JOURNAL OF CLINICAL MICROBIOLOGY, Aug. 1992, p. 2158-2164
Vol. 30, No. 8
0095-1137/92/082158-07$02.00/0 Copyright © 1992, American Society for Microbiology
Retrospective Evaluation of Two Latex Agglutination Tests for Detection of Circulating Antigens during Invasive Candidosis PHILIPPE
HERENT,1 DIRK STYNEN,2 FERNANDO HERNANDO,3 JEANINE FRUIT,1
AND DANIEL POULAINl* Institut National de la Sante et de la Recherche Medicale, Unite 42, 369 rue Jules Guesdes, F-59650 Villeneuve d'Ascq, and Laboratoire de Parasitologie et Mycologie, Faculte de Me6decine, Place de Verdun, F-59000 Lille, France'; Sanofi Diagnostics Pasteur, B-3600 Genk, Belgium2; and Departamento de Microbiologia e Inmunologia, Facultad de Ciencias, Universidad del Pais Vasco, E-48080 Bilbao, Spain3
Received 3 February 1992/Accepted 4 May 1992
Two latex agglutination tests for the detection of Candida antigens, Pastorex Candida (Sanofi Diagnostics Pasteur, Marnes-la-Coquette, France) and Cand-Tec (Ramco Laboratories, Inc., Houston, Tex.), were applied to 79 serum samples from 19 patients who were retrospectively selected on the basis of mycological and clinical evidence of C. albicans infection and the availability of serial serum samples taken near the date of a positive culture. The specificity in 60 control individuals was 100% for Pastorex and 98.3% for Cand-Tec. The tests scored positive for 10 (52.6%) and 9 (47.4%) patients, respectively. Pastorex detected antigen in only 3 of 12 patients (25%) with positive antibody detection tests, but was positive for all 7 patients (100%Y) who produced no or a low antibody response, suggesting that the test performs better in the absence of antibodies. However, the sensitivity of Pastorex also increased with the number of samples available per patient, which was lower for high-antibody-responder patients (2.8 versus 5.7). If the patients who provided only one or two serum samples were eliminated, the sensitivity of Pastorex rose to 76.9%. For the Cand-Tec, the sensitivity was not related to the presence of antibodies, nor was it related to the number of samples per patient. The observed antigenemia was transient with both Pastorex and Cand-Tec. Only 12.5% of the positive reactions occurred on the same serum sample, confirming that the two tests react with different antigens. A positive antigen test preceded other diagnostic indications for 6 of 10 Pastorex-positive patients and 5 of 9 Cand-Tec-positive patients. In the past 15 years, the incidence of hospital infections caused by yeasts of the genus Candida has shown a steady increase (1, 31, 34). Since the clinical symptoms of deep candidosis are nonspecific (4), the mycological laboratory can play an essential role by providing assistance to the clinician with the establishment of a diagnosis. Furthermore, this biological diagnosis must be made early, since timely initiation of antifungal therapy is essential for a favorable prognosis (1, 31). To reach these goals, three types of biological procedures are available, namely, mycological examination and tests for detection of antibodies and antigen; each of these has its limitations. The subject has been extensively discussed in the literature (2, 9, 15, 27). It is generally agreed that these three types of tests may complement each other and that mycological and serological surveys of high-risk patients, defined according to clinical data, are necessary. The aim of this study was to evaluate the diagnostic value of two commercially available latex agglutination tests, i.e., Cand-Tec (Ramco Laboratories, Inc., Houston, Tex.) which has been available for several years (7, 11, 12), and Pastorex Candida (Sanofi Diagnostics Pasteur, Marnes-la-Coquette, France), which was introduced more recently (13, 23). This evaluation was made in the context of a routine mycological laboratory to which mycological and serological samples are sent at the initiative of clinicians. It involved 19 patients who were retrospectively selected on the basis of mycological and clinical evidence of deep C. albicans infection and the availability of several *
serial serum samples taken near the date of a positive culture of samples from normally sterile sites. Our routine serological procedure involves the detection of antibodies by cocounterimmunoelectrophoresis (Co-CIE) and an indirect immunofluorescence assay (IFA) (29) plus antigen detection by the Cand-Tec test in cases of nonsignificant antibody levels (12). The objective of the present analysis was therefore to assess the advantages, if any, of Pastorex Candida compared with those of Cand-Tec.
