On-site testing would diminish time, costs, and risks involved in handling of highly infectious materials.
Crimean-Congo hemorrhagic fever (CCHF) is a highly contagious viral tick-borne disease with case-fatality rates as high as 50%. We describe a collaborative evaluation of the characteristics, performance, and on-site applicability of serologic and molecular assays for diagnosis of CCHF. We evaluated ELISA, immunofluorescence, quantitative reverse transcription PCR, and low-density macroarray assays for detection of CCHF virus using precharacterized archived patient serum samples. Compared with results of local, in-house methods, test sensitivities were 87.8%–93.9% for IgM serology, 80.4%–86.1% for IgG serology, and 79.6%–83.3% for genome detection. Specificity was excellent for all assays; molecular test results were influenced by patient country of origin. Our findings demonstrate that well-characterized, reliable tools are available for CCHF diagnosis and surveillance. The on-site use of such assays by health laboratories would greatly diminish the time, costs, and risks posed by the handling, packaging, and shipping of highly infectious biologic material.
Crimean-Congo hemorrhagic fever (CCHF) is a tick-borne zoonotic disease caused by a virus (CCHFV) belonging to the
Sporadic cases of CCHF and community and nosocomial outbreaks have been increasingly reported, and the disease’s geographic distribution is the most extensive among tick-borne diseases. Currently, CCHFV is enzootic in southeastern Europe (Bulgaria, Albania, Kosovo, and Greece), southern Russia, and several countries in the Middle East, Africa, and Asia (
Although viral isolation is the standard for CCHF diagnosis, because it has to be done in high-containment biosafety level 4 facilities, the number of laboratories that can perform this technique is limited. Moreover, because cell cultures lack sensitivity and usually only detect the relatively high viremia level encountered during the first 5 days of illness, viral isolation is not without error or uncertainty. As a consequence, reference laboratories have been using the best available practicable methods to determine the presence or absence of infection (
In this context, a working group of experts from reference laboratories was constituted under the initiative of the European Network for Diagnostics of Imported Viral Diseases to take part in a multicenter study of CCHF diagnostic tests. The aim of this study was to evaluate and compare the performance of, and review the operational characteristics of, available CCHF diagnostic tests by using panels of well-characterized, archived serum samples from patients from geographically diverse settings.
Experts from 5 institutions participated in this study: Bundeswehr Institute of Microbiology, Munich, Germany; Department of Microbiology, Aristotle University of Thessaloniki, Thessaloniki, Greece; Center for Vectors and Infectious Diseases Research, National Institute of Health, Águas de Moura, Portugal; Institute of Microbiology and Immunology, Medical Faculty, Ljubljana, Slovenia; and Center for Microbiological Preparedness, Swedish Institute for Infectious Disease Control, Solna, Sweden. Diagnostic methods that could be performed in standard laboratory facilities were selected on the basis of a systematic review of the literature and the experiences of the members of the working group. During April 2010, an extensive search of available CCHF diagnostic tools was performed by using both generic (Google) and scientific (PubMed) Internet-based search engines. To meet the selection criteria, assays had to be commercially available or in the prerelease phase at the time of our assessment (or have quality assessed reagents and well-defined protocols for noncommercial assays); yield early and rapid results (within 5 hours); not require the purchase of specific equipment; and have demonstrated sufficient scope to detect diverse CCHFV variants or antibodies. The reporting of results was conducted according to Standards for Reporting of Diagnostic Accuracy criteria (
| Laboratory | Reference IgM serologic tests | Reference IgG serologic tests | Reference molecular tests |
|---|---|---|---|
| 1 | In-house IFA (CCHFV-infected cells) | In-house IFA (CCHFV-infected cells) | Nested RT-PCR ( |
| 2 | In-house IgM capture ELISA (CCHFV strain IbAr10200 antigen) | In-house sandwich ELISA (CCHFV strain IbAr10200 antigen) | Nested RT-PCR ( |
| 3 | In-house IFA (CCHFV strain ArD39554 infected cells) | In-house IFA (CCHF strain ArD39554 infected cells) | qRT-PCR ( |
| 4 | In-house IFA (CCHFV infected cells) | In-house IFA (CCHF infected cells) | qRT-PCR ( |
| 5 | Testing performed by external reference laboratory | Testing performed by external reference laboratory | qRT-PCR ( |
*CCHF, Crimean-Congo hemorrhagic fever; IFA, immunofluorescence assay; CCHFV, CCHF virus; RT-PCR, reverse transcription PCR; qRT-PCR, quantitative RT-PCR.
