Diagnostic test accuracy of anti-glycopeptidolipid-core IgA antibodies for Mycobacterium avium complex pulmonary disease: systematic review and meta-analysis

Abstract

Currently, an anti-glycopeptidolipid (GPL)-core IgA antibody assay kit for diagnosing Mycobacterium avium complex (MAC) is commercially available. We conducted this systematic review and meta-analysis to reveal the precise diagnostic accuracy of anti-GPL-core IgA antibodies for MAC pulmonary disease (MAC-PD). We systematically searched reports that could provide data for both sensitivity and specificity by anti-GPL-core IgA antibody for clinically diagnosed MAC-PD. Diagnostic test accuracy was estimated using the bivariate model. Of the 257 articles that we had found through primary search, we finally included 16 reports consisted of 1098 reference positive subjects and 2270 reference negative subjects. The diagnostic odds ratio was 24.8 (95% CI 11.6-52.8, I² = 5.5%) and the area under the hierarchical summary receiver operating characteristic curves was 0.873 (95% CI 0.837-0.913). With a cutoff value of 0.7 U/mL, the summary estimates of sensitivity and specificity were 0.696 (95% CI 0.621-0.761) and 0.906 (95% CI 0.836-0.951), respectively. The positive and negative likelihood ratios were 7.4 (95% CI 4.1-13.8) and 0.34 (95% CI 0.26-0.43), respectively. The demanding clinical diagnostic criteria may be a cause of false positive of the index test. The index test had good overall diagnostic accuracy and was useful to ruling in MAC-PD with the cutoff value.

Diagnostic test accuracy of anti-glycopeptidolipid-core IgA antibodies for Mycobacterium avium complex pulmonary disease: systematic review and meta-analysis

OPEN

R

P

Yuji Shibata, Nobuyuki Horita, MasakiYamamoto, ToshinoriTsukahara, Hideyuki Nagakura, KenTashiro, HirokiWatanabe, Kenjiro Nagai, Kentaro Nakashima, Ryota Ushio,Misako Ikeda, Atsuya Narita, Akinori Kanai, Takashi Sato & Takeshi Kaneko

Currently, an anti-glycopeptidolipid (GPL)-core IgA antibody assay kit for diagnosing Mycobacterium avium complex (MAC) is commercially available. We conducted this systematic review and meta-analysis to reveal the precise diagnostic accuracy of anti-GPL-core IgA antibodies for MAC pulmonary disease (MAC-PD). We systematically searched reports that could provide data for both sensitivity and =

may be a cause of false positive of the index test. The index test had good overall diagnostic accuracy

Mycobacterium (M.) avium complex (MAC) is the most common type of non-tuberculosis mycobacterium and oen causes chronic pulmonary disease in both immunocompromised and immunocompetent persons. MAC infection usually presents as pulmonary disease but can involve other organs1. Because the worldwide prevalence of MAC has greatly increased during the past three decades2, nowadays, clinicians oen need to diagnose MAC pulmonary disease (PD) in daily practice. Even though we have limited evidence of the eectiveness of antibiotics treatment for MAC, rapid and accurate diagnosis is crucial for deciding treatment and follow-up plans3,4.

The microbiological diagnostic criteria of MAC-PD in the American Thoracic Society and Infectious Diseases Society of America 2007 Statement (ATS/IDSA 2007 statement) requires two culture-positive sputum samples or one culture positive bronchial lavage4. However, the diagnosis of MAC by culture has some limitations. First, the sputum culture is not sufficiently sensitive especially for solitary-nodule cases. Second, the mycobacterium culture needs a long incubation time. Third, we should consider the risk of contamination because MAC is a ubiquitous bacterium found in the environment.

Glycopeptidolipid (GPL) core, which is composed of fatty acid, three amino acids, and rhamnose, is a common major cell wall component of the M. avium and M. intracellulare5. The majority of the other mycobacterium

Yokohama, Japan. Correspondence and requests for materials should be addressed to N.H. (email: nobuyuki_ [email protected])

SCIENTIFIC REPORTS

1

www.nature.com/scientificreports/

species do not have this core component except for rapidly growing mycobacteria (RGM), namely M. abscessus, M. chelonae, and M. fortuitum5. After comparing diagnostic power of IgG, IgA, and IgM antibodies for anti-GPL-core, IgA antibody was selected as a candidate target of enzyme immune assay to diagnose MAC-PD6.

