Comparison of Five Serologic Tests for Diagnosis of Acute Infections by Chlamydia pneumoniae

Transcription

1 CLINICAL AND DIAGNOSTIC LABORATORY IMMUNOLOGY, Sept. 2000, p Vol. 7, No X/00/$ Copyright 2000, American Society for Microbiology. All Rights Reserved. Comparison of Five Serologic Tests for Diagnosis of Acute Infections by Chlamydia pneumoniae KENNETH PERSSON 1 * AND JENS BOMAN 2 Department of Clinical Microbiology, Malmö University Hospital, S Malmö, 1 and Department of Clinical Virology, The University Hospital of Umeå, S Umeå, 2 Sweden Received 24 January 2000/Returned for modification 5 April 2000/Accepted 14 June 2000 Serology is often used to diagnose acute infections by Chlamydia pneumoniae. In this study paired sera from patients with acute respiratory tract infection during an epidemic of C. pneumoniae infections were examined by five different antibody tests. These tests were the complement fixation (CF) test, the microimmunofluorescence (MIF) test, a recombinant enzyme immunoassay (reia) (Medac) based on a recombinant lipopolysaccharide of chlamydia and measuring antibodies to a common chlamydial antigen, and two tests that utilize preparations of C. pneumoniae organisms, the SeroCp-EIA (Savyon) (with preserved lipopolysaccharide) and the LOY-EIA (Labsystems) (without this antigen). Both of the last two tests should measure specific antibodies to C. pneumoniae, although cross-reacting antibodies may also be detected by the SeroCp-EIA. Acute infection of C. pneumoniae was serologically confirmed in 44% of the cases by at least two different tests. Using an expanded gold standard, i.e., the presence of significant reactions in at least two tests, the sensitivity of the CF test was 69%, that of the MIF test was 88%, that of the reia was 89%, that of the LOY-EIA was 96%, and that of the SeroCp-EIA was 92%. Specificity was high for all methods, but adjustments of diagnostic criteria were made to several of the tests. The basis for these adjustments and supportive data are presented. Infections of C. pneumoniae were detected in patients from 8 to 83 years of age. Two peaks in the incidence of such infections were observed: one among young teenagers and a second in adults 30 to 45 years of age, corresponding to parents of young teen-agers. The tests were equally sensitive in different age groups. Reinfections seemed to be rare. Serological diagnosis has been important to unravel the clinical manifestations of acute infections by Chlamydia pneumoniae. Isolation of the organism or detection of its nucleic acid by PCR has extended the diagnostic arsenal. These methods can provide early diagnosis, which might be helpful for accurate treatment. Serology based on analysis of paired sera will detect acute infection even after antibiotic treatment has been started and might be useful to discriminate between infections by C. pneumoniae and Chlamydia psittaci. The complement fixation (CF) test based on the common lipopolysaccharide (LPS) antigen of chlamydia has been used for many years to detect acute infections by C. psittaci. After C. pneumoniae was recognized and infection by this new agent could be confidently diagnosed by the microimmunofluorescence (MIF) test, many cases previously detected by the CF test and thought to be cases of ornithosis were found to in fact be infections by C. pneumoniae (12, 18). Although the CF test can detect acute infections by C. pneumoniae, its sensitivity has been considered low for such infections, especially in reinfection (13). The MIF test has been of paramount importance to detect acute infections by C. pneumoniae and to describe the prevalence of such infections. The performance of the test depends on several factors, including the antigen preparations used and the experience of the person reading the test. The test has been questioned for different reasons. Some have found it unspecific during acute infections due to cross-reactive antibodies (5, 17). Others have questioned its ability to discriminate acute infections, either by being nonreactive where * Corresponding author. Mailing address: Dept. Clinical Microbiology, Malmö University Hospital, S Malmö, Sweden. Phone: Fax: other tests suggest infection (3, 7, 9) or by identifying cases which cannot be confirmed by other means or which seem unlikely for other reasons (10, 11, 14, 15). Serological tests in an enzyme-linked immunoassay (ELISA) format might overcome some of the potential problems with the