Because no reference test for CCHV is universally accepted, patients with clinically suspected CCHF were confirmed on the basis of results of serologic and molecular diagnostic tests that were in use in the CCHF reference laboratories at the time of the study (
Tests were performed according to the manufacturers’ instructions or according to validated protocols provided by the developer. Working group members tested the selected assays in duplicate on their respective sample panels within their facilities. Results of qualitative assays (IFAs and low-cost, low-density [LCD] arrays) were read by 2 independent readers. Results were collected at the end of each testing session by using a standard report datasheet and combined into a final database.
Results obtained for each selected diagnostic tool were compared in a 2×2 table with results from the reference in-house diagnostic to estimate indices of sensitivity, specificity, and corresponding 95% CIs. In addition, test results were compared with the results of a composite reference standard (positive if detection of specific CCHFV genome or specific CCHF IgM or IgG antibodies by in-house reference methods; negative otherwise) to confirm the specificity estimates and corresponding 95% CIs. Statistical analysis was performed in STATA/SE version 12.0 software (StataCorp, College Station, TX, USA). CIs were calculated by using binomial exact methods. A univariate analysis was conducted by using the Fisher exact test to identify factors influencing test sensitivity (i.e., patient country of origin, severity of disease, number of days after illness onset that sample was collected, and sample storage time before testing); p<0.05 was considered significant. A multinomial exact logistic regression analysis was performed to identify independent factors influencing sensitivity, including all variables associated with sensitivity in the univariate analysis (p<0.1). Data on operational characteristics of each test (i.e., ease-of-use, level of staff training required, time, ease of interpretation) were gathered through a questionnaire. Tests were scored on operational characteristics.
Six diagnostic assays met the criteria for inclusion in the study. For specific CCHF serodiagnosis, a commercial IgM and IgG ELISA (Vector-Best, Novosibirsk, Russia) and a commercial IgM and IgG IFA (Euroimmun, Luebeck, Germany) were selected. For detection of the CCHFV genome, a commercial real-time RT-PCR (Altona-Diagnostics, Hamburg, Germany) and a low-cost, low-density macroarray (
| Characteristics | IgM ELISA | IgG ELISA | IgM IFA | IgG IFA | qRT-PCR | LCD array | |
|---|---|---|---|---|---|---|---|
| Assay (manufacturer, location or reference) | VectoCrimea-CHF ELISA (Vector-Best, Novosibirsk, Russia) | Crimean-Congo Fever Mosaic 2 IFA (Euroimmun, Luebeck, Germany) | RealStar CCHFV RT-PCR Kit 1.2 (Altona-Diagnostics, Hamburg, Germany) | CCHF2006 1.5 LCD Kit ( | |||
| Reference no. | D-5054 | D-5056 | FI 279a-1010-2M | FI 279a-1010-2G | 181203 | NA | |
| Target | CCHFV-specific IgM | CCHFV-specific IgG | CCHFV-specific IgM | CCHFV-specific IgG | CCHFV S segment | CCHFV S segment | |
| Shelf life, mo | 9 | 9 | 18 | 18 | 12 | Unknown | |
| Storage temperature, °C | 2 to 8 | 2 to 8 | 2 to 8 | 2 to 8 | −15 to −25 | 2 to 8, −20 | |
| Quoted accuracy, % | |||||||
| Sensitivity | 100 | 100 | 97.2 | 89.5 | Unknown | 100 | |
| Specificity | 100 | 100 | 97.5 | 100 | Unknown | 100 | |
| Sample type | Serum, plasma | Serum, plasma | Serum, plasma | Serum, plasma | RNA extract from serum or blood | RNA extract from serum or blood | |
| Sample volume, µL | 10 | 10 | 5 | 5 | 10 | 10 | |
| Minimum kit format (no. reactions) | 12 strips × 8 tests (96) | 12 strips × 8 tests (96) | 10 slides × 5 tests (50) | 10 slides × 5 tests (50) | 8 tubes × 12 tests (96) | 4 slides × 8 tests (32) | |
| Price, Euros† | |||||||
| Per kit | 139.2 | 139.2 | 326 | 326 | 1,200 | Unknown | |
| Per reaction | 1.45 | 1.45 | 6.51 | 6.51 | 12.50 | Unknown | |
| Estimated run time, min | 175 | 175 | 70 | 70 | 58 | 175 | |
*CCHFV, Crimean-Congo hemorrhagic fever virus; IFA, immunofluorescence assay; qRT-PCR, quantitative reverse transcription PCR; LCD, low-cost, low-density; NA, not applicable; S segment, small segment. †Does not include shipping costs.
