Figures
Abstract
Background
A new lateral flow immunoassay (LFA) for the detection of cryptococcal antigen was developed.
Objective
We aimed to systematically review all relevant studies to evaluate the diagnostic accuracy of the cryptococcal antigen LFA on serum, CSF and urine specimens.
Methods
We searched public databases including PubMed, Web of Science, Elsevier Science Direct and Cochrane Library for the English-language literature published up to September 2014. We conducted meta-analyses of sensitivity, specificity, positive likelihood ratio (PLR), negative likelihood ratio (NLR) and diagnostic odds ratios (DOR) and SROC of LFA in serum and CSF, respectively. The sensitivity of LFA in urine was also analyzed. Subgroup analyses were carried out to analyze the potential heterogeneity.
Results
12 studies were included in this study. The pooled sensitivity and specificity values of LFA in serum were 97.6% (95% CI, 95.6% to 98.9%) and 98.1% (95% CI, 97.4% to 98.6%), respectively. The average PLR of LFA in serum was 43.787 (95% CI, 22.60–84.81) and the NLR was 0.03 (95% CI, 0.01–0.09). The pooled DOR was 2180.30 (95% CI, 868.92–5471.00) and the AUC was 0.9968. The pooled sensitivity and specificity values of LFA in CSF were 98.9% (95% CI, 97.9% to 99.5%) and 98.9% (95% CI, 98.0% to 99.5%), respectively. The average PLR of LFA in serum was 48.83 (95% CI, 21.59–110.40) and the NLR was 0.02 (95% CI, 0.01–0.04). The pooled DOR was 2931.10 (95% CI, 1149.20–7475.90) and the AUC was 0.9974. The pooled sensitivity value of LFA in urine was 85.0% (95% CI, 78.7% to 90.1%)
Citation: Huang H-R, Fan L-C, Rajbanshi B, Xu J-F (2015) Evaluation of a New Cryptococcal Antigen Lateral Flow Immunoassay in Serum, Cerebrospinal Fluid and Urine for the Diagnosis of Cryptococcosis: A Meta-Analysis and Systematic Review. PLoS ONE 10(5): e0127117. https://doi.org/10.1371/journal.pone.0127117
Academic Editor: Robert L. Schmidt, University of Utah Health Sciences Center, UNITED STATES
Received: December 30, 2014; Accepted: April 10, 2015; Published: May 14, 2015
Copyright: © 2015 Huang et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited
Data Availability: All relevant data are within the paper and its Supporting Information files.
Funding: This work was supported by the National Science Foundation of China (NSFC81170003, 81370109); Projects from STCSM (12PJD004, 134119a6400, 12JC1402300 and XYQ2011006). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing interests: The authors have declared that no competing interests exist.
Introduction
Cryptococcosis is a worldwide distributed mycosis caused by Cryptococcus neoformans and Cryptococcus gattii [1]. C.neoformans mostly infects AIDS patients particularly in sub-Saharan Africa as well as immunocompromised group(e.g.organ transplanted patients) [2]. In contrast, C.gattii is responsible for infecting immunocompetent individuals on Vancouver Island and surrounding areas [3]. Cryptococcosis is mainly caused by inhalation of pathogens through the respiratory tract. The central nervous system infection is the main clinical manifestation of cryptococcosis, so cryptococcal meningitis or meningoencephalitis is the main reason of high fatality rate. The increasing incidence of cryptococcosis is associated with use of broad-spectrum antibiotics, HIV infection, tumor radiotherapy and chemotherapy, organ transplantation, increased use of corticosteroids and immunosuppressants.
The principle of cryptococcal antigen test is to detect cryptococcal capsular polysaccharide antigen in serum or body fluids [4]. Latex agglutination method (LA) is the most popular one which is more sensitive than culture. However, LA takes much labor and the judgment results may be subjective. In order to make up with the drawbacks of LA, some labs in the USA began using enzyme-linked immunoassay (EIA). It can be automated and can analyze results objectively.