MATERLILS AND METHODS Human sera. (i) Sera from patients with C. albicans infection. The following criteria were applied to laboratory and clinical files as retrospective selection rules: (i) selection of patients having presented evidence of the presence of C. albicans, which was isolated at least once in large quantities from normally sterile sites (blood, catheter, drain, wound, peritoneal cavity, gastric or lung biopsy specimens, bile, pancreatic cyst, pericardial fluid); (ii) availability of serial serum samples for the corresponding patients taken near the date of positive cultures; and (iii) subsequent analysis of the selected patients' clinical files (together with the physicians in charge of the patients), leading to consideration of candidosis as highly probable once it was integrated into the underlying conditions (e.g., fever resistant to antibacterial agents and response to antifungal therapy). Application of this procedure for the period between February 1988 and May 1989 led to the selection of 73 serum samples from 18 patients from the surgery, intensive care, and hematology wards of the Lille University Hospital.
Corresponding author. 2158
VOL. 30, 1992
Details concerning patients with C. albicans infection are given in Tables 1 and 2. In Tables 1 and 2, the data are for 19 patients, not 18 patients, since 1 patient who underwent two episodes of invasive candidosis in a 1-year interval (separated by a total recovery) was treated as two different patients (patients A9 and B5). (ii) Control sera. Four groups of control individuals (a total of 60 individuals) were tested in order to determine the specificity of the methods. The first group consisted of 10 members of the laboratory staff, the second group included 20 surgery patients without any sign of clinical candidosis, the third group encompassed 10 patients with immunologically confirmed aspergillosis and 2 patients with allergic bronchopulmonary aspergillosis (3), and the fourth group included 18 patients in acute phases of parasitic diseases, as follows: 13 patients with malaria and 1 patient each with fascioliasis, schistosomiasis, filariasis, echinococciasis, and amebiasis. Mycological examinations. Cultures were grown on Sabouraud glucose agar containing kanamycin (800 ,g/ml). The isolates were identified as C. albicans on the basis of germ tube (20) and chlamydosporulation (32) tests. Antibody detection. Two methods for antibody detection were used. (i) Co-CIE. Co-CIE (bio Merieux, Marcy l'Etoile, France) has been described in detail before (29). It essentially consists of a counterimmunoelectrophoresis on cellulose acetate membranes in which three 20-,ul serum deposits, placed at the anode, are run against a line of antigen consisting of 15 ,ul of a 25-mg/ml solution of somatic antigen from C. albicans VW32 placed at the cathode. Patient sera are run side by side with a rabbit anti-VW32 germ tube antiserum to detect precipitating antibodies that present a line of identity with anti-germ tube antiserum. This line of identity is named the cospecific precipitin line (CSPL). Interpretation of Co-CIE results refers to the presence or absence of CSPL and quantitative evolution, namely, migration toward the cathode and variation in intensity for successive sera from the same patient run side by side with the anti-germ tube antiserum. (ii) IFA. IFA was performed against unfixed blastoconidia of a C. albicans VW32 serotype A cloned strain (bioMerieux,) incubated with doubling dilutions (initial dilution, 1/100) of patient sera in phosphate-buffered saline (PBS). Antibody binding was revealed by incubation with a fluorescent goat anti-human immunoglobulin antiserum (1/100 in PBS containing Evans blue at a final dilution of 1/50,000); Sanofi Diagnostics Pasteur). Results were expressed as the reciprocal titer, i.e., the inverse of the highest serum dilution that gave strong fluorescence of all blastoconidia. Standardization of this test was achieved by the use of a pool of sera from patients with candidosis (bio-Merieux). A titer equal to or greater than 400 was considered indicative of candidosis. Relying on the results obtained with Co-CIE and IFA, the following conclusions were assigned to the serum samples: N, no immunological suspicion of candidosis (no CSPL and IFA < 400); S, immunologically suspected candidosis (presence of CSPL and/or IFA > 400); C, immunologically confirmed candidosis (strong CSPL and IFA > 400). Detection of circulating antigens. Both antigen detection kits were used according to the manufacturer's instructions. The Cand-Tec test (Ramco) uses latex beads coated with a polyclonal rabbit anti-Candida serum against an unknown antigen (12). The test is positive when serum, which is diluted 1/4, agglutinates the latex beads. Serial dilutions of serum were tested to determine the antigen titer, which is