The serum panel constituted for the evaluation of the serologic tests consisted of 66 stored serum samples from acute-phase CCHF patients (those who recovered or died) and patients with confirmed CCHF diagnosis who had an early recovery; 32 samples from febrile patients who had symptoms compatible with CCHFV infection; and 41 samples from healthy persons. Molecular tests were evaluated by using a panel of RNA extracts from acute-phase patient serum samples: 54 samples from patients with confirmed CCHF diagnosis, 16 samples from febrile patients who had symptoms compatible with CCHFV infection, and 5 samples from healthy persons. Characteristics of the patient population and the sample panels are shown in
| Characteristics | Sample panel 1, serology, no. (%) | Sample panel 2, genome detection, no. (%) |
| CCHFV positive, n = 66 | CCHFV negative, n = 73 | CCHFV positive, n = 54 | CCHFV negative, n = 21 | ||
|---|---|---|---|---|---|
| Conditions | |||||
| CCHF | 66 (100.0) | 0 | 54 (100.0) | 0 | |
| Brucellosis | 0 | 2 (2.7) | 0 | 2 (9.5) | |
| Chikungunya | 0 | 1 (1.4) | 0 | 1 (4.8) | |
| HFRS | 0 | 13 (17.8) | 0 | 7 (33.3) | |
| Leptospirosis | 0 | 13 (17.8) | 0 | 3 (14.3) | |
| Q fever | 0 | 1 (1.4) | 0 | 1 (4.8) | |
| Tularemia | 0 | 1 (1.4) | 0 | 1 (4.8) | |
| West Nile fever | 0 | 1 (1.4) | 0 | 1 (4.8) | |
| Healthy or non-CCHF | 0 | 41 (56.2) | 0 | 5 (23.8) | |
| Patient country of origin | |||||
| Albania | 9 (13.6) | 0 | 8 (14.8) | 0 | |
| Germany | 0 | 23 (31.5) | 0 | 7 (33.3) | |
| Greece | 0 | 9 (12.3) | 0 | 9 (42.9) | |
| Iran | 32 (48.5) | 0 | 31 (57.4) | 0 | |
| Kosovo | 21 (31.8) | 20 (27.4) | 7 (13.0) | 4 (19.0) | |
| Portugal | 0 | 20 (27.4) | 0 | 0 | |
| Sub-Saharan Africa | 4 (6.1) | 0 | 0 | 0 | |
| Turkey | 0 | 1 (1.4) | 8 (14.8) | 1 (4.8) | |
| CCHF disease severity | |||||
| Moderate | 49 (74.2) | 0 | 37 (68.5) | 0 | |
| Severe | 11 (16.7) | 0 | 2 (3.7) | 0 | |
| Fatal | 3 (4.5) | 0 | 6 (11.1) | 0 | |
| Asymptomatic | 3 (4.5) | 0 | 7 (13.0) | 0 | |
| Unknown | 0 | 0 | 2 (3.7) | 0 | |
| Length of illness, d | |||||
| <15 d | 46 (69.7) | 0 | 43 (79.6) | 0 | |
| ≥15 d | 12 (18.2) | 0 | 0 | 0 | |