MIF test. Three new tests have been evaluated in this study and compared to the MIF and CF tests for the serological diagnosis of acute infections by C. pneumoniae. One of the new tests is based on an acylated recombinant lipopolysaccharide from chlamydia (recombinant enzyme immunoassay [reia]) and has been previously evaluated using the same diagnostic criteria (19). The other tests utilize preparations of C. pneumoniae organisms with (SeroCp-EIA) or without (LOY-EIA) LPS. MATERIALS AND METHODS Patients. During an epidemic of infections by C. pneumoniae, 261 consecutive patients with cases of acute respiratory tract infection were chosen for the study. Patients were eligible if paired sera were available and the symptoms indicated respiratory tract infection. Most patients were treated as outpatients. Twentyone cases were excluded from the final assessment, as all tests could not be performed due to lack of serum. Therefore, 240 cases were finally studied. The study group consisted of 103 males and 137 females. The median age for the males was 31 years, with a range of 1 to 90 years, and that of the females was 40 years, with a range of 6 to 75 years. The sera were collected from late 1994 to the first months of Initially, the CF and MIF tests were used to detect cases of acute infection. Later the sera were also examined by reia, LOY-EIA, and SeroCp-EIA. CF test. The CF test was performed by standard procedures. The chlamydial antigen for this test was obtained from the State Serum Institute (Copenhagen, Denmark). A fourfold titer rise or a titer of 64 or higher was considered diagnostically significant. MIF test. The MIF test was performed by previously described methods (4, 23). Slides with antigens of Chlamydia trachomatis, C. psittaci, and C. pneumoniae were obtained from Labsystems OY (Helsinki, Finland). In this test the C. pneumoniae and C. trachomatis antigens have been treated to remove the LPS antigen, which is, however, retained in the C. psittaci antigen. Chlamydial immunoglobulin G (IgG), IgA, and IgM antibodies were determined. All IgM reactions were confirmed after the IgG antibodies had been removed with RF 739

2 740 PERSSON AND BOMAN CLIN. DIAGN. LAB. IMMUNOL. Absorbent (Behringwerke, Marburg, Germany). Fourfold titer rises of IgG and/or IgA and/or a titer of 16 in IgM was considered diagnostically significant. High titers of IgG or IgA that did not change between acute- and convalescentphase sera were not considered diagnostic per se and will be discussed below. reia. The reia kits were kindly provided by the Medac Company (Wedel, Germany). The test is based on a chemically modified recombinant LPS from chlamydia (6). IgG, IgA, and IgM antibodies to the chlamydial LPS are determined separately with this test. IgG antibodies are removed before measuring IgM antibodies to minimize the risk of false- reactions due to rheumatoid factor. Serum pairs with differences in optical density values for IgG or IgA of more than 0.3 or with values of 2.0 were titrated in two, three, or four steps to determine the titer differences. Titers were calculated according to the manufacturer s instructions. A threefold change of titer of either IgG or IgA or a twofold change of both IgG and IgA was considered diagnostically significant (19). IgM titers of 200 or more were accepted as a diagnostic sign, although a titer of 50 should be considered according to the manufacturer. LOY-EIA. The LOY-EIAs (Labsystems OY, Helsinki, Finland) for C. pneumoniae IgG, IgA, and IgM antibodies are indirect solid-phase EIAs based on an antigen from C. pneumoniae devoid of LPS. The analysis with the LOY-EIA was performed by Tamara Tuuminen, Labsystems OY, Helsinki, Finland, without knowledge of the results from the other tests. The results for IgG and IgA are expressed as enzyme immunounits (EIU), which are calculated as: (A sample A blank )/(A calibrator A blank ) n. The assays are calibrated to correspond to the inverted titers of the MIF assay (Labsystems OY), with n 130 for the IgG EIA and n 30 for the IgA EIA. A level of 30 EIU for IgG is considered the limit of detection (corresponding to a titer of 32 in the MIF assay), and a level of 8 EIU in the IgA EIA is the corresponding limit of detection (titer of 8 in the MIF assay). The criterion for a diagnostically significant change of EIU values for IgG and IgA is a 1.5-fold change of EIU in the zones below 130 EIU (for IgG) and 50 EIU (for IgA). When the first sample