1374-81. Crossref

23. McCain GA, Bell DA, Chodirker WB, Komar RR. Antibody to extractable nuclear antigen in the rheumatic diseases. J Reumatol 1978;5:399-406.

24. Hamburger M, Hodes S, Barland P. The incidence and clinical significance of antibodies to extractable nuclear antigens. Am J Med Sci 1977;273:21-8. Crossref

25. Clark G, Reichlin M, Tomasi TB Jr. Characterization of a soluble cytoplasmic antigen reactive with sera from patients with systemic lupus erythematosus. J Immunol 1969;102:117-22.

26. Tan EM, Kunkel HG. Characteristics of a soluble nuclear antigen precipitating with sera of patients with systemic lupus erythematosus. J Immunol 1966;96:464-71.

27. Bossuyt PM, Reitsma JB, Bruns DE, et al. STARD 2015: an updated list of essential items for reporting diagnostic accuracy studies. BMJ 2015;351:h5527. Crossref

28. Jaskowski TD, Schroder C, Martins TB, Mouritsen CL, Litwin CM, Hill HR. Screening for antinuclear antibodies by enzyme immunoassay. Am J Clin Pathol 1996;105:468-73. Crossref

29. Damoiseaux J, Vaessen M, Knapen Y, et al. Evaluation of the FIDIS vasculitis multiplex immunoassay for diagnosis and follow-up of ANCA-associated vasculitis and Goodpasture’s disease. Ann N Y Acad Sci 2007;1109:454-63. Crossref

30. Bonilla E, Francis L, Allam F, et al. Immunofluorescence microscopy is superior to fluorescent beads for detection of antinuclear antibody reactivity in systemic lupus erythematosus patients. Clin Immunol 2007;124:18-21. Crossref

31. Tozzoli R, Bonaguri C, Melegari A, Antico A, Bassetti D, Bizzaro N. Current state of diagnostic technologies in the autoimmunology laboratory. Clin Chem Lab Med 2013;51:129-38. Crossref

32. Op De Beéck K, Vermeersch P, Verschueren P, et al. Antinuclear antibody detection by automated multiplex immunoassay in untreated patients at the time of diagnosis. Autoimmun Rev 2012;12:137-43. Crossref

33. Agmon-Levin N, Damoiseaux J, Kallenberg C, et al. International recommendations for the assessment of autoantibodies to cellular antigens referred to as anti-nuclear antibodies. Ann Rheum Dis 2014;73:17-23. Crossref

34. Copple SS, Giles SR, Jaskowski TD, Gardiner AE, Wilson AM, Hill HR. Screening for IgG antinuclear autoantibodies by HEp-2 indirect fluorescent antibody assays and the need for standardization. Am J Clin Pathol 2012;137:825-30. Crossref

Breast cancer is the most common malignant tumour and the leading cause of cancer-related deaths among females in developing countries.1 Breast cancer displays heterogeneity: it comprises distinct pathologies and histological features and can have different chemotherapy responses and clinical outcomes.2 The identification of tumour-related factors that can predict tumour behaviour is important. Predictive factors can help identify as early as possible not only patients who have a high risk of recurrence and metastasis, but also patients who can benefit from different types of adjuvant therapy.3 Conventional predictive factors of high risk of recurrence and metastasis include relatively large (>5 cm) tumour size and high nuclear grade; negativity for oestrogen receptor and progesterone receptor; human epidermal growth factor receptor 2 (HER2) overexpression; and increased lymph node involvement at the time of breast cancer diagnosis.4 Recent advances in genetic profiling of tumours have extended our understanding of breast cancer biology and have allowed the use of several prognostic gene signatures to select patients at highest risk of early recurrence and those who may benefit from certain adjuvant treatment.2 5 6 However, despite receiving standard treatments routinely guided by predictive factors, more than 30% of breast cancer patients develop metastatic disease and have poor survival.7 8 As such, it is essential and urgent to identify reliable predictive factors to assist in diagnosis, staging, evaluation of recurrence risk, and development of new treatment modalities.