ANTIGENEMIA IN CANDIDOSIS
2159
defined as the reciprocal of the highest serum dilution that still causes agglutination. In the Pastorex Candida test (Sanofi Diagnostics Pasteur), the latex beads were sensitized with a monoclonal antimannan antibody (23). Serum samples were incubated with a serum treatment solution at 100°C. After centrifugation, the supematants were incubated with the latex beads. A serum sample scored positive when undiluted supernatant agglutinated the latex beads. Serial dilutions of the supernatant were tested so that we could obtain semiquantitative results. RESULTS The patients were divided into two groups according to their antibody responses as analyzed by Co-CIE and IFA. Group A consisted of 12 patients with immunologically confirmed candidosis, as established by these routine procedures. The average number of serum samples per patient in this group was 2.8 + 1.3 (Table 1). Group B included seven patients who failed to produce an antibody response, which was considered indicative of infection, but they were judged to suffer from an invasive candidosis on clinical and mycological grounds. The average number of serum samples per patient in this group was 5.7 ± 3.1 (Table 2). For group A patients (Table 1), Cand-Tec was positive at a significant titer for 5 of 12 patients, whereas Pastorex Candida was positive for 3 patients. For group B patients (Table 2), four of seven patients were positive by Cand-Tec; Pastorex Candida yielded positive results for all seven patients. When the results for both patient groups were combined, 9 of 19 cases scored positive by Cand-Tec, as opposed to 10 of 19 by Pastorex Candida. Patients were often positive by only one test (Tables 1 and 2). Five patients (patients A6, All, B3, B4, and B6) were negative by Cand-Tec but positive by Pastorex Candida, while four patients (patients A3, A4, A9, and A12) remained negative by Pastorex Candida but were positive by CandTec. Both tests failed to detect antigen in five patients (patients Al, A2, A5, A8, and A10). A remarkable feature shared by both tests in both groups of patients was that when more than one serum sample was available per patient, antigenemia was always transient. For the serum samples tested and sent at the initiative of clinicians, the minimum delay observed for the appearance of detectable antigen by Cand-Tec and Pastorex were, respectively, 4 days (patient B7) and 1 day (patient A6); conversely, it was possible to observe the disappearance of detectable antigens in both tests within 48 h (patient B5, Cand-Tec; patient B7, Pastorex). The longest antigenemia period detected without any intervening negative test was of 47 days and concerned the Pastorex Candida in patient B5. Figure 1 shows the sensitivities of both tests as a function of the number of serum samples tested per patient. The sensitivity of the Pastorex Candida increased when there were increasing numbers of serum samples available for each patient. In the entire patient population, the sensitivity of Pastorex was 52.6%. When data for the patients who provided only one or two samples were omitted, the sensitivity increased to 76.9%, eventually attaining 100% in patients who provided five or more serum samples. The Cand-Tec test did not show such a correlation. Both tests were negative for samples from the healthy controls and from surgery patients (Table 3). In patients with different fungal and parasite infections, one sample was positive by Cand-Tec. This sample came from a patient with aspergillosis. For this patient, the Candida serology tests,
2160
HERENT ET AL.