| Asymptomatic | 3 (4.5) | 0 | 7 (13.0) | 0 | |
| Unknown | 5 (7.6) | 0 | 4 (7.4) | 0 | |
| Sample storage time, y | |||||
| <10 | 49 (74.2) | 64 (87.7) | 54 (100.0) | 10 (47.6) | |
| ≥10 | 17 (25.8) | 0 | 0 | 2 (9.5) | |
| Unknown | 0 | 9 (12.3) | 0 | 9 (42.9) | |
| Sample storage temperature, °C | |||||
| −80 | 34 (51.5) | 49 (67.1) | 9 (16.7) | 4 (19.1) | |
| −70 | 32 (48.5) | 0 | 45 (83.3) | 9 (42.9) | |
| −20 | 0 | 24 (32.9) | 0 | 8 (38.1) |
*CCHF, Crimean-Congo hemorrhagic fever; CCHFV, CCHF virus; HFRS, hemorrhagic fever with renal syndrome.
A total of 138 and 90 samples from the collected patient serum panels were tested for CCHFV-specific IgM by the Vector-Best ELISA and the Euroimmun IFA, respectively. Because of limited sample amounts, IFA could not be performed on all collected samples. When compared with the reference IgM serology tests, the sensitivity of the IgM ELISA ranged from 75.0% to 100.0% for different laboratories, with an overall sensitivity of 87.8% (95% CI 78.6%–96.9%). For the IgM IFA, sensitivity ranged from 75.0% to 100.0%, with an overall sensitivity of 93.9% (95% CI 85.8%–100.0%). Overall specificity was estimated to be 98.9% (95% CI 96.7%–100.0%) for the IgM ELISA and 100% for the IgM IFA (
| Parameter | IgM serology | IgG serology | Genome detection | |||||
|---|---|---|---|---|---|---|---|---|
| ELISA | IFA | ELISA | IFA | qRT-PCR | LCD array | |||
| No. samples tested | 138 | 90 | 137 | 92 | 71 | 70 | ||
| No. true positive | 43 | 31 | 41 | 31 | 39 | 40 | ||
| No. false negative | 6 | 2 | 10 | 5 | 10 | 8 | ||
| No. true negative | 88 | 57 | 86 | 56 | 21 | 21 | ||
| No. false positive | 1 | 0 | 0 | 0 | 1 | 1 | ||
| Sensitivity, % (95% CI) | 87.8 (75.2–95.3) | 93.9 (79.8–99.3) | 80.4 (66.9–90.2) | 86.1 (70.5–95.3) | 79.6 (65.7–89.8) | 83.3 (69.8–92.5) | ||
| Specificity, % (95% CI) | 98.9 (93.9–100.0) | 100.0 (93.7–100.0)† | 100.0 (95.8–100.0) | 100.0 (93.6–100.0) | 95.5 (77.2–99.9) | 95.5 (77.2–99.9) | ||
*CCHF, Crimean-Congo hemorrhagic fever; IFA, immunofluorescent assay; qRT-PCR, quantitative reverse transcription PCR; LCD, low-cost, low-density. †One-sided 95% CI.