Cluster of differentiation 147 (CD147), a transmembrane glycoprotein that belongs to the immunoglobulin superfamily, can promote tumour invasion and metastasis, and mediate breast cancer drug resistance.9 10 11 12 13 Expression of CD147 is significantly correlated with axillary lymph node involvement; tumour, node, and metastasis staging; and shorter progression-free survival and overall survival.14 Previous data demonstrated that CD147 exists in both membrane-bound and soluble forms in many solid tumours, and soluble CD147 (sCD147) can be detected in the conditioned medium of tumour cells and peripheral blood of cancer patients.15 16 17 Overexpression of the CD147 gene in human breast cancer cells can increase the sCD147 level, indicating that sCD147 release is correlated with the degree of CD147 expression in tumour cells.15 16 17 Full-length CD147 may be exported into the microenvironment from tumour cells by microvesicle shedding or by matrix metalloproteinase (MMP)–dependent cleavage, thereby stimulating MMP expression in fibroblasts.18 19 20 In turn, sCD147 derived from tumour cells acts in a paracrine fashion on stromal cells that are both adjacent and distant to tumour sites, so as to further stimulate the production of MMPs and CD147. This additional CD147 consequently contributes to tumour angiogenesis, tumour growth, and metastasis.16 21 Importantly, several studies investigating the role of sCD147 level in patients with tumours have suggested that sCD147 may offer a useful approach in diagnosis, as it is correlated with disease severity.15 22 However, little is known about the level of sCD147 in patients with breast cancer. Furthermore, the biological and clinical significance of sCD147 in breast cancer has not been investigated.

In this study, we measured plasma sCD147 levels in patients with breast cancer and evaluated the results with respect to clinicopathological factors. We aimed to demonstrate the association between plasma sCD147 levels with tumour size, lymph node metastasis, recurrence risk, and chemoresistance in breast cancer patients.

The results of this study are presented in accordance with the reporting recommendations for tumour marker prognostic studies.23 We conducted the study between February 2014 and February 2017 in the Affiliated Xiangya Hospital of Central South University in Changsha of Hunan Province, China. We collected peripheral blood samples from consecutive patients with breast cancer, including primary breast cancer, during their first hospital admission. To be eligible for this study, patients had to be adult females who had no other malignant diseases or severe systemic diseases, especially rheumatic, inflammatory, and cardiovascular diseases. The peripheral blood of consecutive patients with palpable benign breast masses, including fibroadenoma and adenopathy, was also collected to serve as control samples during the same period. All blood samples were centrifuged at 3000 rpm at 4°C for 5 minutes, and the plasma samples were stored at -70°C for later plasma sCD147 testing. All the patients’ clinicopathological findings were supplied by the Xiangya Hospital of Central South University. Breast cancer subtypes were identified according to the St Gallen Consensus 2013 classification system.24 Recurrence risk of breast cancer was evaluated according to the St Gallen Consensus 2007 criteria.25

The association between chemotherapy response and plasma sCD147 level was retrospectively analysed. The patients included in this analysis had to meet all of the following criteria: (1) had a confirmed diagnosis of invasive ductal breast carcinoma by pathology and had consented to undergo neoadjuvant chemotherapy; (2) had operable breast cancer consisting of a large tumour (>2 cm) that fulfilled the criteria for breast conserving surgery except tumour size, or triple-negative breast cancer (TNBC; ie, negative for oestrogen/progesterone receptors and HER2) with small (T1 stage) tumours; (3) had received no previous treatment; (4) had received only four cycles of pirarubicin-cyclophosphamide/epirubicin-cyclophosphamide (AC/EC)–based neoadjuvant chemotherapy before surgery; and (5) had complete hospital records that included evaluation of chemotherapy efficacy. Clinical response to AC/EC-based chemotherapy was evaluated by the decrease in tumour size and classified according to response evaluation criteria in solid tumours (RECIST criteria).26 Patients with complete remission or partial remission were classified as chemotherapy-sensitive, whereas patients with stable disease or progressive disease were classified as chemotherapy-resistant.

The concentrations of plasma sCD147 were measured by enzyme-linked immunosorbent assay (ELISA). Plasma sCD147 levels were assessed using the EMMPRIN/CD147 ELISA kit (R&D Systems, Minneapolis [MN], US) according to the manufacturer’s protocol. The concentration of the sample in each ELISA well was determined by interpolation from a standard curve. Each sample was tested in duplicate.