In 2009, Immuno-Mycologics (IMMY) invented a new cryptococcal antigen lateral flow immunoassay (LFA) for diagnosis of cryptococcal infection. As LA, it provides qualitative and semi-quantitative experimental results, and the problem of subjective judgment still persists. Despite of that, LFA has some advantages over the former two assays. It is a rapid diagnostic test as the reaction time is less than 10 minutes. LFA is stable at room temperature and has low requirement for laboratory equipment, demonstrating its usefulness as a point-of-care assay for diagnosis of cryptococcosis in developing countries.
Some studies have been conducted to evaluate the performance of these methods. As the results reported by different authors showed variance for the diagnostic role of different antigen detection methods, the real sensitivities and specificities of such methods remain unclear. Hence, we performed this meta-analysis aiming to systematically review all relevant studies to evaluate the diagnostic accuracy of the cryptococcal antigen LFA on serum, CSF and urine specimens.
Materials and Methods
Data Sources and Searches
Two investigators (HRH LCF) searched several public databases, including PubMed, Web of Science, Elsevier Science Direct and Cochrane Library for the English articles published till September 2014 respectively.”Cryptococcal antigen test”,”lateral flow assay”,”cryptococcal antigen lateral flow assay”,”LFA”,”cryptococcosis” and”diagnosis” were the key words. An expanded hand search of references of the relevant articles was also performed. We not only collected data from published full-text papers, but also from meeting and conference abstracts. We contacted the authors to obtain the unpublished data by e-mail.
Study Selection
Two investigators (HRH LCF) first independently screened articles by title and abstract to produce a list of articles for full text review, then resolved the differences by discussion. We included studies of adult patients admitted to the hospital suspected with cryptococcosis (e.g. HIV-infected persons suspected with meningitis) that provided data that could be used to construct a two-by-two cross-tabulation for LFA against a reference test. All reference standards have blood culture and/or LA and/or EIA included. We excluded studies in which LFA was applied to samples other than serum, CSF and urine. We excluded reviews, duplicated studies which only described the LFA. Studies that reported unavailable data were excluded as well. If the same authors had published two or more studies using the same or overlapping dataset, only one study was included.
Quality Assessment and Publication Bias
Two investigators (HRH LCF) assessed the risk of bias in each study by using the QUADAS-2 tool respectively [5]. Discrepancies were solved by discussion to get a consensus assessment and there was an adjudicator in case of persistent disagreement. We performed Deek’s funnel plot asymmetry test to detect publication bias [6].
Data Extraction
Two investigators (HRH LCF) extracted the data independently using the pre-designed data extraction protocol. Extracted data included study details (e.g., the first author’s name, year of publication, design of studies, nationality, sample size, article type, etc) and the patient’s characteristics. We contacted authors of conference abstracts to obtain further information for clarification. Discrepancies were resolved by discussing or by contacting the original investigators. Data was extracted from studies directly, or estimated using common mathematical methods.
Statistical Analysis
We calculated pooled estimates of the sensitivity, specificity and its 95% confidence interval (95% CI) for each study by using the DerSimonian and Laid method in the random effect model [7]. We also reported the positive and negative likelihood ratio (PLR and NLR) and the diagnostic odds ratio (DOR). The summary receiver operating characteristic (SROC) curve was used to represent the performance of the assay in each sample types. The area under the curve (AUC) and an index Q were discussed as potentially useful summaries of the curve [8]. We assessed statistical heterogeneity of the studies by the chi-square test, expressed with the I2 index [9]. A significant χ2 (p < 0.10) or I2-statistic (I2 > 50) indicated heterogeneity across studies. We performed threshold analysis and subgroup meta-analysis when heterogeneity was detected. Analyses were performed using the Meta-DiSc software v.1.4 and the Review Manager5.1software. All p values were two-sided and P<0.05 were considered statistically significant.
Results
Search Results
We had 719 potentially relevant papers from database search. After the duplicates were removed, there were 293 reports left. We excluded 261 irrelevant reports by reviewing abstract. With further screening of full texts, 19 studies were excluded for various reasons. At the end, we had 12 separate studies included in this meta-analysis. Detailed results of the literature search, selection, and reasons for exclusion are summarized in the flow diagram (S1 Fig).