J. CLIN. MICROBIOL.
TABLE 1. Patient information, culture data, and results of antibody and antigen testing in patients with an antibody response to C. albicans (group A) Patient
Al A2
Sex'
M M
Age (yr)
63
Underlying condition Pancreas carcinoma
86
Site
ofn'
Bile Catheter
sera were Co-CIEd Conclusionf yFo IFCCand-Tec collectedc
-2 8
4 3
1,600 400
C C
2 (NS)
-10 15
0 3
800
N C
2 (NS) -
M
27
Ileostomy Jejunostomy
Blood (day 0) Drain (day 6)
1 -7
4 3
1,600 400
C C
4
A4
M
51
Oesophagal neoplasm
Drain
1 12 18
3 4 4
400 400 800
C C C
4
AS
M
62
Duodenal perforation
Peritoneal cavity
-1 13
0 4
100 1,600
N C
2 (NS)
A6
M
18
Liver transplant
Gastric biopsy
-5 -4 16
2 3 4
200 200 800
S S C
2 (NS)
-18 -12 -4 4
0 2 3 3
200 400 400 400
N S C C
8 2 (NS) 4 2 (NS)
0 7
2 3
400 800
S
-
C
-
-17 -1 4
2 2 3
100 400 400
10 21 34 >57
2 2 2 4
200 400 400 1,600
-5 0 3 5 14
2 3 3 3 3
400 800 800 800 800
0
2
400
M
43
Non-Hodgkin's malignant lymphoma Pneumopathy
Lung biopsy
A8
M
49
Chronic pancreatitis
Pancreatic cyst
A9
A10
All
A12
F
M
M
M
69
64
46
70
Liver cancer
Sternal resynthesis
Bronchopneumopathy
Duodenostomy Gastrostomy
Blood
Pericardial fluid (day 0) Median wound (day 27) Blood
Peritoneal cavity
S
-
-
1
2 (NS)
C
2 (NS) 2 (NS) 4
S
2 (NS)
S
-
_
A3
A7
Pastorex
2
S S C
C
-
C
-
1
C
-
-
C
-
-
C
-
-
C
M, male; F, female. Unless otherwise indicated, yeasts were isolated on day 0. Indicated as the days before (negative numbers) or after the first isolation of C. albicans. d 0, absence of a precipitin line; 1, nonspecific precipitin line; 2, cospecific precipitin line; 3, intense cospecific precipitin line; 4, precipitin line. e Reciprocal titer. NS, not significant. -, negative. f N, no immunological suspicion of candidosis; S, immunologically suspected candidosis; C, immunologically confirmed candidosis.