The Mann-Whitney U test was used to compare levels of plasma sCD147 in different groups according to variable clinicopathological factors. The Chi squared contingency test with Yates correction was used to determine the relationship between clinicopathological factors of breast cancer patients and lymph node status or chemotherapy sensitivity. Binary logistic regression was used to assess clinicopathological factors (plasma sCD147, tumour size, and HER2) that were associated with lymph node metastasis or chemoresistance in invasive breast cancer. All multivariable logistic regression models used backward stepwise procedures, and only datasets complete for every outcome analysed were used. Receiver operating characteristic (ROC) curve analysis was performed to calculate the area under the curve and evaluate the optimal cut-off point, which was given by the maximum of the Youden index. Statistical significance was set at P<0.05. The GraphPad Prism 6 software (GraphPad Software, La Jolla [CA], US) and SPSS (Windows version 19.0; IBM Corp, Armonk [NY], US) were used for statistical analysis.

Among all eligible patients with complete records, 165 had benign breast disease (age range, 22-68 years) and 308 had primary breast cancer (age range, 24-77 years). There was no significant difference in age between the two groups (P=0.381). Breast cancer patients comprised 11 with ductal carcinoma in situ and 297 with invasive ductal carcinoma. Retrospective analysis of the association of plasma sCD147 level with response to neoadjuvant chemotherapy included 175 patients who met all the inclusion criteria (Fig)—luminal A in 39, luminal B in 70, HER2-positive in 28, and TNBC in 38. In all, 170 patients had T2-T4 tumours and five had T1 TNBC tumours. Using the RECIST criteria, we assigned the 175 patients to two groups: chemotherapy-sensitive (n=126) and chemotherapy-resistant (n=49).

According to ELISA results, plasma sCD147 levels were significantly higher in patients with primary breast cancer than in those with benign breast disease (median [interquartile range; IQR], 8629.81 pg/mL [7426.33-10 309.20 pg/mL] vs 7625.99 pg/mL [6739.20-9140.04 pg/mL]; P=0.001). However, there was no significant difference in plasma sCD147 levels between patients with invasive breast cancer and those with ductal carcinoma in situ (8618.91 pg/mL [7404.81-10 358.50 pg/mL] vs 9185.79 pg/mL [7671.15-9626.47 pg/mL]; P=0.787). Regarding cancer subtypes of the 297 patients with invasive breast carcinoma, median (IQR) plasma sCD147 levels were significantly higher in patients with HER2-positive breast cancer (10 042.34 pg/mL [7772.01-11 058.48 pg/mL]) than in those with luminal A tumours (7991.05 pg/mL [7101.72-10 237.4 pg/mL]; P=0.007), luminal B tumours (8629.81 pg/mL [7200.45-9953.32 pg/mL]; P=0.017), and TNBC tumours (8585.16 pg/mL [7884.27-10 545.51 pg/mL]; P=0.027).

The association between plasma sCD147 level and clinicopathological factors in patients with invasive breast cancer is summarised in Table 1. Plasma sCD147 levels increased with tumour size: median (IQR) levels were significantly higher in patients with locally advanced (stage T3-T4) than those with early (stage T1-T2) breast cancer (10 093.26 pg/mL [7974.73-11 451.21 pg/mL] vs 8561.45 pg/mL [7169.41-9952.90 pg/mL]; P=0.001). Plasma sCD147 levels were also elevated in patients with lymph node metastasis compared with those without (median [IQR], 9991.42 pg/mL [8154.61-11 452.84 pg/mL] vs 7814.78 pg/mL [6936.82-9516.85 pg/mL]; P<0.001). In addition, plasma sCD147 levels were significantly higher in patients with a high risk of recurrence than in those with a medium risk (median [IQR], 10 093.26 pg/mL [8135.35-11 679.71 pg/mL] vs 8134.68 pg/mL [7151.41-9616.68 pg/mL]; P<0.001). Although plasma sCD147 levels were elevated for the HER2- positive breast cancer subtype as compared with other breast cancer subtypes, there was no significant difference between HER2-positive and HER2-negative patients (median [IQR], 9254.34 pg/mL [7157.63-11 199.38 pg/mL] vs 8568.83 pg/mL [7448.34-10 070.42 pg/mL]; P=0.160).

Because plasma sCD147 was associated with lymph node status and recurrent risk, we speculated that plasma sCD147 may be a predictor for lymph node metastasis of breast cancer. Univariate analysis showed that tumour size and HER2 status may be involved in lymph node metastasis (Table 2). We subsequently used binary logistic regression analysis to identify clinicopathological factors associated with lymph node metastasis in invasive breast cancer. Our data showed that plasma sCD147 (P<0.001), HER2-positive tumours (P=0.001), and tumour size T3-T4 (P=0.005) were independent predictors of lymph node metastasis of breast cancer (Table 3). When we analysed ROC curves to evaluate use of plasma sCD147 as a diagnostic biomarker for lymph node metastasis, the area under the curve was 0.745 (95% confidence interval, 0.676-0.813) and the optimal cut-off point of plasma sCD147 was 8577 pg/mL, which provided a sensitivity of 70.9% and a specificity of 61.7%.