Characteristics of Eligible Studies
We identified three conference abstracts and nine full-length published articles till September 2014 [10–21]. Two abstracts and six full-length articles reported data on serum specimens. One abstract and five full-length articles included data on CSF specimens and three full-length articles included data on urine specimens. We included a total of 4622 specimens and the sample size ranged from 55 to 1000. We included prospective, retrospective, cross-sectional, cohort and case-control studies. The population evaluated in our meta-analysis consisted patients who were suspected for having cryptococcosis such as cryptococcal meningitis. A part of the populations was HIV-infected. The characteristics of the included studies are summarized in Table 1.
Risk of Bias Assessments
A summary of the risk of bias assessment results are shown in S2 Fig. Each of the 11 components according to QUADAS-2 criteria was graded as “yes”,”unclear” or “no”, which meant “low risk of bias”, “uncertain of bias” and “high risk of bias” respectively, based on the methods reported in each study. All studies met the requirement as an acceptable reference standard. In a number of studies, the assessment of risk of bias was affected by unclear reporting. Few studies implied blinding of the index test and reference test. Non-publication bias was detected by Deek’s funnel plot asymmetry test (P = 0.73).
Data Synthesis and Meta-analysis
We classified studies into three categories according to the specimen types. Forest plots of sensitivity and specificity showed each study and the overall studies, respectively (Fig 1). For each study of serum specimens, sensitivity and specificity estimation with respect to the reference standard ranged from 90% to 100% and 93% to 100%, respectively. The pooled sensitivity and specificity values of the random effect model were 97.6% (95% CI, 95.6% to 98.9%) and 98.1% (95% CI, 97.4% to 98.6%), respectively (Table 2). The average PLR of LFA in serum was 43.79 (95% CI, 22.60–84.81) and the NLR was 0.03 (95% CI, 0.01–0.09). The pooled diagnostic odds ratio (DOR) was 2180.30 (95% CI, 868.92–5471.00) and the area under the curve (AUC) was 0.9968 (Fig 2).
Studies were classified into three categories according to the specimen types. Forest plots of sensitivity and specificity showed each study and the overall studies, respectively.
Abbreviations: AUC, area under curve; SE, standard error; Q*, Q-statistic.
For CSF specimens, sensitivity and specificity estimation with respect to the reference standard in each study ranged from 96% to 100% and 93% to 100%, respectively (Fig 1). The pooled sensitivity and specificity values of the random effect model were 98.9% (95% CI, 97.9% to 99.5%) and 98.9% (95% CI, 98.0% to 99.5%), respectively (Table 2). The average PLR of LFA in serum was 48.83 (95% CI, 21.59–110.40) and the NLR was 0.02 (95% CI, 0.01–0.04). The pooled diagnostic odds ratio (DOR) was 2931.10 (95% CI, 1149.20–7475.90) and the area under the curve (AUC) was 0.9974 (Fig 3).
Abbreviations: AUC, area under curve; SE, standard error; Q*, Q-statistic.
For urine specimens, sensitivity estimation with respect to the reference standard in each study ranged from 70% to 98% (Fig 1). The pooled sensitivity value of the random effect model was 85.0% (95% CI, 78.7% to 90.1%) (Table 2).
We found significant heterogeneity for all the test performances because I-square values were above 50% except the sensitivity in CSF which was 18.90%. In subgroup analyses (Table 3), the sensitivity of subgroups were all above 90% and the specificity were all above 95% except the EIA reference subgroup in urine specimen. Its sensitivity was 83.1%. In the analysis of potential influence of heterogeneity, the I-square value of the EIA reference subgroups in serum and urine were both over 50%. Although the I-square value of the LA reference subgroup in CSF specimen was 50.2%, the sensitivity and specificity of it were 96.9% (95% CI, 92.9% to 99.0%) and 98.7% (95% CI, 95.3% to 99.8%), respectively, which were relatively high.
Discussion
Many studies [10–21] have reported on the diagnostic value of the lateral flow immunoassay in detecting cryptococcal antigen. These studies, however, have demonstrated mixed results due to small sample sizes or low statistical power. In our meta-analysis, we combined 12 separate studies, consisting of 4622 samples to evaluate the diagnostic accuracy of the LFA for cryptococcal infection on serum, CSF and urine specimens. We found that the pooled sensitivity was 97.6% (95% CI, 95.6% to 98.9%) and the pooled specificity was 98.1% (95% CI, 97.4% to 98.6%) for serum specimen, which predicts LFA in serum is a very good method for diagnosis of cryptococcosis. For CSF specimen, the pooled sensitivity and specificity were both higher than those in serum. We speculated that the possible reason was because the majority of patients were suspected with cryptococcal meningitis. It also suggests us to conduct a lumbar puncture examination to identify cryptococcal infection on suspected patients. The pooled sensitivity of LFA in urine was 85.0% (95% CI, 78.7% to 90.1%). The pooled specificity was not estimated because of unavailable data. The sensitivity in urine was lower compared to that in serum or CSF. Nevertheless, urine specimen is easier to collect than cerebrospinal fluid, it provides us a convenient way to screen the suspected patients.