shows an EIU value of 130 or more for IgG and 50 for IgA, a 1.3-fold change is considered significant. The results of the C. pneumoniae IgM EIA are expressed as a signal/cutoff (S/CO) ratio after subtraction of the blank value. Samples with a S/CO ratio of less than 0.5 are considered negative samples, those with a S/CO ratio of more than 1.1 are considered, and samples with a S/CO ranging from 0.5 to 1.1 are considered equivocal. SeroCp-EIA. Kits for the SeroCp-EIA (Savyon Diagnostics Ltd., St Ashdod, Israel) were kindly provided by the local representative and the manufacturer. Kits for detection of IgG, IgA, and IgM antibodies were used according to the recommendations of the manufacturer and as delineated in the kit inserts. A significant titer change of IgG and/or IgA antibodies was set to be a twofold titer change after consulting the company. After a preliminary assessment of the IgM results, an adjustment of the criterion for a significant IgM titer was introduced. A significant IgM reaction was considered to be twice the cutoff value and not 1.1 times the cutoff at a dilution of 1:105 as the insert of the manufacturer suggested. The reasons for changing the criterion are discussed below. RESULTS Diagnostically significant reactions were found in 118 cases by at least one test (Table 1). When MIF was used as the gold standard, the sensitivity for the CF test was 69% and those for the three different ELISAs were 87 to 95%, with specificities of between 91 and 93% (Table 2). Some cases had significant titer changes detected by the ELISAs but not by MIF. It may be possible that the MIF test is not the most sensitive test and that the gold standard ought to be changed. Therefore, an expanded gold standard was adopted, where a true case was defined as one for which there were significant reactions by at least two tests. Such confirmed reactions were found in 106 cases, while 12 cases (10%) were by only one test and were considered to have false- reactions. Resolved sensitivities, specificities, and predictive values are presented in Table 3. The sensitivity was lowest for the CF test at 69%, while it ranged between 87 and 96% for the other tests. The LOY-EIA was somewhat more sensitive than MIF and reia but not significantly different from the SeroCp-EIA. The specificities were high at 99% for all tests except the SeroCp-EIA, which had some unconfirmed IgM reactions also with the modified criteria used. The patients with confirmed cases consisted of 45 males with a median age of 32 years (range, 8 to 83 years) and 61 females with a median age of 40 years (range, 12 to 75 TABLE 1. Diagnostically significant reactions of different tests No. of tests Positive tests No. of patients Accumulated cases 5 CF, MIF, reia, LOY-EIA, SeroCp-EIA CF, MIF, LOY-EIA, SeroCp-EIA 2 CF, MIF, reia, SeroCp-EIA 2 CF, MIF, reia, LOY-EIA 1 CF, reia, LOY-EIA, SeroCp-EIA 5 MIF, reia, LOY-EIA, SeroCp-EIA 18 Total CF, MIF, LOY-EIA 1 CF, LOY-EIA, SeroCp-EIA 1 MIF, reia, LOY-EIA 1 MIF, LOY-EIA, SeroCp-EIA 5 MIF, reia, SeroCpEIA 1 reia, LOY-EIA, SeroCp-EIA 2 Total MIF, LOY-EIA 2 MIF, reia 1 reia, LOY-EIA 3 LOY-EIA, SeroCp-EIA 1 Total LOY-EIA 1 CF 1 reia 2 MIF 1 SeroCp-EIA 7 Total Cases not included a a There was not enough serum for all tests. years). The rate of infections by C. pneumoniae in males was 45 of 103 (44%), and that in females was 61 of 137 (45%). The age-specific distribution of serologically confirmed acute infections is shown in Fig. 1. Two peaks were evident: one in young teenagers and a second in adults 30 to 45 years of age, corresponding to parents of teenagers. As seen in Table 1, most cases (56%) had reactions in all five tests, while an additional 26% of the cases had significant reactions in four tests, 11% had significant serological signs in three tests, and 7% were in two different tests. CF test and reia. The CF test and the reia are both based on the chlamydial LPS antigen. The sensitivity of the CF test was 69%, and that of the reia was 87%. One or both of the tests reacted in 97 of 106 (92%) of the confirmed TABLE 2. Sensitivity and specificity of the different serological tests, using MIF as the gold standard, in acute infections by C. pneumoniae Test Sensitivity (%) Specificity (%) CF 65/94 (69) 138/146 (95) reia 84/94 (89) 136/146 (95) LOY-EIA 89/94 (95) 133/146 (91) SeroCp-EIA 87/94 (93) 136/146 (93)