Currently, an anti-GPL-core IgA antibody assay kit for MAC is commercially available5. Some studies have reported the diagnostic test accuracy of the test722. However, there was considerable inconsistency in the results of these studies. We conducted this systematic review and meta-analysis to reveal the precise diagnostic accuracy of anti-GPL-core IgA antibodies for MAC-PD.

Methods

Study registration. The protocol of the current systematic review was draed following the Preferred Reporting Items for Systematic Reviews and Meta- Analyses (PRISMA) statement and the Cochrane Handbook for Diagnostic Test Accuracy Reviews23,24. It has been registered with the international prospective register of systematic reviews (PROSPERO) as number CRD4201603544925.

Eligibility criteria. Type of studies and participants. We included both one-gate and two-gate studies24.

A so-called cohort study that included only MAC-PD-suspected cases had one gate. A so-called case-control study that included both MAC-PD cases and non-MAC-PD cases had two gates. In addition, a study that included MAC-PD-suspected cases and MAC-PD cases and a study that included both MAC-PD-suspected cases and non-MAC-PD had two gates. The two-gate studies had a high risk of bias concerning patient selection26.

Index test. We evaluated anti-GPL-core IgA antibody for MAC-PD as an index test. Along with a commercialized kit, Capilia MAC Ab ELISA (TAUNS, Shizuoka, Japan), in-house assays were also allowed. Sensitivity and specicity were evaluated using a cuto value of 0.7U/mL27.

Reference test. Reference diagnoses based on ATS/IDSA 2007 statement and American Thoracic Society 1997 statement were preferred4,28. However, other clinical diagnoses were also accepted.

Outcomes. After extracting the number of subjects with true positives/false negatives/false positives/true negatives, we composed a two-by-two contingency table. Overall diagnostic test accuracy was assessed using a diagnostic odds ratio (DOR), and the area under the hierarchical summary receiver operating characteristic (HSROC) curves (AUC). The summary estimates of sensitivity and specicity, positive likelihood ratio (PLR), and negative likelihood ratio (NLR) were also assessed. Positive predictive value and negative predictive value were obtained across the pretest probability ranging from 0% to 100%.

Literature search strategy. In the electronic search, we systematically searched Pubmed, EMBASE, the Cochrane Library, and Web of Science on February 18th, 2016. We used the following search formula for Pubmed without limitation: ((Mycobacterium avium complex) OR (Mycobacterium avium-complex) OR MAC OR MAC-PD OR (non-tuberculosis Mycobacterium)) AND (glycopeptidolipid OR anti-glycopeptidolipid OR anti-glycopeptidolipid-core OR GPL OR anti-GPL OR Capilia OR tauns OR (EIA kit) OR (ELISA kit) OR (enzyme immunoassay kit)) AND (sensitivity OR specicity OR predictive value OR likelihood OR true positive OR true negative OR false positive OR false negative OR diagnostic OR diagnosis). We used similar search formulas for EMBASE, the Cochrane Library, and Web of Science (Supplementary Text 1).

We hand-searched published reviews and included original studies.

Study selection. Two investigators (YS, NH) independently screened the candidate reports by reading only the title and abstract. Then, the two investigators independently scrutinized the full text of reports that had not been excluded by at least one investigator. Duplicate use of the same data was carefully assessed. The nal inclusion was determined by discussion between the two investigators.

Data extraction. The two investigators (YS, NH) independently extracted the data from the original studies. These data were cross-checked. If necessary, an investigator (NH) tried to contact the author of original reports by e-mail.

Quality assessment for bias and applicability. The two investigators (YS, NH) independently scored the seven domains of the Revised Tool for the Quality Assessment of Diagnostic Accuracy Studies (QUADAS-2)26.

When a study might include denite-MAC patients under treatment, the study was scored for high risk of bias concerning patient selection. The nal score was determined aer discussion between the two investigators. A study that had no domain with a high risk of bias and no domain with high applicability concerns was regarded as a high-quality study.

Statistical analysis and quantitative synthesis. Data synthesis and interpretation. We obtained the DOR using a DerSimonian-Laird random-model and the AUC using Hollings proportional hazard model. Based on Jones criteria29, we interpreted AUC >0.97, 0.930.96, 0.750.92, and 0.50.75 as excellent, very good, good, and reasonable, respectively. We obtained a paired forest plot, HSROC curve, and the summary estimate of the sensitivity and the specicity using the bivariate model24. PLR and NLR were calculated from the summary estimates of sensitivity and specicity. Following Grimess criteria30, we interpreted PLR in the range of 25, 510, and >10 as representing small, moderate, and large increases of probability when the index test was positive. We also interpreted the NLR in the range of 0.20.5, 0.20.1, and <0.1 as representing small, moderate, and large decreases of probability when the index test was negative. The PPV and the NPV that were calculated from the

SCIENTIFIC REPORTS

2

www.nature.com/scientificreports/

summary estimate of sensitivity and specicity were presented as variables depending on the pretest probability of MAC-PD ranging from 0% to 100%.