Subgroup analyses were performed for further description of the results. We found that the LFA was highly unanimous with the cryptococcal LA on serum specimens. However, in the EIA subgroup, the sensitivity was relatively lower than the specificity and the I-square value was 89.00%, which indicates significant heterogeneity. The performance of LFA was highly consistent with EIA and fungal culture method on CSF specimens. Although I-square value of the LA reference subgroup in CSF specimen was 50.2%, we still considered a little heterogeneity because of p>0.1. For urine specimens, there was heterogeneity between studies because of two different reference standards. The test performance of LFA was more consistent with culture than with EIA. When urine was evaluated, the LFA indicated higher sensitivity (93.5%) than EIA.
Our study had some limitations. Firstly, it is hard to ensure that no data was missed although we tried to collect all relevant papers. Secondly, we included only English articles and published papers which may lead to bias. Thirdly, significant between-study heterogeneity was detected in the current meta-analysis which may weaken conclusions’ reliance. Different diagnostic criteria may contribute to the heterogeneity among the selected studies. Despite the poor sensitivity, culture remains the gold standard one.
In summary, this meta-analysis demonstrates a very high accuracy of LFA in serum and CSF for the diagnosis of cryptococcosis for patients at risk. LFA in urine can be a promising sample screening tool for early diagnosis of cryptococcosis. Further studies testing CrAg in urine for diagnosis of cryptococcosis need to be performed.
Supporting Information
S1 Fig. Flow diagram of selection of studies for the meta-analysis.
https://doi.org/10.1371/journal.pone.0127117.s003
(TIF)
S2 Fig. Summary of the methodological quality assessment of the included studies according to QUADAS-2 criteria.
https://doi.org/10.1371/journal.pone.0127117.s004
(TIF)
Author Contributions
Conceived and designed the experiments: JFX HRH. Performed the experiments: HRH LCF. Analyzed the data: HRH LCF JFX. Contributed reagents/materials/analysis tools: LCF JFX. Wrote the paper: HRH BR JFX.
References
- 1. Perfect JR, Dismukes WE, Dromer F, Goldman DL, Graybill JR, Hamill RJ, et al. Clinical practice guidelines for the management of cryptococcal disease: 2010 update by the Infectious Diseases Society of America. Clin Infect Dis. 2010;50(3): 291–322. pmid:20047480
- 2. Park BJ, Wannemuehler KA, Marston BJ, Govender N, Pappas PG, Chiller TM. Estimation of the current global burden of cryptococcal meningitis among persons living with HIV/AIDS. AIDS. 2009;23(4): 525–530. pmid:19182676
- 3. Kidd SE, Hagen F, Tscharke RL, Huynh M, Bartlett KH, Fyfe M, et al. A rare genotype of Cryptococcus gattii caused the cryptococcosis outbreak on Vancouver Island (British Columbia, Canada). Proc Natl Acad Sci USA. 2004;101(49): 17258–17263. pmid:15572442
- 4. Saha DC, Xess I, Biswas A, Bhowmik DM, Padma MV. Detection of Cryptococcus by conventional, serological and molecular methods. J Med Microbiol. 2009;58(8): 1098–1105.