3 VOL. 7, 2000 SERODIAGNOSIS OF C. PNEUMONIAE 741 TABLE 3. Performance of the different tests for the diagnosis of acute infections by C. pneumoniae using an expanded gold standard (i.e., with at least two tests showing significant reactions) Test Sensitivity (%) Specificity (%) Predictive value (%) Positive Negative CF 73/106 (69) 133/134 (99) 73/74 (99) 133/166 (80) MIF 93/106 (88) 133/134 (99) 93/94 (99) 133/146 (91) reia 94/106 (89) 132/134 (99) 92/94 (98) 132/146 (90) LOY-EIA 102/106 (96) 133/134 (99) 102/103 (99) 133/137 (97) SeroCp-EIA 97/106 (92) 127/134 (95) 97/104 (93) 127/136 (93) TABLE 4. Significant reactions of different antibody classes Antibody class(es) No. of significant reactions (%) in: CF a MIF b reia LOY-EIA SeroCp-EIA IgG, IgA, IgM 35 (36) 30 (29) 7 (7) IgG, IgM 74 (100) 41 (44) 5 (5) 24 (23) 21 (20) IgA, IgM 18 (19) 16 (16) 15 (14) IgG, IgA 17 (18) 7 (7) 4 (4) IgM 44 (47) 9 (10) 19 (18) 49 (47) IgG 9 (9) 5 (5) 4 (4) 3 (3) IgA 7 (7) 3 (3) 5 (5) Total Unconfirmed c a The CF reaction may be due to IgG or IgM or both, but the antibody class cannot be differentiated by the test. b Only MIF results for IgG and IgM have been included, as this is the routine procedure; no additional case was detected by IgA, however. c Reactions that were significant by only one test and thus were unconfirmed. cases. In 67 of the 73 cases by the CF test (92%), the reia was also. One unconfirmed significant reaction occurred by the CF test, and two such reactions were observed with the reia. In the reia significant antibody reactions were seen most often in IgA (Tables 4 and 5). A titer of 200 was chosen as the cutoff for significant IgM reactions. The manufacturer suggests titers over 50 as and diagnostically significant. With a cutoff value for IgM reactions at 100 in this material, an additional 18 cases would have been obtained, none of which was accompanied by significant changes of IgG or IgA in the same test and all of which were unconfirmed by any other test. Two more cases had IgM titers of between 100 and 200 and were also by other tests and would thus have increased sensitivity for the reia. Decreasing the cutoff limit to 50 as suggested by the insert would have increased the number of true cases obtained by the test by 8 but would also have detected 19 unconfirmed cases. MIF test, LOY-EIA, and SeroCp-EIA. In the MIF test used here, the LPS antigen has been removed from the C. pneumoniae and C. trachomatis antigens but not from the C. psittaci antigen. The LOY-EIA is based on organisms of C. pneumoniae devoid of the LPS antigen, while the SeroCp-EIA uses elementary bodies containing LPS. These three tests therefore measure specific antibodies to C. pneumoniae but may pick up cross-reactive antibodies to the LPS, depending on the preparation of antigen. The sensitivity of the MIF test was 88%, and that of the LOY-EIA was 96%. The difference was statistically significant (chi-square test with Yates correction, P 0.04). The sensitivity of the LOY-EIA was also higher than that of the reia. As seen in Table 5, the LOY-EIA most often showed significant reactions in IgM and IgG, while diagnostic changes were less common with IgA. The sensitivity of the SeroCp-EIA was 92%, and this test showed significant reactions most often in IgM (89%) and much less often in IgG and IgA (36 and 31%, respectively). Seven unconfirmed reactions occurred, with IgA and IgM being the antibody classes involved (Table 6). It was decided that the criteria for the SeroCp-EIA would be a twofold change in IgG and/or IgA and/or a titer of 200 for IgM. Had a titer change of 1.5 times been used instead for IgG and IgA, another four cases confirmed by other methods would have been picked up, although four unconfirmed cases would also have become. If the cutoff for IgM had been lowered to 100 in the SeroCp-EIA as suggested by the insert, another four true cases would have been detected but 25 cases unconfirmed by other tests would also have occurred. There was agreement between the MIF test and the SeroCp-EIA in 223 of 240 cases and between the MIF test and the LOY-EIA in 222 of 240 cases. When the reia or the MIF test was combined with the LOY-EIA, all cases would have been detected (Table 1). Other combinations would have detected between 100 and 105 of the 106 cases. The correlation between the IgG titers found in the MIF test and those determined by LOY- EIA was rather good (r s 0.82 [Spearman s correlation test]). FIG. 1. Rates of serologically confirmed cases of acute infections by C. pneumoniae in different age groups.