Sensitivity analysis. As part of the sensitivity analysis, subgroup analysis was conducted focusing on reports that evaluated Capilia MAC Ab ELISA (cuto: 0.70U/mL), high-quality reports, and reports evaluating RA cases5,26.

Heterogeneity. We evaluated the heterogeneity using I2 statistics. I2< 40% is usually considered not important heterogeneity31.

Soware. We used commands of the statistics soware R as follows: madauni command for DOR, phm command for AUC, and reitsma command for the HSROC curve and the summary estimates of sensitivity and specicity32,33.

Results

Study search and study characteristics. Of the 257 articles that we had found through primary search, 86, 122, and 33 were excluded through removal of duplication, screening, and full-article reading, respectively (Fig.1). Notably, Kitada et al. reported a number of reports from a hospital, some of which were excluded due to a possible overlap of included subjects. Our hand search found no eligible article.

We nally included 16 reports, comprising 10 full-length articles and six conference abstracts, all of which were written in English (Table1)722. We obtained non-published data concerning four reports through e-mail communication with the authors1619. Among the 16 reports, 13 were from Japan and one each was from of Korea, the USA, and Taiwan. The Capilia kit was used in 14 studies. The cuto value of 0.70U/mL was used in 15 studies. The diagnostic criteria of ATS/IDSA 2007 were used in 10 studies. Four studies used the one-gate cohort approach, all of which were regarded as high-quality studies (Supplementary Figure 1). Three studies included only subjects with rheumatoid arthritis (RA).

The number of participants in each study ranged from 18 to 906 with a median of 143. The total number of subjects was 3368. This total consisted of 1098 reference positive subjects and 2270 reference negative subjects. Across the 16 studies, the sensitivity ranged from 0.20 to 1 with a median of 0.75, and the specicity ranged from 0.33 to 1 with a median of 0.92 (Fig.2).

Overall diagnostic accuracy. Using data from all 16 studies consisting of 1098 MAC-PD subjects and 2270 non-MAC-PD subjects, DOR was 24.8 (95% condence interval (95% CI) 11.652.8, I2 = 5.5%) and AUC was 0.873 (95% CI 0.8370.913) (Table2 and Fig.3A). In accordance with a criterion of Jones et al., AUC of 0.873 was categorized as good.

As sensitivity analyses, we used two subgroups. When focusing on 14 studies that used the Capilia kit and the cuto value of 0.7 U/mL, the DOR was 23.1 (95% CI 10.750.1, I2 = 7.2%) and the AUC was 0.874 (95% CI 0.8380.913) (Table2 and Fig.3B). Four high-quality studies yielded the DOR of 17.4 (95% CI 3.587.1, I2 = 31.9%) and the AUC of 0.853 (95% CI 66.51.000). (Table2 and Fig.3C). Based on three reports that included only RA cases, the DOR was 200.1 (95% CI 53.0754.9 I2= 0%) and the AUC was 0.946 (0.8980.999) (Table2 and Fig.3D).

The summary estimates of sensitivity and specicity were calculated from 14 studies that used the Capilia kit with the cuto value of 0.7 U/mL. The summary estimates of sensitivity and specicity were 0.696 (95% CI 0.6210.761) and 0.906 (95% CI 0.8360.951), respectively. (Table2 and Fig.3B)

According to a sensitivity analysis using four high-quality studies, the summary estimates of sensitivity and specicity were 0.646 (95% CI 0.5190.756) and 0.918 (95% CI 0.7060.981), respectively (Table2 and Fig.3C). The summary estimates sensitivity of 0.790 (95% CI 0.3010.971) and specicity of 0.979 (95% CI 0.8730.997) were obtained from three reports including only RA cases.

Positive and negative likelihood ratios. Based on the summary estimate of sensitivity and specicity calculated from 14 reports using the Capilia kit with the cuto value of 0.7 U/mL, PLR and NLR were 7.4 (95% CI 4.113.8) and 0.34 (95% CI 0.260.43). According to Grimes criteria, a positive index test moderately increases the probability of MAC-PC, while a negative index test suggests a small decrease of the probability (Table2 and Fig.4).