- 5. Whiting PF, Rutjes AW, Westwood ME, Mallett S, Deeks JJ, Reitsma JB, et al. QUADAS-2: a revised tool for the quality assessment of diagnostic accuracy studies. Ann Intern Med. 2011;155(8): 529–536. pmid:22007046
- 6. Deeks JJ, Macaskill P, Irwig L. The performance of tests of publication bias and other sample size effects in systematic reviews of diagnostic test accuracy was assessed. J Clin Epidemiol. 2005;58(9): 882–893. pmid:16085191
- 7. DerSimonian R, Laird N. Meta-analysis in clinical trials. Control Clin Trails. 1986;7: 177–188. pmid:3802833
- 8. Walter S. Properties of the summary receiver operating characteristic (SROC) curve for diagnostic test data. Stat Med. 2002;21: 1237–1256. pmid:12111876
- 9. Higgins JP, Thompson SG, Deeks JJ, Altman DG. Measuring inconsistency in meta-analyses. BMJ. 2003;327: 557–560. pmid:12958120
- 10. Jarvis JN, Percival A, Bauman S, Pelfrey J, Meintjes G, Williams GN, et al. Evaluation of a Novel Point-of-Care Cryptococcal Antigen Test on Serum, Plasma, and Urine From Patients With HIV-Associated Cryptococcal Meningitis. Clin Infect Dis. 2011;53: 1019–1023. pmid:21940419
- 11. Lindsley MD, Mekha N, Baggett HC, Surinthong Y, Autthateinchai R, Sawatwong P, et al. Evaluation of a Newly Developed Lateral Flow Immunoassay for the Diagnosis of Cryptococcosis. Clin Infect Dis. 2011;53(4): 321–325. pmid:21810743
- 12. Binnicker MJ, Jespersen DJ, Bestrom JE, Rollins LO. Comparison of Four Assays for the Detection of Cryptococcal Antigen. Clin Vaccine Immunol. 2012;19: 1988–1990. pmid:23081814
- 13. McMullan BJ, Halliday C, Sorrell TC, Judd D, Sleiman S, Marriott D, et al. Clinical Utility of the Cryptococcal Antigen Lateral Flow Assay in a Diagnostic Mycology Laboratory. PLoS One. 2012;7(11): e49541. pmid:23166705
- 14.
Clarke S, Gibb R. Evaluation of the IMMY CrAg lateral flow assay for detection of cryptococcal antigen. In: Aust Soc Microbiol. San Francisco; 2012.
- 15.
Vijayan T, Bauman S, Chiller S, Klausner J. Test performance of a novel lateral-flow assay to detect cryptococcal disease. In: ID Week. San Diego; 2012.
- 16.
Rolfes M, Butler E, von Hohenberg M, Nabeta H, Kwizera R, Rajasingham R, et al. Evaluation of a novel point-of-care lateral flow assay to detect cryptococcal antigen in plasma and CSF. In: 19 th Conference on Retrovirus and Opportunistic Infections. Seattle; 2012.
- 17. Escandón P, Lizarazo J, Agudelo CI, Chiller T, Castañeda E. Evaluation of a rapid lateral flow immunoassay for the detection of cryptococcal antigen for the early diagnosis of cryptococcosis in HIV patients in Colombia. Med Mycol. 2013;51(7): 765–768. pmid:23611420
- 18. Hansen J, Slechta ES, Gates-Hollingsworth MA, Neary B, Barker AP, Bauman S, et al. Large-Scale Evaluation of the Immuno-Mycologics Lateral Flow and Enzyme-Linked Immunoassays for Detection of Cryptococcal Antigen in Serum and Cerebrospinal Fluid. Clin Vaccine Immunol. 2013;20(1): 52–55. pmid:23114703
- 19. Rugemalila J, Maro VP, Kapanda G, Ndaro AJ, Jarvis JN. Cryptococcal antigen prevalence in HIV-infected Tanzanians: a cross-sectional study and evaluation of a point-of-care lateral flow assay. Trop Med Int Health. 2013;18(9): 1075–1079. pmid:23937699
- 20. Boulware DR, Rolfes MA, Rajasingham R, Hohenberg MV, Qin ZP, Taseera K, et al. Multisite Validation of Cryptococcal Antigen Lateral Flow Assay and Quantification by Laser Thermal Contrast. Emerg Infect Dis. 2014;20(1): 45–53. pmid:24378231
- 21. Kabanda T, Siedner MJ, Klausner JD, Muzoora C, Boulware DR. Point-of-care diagnosis and prognostication of cryptococcal meningitis with the cryptococcal antigen lateral flow assay on cerebrospinal fluid. Clin Infect Dis. 2014;58(1): 113–116. pmid:24065327