4 742 PERSSON AND BOMAN CLIN. DIAGN. LAB. IMMUNOL. TABLE 5. Significant antibody reactivity of each antibody class (confirmed reactions only) Antibody class No. of significant reactions/total (%) in: MIF a reia LOY-EIA SeroCp-EIA IgG 50/93 (54) 60/94 (64) 65/102 (64) 35/97 (36) IgM 84/93 (90) 67/94 (71) 89/102 (87) 86/97 (89) IgA 76/94 (81) 55/102 (54) 30/97 (31) a Only IgG and IgM results are included, as this is the standard procedure. TABLE 7. Diagnostic antibody reactions of the different tests in the Test (criterion) No. with /all cases (%) No. with / all cases in the same test (%) No. with / no. IgM in the same test (%) CF (titer of 64) 55/106 (52) 55/73 (75) MIF (IgM ) 64/106 (60) 64/93 (69) 64/84 (76) reia (IgM ) 52/106 (49) 52/94 (55) 52/67 (78) LOY-EIA (IgM ) 64/106 (60) 64/102 (63) 64/89 (72) SeroCp-EIA (IgM ) 75/106 (71) 75/97 (77) 75/86 (87) Single-serum testing. With the diagnostic criteria chosen, testing of the showed a significant antibody reaction in only 48 to 71% of the true cases in the different tests (Table 7). In the cases of the MIF test, reia, and LOY-EIA, this meant significant IgM reactions. Such reactions in the first serum were detected in 72 to 87% of all IgM- cases. High MIF titers and acute C. pneumoniae infection. High titers, of 256 or more, that were unchanged between acute- and convalescent-phase sera were detected in 21 cases without confirmed acute C. pneumoniae infection. Prior sera were available in 16 of these cases and had been collected on average 6 years before the present disease (range, 1 to 17 years). All of these cases had prior antibody titers. Eleven of the 16 cases had prior titers of 128 or more, and 5 cases had titers of 32 or 64. Among the 106 cases, 18 had high unchanging titers of IgG antibodies to C. pneumoniae. Of these 18 cases, 13 had IgM antibodies in MIF and 5 had diagnostic findings by other tests. Four of these 13 cases with IgM antibodies where prior sera were available had no or low antibody titers in previous sera. Thus, only 5 of 26 cases with a stable titer of 256 or more in MIF had current infection by C. pneumoniae confirmed by other tests, while 21 cases with high titers remained unconfirmed by other tests and also had prior antibodies in most instances. Thirteen cases had stable IgG titers of 512 or more, of which three had serological signs of acute infection by C. pneumoniae by other tests. Previous sera were available in seven cases. Among the three cases with reactions in other tests, previous sera were available in two cases, of which one lacked prior antibody and the other had a low titer of 32 in prior serum. Previous sera from 5 of the 10 cases with high MIF IgG titers only all had antibody titers of 128 or more in prior sera. Reinfections. The CF test has been considered less sensitive for reinfections by C. pneumoniae. It is reasonable to assume that primary infections are more common in teenagers than among adults. The proportions of cases by the CF test and the LOY-EIA were compared for different age groups (Table 8). There was no statistically significant difference between these proportions (chi-square test). IgM antibodies are not expected in reinfections. In this study there was no significant difference in the rate of such antibodies detected by the LOY-EIA test in different age groups (Table 8). Of the 106 cases with serological signs of acute infection by C. pneumoniae, 65 (61%) had IgG antibody titers of 32 or more in the. Titers of 128 or more were detected in 25 (38%) of these 65 cases. We were able to examine previously collected sera from 32 of the 106 cases with acute infection. These sera had been collected on average 7 years earlier, with a range of 1 to 19 years. Most of these prior 32 sera were antibody negative, and only 3 had low titers of 32 or less. Of the 32 cases with prior sera, 9 had a titer of 64 or more in the acute-phase serum. Only three of these nine cases had prior antibodies at a low titer of 32 or less, and six cases were antibody negative. Reinfection was therefore not demonstrated in any case with titers of 64 or more in prior sera. DISCUSSION During an epidemic of acute respiratory tract infections by C. pneumoniae, the diagnosis could be serologically confirmed (i.e., by at least two tests) in 106 of 240 cases of acute respiratory tract infection (44%). In addition, 12 (10%) unconfirmed (i.e., in only one test) significant serological reactions were observed. This is an unusually high frequency of acute infections by C. pneumoniae but reflects the epidemiological situation at that time. Currently, the rate of serological diagnoses of C. pneumoniae infections is 2% in the same area. The spectrum of