A total of 137 and 92 samples from the collected patient serum panel were tested for CCHFV-specific IgG by the Vector-Best ELISA and the Euroimmun IFA, respectively. When compared with the reference IgG serology tests, the estimated sensitivity for the IgG ELISA ranged from 75.0% to 100.0%, with an overall sensitivity of 80.4% (95% CI 69.5%–91.3%). For the IgG IFA, sensitivity ranged from 40.0% to 100.0%, with an overall sensitivity of 86.1% (95% CI 74.8%–97.4%). Specificity was estimated to be 100% for both assays (
A total of 71 and 70 samples, respectively, from the collected panel of serum RNA extracts were tested for the presence of the CCHF genome by the Altona-Diagnostics CCHFV qRT-PCR and the CCHF LCD array. When compared with the results of the reference genome detection methods, sensitivity ranged from 42.9% to 100%, with an overall sensitivity of 79.6% (95% CI 68.3%–90.9%) for the qRT-PCR and from 25.0% to 100% with an overall sensitivity of 83.3% (95% CI 72.8%–95.5%) for the LCD array. Both assays demonstrated a specificity of 95.5% (95% CI 90.6%–100%) (
The influence of several patient and sample characteristics on the sensitivity of the selected assays was analyzed by univariate analysis (
Scores for operational characteristics are summarized in
| Operational characteristic | Mean score | |||
|---|---|---|---|---|
| VectoCrimea-CHF ELISA† | Crimean-Congo Fever Mosaic 2 IFA‡ | RealStar CCHFV RT-PCR Kit 1.2§ | CCHF2006 1.5 LCD Kit ( | |
| Equipment¶ | ||||
| Maintenance of equipment required | 0.8 | 0.3 | 0.3 | 0.3 |
| Training for equipment required | 0.8 | 0.7 | 0.3 | 0.5 |
| Additional equipment required | 0.8 | 1.0 | 0.5 | 0.8 |
| Technique | ||||
| Clarity of instructions¶ | 1.0 | 0.7 | 1.0 | 1.0 |
| Technical training required¶ | 0.8 | 0.3 | 0.3 | 0.3 |
| Technical complexity# | 2.0 | 1.7 | 1.5 | 1.3 |
| Interpretation | ||||
| Training required for result interpretation¶ | 1.0 | 0.7 | 0.8 | 0.8 |
| Ease of interpretation of results# | 1.5 | 1.3 | 1.5 | 1.5 |
| Total score | 8.5/10 | 6.7/10 | 6.0/10 | 6.3/10 |
*CCHF, Crimean-Congo hemorrhagic fever; IFA, immunofluorescent assay; CCHFV, CCHF virus; RT-PCR, reverse transcription PCR; LCD, low-cost, low-density. †Vector-Best, Novosibirsk, Russia. ‡Euroimmun, Luebeck, Germany. §Altona-Diagnostics, Hamburg, Germany. ¶A score of 1 was attributed when instructions were sufficiently clear or when no specific training, no additional equipment, or no regular equipment maintenance were necessary. #Score was attributed according to the degree of simplicity of the technique or interpretation: 2 if it was considered easy, 1 if it was considered acceptable, and 0 if it was considered difficult.
A number of published studies have described major epidemics, community and nosocomial outbreaks, and the ecology of CCHF (
A strong laboratory capacity, in particular standardized approaches for diagnostic methods and assay validation, has been identified as a short-term priority in CCHF-endemic areas and regions where CCHFV could be expected to circulate (
We assessed the performances of 2 commercially available IgM and IgG serologic tests, the Vector-Best CCHF ELISA and the Euroimmun CCHF IFA, and 2 assays for viral genome detection, the Altona-diagnostics CCHF qRT-PCR and a CCHF LCD array. The IgM and IgG ELISAs showed a sensitivity of 88% and 80%, respectively, lower than the numbers given by the manufacturer. These assays were validated by the manufacturer by using serum panels from CCHF cases originating from southwestern Russia (S. Suchkov, pers. comm.). Therefore, lower sensitivity for certain serum samples tested in this study may reflect antigenic variation among CCHFVs circulating in other countries.
Observed sensitivities of the IgM and IgG IFA were 93.9% and 86.1%, respectively. Although these estimates are higher than those observed for the IgM and IgG ELISAs, these results may have had a sampling bias because not all serum samples tested by ELISA could be tested by IFA. This bias was, however, minimized, because the tested serum panel included 15/16 false-negative samples observed for the ELISAs.