Table 1 shows that plasma sCD147 levels in the chemotherapy-resistant group were significantly higher than those in the chemotherapy-sensitive group (median [IQR], 10 093.26 pg/mL [7974.73-11 261.88 pg/mL] vs 8585.16 pg/mL [7789.74-9868.87 pg/mL]; P=0.040). Univariate analysis revealed that tumour size and HER2 status may be involved in chemotherapy response (Table 4). Binary logistic regression analysis demonstrated that plasma sCD147 was not an independent predictor for chemotherapy response of breast cancer patients, but tumour size of T3-T4 was (P=0.001) [Table 5].

The tumour microenvironment plays a proactive role in malignant disease progression, including the transition from ductal carcinoma in situ to invasive cancer, tumour cell proliferation, dissemination, and metastasis.27 CD147 has been found to be overexpressed in breast cancer, associated with tumour size and staging, and predictive of poor prognosis.28 29 30 31 Tumour cells express molecules, either secreted or presented on the cell surface, that interact with surrounding stromal cells. Soluble CD147 may be released from membrane-associated CD147 as a result of both MMP proteolytic activity and microvesicle shedding in the tumour microenvironment. Soluble CD147 may then act in a paracrine fashion on stromal cells to further trigger production of MMPs and CD147; the latter contributes to tumour angiogenesis, tumour growth, and metastasis.16 19 21

Wu et al15 reported that serum sCD147 enhances the secretion of MMP-2 from hepatocellular carcinoma cells by activating extracellular signal-regulated kinase and focal adhesion kinase, as well as phosphoinositide-3-kinase/Akt signalling, indicating that sCD147 may contribute to hepatocellular carcinoma progression. Moreover, serum sCD147 was elevated in patients with hepatocellular carcinoma compared with healthy individuals, and sCD147 level was associated with tumour size and Child-Pugh score.15 Gross et al22 also reported that sCD147 levels were elevated in patients with multiple myeloma, and elevated levels were associated with refractory disease and shortened progression-free survival, indicating that sCD147 may be a new prognostic factor for patients with multiple myeloma.

A previous study demonstrated that CD147 was overexpressed in human breast cancer.10 In this study, we measured plasma sCD147 levels by ELISA and found that plasma sCD147 levels were significantly elevated in breast cancer patients compared with control patients who had benign breast diseases. We also found that plasma sCD147 was significantly elevated in lymph node metastasis in breast cancer patients. Taken together, these data show that plasma sCD147 may be released from tumour cells and promote lymph node metastasis of breast cancer. Some studies have reported that sCD147 has been detected in patients with inflammatory diseases31 or cardiovascular diseases.32 33 To eliminate interference from other diseases and conditions, we excluded patients with inflammatory or cardiovascular diseases and ensured patients in each group had a similar age distribution.

Previous studies have shown that membrane-bound CD147 may correlate with HER2 expression. Yan et al34 reported that CD147 induces angiogenesis by stimulating vascular endothelial growth factor production, invasiveness by stimulating MMP production, and multidrug resistance by hyaluronan-mediated upregulation of HER2 signalling. Xue et al30 reported that CD147 expression was positively correlated with HER2 overexpression. In a recent study, CD147 knockdown was shown to improve the antitumour efficacy of trastuzumab in HER2-positive breast cancer cells.35 Although we found that plasma sCD147 levels were significantly higher in the HER2-positive breast cancer subtype than in luminal A, luminal B, and TNBC subtypes, plasma sCD147 had no association with expression of HER2 or oestrogen/progesterone receptors in breast cancer. The reason for this finding is that there are four breast cancer subtypes—luminal A, luminal B, HER2-positive, and TNBC—according to oestrogen/progesterone receptor, HER2, and Ki67 status. The luminal B subtype includes some breast cancers that are positive for oestrogen/progesterone receptor and HER2. Hence, patients who are HER2-positive (Table 1) include those with HER2-positive subtype and also luminal B subtype; plasma sCD147 levels in patients who were ‘HER2-positive’ were different from those with a HER2-positive subtype.

