Previous Article | Next Article ![]()
Clinical and Diagnostic Laboratory Immunology, November 2003, p. 1019-1024, Vol. 10, No. 6
1071-412X/03/$08.00+0 DOI: 10.1128/CDLI.10.6.1019-1024.2003
Copyright © 2003, American Society for Microbiology. All Rights Reserved.
Craig Laferriere,5 Dominique Wauters,5 Moon H. Nahm,6 Mark F. Schinsky,1,
Brian D. Plikaytis,1 and George M. Carlone1
Centers for Disease Control and Prevention, Atlanta, Georgia,1 U.S. Food and Drug Administration, Bethesda, Maryland,2 University of London, London, United Kingdom,3 National Public Health Institute, Helsinki, Finland,4 Glaxo SmithKline Biologicals, Rixensart, Belgium,5 University of Rochester, Rochester, New York6
Received 24 June 2003/ Returned for modification 23 July 2003/ Accepted 18 August 2003
|
|
|---|
50% killing) for each serum sample. The percentage of sera within one or two dilutions of the calculated median OPA titer was determined for each laboratory and for each serotype. In general, laboratories were capable of detecting OPA titers within one or two dilutions of the median for at least 75 and 88%, respectively, of the sera tested. The level of agreement with the median OPA titers varied depending on the participating laboratory (overall agreement = 0.8 [99% confidence interval = 0.75 to 0.85]). All OPA median titers reported for quality control sera were within one dilution of the expected titer. We conclude that this OPA can be done in multiple laboratories with a high degree of interlaboratory reproducibility. |
|
|---|
Efforts have been made for the identification and standardization of laboratory correlates of protection that can aid vaccine efficacy trials in the estimation of vaccine-induced protection. Currently, a highly standardized enzyme-linked immunosorbent assay (ELISA) is available (www.vaccine.uab.edu) for the evaluation of infant sera. Several modifications to the protocol described by Quartaert et al. (20) allowed for the measurement of Ps-specific antibodies in children and adults (6, 19, 18). Adults can have cross-reactive antibodies, which confound the measurements of immunoglobulin G (IgG) antibodies by ELISA, especially if absorption with a nonrelevant serotype is not performed prior to testing (6, 7, 26). These cross-reactive antibodies are more prevalent in prevaccination sera than in postvaccination sera (6). Due to the lower specificity previously observed with ELISA-based assays, efforts were centered on the identification of alternative assays that can measure the function of the anti-capsular antibodies and serve as correlates of protection against disease. In the case of pneumococcus, opsonophagocytosis is the primary mechanism for protection in the host (25). Therefore, standardization and validation of assays measuring opsonophagocytic activity are of high importance for Pnc vaccine evaluation.
This study describes the results of a multilaboratory evaluation of a viability opsonophagocytosis assay (OPA). The viability OPA was previously standardized for the use of HL-60 granulocytes as the effector cells (21). Functional assays are inherently variable. In this study, we calculate the interlaboratory variability of the viability OPA and we determine if this OPA is suitable for multilaboratory comparisons of functional antibodies induced by Ps-based Pnc vaccines.
|
|
|---|
50% compared to the in-plate complement controls was reported as the OPA titer. Worksheets with OPA results were sent to the CDC for final analysis. All test plates included a positive control serum (R21654-34301107) and Sandoglobulin (purified IgG, lot no. 441x9605; Sandoz Pharmaceuticals Co., East Hanover, N.J.). Control serum and Sandoglobulin (6% [wt/vol] concentration) were provided by the CDC along with a list of their expected OPA titers (median ± dilution range) and initial testing dilution for each of the serotypes tested. Laboratory 3 modified the assay by automating the colony counting procedure. This modification is now widely used by many other laboratories for the determination of viability colony counts. Statistical analysis. The median OPA titer was calculated for each serum sample and serotype. Once median OPA titers were obtained, results from all laboratories were compared to the median values (median). The percentage of serum samples in agreement with the median titer and the percentage of serum samples within one or two dilutions from the median titer were calculated for each serotype and for each participating laboratory. The level of agreement to the median was determined in 2 by 2 tables with the formula a/(a + c), where a is the number of sera at or above the given titer by both the test laboratory and the median and c is the number of sera below the given titer by the test laboratory but at or above the given titer by the median. The 95 and 99% confidence intervals (CI) were calculated with the aid of the Sigma Plot software program, version 3.0 (Jandel Scientific, San Rafael, Calif.). OPA titers of <8 were reported as a titer of 4 for calculation purposes. Therefore, a titer of 8 was defined as the lower limit of quantification. Outliers were identified if OPA titers were more than 3 dilutions from the median titer. However, all data submitted were included in the analysis. No data points were edited or excluded.