Positive and negative predictive values. We calculated PPV and NPV from the summary estimate of sensitivity and specicity calculated from 14 reports using the Capilia kit with the cuto value of 0.7 U/mL, i.e. a sensitivity of 0.696 and a specicity of 0.906 (Fig.4). This gure indicates that the positive test has a stronger impact than the negative test.

Positive rate in non-MAC-PD subjects. Some included studies provided data concerning the positive rate in non-MAC-PD subjects. The anti-GPL-core IgA antibody was positive in 27100% of RGM cases, while the test was negative in 93100% of pulmonary-healthy controls, 90100% of M. tuberculosis cases, and 97100% of non-MAC non-RGM NTM-PD cases (Table3). The pooled positive rate for rapid growing Mycobacterium was 0.64 (95% CI 0.271.00) (Supplementary Figure 2).

Discussion

To the best of our knowledge, the current report is the rst systematic review and meta-analysis evaluating the diagnostic test accuracy of anti-GPL-core IgA antibody for MAC-PD. We think the results from our study are

SCIENTIFIC REPORTS

3

www.nature.com/scientificreports/

Figure 1. Preferred Reporting Items for Systematic Reviews and Meta- Analyses ow chart for study search.

robust for various reasons. First, we included a sufficient number of studies and subjects. Second, we used the recently recommended hierarchical meta-analysis approach. Third, across all analyses, the observed heterogeneities were not important (<40%). Fourth, a sensitivity analysis revealed consistent results.

SCIENTIFIC REPORTS

4

www.nature.com/scientificreports/

Author (Year) Country Gates Patient background

Reporttype Facility Assay

Cuto(U/mL) Reference criteria Subjects Quality

Hamaguchi7 Japan 2 MAC-PD s/o, MAC

PD CA A Red Cross Hp 0.7 ATS 136 Low

Hirose8 Japan 1 MAC-PD s/o,

screening (all had RA) FA

A rheumatoidclinic Capilia 0.7 ATS/IDSA 2007 369 High

Joeng9 Korea 2 MAC-PD, MAB-PD,

TB, HC FA A medical center Capilia 0.7 ATS/IDSA 2007 120 Low

Kamiya10 Japan 2 MAC-PD, non-MAC

NTM, TB, LK CA

A secondaryreferral Hp Capilia 0.7 ATS 69 Low

Kitada11

Japan

2

MAC-PD, MAC colonization, MKA, TB, HC

FA

A teaching Hp

In-house EIA

0.072

ATS 1997

350

Low

Kitada12 USA 2 MAC-PD s/o, HC FA A research Hp Capilia 0.7 ATS/IDSA 2007 152 Low

Kitada13 Japan 2 MAC-PD, MKA-PD,

TB, HC FA A teaching Hp Capilia 0.7 ATS/IDSA 2007 906 Low

Kobashi14

Japan

2

MAC-PD, MAC-PD s/o, TB, non-MAC NTM, PD, HC

FA

Uni and affiliated Hps

Capilia

0.7

ATS/IDSA 2007

150

Low

Komazaki15 Japan 2 MAC-PD, MAC r/o (all

had RA) FA A uni Hp Capilia 0.7 ATS/IDSA 2007 34 Low

Koreeda16 Japan 2 MAC-PD, MAB-PD,

other NTM CA

Uni andaffiliated Hps Capilia 0.7 42 Low

Marukawa17 Japan 2 MAC-PD, MACctm,

TB, other PD, HC CA A medical center Capilia 0.7 400 Low

Nishimura18 Japan 1 MAC-PD s/o,

asymptomatic CA A teaching Hp Capilia 0.7 ATS/IDSA 2007 18 High

Numata19 Japan 1 MAC-PD s/o FA A uni Hp Capilia 0.7 ATS/IDSA 2007 296 High

Shimizu20 Japan 1 MAC-PD s/o FA A Red Cross Hp Capilia 0.7 ATS/IDSA 2007 66 High

Shu21 Taiwan 2 MAC-PD, MACctm,

RGM, MKA, TB, HC FA A uni Hp Capilia 0.7 ATS/IDSA 2007 197 Low

Watanabe22

Japan

2

MAC-PD, non-MAC NTM, abnormal CT, normal chest (All had RA)