It is essential to establish predictive factors to allow evaluation of the recurrence risk of breast cancer, so that optimal adjuvant therapy can be selected for individual patients.3 36 Larger tumour size at diagnosis, high proliferation factors, absence of oestrogen/progesterone receptors and HER2 overexpression, and lymph node metastasis are related to a high risk of recurrence and poor survival, and are commonly recognised as prognostic and predictive factors for breast cancer recurrence risk.4 37 38 Consistent with these findings, we found that plasma sCD147 levels were significantly increased in patients with locally advanced lymph node metastasis and a high risk of breast cancer recurrence. We also found that plasma sCD147 was positively associated with tumour size, lymph node metastasis, and high recurrence risk of invasive breast cancer.

Lymph node status, which confers different strategies for patients at different tumour stages, is critical information for the treatment of breast cancer, and the accurate prediction of lymph node status is a prerequisite for treatment decision. Our binary logistic regression analysis showed that plasma sCD147, HER2 positive subtype, and tumour size (T3-T4) were independent predictors for lymph node metastasis of breast cancer patients. Taken together, these data suggest that plasma sCD147 may be a new factor for the evaluation of breast cancer recurrence risk. Our ROC analysis demonstrated that plasma sCD147 could be a biomarker for distinguishing breast cancer patients with lymph node metastasis from those without; however, the sensitivity and specificity were not high (70.9% and 61.7%, respectively). The relatively low sensitivity and specificity suggest that using plasma sCD147 as the sole biomarker may result in substantial numbers of false positives and false negatives. Therefore, it may be necessary to investigate whether the combination of plasma sCD147 and other biomarkers can improve efficacy.

According to the data of 303 patients who were followed up for 3 to 38 months (median, 20 months), 11 patients had relapse: two had local recurrences and nine had distant metastases. The mean time of recurrence/metastasis was 23.6 months, with no difference between patients with relapse and those without (Table 6). We were not able to investigate the relationship between plasma sCD147 and disease-free survival or overall survival, because of the short median follow-up period.

Previous data have shown that CD147 is one of the apoptosis-related proteins and it may mediate adriamycin chemoresistance in breast cancer by affecting the cellular localisation and dimerisation of the protein ABCG2 (ATP-binding cassette subfamily G member 2).10 In this study, we studied the relationship between plasma sCD147 and chemotherapy response in invasive breast cancer. All patients were given four cycles of AC/EC-based chemotherapy. We also found that plasma sCD147 levels were significantly higher in the chemotherapy-resistant group than in the chemotherapy-sensitive group, and such levels were positively associated with chemotherapy resistance. Although our data also showed that plasma sCD147, tumour size (T3-T4), and HER2 positive subtype may be involved in chemotherapy response, binary logistic regression demonstrated that tumour size (T3-T4) was an independent predictor for chemotherapy response of breast cancer patients, but plasma sCD147 was not. Owing to the small number of cases in the chemotherapy-resistant group, the statistical analysis of data may be underpowered.

In addition to the small sample of study and short median follow-up period, there were other limitations in this study. This study was conducted in one centre, and the researchers who extracted the data and conducted the analysis were not blinded to the study hypothesis. There may have increased selection and information bias. Furthermore, as the design of this study was relatively simple, there may be insufficient control for potential confounding factors in the multivariable analysis.

In conclusion, our study found that plasma sCD147 levels were elevated in breast cancer patients compared with controls with benign breast disease, and plasma sCD147 level was associated with tumour size, lymph node metastasis, high recurrence risk, and AC/EC-based chemoresistance. Moreover, our study supports that plasma sCD147 is an independent predictive factor for lymph node metastasis and is a feasible marker to distinguish breast cancer patients with lymph node metastasis from patients without.

Concept or design: LL Tang, LQ Liao.
Acquisition of data: YJ Liu, YH Kuang, SM Wang, GJ Yan.
Analysis or interpretation of data: LL Tang, LQ Liao.
Drafting of the article: YH Kuang, LQ Liao.
Critical revision for important intellectual content: YH Kuang, LQ Liao.
YH Kuang, YJ Liu, and LL Tang contributed equally to this study.

This study was supported by two grants from the National Natural Science Foundation of China (No. 81101654, awarded to LQ Liao, and No. 81573049, awarded to YH Kuang).