cases would represent both primary infections and reinfections, as the youngest child with C. pneumoniae infection was 8 years old and the oldest patient was 83 years of age. The use of a proper gold standard is critical. Serological tests should preferably be confirmed by tests that demonstrate the organism, like cell culture or PCR, although these methods also have limitations. In this study this was not possible, as no material was available for such tests. New serological tests may, however, be compared to established ones like the MIF and CF tests, although the latter is considered less sensitive. In this TABLE 6. Summary of unconfirmed reactions by different tests Antibody class(es) No. of unconfirmed reactions by: CF MIF reia LOY-EIA SeroCp-EIA IgG, IgA 1 IgA, IgM 1 IgG 1 IgA 1 1 IgM 1 5 Total TABLE 8. Proportions of cases by the CF test and the LOY-EIA in different age groups Age group (yr) No. by CF/no. by LOY-EIA No. LOY-EIA IgM / total LOY-EIA (%) 20 16/28 (1/1.8) 26/28 (93) /33 (1/1.2) 28/33 (85) 41 31/41 (1/1.3) 35/41 (85) Total 73/102 (1/1.4) 89/102 (87)

5 VOL. 7, 2000 SERODIAGNOSIS OF C. PNEUMONIAE 743 study a commercial MIF test which should be generally available was used. We have previously evaluated reia using exactly the same diagnostic criteria for IgG and IgA as in this study and found good agreement with MIF, although false reactions were observed for parvovirus infections. Therefore, we lacked previous experience only with the LOY- EIA and the SeroCp-EIA. An expanded gold standard was finally adopted, where significant reactions in at least two tests constituted a true- case. Of the 106 cases, 100 were confirmed by MIF and/or CF, which are both established methods. Another five cases were confirmed by reia, which we have evaluated previously and found to have a good specificity with the criteria used. Thus, only 1 of 106 true- cases was included where LOY-EIA and SeroCp-EIA confirmed one another with other tests being negative. The criteria for serodiagnosis of acute chlamydial infection are important. For the CF test, a fourfold titer change or a titer of 64 or more was considered significant. In the case of MIF, a fourfold titer change of IgG and/or IgA antibodies and/or an IgM titer of at least 16 was considered diagnostic. High titers of IgG antibodies have been suggested as a marker for current infection by C. pneumoniae (13, 21), but this has been questioned by several authors (10, 11, 14, 15). In a previous study, titers of 512 or more in our test system did not predict acute respiratory infections, as similar rates of such titers were observed among patients and controls (19). In this study, with a prevalence of C. pneumoniae infections of 44%, only 5 of 26 cases (19%) with an IgG titer of 256 or more in the MIF test had significant antibody reactions by other tests. Among the 21 cases with high IgG titers in the MIF test but lacking significant reactions by other tests, all cases had previous antibodies when prior sera were available (16 sera). Thus, only about 20% of cases with high IgG titers in the MIF test represented current infection as indicated by the other tests, while the majority had high titers in prior sera, probably reflecting previous infection, and lacked diagnostic signs of acute infections by other tests. Diagnostic criteria in the reia for infections of C. pneumoniae or C. psittaci have been established in a previous study (19). It became obvious during this study that the significance of IgM titers is problematic, and a titer of 200 or more was adopted as a diagnostic sign of acute infection. The use of a lower titer would have affected specificity rather than sensitivity. This adjustment of diagnostic criteria therefore seemed justified. The diagnostic criteria for the LOY-EIA were suggested by the manufacturer, and those for SeroCp-EIA were decided after consultation with the manufacturer. The criterion for a IgM reaction in the SeroCp-EIA differs from that given in the insert of the kit. Had a lower titer been used, several additional unconfirmed cases would have occurred. Reinfections would be assumed to be more common at older ages, but the CF test seemed to perform equally well for older patients. In fact, the LOY-EIA demonstrated equal proportions of IgM- cases in different age groups. Reinfection has been suggested to be indicated by a titer rise of IgG antibodies without an IgM response in patients with prior antibodies (13). In this study, 65 cases of C. pneumoniae infection (61%) had IgG antibodies by the MIF test in the acutephase serum and might be reinfections. Where prior sera were available, only 3 of 32 such sera had low-titer antibodies, and 29 were