FA

Uni and tertiary referral Hps

Capilia

0.7

ATS

63

Low

Table 1. Characteristics of included studies. <Gates> A so-called cohort study that included only MACPD-suspected cases had one gate. A so-called case-control study that included both MAC-PD cases and non-MAC-PD cases had two gates. <Patients background> MAC: M. avium complex. MACctm: MAC contamination. MAC s/o: suspected diagnosis of MAC. MAB: M. abcessus. MKA: M. kansasii. RGM: rapidly growing mycobacteria. TB: M. tuberculosis. NTM: non-tuberculosis mycobacterium. RA: Rheumatoid arthritis. LK: lung cancer. PD: pulmonary disease. HC: healthy control. <Report type> FA: Full-length article. CA: Conference abstract. All were written in English. <Facility> Uni: university. Hp: hospital. <Assay> EIA: enzyme immunoassay. <Reference criteria> ATS: the American Thoracic Society. IDSA: the Infectious Diseases Society of America. <Quality> A study that had no domain with high risk of bias and no domain with high applicability concerns was regarded as a high-quality study.

Figure 2. The paired forest plot by anti-glycopeptidolipid-core IgA antibody for M. avium complex pulmonary disease. TP: true positive. FP: false positive. FN: false negative. TN: true negative.

SCIENTIFIC REPORTS

5

www.nature.com/scientificreports/

All reports Capilia (cuto value 0.7U/mL) High-quality reports Rheumatoid arthritis

Studies 16 14 4 3

MAC-PD reference positive 1098 964 67 38

MAC-PD reference negative 2270 1918 682 432

Diagnostic odds ratio 24.8 (11.652.8)

I2 =5.5%

23.1 (10.750.1) I2 =7.2%

17.4 (3.587.1) I2 =31.9%

200.1 (53.0754.9)

I2 =0%

AUC 0.873 (0.8370.913) 0.874 (0.8380.913) 0.853 (0.6651.000) 0.946 (0.8980.999)

Sensitivity Not available 0.696 (0.6210.761) 0.646 (0.5190.756) 0.790 (0.3010.971)

Specicity Not available 0.906 (0.8360.951) 0.918 (0.7060.981) 0.979 (0.8730.997)

Positive likelihood ratio Not available 7.4 (4.113.8) 7.9 (2.233.7) 37.6 (4.8253.5)

Negative likelihood ratio Not available 0.34 (0.260.43) 0.39 (0.270.57) 0.21 (0.030.73)

Table 2. Summary of diagnostic accuracy by anti-glycopeptidolipid-core IgA antibody assay for M. avium complex-pulmonary disease (MAC-PD). Brackets indicate 95% condence interval. High quality reports: A study that had no domain with high risk of bias and no domain with high applicability concerns was regarded as a high-quality study. AUC: area under hierarchical summary receiver operating characteristics curve. Main outcomes concerning diagnostic accuracy are written in italics. The others are results from sensitivity analyses.

Figure 3. Hierarchical summary receiver operating characteristic curves.

Approximately 8% of reference negative subjects had a positive result as regards the anti-GPL-core antibody assessed with the Capilia kit. There are two plausible explanations for a subject with the positive anti-GPL-core IgA antibody assay and the negative clinical diagnosis, usually by ATS/IDSA 2007 statement. First, all MAC,

SCIENTIFIC REPORTS

6

www.nature.com/scientificreports/

Figure 4. Predictive values. PPV: positive predictive values. NPV: negative predictive values. PoTP: post-test probability.

Pulmonary-healthy control M. tuberculosis RGM

M. abcessus and other RGM commonly have GPL antigen in their cell walls5. Thus, the anti-GPL-core antibody assay intrinsically has cross-reactivity for MAC and RGM. This is a serious pitfall when interpreting the results of the anti-GPL-core antibody assay because dierent antibiotic regimens should be prescribed to treat MAC and RGM3,4. When the anti-GPL IgA antibody is positive, we should carefully consider the possibility of RGM infection, especially M. abcessus9. While pulmonary imaging studies for MAC-PD commonly present multiple discrete pulmonary nodules34, common CT ndings of RGM-PD are bilateral bronchiectasis, bronchiolitis, and upper lobe cavities35. Second, the universally used diagnostic criteria by ATS/IDSA 2007 statement, which is simpler than the previous American Thoracic Society 1997 statement, are still demanding4. The ATS/IDSA 2007 diagnostic criteria overlook some of true MAC-PD population. Although the culture isolation of the microbe has been the gold standard diagnostic method of infectious disease, the sero-diagnosis by anti-GPL-core IgA antibody, for example combining one MAC culture positive plus anti-GPL-core IgA antibody positive, can be used for diagnostic criteria5.