The authors have no conflicts of interest to disclose. All authors had full access to the data, contributed to the study, approved the final version for publication, and take responsibility for its accuracy and integrity.

The research protocols for the use of human tissue were approved by and conducted in accordance with the policies of the Institutional Review Boards at Central South University (Ref No. 201403152), which were formulated based on the 1964 Helsinki Declaration and its later amendments. Written informed consent was obtained from all participants.

1. Torre LA, Bray F, Siegel RL, et al. Global cancer statistics, 2012. CA Cancer J Clin 2015;65:87-108. Crossref

2. Rivenbark AG, O’Connor SM, Coleman WB. Molecular and cellular heterogeneity in breast cancer: challenges for personalized medicine. Am J Pathol 2013;183:1113-24. Crossref

3. Goldhirsch A, Wood WC, Coates AS, et al. Strategies for subtypes—dealing with the diversity of breast cancer: highlights of the St. Gallen International Expert Consensus on the Primary Therapy of Early Breast Cancer 2011. Ann Oncol 2011;22:1736-47. Crossref

4. Soerjomataram I, Louwman MW, Ribot JG, et al. An overview of prognostic factors for long-term survivors of breast cancer. Breast Cancer Res Treat 2008;107:309-30. Crossref

5. Adaniel C, Jhaveri K, Heguy A, et al. Genome-based risk prediction for early stage breast cancer. Oncologist 2014;19:1019-27. Crossref

6. Weigelt B, Peterse JL, van ’t Veer LJ. Breast cancer metastasis: markers and models. Nat Rev Cancer 2005;5:591-602. Crossref

7. Redig AJ, McAllister SS. Breast cancer as a systemic disease: a view of metastasis. J Intern Med 2013;274:113-26. Crossref

8. O’Shaughnessy J. Extending survival with chemotherapy in metastatic breast cancer. Oncologist 2005;10 Suppl 3:20-9. Crossref

9. Kuang YH, Chen X, Su J, et al. RNA interference targeting the CD147 induces apoptosis of multi-drug resistant cancer cells related to XIAP depletion. Cancer Lett 2009;276:189-95. Crossref

10. Zhou S, Liao L, Chen C, et al. CD147 mediates chemoresistance in breast cancer via ABCG2 by affecting its cellular localization and dimerization. Cancer Lett 2013;337:285-92. Crossref

11. Yang JM, Xu Z, Wu H, et al. Overexpression of extracellular matrix metalloproteinase inducer in multidrug resistant cancer cells. Mol Cancer Res 2003;1:420-7.

12. Marieb EA, Zoltan-Jones A, Li R, et al. Emmprin promotes anchorage-independent growth in human mammary carcinoma cells by stimulating hyaluronan production. Cancer Res 2004;64:1229-32. Crossref

13. Nabeshima K, Iwasaki H, Koga K, et al. Emmprin (basigin/CD147): matrix metalloproteinase modulator and multifunctional cell recognition molecule that plays a critical role in cancer progression. Pathol Int 2006;56:359-67. Crossref

14. Zhao S, Ma W, Zhang M, et al. High expression of CD147 and MMP-9 is correlated with poor prognosis of triple-negative breast cancer (TNBC) patients. Med Oncol 2013;30:335. Crossref

15. Wu J, Hao ZW, Zhao YX, et al. Full-length soluble CD147 promotes MMP-2 expression and is a potential serological marker in detection of hepatocellular carcinoma. J Transl Med 2014;12:190. Crossref

16. Tang Y, Kesavan P, Nakada MT, et al. Tumor-stroma interaction: positive feedback regulation of extracellular matrix metalloproteinase inducer (EMMPRIN) expression and matrix metalloproteinase-dependent generation of soluble EMMPRIN. Mol Cancer Res 2004;2:73-80.

17. Bordador LC, Li X, Toole B, et al. Expression of emmprin by oral squamous cell carcinoma. Int J Cancer 2000;85:347-52.