antibody negative. Reinfection could not be confirmed in any case with a prior titer of 64 or more and seemed to be a rare event. In contrast, reinfections have been considered rather common in several reports (1, 2, 8). In those studies, patients with reinfections had milder clinical symptoms than those with primary infection or were asymptomatic. The patients in our study were all symptomatic, and all had sought medical attention for their symptoms. Recently, the reia has been evaluated by three other groups. For single sera the test performance was poor in comparison to culture (16). Even with paired sera, the test failed to detect acute infection in the majority of the cases. Another study reported rather good agreement between different serological tests and detection of C. pneumoniae by PCR or culture (22). A poor correlation was found between reia and PCR in a third study, but only a few cases were detected (20). The MIF test has been considered difficult to perform and has an element of subjective assessment. In this study, all but one reaction suggesting acute infections in the MIF test could be confirmed by other tests. In fact, 84 of the 94 cases by MIF could be confirmed by two or three other tests, and 9 cases were confirmed by one of the other tests, while one reaction remained unconfirmed. The MIF cases were confirmed by an independent LPS-based test in 86 of the 94 cases. The usefulness of serology for diagnosis of acute infections by C. pneumoniae has been questioned. Several authors have not been able to consistently find an antibody response in cases with C. pneumoniae infection demonstrated by culture or PCR (3, 7, 9). Other studies have reported antibody responses regularly in culture- or PCR- cases (4, 24). These differing results cannot be explained at present. Differences in the patients studied may be one factor, as some studies have involved children with chronic conditions like asthma, and other studies have focused on acute respiratory infections in adults. Acute infections are expected to generate an antibody response with titer changes, while detection of C. pneumoniae in long-term asymptomatic cases may be unaccompanied by a significant antibody titer change. It is hoped that these differences will be resolved in the future. We have studied only patients with acute symptomatic respiratory infections. Currently, there is a growing interest in the possible existence of persistent infection of C. pneumoniae and its link to coronary artery disease. Serological markers have been suggested for such persistent infections based on epidemiological studies. The criteria used in this study to diagnose acute infections were based on a dynamic change of antibody titers. Serological markers for latent or persistent infection may be validated in patients where C. pneumoniae DNA from atherosclerotic plaques in arteries or from mononuclear cells in the blood has been detected. In conclusion, three new ELISAs based on different chlamydia antigens are now commercially available. The tests perform well in comparison to the CF and MIF tests. Diagnostic criteria were assessed, and adjustments were adopted for several tests in order to minimize unconfirmed reactions but still preserve high sensitivity. The validation of diagnostic criteria for serodiagnosis of infections by C. pneumoniae can be done by judiciously comparing the results of different serological tests. Using adjusted diagnostic criteria, only a few reactions remained uncorroborated. Although direct detection methods for C. pneumoniae are gaining importance, there will be a continuing interest in serological methods for both C. pneumoniae and C. psittaci. The new tests assessed in this study all seem to be useful, but findings should preferably be confirmed by an established test like the MIF test until further experience has been obtained. ACKNOWLEDGMENTS We are indebted to Tamara Tuuminen, Labsystems OY, Helsinki, Finland, for running all tests with the LOY-EIA and to the Savyon and Medac companies for providing kits for this study.