Kitada et al. proposed a cuto value for the anti-GPL-core IgA antibody of 0.7U/mL27. This is because this cuto can provide high specicity, though it is not sensitive. The summary estimate of sensitivity was 69% in our analysis (Table2), which means that almost a third of MAC-PD cases were overlooked by the anti-GPL-core IgA antibody assay. The serum enzyme immunoassay requires a sufficient quantity of targeted antibody to identify the MAC-infected individual. However, non-extensive pulmonary lesions, especially solitary nodules, and an immunocompromised state, namely an HIV-infected state and use of immunosuppressants, lead to a low serum antibody level13.

We need to discuss the limitations of our study. First, most of the included studies had a high risk of bias due to the two-gate study design. Because the prevalence of MAC is not high, it is difficult to recruit a large number of patients with MAC in a cohort study. Nonetheless, we believe the results of our analysis are reliable. This is because the high-quality studies based on sensitivity subgroup analysis that included only one-gate studies yielded consistent results (Table2 and Fig.3C). Second, most of the included studies were from Japan and only three were from other countries (Table1). To reconrm the diagnostic test accuracy of the sero-diagnosis for MAC, Kitada et al. evaluated 152 subjects in the USA and revealed that the sensitivity and specicity were 70.1% and 93.9%, respectively12. Third, subgroup analysis focusing on RA cases revealed higher accuracy than results from overall reports (Table2 and Fig.3D). However, the analysis was not conclusive due to the lack of a

Non-MAC

NTM collectively

Joeng9 0/20 (0%) 0/20 (0%) 28/40 (70%) 28/40 (70%)

Kamiya10 0/7 (0%) 1/4 (25%)

Kitada13 4/126 (3%) 4/77 (5%) 3/30 (10%)

Kitada12 3/52 (6%)

Kobashi14 0/20 (0%) 0/18 (0%) 0/9 (0%)

Komazaki15 0/45 (0%)

Koreeda16 4/4 (100%) 0/2 (0%) 4/6 (67%)

Numata19 0/12 (0%) 1/5 (20%)

Shu21 3/42 (7%) 5/48 (10%) 7/26 (27%) 0/14 (0%) 7/40 (18%)

Watanabe22 0/30 (0%) 0/3 (0%)

Table 3. Positive rate in non-MAC subjects. MAC: M. avium complex. RGM: rapidly growing mycobacterium including M. abcessus. NTM: non-tuberculosis mycobacterium.

Non-MAC non-RGM NTM

SCIENTIFIC REPORTS

7

www.nature.com/scientificreports/

plausible biological explanation and the limited number of included reports and subjects. Fourth, it is not likely that through checkup including BALF was done for every non-MAC persons to completely deny the possibility of MAC due to ethical regulation.

In conclusion, we conducted the rst systematic review and meta-analysis concerning the diagnostic test accuracy of anti-GPL-core IgA antibody for MAC-PD. According to our analysis using data of 16 reports and 3368 subjects. The DOR was 24.8 and the AUC was 0.873. The summary estimates of sensitivity and specicity were 0.691 and 0.919, respectively. Considering the demanding clinical diagnostic criteria of the ATS/IDSA 2007 statement, the true specicity of the anti-GPL-core IgA antibody may be higher. A positive index test moderately increases the probability of MAC-PC, while a negative index test suggests a small decrease of the probability. This test is useful to ruling in MAC-PD.

References

1. Inderlied, C. B., Kemper, C. A. & Bermudez, L. E. The Mycobacterium avium complex. Clin Microbiol Rev 6, 266310 (1993).2. Kendall, B. A. & Winthrop, K. L. Update on the epidemiology of pulmonary nontuberculous mycobacterial infections. Semin Respir Crit Care Med 34, 8794, doi: 10.1055/s-0033-1333567 (2013).

3. Kasperbauer, S. H. & Daley, C. L. Diagnosis and treatment of infections due to Mycobacterium avium complex. Semin Respir Crit Care Med 29, 569576, doi: 10.1055/s-0028-1085708 (2008).

4. Griffith, D. E. et al. An official ATS/IDSA statement: diagnosis, treatment, and prevention of nontuberculous mycobacterial diseases. Am J Respir Crit Care Med 175, 367416, doi: 10.1164/rccm.200604-571ST (2007).

5. Kobayashi, K. Serodiagnosis of Mycobacterium avium complex disease in humans: translational research from basic mycobacteriology to clinical medicine. Jpn J Infect Dis 67, 329332 (2014).