18. Taylor PM, Woodfield RJ, Hodgkin MN, et al. Breast cancer cell-derived EMMPRIN stimulates fibroblast MMP2 release through a phospholipase A(2) and 5-lipoxygenase catalyzed pathway. Oncogene 2002;21:5765-72. Crossref

19. Sidhu SS, Mengistab AT, Tauscher AN, et al. The microvesicle as a vehicle for EMMPRIN in tumor-stromal interactions. Oncogene 2004;23:956-63. Crossref

20. Egawa N, Koshikawa N, Tomari T, et al. Membrane type 1 matrix metalloproteinase (MT1-MMP/MMP-14) cleaves and releases a 22-kDa extracellular matrix metalloproteinase inducer (EMMPRIN) fragment from tumor cells. J Biol Chem 2006;281:37576-85. Crossref

21. Hanata K, Yamaguchi N, Yoshikawa K, et al. Soluble EMMPRIN (extra-cellular matrix metalloproteinase inducer) stimulates the migration of HEp-2 human laryngeal carcinoma cells, accompanied by increased MMP-2 production in fibroblasts. Arch Histol Cytol 2007;70:267-77. Crossref

22. Gross Z, Udd K, Ghermezi M, et al. Serum CD147 levels are increased in multiple myeloma patients and elevated levels are associated with refractory disease and shortened progression free survival. Am Soc Hematology 2016;128:5652.

23. McShane LM, Altman DG, Sauerbrei W, et al. Reporting recommendations for tumor marker prognostic studies (REMARK). J Natl Cancer Inst 2005;97:1180-4. Crossref

24. Goldhirsch A, Winer EP, Coates AS, et al. Personalizing the treatment of women with early breast cancer: highlights of the St Gallen International Expert Consensus on the Primary Therapy of Early Breast Cancer 2013. Ann Oncol 2013;24:2206-23. Crossref

25. Goldhirsch A, Wood WC, Gelber RD, et al. Progress and promise: highlights of the international expert consensus on the primary therapy of early breast cancer 2007. Ann Oncol 2007;18:1133-44. Crossref

26. Eisenhauer EA, Therasse P, Bogaerts J, et al. New response evaluation criteria in solid tumours: revised RECIST guideline (version 1.1). Eur J Cancer 2009;45:228-47. Crossref

27. Liotta LA, Kohn EC. The microenvironment of the tumour-host interface. Nature 2001;411:375-9. Crossref

28. Dalberg K, Eriksson E, Enberg U, et al. Gelatinase A, membrane type 1 matrix metalloproteinase, and extracellular matrix metalloproteinase inducer mRNA expression: correlation with invasive growth of breast cancer. World J Surg 2000;24:334-40. Crossref

29. Reimers N, Zafrakas K, Assmann V, et al. Expression of extracellular matrix metalloproteases inducer on micrometastatic and primary mammary carcinoma cells. Clin Cancer Res 2004;10:3422-8. Crossref

30. Xue S, Li SX, Wu ZS, et al. Expression of CD147, matrix metalloproteinases and transforming growth factor beta1 in breast cancer [in Chinese]. Zhonghua Bing Li Xue Za Zhi 2009;38:524-8.

31. Yanaba K, Asano Y, Tada Y, et al. Increased serum soluble CD147 levels in patients with systemic sclerosis: association with scleroderma renal crisis. Clin Rheumatol 2012;31:835-9. Crossref

32. Major TC, Liang L, Lu X, et al. Extracellular matrix metalloproteinase inducer (EMMPRIN) is induced upon monocyte differentiation and is expressed in human atheroma. Arterioscler Thromb Vasc Biol 2002;22:1200-7. Crossref

33. Schmidt R, Bultmann A, Fischel S, et al. Extracellular matrix metalloproteinase inducer (CD147) is a novel receptor on platelets, activates platelets, and augments nuclear factor kappaB-dependent inflammation in monocytes. Circ Res 2008;102:302-9. Crossref

34. Yan L, Zucker S, Toole BP. Roles of the multifunctional glycoprotein, emmprin (basigin; CD147), in tumour progression. Thromb Haemost 2005;93:199-204.

35. Xiong L, Ding L, Ning H, et al. CD147 knockdown improves the antitumor efficacy of trastuzumab in HER2-positive breast cancer cells. Oncotarget 2016;7:57737-51. Crossref

36. Cianfrocca M, Goldstein LJ. Prognostic and predictive factors in early-stage breast cancer. Oncologist 2004;9:606-16. Crossref

37. Harris L, Fritsche H, Mennel R, et al. American Society of Clinical Oncology 2007 update of recommendations for the use of tumor markers in breast cancer. J Clin Oncol 2007;25:5287-312. Crossref

38. Taneja P, Maglic D, Kai F, et al. Classical and novel prognostic markers for breast cancer and their clinical significance. Clin Med Insights Oncol 2010;4:15-34. Crossref