The participating laboratories were as follows: Respiratory Diseases Branch, Immunology Section, CDC; Center for Biologics Evaluation and Research, U.S. Food and Drug Administration, Bethesda, Md.; U.S. Pneumococcal Reference Laboratory, University of Rochester, Rochester, N.Y., presently located at the University of Alabama, Birmingham; National Public Health Institute, Helsinki, Finland; and Glaxo SmithKline Biologicals, Rixensart, Belgium.
|
|
|---|
50% killing compared to the complement control wells (no antibody source added). Table 1 gives the median OPA titers by serotype for the 24 standard serum samples (pre- and postvaccination). The expected median OPA titers for the quality control reference serum R21654-34301107 and Sandoglobulin lot no. 441x9605 were determined prior to the initiation of the study (n = 6 independent duplicate assays per serotype). The expected OPA titers for the reference serum were 128 (serotype 4), 256 (serotypes 6B, 9V, 14, 18C, and 19F) and 2,048 (serotype 23F), and the expected OPA titers for Sandoglobulin were 128 (serotype 4), 256 (serotype 9V), 512 (serotypes 18C and 19F), and 1,024 (serotypes 6B, 14, and 23F). The variability of the expected median OPA titers was within 1 dilution, except for Sandoglobulin (serotype 6B, ±2 dilutions). Observed OPA titers were determined by the participant laboratories (n = 6 OPA titers per serotype submitted by laboratories 1, 2, 4, and 5). Observed OPA titers were within one or two dilutions of the expected median OPA titer for both controls, except for the reference serum for serotypes 4 and 9 (
3 dilutions apart). These two controls allowed for the assurance that HL-60 cell differentiation was optimal and/or other assay parameters were optimal for the calculation of OPA titers in test sera (12 paired sera). |
View this table: [in a new window] |
TABLE 1. Median OPA titers for 24 quality control adult sera
|
3 dilutions from the median OPA titer) in sera with higher OPA titers. |
View this table: [in a new window] |
TABLE 2. Percentage of sera reported by each participating laboratory within one and two dilutions of the calculated median OPA titer
|
![]() View larger version (29K): [in a new window] |
FIG. 1. Reported OPA titers (log2) for each quality control serum (a prevaccination serum is followed by its corresponding postvaccination serum) for Pnc serotypes 6B (A), 14 (B), and 23F (C).
|
8), were expected among the 12 postvaccination serum samples. However, as shown in Fig. 1 for serotype 23F, we also found a high proportion of postvaccination sera with OPA titers of 4, especially for serotypes 4 and 23F (42 and 50%, respectively).
The median OPA titer for each serum sample was compared to the laboratory's reported OPA titer for each serum. The total number of OPA titers in agreement or disagreement at a given cutoff titer was calculated. Laboratories varied in the level of agreement to the median OPA titers (Fig. 2). Laboratory 1 had lower agreements for serotypes 6B (0.33), 9V (0.13), and 14 (0.50). Laboratory 2 had a lower agreement for serotype 14 (0.71). Laboratory 3 had lower agreements for serotypes 6B (0.39) and 19F (0.65). Laboratory 4 had a lower agreement for serotype 18C (0.55). Laboratory 5 reported OPA titers with levels of agreement of
0.89 for all the serotypes evaluated. The mean interlaboratory level of agreement for serum samples with OPA titers above the minimum level of quantification was calculated to be 0.8 (99% CI = 0.75 to 0.85).
![]() View larger version (21K): [in a new window] |
FIG. 2. Agreement (mean ± 95% CI) of OPA titers compared to the median OPA titers given in Table 1. The level of agreement to the median was determined in 2 by 2 tables with the formula a/(a + c), where a is the number of sera at or above the given titer by both the test laboratory and the median and c is the number of sera below the given titer by the test laboratory but at or above the given titer by the median.
|
![]() View larger version (20K): [in a new window] |
FIG. 3. Pearson's product moment correlation coefficient (r = 0.79) between log2-transformed data for median OPA titers obtained in this study and the ELISA IgG concentrations (in micrograms/milliliter) assigned to a subset of 7 sera through a multilaboratory evaluation of ELISA methods (18). The reference ELISA concentrations for serum samples 380773 (96-730), 380807 (96-734), 380964 (96-738), 380828 (96-742), 380638 (96-744), 380808 (96-748), and 380860 (96-752) were obtained from the database, as reported to CDC by the participating laboratories. This database was used in the ELISA previously described by Plikaytis et al. (18). The correlation coefficient (r = 0.76, slope = 1.19) between continuous OPA titers (laboratory 5) and ELISA IgG concentrations (n = 49) was similar to the coefficient shown in this figure for median OPA titers (discontinuous data).