6. Kitada, S. et al. Serodiagnosis of pulmonary disease due to Mycobacterium avium complex with an enzyme immunoassay that uses a mixture of glycopeptidolipid antigens. Clin Infect Dis 35, 13281335, doi: 10.1086/344277 (2002).

7. Hamaguchi, N. et al. Usefulness of serum glycopeptidolipid core immunoglobulin A antibody in mycobacterium avium complex infection. Eur Respir J 44, 20 (2014).

8. Hirose, W. et al. Diagnostic performance of measuring antibodies to the glycopeptidolipid core antigen specic to Mycobacterium avium complex in patients with rheumatoid arthritis: results from a cross-sectional observational study. Arthritis Res Ther 17, 273, doi: 10.1186/s13075-015-0787-y (2015).

9. Jeong, B. H. et al. Serodiagnosis of Mycobacterium avium complex and Mycobacterium abscessus complex pulmonary disease by use of IgA antibodies to glycopeptidolipid core antigen. J Clin Microbiol 51, 27472749, doi: 10.1128/JCM.00702-13 (2013).

10. Kamiya, Y. et al. erodiagnosis of mycobacterium avium-complex pulmonary disease using anti-glycopeptidolipid core antigen IgA antibody. Eur Respir J 42, 40 (2013).

11. Kitada, S. et al. Use of glycopeptidolipid core antigen for serodiagnosis of mycobacterium avium complex pulmonary disease in immunocompetent patients. Clin Diagn Lab Immunol 12, 4451, doi: 10.1128/CDLI.12.1.44-51.2005 (2005).

12. Kitada, S. et al. Serodiagnosis of Mycobacterium avium complex pulmonary disease in the USA. Eur Respir J 42, 454460, doi: 10.1183/09031936.00098212 (2013).13. Kitada, S. et al. Validation of a commercial serodiagnostic kit for diagnosing pulmonary Mycobacterium avium complex disease. Int J Tuberc Lung Dis 19, 97103, doi: 10.5588/ijtld.14.0564 (2015).

14. Kobashi, Y., Mouri, K., Obase, Y., Kato, S. & Oka, M. Serological assay by use of glycopeptidolipid core antigen for Mycobacterium avium complex. Scand J Infect Dis 45, 241249, doi: 10.3109/00365548.2012.714904 (2013).

15. Komazaki, Y. et al. Serodiagnosis of Mycobacterium avium complex pulmonary disease in rheumatoid arthritis. Respiration 87, 129135, doi: 10.1159/000354791 (2014).

16. Koreeda, Y. et al. Serum IgA antibodies against glycopeptidolipid core antigen in patients with pulmonary disease due to nontuberculous mycobacteria other than mycobacterium avium-complex. Repirology 18, 161 (2013).

17. Marukawa, M., Hirabae, A., Suzaki, N. & Nagata, T. Evaluation of an enzyme immunoassay kit for the serodiagnosis of mycobacterium avium complex pulmonary disease. Respirology 20, 129 (2015).

18. Nishimura, N. et al. Clinical utility of serum anti-mycobacterium avium complex (MAC) antibody for diagnosis of MAC lung disease in asymptomatic patients with radiologically suspected MAC lung disease: A pilot study. Eur Respir J 46, 6 (2015).

19. Numata, T. et al. Clinical efficacy of anti-glycopeptidolipid-core IgA test for diagnosing Mycobacterium avium complex infection in lung. Respirology 20, 12771281, doi: 10.1111/resp.12640 (2015).

20. Shimizu, Y. et al. Serum glycopeptidolipid core IgA antibody levels in patients with chest computed tomography features of mycobacterium aviumintracellulare complex pulmonary disease. J Biol Regul Homeost Agents 28, 399405 (2014).

21. Shu, C. C. et al. Sero-diagnosis of Mycobacterium avium complex lung disease using serum immunoglobulin A antibody against glycopeptidolipid antigen in Taiwan. PLoS One 8, e80473, doi: 10.1371/journal.pone.0080473 (2013).

22. Watanabe, M. et al. Serodiagnosis of Mycobacterium avium-complex pulmonary disease with an enzyme immunoassay kit that detects anti-glycopeptidolipid core antigen IgA antibodies in patients with rheumatoid arthritis. Mod Rheumatol 21, 144149, doi: 10.1007/s10165-010-0368-5 (2011).23. Moher, D., Liberati, A., Tetzla, J., Altman, D. G. & Group, P. Preferred reporting items for systematic reviews and meta-analyses: the PRISMA statement. BMJ 339, b2535 (2009).