6 744 PERSSON AND BOMAN CLIN. DIAGN. LAB. IMMUNOL. REFERENCES 1. Aldous, M. B., J. T. Grayston, S. P. Wang, and H. M. Foy Seroepidemiology of Chlamydia pneumoniae TWAR infection in Seattle families, J. Infect. Dis. 166: Berdal, B. P., O. Scheel, A. R. Ögaard, T. Hoel, T. J. Gutteberg, and G. Ånestad Spread of subclinical Chlamydia pneumoniae infection in a closed community. Scand. J. Infect. Dis. 24: Black, C. M., P. I. Fields, T. O. Messmer, and B. P. Berdal Detection of Chlamydia pneumoniae in clinical specimens by polymerase chain reaction using nested primers. Eur. J. Clin. Microbiol. Infect. Dis. 13: Boman, J., A. Allard, K. Persson, M. Lundborg, P. Juto, and G. Wadell Rapid diagnosis of respiratory Chlamydia pneumoniae infection by nested touchdown polymerase chain reaction compared with culture and antigen detection by EIA. J. Infect. Dis. 175: Bourke, S. J., D. Carrington, C. E. Frew, R. D. Stevenson, and S. W. Banham Serological cross-reactivity among chlamydial strains in a family outbreak of psittacosis. J. Infect. 19: Brade, L., H. Brunnemann, M. Ernst, Y. Fu, O. Holst, P. Kosma, H. Näher, K. Persson, and H. Brade Occurrence of antibodies against chlamydial lipopolysaccharide in human sera as measured by Elisa using an artificial glycoconjugate antigen. FEMS Immunol. Med. Microbiol. 8: Chirgwin, K., P. M. Roblin, M. Gelling, M. R. Hammerschlag, and J. Schachter Infection with Chlamydia pneumoniae in Brooklyn. J. Infect. Dis. 163: Ekman, M. R., J. T. Grayston, R. Visakorpi, M. Kleemola, C. C. Kuo, and P. Saikku An epidemic of infections due to Chlamydia pneumoniae in military conscripts. J. Infect. Dis. 17: Emre, U., P. M. Roblin, M. Gelling, W. Dumornay, M. Rao, and M. R. Hammerschlag The association of Chlamydia pneumoniae infection and reactive airway disease in children. Arch. Pediatr. Adolesc. Med. 148: Gaydos, C. A., P. M. Roblin, M. R. Hammerschlag, C. L. Hyman, J. J. Eiden, J. Schachter, and T. C. Quinn Diagnostic utility of PCR-enzyme immunoassay, culture, and serology for detection of Chlamydia pneumoniae in symptomatic and asymptomatic patients. J. Clin. Microbiol. 32: Gaydos, C. A., J. J. Eiden, D. Oldach, L. M. Mundy, P. Auwaerter, M. L. Warner, E. Vance, A. A. Burton, and T. Quinn Diagnosis of Chlamydia pneumoniae infection in patients with community-acquired pneumonia by polymerase chain reaction enzyme immunoassay. Clin. Infect. Dis. 19: Grayston, J. T., C. Mordhorst, A. L. Bruu, S. Vene, and S. P. Wang Countrywide epidemics of Chlamydia pneumoniae, strain TWAR, in Scandinavia, J. Infect. Dis. 159: Grayston, J. T., L. A. Campell, C. C. Kuo, C. H. Mordhorst, P. Saikku, D. H. Thom, and S. P. Wang A new respiratory tract pathogen: Chlamydia pneumoniae strain TWAR. J. Infect. Dis. 161: Hyman, C. L., P. M. Roblin, C. A. Gaydos, T. C. Quinn, J. Schachter, and M. R. Hammerschlag Prevalence of asymptomatic nasopharyngeal carriage of Chlamydia pneumoniae in subjectively healthy adults: assessment by polymerase chain reaction-enzyme immunoassay and culture. Clin. Infect. Dis. 20: Kern, D. G., M. A. Neill, and J. Schachter A serological study of Chlamydia pneumoniae in Rhode Island: evidence of serological cross-reactivity. Chest 104: Kutlin, A., N. Isomura, U. Emre, P. M. Roblin, and M. R. Hammerschlag Evaluation of chlamydia immunoglobulin M (IgM), IgG, and IgA relisas Medac for the diagnosis of Chlamydia pneumoniae infection. Clin. Diagn. Lab. Immunol. 4: Ozanne, G., and J. Lefebvre Specificity of the microimmunofluorescence assay for the serodiagnosis of Chlamydia pneumoniae infections. Can. J. Microbiol. 38: Persson, K., and J. Treharne Diagnosis of infection caused by Chlamydia pneumoniae (strain TWAR) in patients with ornithosis in Southern Sweden Scand. J. Infect. Dis. 21: Persson, K., and S. Haidl Evaluation of a commercial test for the chlamydial lipopolysaccharide (Medac) for the serodiagnosis of acute infections by Chlamydia pneumoniae (TWAR) and Chlamydia psittaci. APMIS 108: Petitjean, J., F. Vincent, M. Fretigny, A. Vabret, J. D. Poveda, J. Brun, and F. Freymuth Comparison of two serological methods and a polymerase chain reaction-enzyme immunoassay for the diagnosis of acute respiratory infections with Chlamydia pneumoniae in adults. J. Med. Microbiol. 47: Thom, D. H., J. T. Grayston, L. A. Campell, C. C. Kuo, V. K. Diwan, and S. P. Wang Respiratory infection with Chlamydia pneumoniae in middleaged and older adult outpatients. Eur. J. Microbiol. Infect. Dis. 13: Verkooyen, R. P., D. Willemse, C. A. M. Hiep-van Casteren, S. A. Mousavi Joulandan, R. J. Snijder, J. M. M. van den Bosch, H. P. T. van Helden, M. F. Peeters, and H. A. Verbrugh Evaluation of PCR, culture, and serology for diagnosis of Chlamydia pneumoniae respiratory infections. J. Clin. Microbiol. 36: Wang, S. P., and J. T. Grayston Immunologic relationship between genital TRIC, lymphogranuloma venereum and related organisms in a new microtiter indirect immunofluorescence test. Am. J. Ophthalmol. 70: Wang, S. P., and J. T. Grayston Chlamydia pneumoniae (TWAR) micro-immunofluorescence antibody studies 1998 update, p In R. S. Stephens, G. I. Byrne, G. Christiansen, I. N. Clarke, J. T. Grayston, R. G. Rank, G. L. Ridgway, P. Saikku, J. Schachter, and W. E. Stamm (ed.), Chlamydial infections. International Chlamydia Symposium, San Francisco, Calif.