|
|
|
|---|
When performing this multilaboratory study for OPA evaluation, the main obstacle encountered by most laboratories was the manipulation of the HL-60 cells. The participating laboratories had the tissue culture capacity; however, ensuring effective differentiation of the HL-60 cells into PMN-like phagocytic cells was difficult for some study participants. HL-60 cells obtained from the American Type Culture Collection under the designation CCL240 were effectively differentiated into PMN-like cells upon the addition of 100 mM DMF. The use of disinfectants which could leave toxic residues in the CO2 incubators, such as formaldehyde, phenol, or sodium hypochlorite, should be avoided. Temperature, CO2 content, and humidity should be constant during cell growth and differentiation. Once the laboratories gained expertise and were familiar with the differentiation protocol with DMF, the study objectives were met in a short period of time and the data were submitted to CDC for analysis. Therefore, we recommend that 100 mM DMF is used for a period of 5 days at 37°C and 5% CO2. Only HL-60 cells that have been actively growing to a density of 4 x 105 to 7 x 105 cells/ml are used for differentiation with DMF. The viability of undifferentiated and differentiated cells is maintained over 90% in RPMI 1640 medium supplemented with 20% fetal calf serum. For this viability OPA, an inoculum of 2 x 105 HL-60 cells/ml should be used for differentiation with a 20% carryover of tissue culture medium from the undifferentiated cell stock. Initially, Romero-Steiner et al. reported a high passage restriction for the differentiation of HL-60 cells (21); however, if the above-mentioned conditions are maintained, there is no longer a passage restriction for the differentiation of this cell line (16). Differentiated cells change from round cells to elongated sickle cell-like cells, and they become phagocytic due to the expression of surface receptors for iC3b, C3b, and Ig G (21). Although monitoring of these receptors is not necessary for this assay, this change in surface markers can be used as an indicator for optimal differentiation. PMN-like HL-60 cells halt the cell division cycle. This is observed in the laboratory as a stable cell count which will eventually start to decline due to cell death. Tissue cultures should be maintained free of mycoplasma contamination, and cultures should be monitored periodically for the presence of mycoplasma.
Other laboratories had difficulty in growing some of the Pnc strains that were sent by CDC (in particular, serotypes 19F and 18C), therefore requiring additional shipments of strains. In general, once all participants had all the materials needed and had gained preliminary experience with the assay conditions, the level of difficulty decreased and the completion of the study became expedited. Most participants found the extended laboratory protocol provided by the CDC useful. This protocol in now available from a website (www.vaccine.uab.edu).
The level of agreement with the median OPA titers varied depending on the participating laboratory. Laboratory 1 had lower levels of agreement (<0.75) with the median OPA titers for 3 of 7 serotypes evaluated. In contrast, laboratory 5 had very high levels of agreement (>0.89) with the median OPA titers for all serotypes tested. It was observed that most laboratories were capable of obtaining OPA titers closer to the median titer if the serum sample was below the level of quantification of the assay. However, as the median OPA titer increased, the reported OPA values diverged more from the median and the level of agreement dropped in some cases. Laboratory 3 had difficulties with the agreement of OPA titers for Pnc serotype 18C. However, participation in this study improved the level of agreement of laboratory 3 for this serotype after completion of the study. Although no clear technical explanation exists for the differences observed (Table 1) with serotypes 6B and 9V, it is possible that these differences may be due to the quantity of capsular Ps during the growth of the target strains (14). The use of the same Pnc reference strains by different testing laboratories is highly recommended to increase assay reproducibility.
This study provides a guide for laboratories that are implementing OPA testing for Pnc vaccine evaluation. The viability OPA can be performed in multiple laboratories with a high degree of interlaboratory reproducibility (75 and 88% of the sera tested were within one or two dilutions of the median titer, respectively). When this viability method for determining OPA titers has been compared to other OPAs, it has been found to be highly sensitive for measuring OPA activity in sera with low antibody concentrations from both infants and adults (12, 23). This OPA highly correlates with the current ELISA IgG method (18). The minimum level of quantification (a titer of 8) of this OPA correlates with the minimum level of ELISA IgG (0.2 µg/ml) conferring protection in infants vaccinated with Ps conjugate vaccine, as recently reported (13). OPA should be used for the evaluation of new Pnc Ps-based vaccines.
We thank Richard R. Facklam, Division of Bacterial and Mycotic Diseases, CDC, for providing the Pnc strains used in this multilaboratory evaluation.
Present address: University of Texas Southwestern Medical Center, Dallas, Tex. ![]()
Present address: School of Medicine, Washington University, St. Louis, Mo. ![]()
|
|
|---|
This article has been cited by other articles:
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Copyright © 2009 by the American Society for Microbiology. For an alternate route to Journals.ASM.org, visit: http://intl-journals.asm.org | More Info»