24. Leeflang, M. M., Deeks, J. J., Takwoingi, Y. & Macaskill, P. Cochrane diagnostic test accuracy reviews. Syst Rev 2, 82, doi: 10.1186/2046-4053-2-82 (2013).25. Booth, A. et al. The nuts and bolts of PROSPERO: an international prospective register of systematic reviews. Syst Rev 1, 2, doi: 10.1186/2046-4053-1-2 (2012).26. Whiting, P. F. et al. QUADAS-2: a revised tool for the quality assessment of diagnostic accuracy studies. Ann Intern Med 155, 529536, doi: 10.7326/0003-4819-155-8-201110180-00009 (2011).

27. Kitada, S. et al. Serodiagnosis of Mycobacterium avium-complex pulmonary disease using an enzyme immunoassay kit. Am J Respir Crit Care Med 177, 793797, doi: 10.1164/rccm.200705-771OC (2008).

28. Diagnosis and treatment of disease caused by nontuberculous mycobacteria. This official statement of the American Thoracic Society was approved by the Board of Directors, March 1997. Medical Section of the American Lung Association. Am J Respir Crit Care Med 156, S125, doi: 10.1164/ajrccm.156.2.atsstatement (1997).

29. Jones, C. M. & Athanasiou, T. Summary receiver operating characteristic curve analysis techniques in the evaluation of diagnostic tests. Ann Thorac Surg 79, 1620, doi: 10.1016/j.athoracsur.2004.09.040 (2005).

30. Grimes, D. A. & Schulz, K. F. Rening clinical diagnosis with likelihood ratios. Lancet 365, 15001505, doi: 10.1016/S0140-6736(05)66422-7 (2005).

31. Higgins, P. J. & Green, S. Cochrane Handbook for Systematic Reviews of Interventions (Version 5.1.0) (2011). Available from: http:// handbook.cochrane.org/front_page.htm. Accessed on December 4th, 2015).

32. Doebler, P. Package mada. Available from: https://cran.r-project.org/web/packages/mada/mada.pdf#search=mada+r. Accessed on November 5th, 2015.

SCIENTIFIC REPORTS

8

www.nature.com/scientificreports/

33. Doebler, P. & Holling, H. Meta-analysis of diagnostic accuracy with mada. Available from: https://cran.r-project.org/web/packages/ mada/vignettes/mada.pdf#search=mada+r+systematic+review+diagnostic+accuracy. Accessed on November 5th, 2015.

34. Prince, D. S. et al. Infection with Mycobacterium avium complex in patients without predisposing conditions. N Engl J Med 321, 863868, doi: 10.1056/NEJM198909283211304 (1989).

35. Han, D. et al. Radiographic and CT ndings of nontuberculous mycobacterial pulmonary infection caused by Mycobacterium abscessus. AJR Am J Roentgenol 181, 513517, doi: 10.2214/ajr.181.2.1810513 (2003).

Acknowledgements

We would like to thank Dr. Y. Koreeda (National Minami-Kyushu Hospital, Kagoshima, Japan), Dr. T. Numata (The Jikei University School of Medicine, Tokyo, Japan), Dr. M. Marukawa (Kagawa Rosai Hospital, Kagawa, Japan), and Dr. N. Nishimura (St Lukes International Hospital, Tokyo, Japan) for providing data concerning their previously published reports.

Author Contributions

Y.S. contributed for study search, quality check, data extraction, and drafting. N.H. work for study search, quality check, data extraction, and analysis as a principal investigator. M.Y. advised as a ucleic acid amplication specialist. T.T., H.N., K.T., H.W., K. Nagai, K. Nakashima, R.U., M.I., A.N. and A.K. provided data interpretation. S.T. and T.K. managed the study. All authors reviewed, revised and nally approved the manuscript.

Additional Information

Supplementary information accompanies this paper at http://www.nature.com/srep

Competing nancial interests: The authors declare no competing nancial interests.

How to cite this article: Shibata, Y. et al. Diagnostic test accuracy of anti-glycopeptidolipid-core IgA antibodies for Mycobacterium avium complex pulmonary disease: systematic review and meta-analysis. Sci. Rep. 6, 29325; doi: 10.1038/srep29325 (2016).

This work is licensed under a Creative Commons Attribution 4.0 International License. The images or other third party material in this article are included in the articles Creative Commons license, unless indicated otherwise in the credit line; if the material is not included under the Creative Commons license, users will need to obtain permission from the license holder to reproduce the material. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/

SCIENTIFIC REPORTS

9