The sensitivity of the selected molecular assays was found to be more modest (79.6% for qRT-PCR and 83.3% for LCD array) than for serologic methods and to be associated with the patient country of origin. This result is consistent with the finding that the application of molecular assays in different settings is hampered by the high diversity of the CCHFV genomes, whereas serologic methods can have a broader use due to cross-reactivities. In particular, the qRT-PCR seems to be less sensitive for patients originating from the Balkans region and Turkey than for patients from other countries compared with in-house reference molecular methods. The in-house methods developed by reference laboratories are optimized for detection of strains circulating in that area, which may result in a lower detection limit when compared with methods that cover a broader spectrum. However, other factors, such as RNA degradation due to freezing and thawing cycles or inhibition of PCR reactions because of inhibitory compounds in the samples, may have contributed to decreased sensitivity of molecular methods compared with serologic methods.
The observed specificity was excellent for all assays, ranging from 95.5% to 100% when compared with the reference method and equal to 100% when compared with a composite reference standard. However, predictive positive and negative values for the different assays could not be calculated because precise prevalence data are not available for most CCHF-endemic areas, and these data cannot be predicted for areas where the virus could emerge.
The CCHF IFA demonstrated a higher complexity in equipment, technique, and interpretation. However, the interpretation of fluorescence patterns may enable trained users to differentiate positive from cross-reactive serum samples, whereas such false positives may not be avoided with the ELISA. The operational characteristics of the molecular assays were comparable. Both methods required regular equipment maintenance and specific training for appropriate use of the equipment; however, this was more apparent for the real-time qRT-PCR method. The LCD array technique was considered to be complex but acceptable. Training for technique and result interpretation was recommended for each method.
All participating laboratories are reference centers for CCHF laboratory diagnosis and surveillance in their respective countries, and some are World Health Organization collaborating centers. The protocols from reference methods in use at each site have been extensively validated previously (
The multicenter design of this study enabled the testing of a large sample size, representative of
Our study has some limitations. Because archived samples were used for the study, specimen quality could have been affected; however, statistical analysis demonstrated that sample storage time and temperature did not influence sensitivity. Also, the use of different sample panels for serologic and molecular testing did not enable calculation of the added value of combining the serologic and molecular CCHF diagnostic methods evaluated.
The results of this study give additional guidance on the type of CCHF diagnostic tools that could be used in different contexts. During a large outbreak, easily interpretable tests for simultaneous analysis of numerous samples, such as the ELISA and real-time qRT-PCR, might be considered useful tools to identify CCHF cases. Methods available in smaller format size and demonstrating a long shelf life, such as the IFA and LCD array, could be used to identify sporadic cases or to confirm single cases as part of a larger outbreak.
This study demonstrates that efficient, well-characterized serologic and molecular assays and protocols are available for CCHF diagnosis. The on-site use of such assays by outbreak assistance laboratories would greatly diminish the risks posed by the handling, packaging, and shipping of highly infectious samples. Moreover, acquiring diagnostic reagents would be more time- and cost-effective for laboratories than would the organization of compliant packaging and shipment of hazardous biologic material to reference laboratories abroad. Nevertheless, laboratory personnel should receive the appropriate training to perform the different assays (e.g., during international workshops or network meetings). Collaborative evaluations of diagnostic methods remain essential to guide decision-making, especially with emerging diseases, where a standard is frequently missing, and laboratory expertise is rare.
Checklist for reporting of studies of diagnostic accuracy and univariate analysis of factors influencing the sensitivity of Crimean-Congo hemorrhagic fever diagnostic assays.
This work was supported by the European Network for Diagnostics of Imported Viral Diseases project, funded by the European Centre for Disease Prevention and Control under the Framework Service (contract ref. no. ECDC/2008/011); and the CCH Fever network (collaborative project), funded by the European Commission under the Health Cooperation Work Programme of the 7th Framework Programme (grant agreement no. 260427).
Dr Vanhomwegen is a research fellow at Institut Pasteur, Paris. Her research interests include the development and validation of tools for rapid detection and characterization of emerging pathogens.