-
4
a
b
c
however, also yielded results compatible with immunologically confirmed candidosis. The other serum samples were all negative by the Pastorex Candida and Cand-Tec tests. DISCUSSION The presence of Candida antigens in serum appears to be specific indicator of invasive disease (2, 9, 15, 22, 24, 28, 33). However, the low and varying antigen concentrations, which often fall below the detection limits of the available a
very
intense cospecific
methods (5, 26, 33), also lead to controversy about the clinical usefulness of antigen detection tests. In this study, we compared the results of culture and of antibody and antigen detection in a group of 19 patients with or without an antibody response. Special emphasis was placed on the role of antigen detection and, more particularly, on the comparison of two commercially available test kits. We found an overall sensitivity of 47.5% for Cand-Tec and 52.6% for Pastorex Candida (Tables 1 and 2). The Cand-Tec test has been the subject of numerous studies which have yielded
2161
ANTIGENEMIA IN CANDIDOSIS
VOL. 30, 1992
TABLE 2. Patient information, culture data, and results of antibody and antigen testing in patients with an antibody response to C albicans (group B) Patient
Sex°
Age (yr)
Underlying condition
Site of yeast
CoyCIE sera were
IF
Conclusio
collected IlAl
Bi
F
23
Acute
lymphoblastic --ia ---
Blood
leukemia
B2
B3
B4
B5
B6
B7
F
M
M
F
M
F
46
21
65
69
43
50
Gastrectomy
Acute lymphobastic leukemia
Ileostomy
Liver cancer
Acute myelomonocytic leukemia
Catheter
Blood
Blood (day 0) Blood (day 1)
Bile
Blood
Drain
-14 -11 -10 3
1 2 1 1
100 100 100
-1 5 14
0 2 2
100 200 200
-1 15 >30
0 0 2
-21 -16 -13 -8 -5 -2 -1 0 8 11 12 28
N N N N
Cand-Tec
Pastorex
-
-
4
4
N S S
4
4
100 200 200
N N
2 (NS)
1 0 1 2 2 2 2 2 2 2 1 0
200 200 200 200 200 200 200 200 200 200 200 100
N N N S S S S S S S S N
-3 2 4 10 48 51 56
1 2 0 1 2 1 0
100 200 200 100 200 200 100
N S N N S S N
-30 -13 1 8 14 >30
0
100 100 100 100
N N N N N N
-9 -7 8 10 14
1 1 1 0 0
400 400 200 200 200
S S
1 1 1 0 0
2 (NS)
S
-
-
-
2 (NS) 2 (NS) 2 (NS) 2 (NS) 2 (NS) 2 (NS)
2 (NS)
1
2 (NS) 4 4 2 (NS) 2 (NS)
1 4 1 2
4
2 (NS) 2 (NS)
2
2 (NS)
1 4
S
N N
1
-
4
F, female; M, male. Unless otherwise indicated, yeasts were isolated on day 0. Indicated as the days before (negative numbers) or after the first isolation of C. albicans. d 0, absence of precipitin line; 1, nonspecific precipitin line; 2, cospecific precipitin line; 3, intense cospecific precipitin line; 4: very intense cospecific precipitin line. I Reciprocal titer. NS, not significant. -, negative. f N, no immunological suspicion of candidosis; S, immunologically suspected candidosis; C, immunologically confirmed candidosis. I
b
I
variable results; the sensitivity of the test has ranged from 19 to 91% (7, 8, 10-12, 16, 28). In an early study of Pastorex Candida, Meulemans et al. (23) found 5 positive patients of 10 autopsy-proven cases, but the number of serum samples per patient tested in that study was only one to three. The number of serum samples per patient is very important, since our data showed that both tests are liable to become positive or negative within 24 to 48 h. In animal models, mannan clearance from the blood has been proven to be very rapid. The antigen accumulated in the liver and
very
the spleen and was also excreted into the urine (18). Because of the efficiency of the clearance mechanism, mannan concentrations remain low and may drop below the detection limit of a reverse passive latex agglutination test. Patients at risk or patients suspected of being at risk should therefore be tested as frequently as possible. Our results with Pastorex Candida confirm that the sensitivity of the test increases with the number of serum samples available per patient. The average number of serum samples provided by the Pastorexnegative patients was 2.3 versus an average of 5.2 serum
2162
HERENT ET AL. 100
J. CLIN. MICROBIOL.
TABLE 3. Result of antibody and antigen testing in control populations
T
80 +
Population (no. of subjects)
Patient no.
Co-Ciea
Serology by IFAb
Antigen detection by Cand-
Tec"
60
+
40
+
Healthy individuals (10)
1
0
200
-
2 3
0 0 0 2
100 100
-
200
-
4-10 Surgery patients (20)
20 +
.4 .5 .3 (5) (13) (9) n samples/patient (n patients/category)
.2 (17)
.6 (3)
1 2-13 1 1 1 1 1
0 0 0 0 0 0 0
200
16-20 Aspergillosis (10)
FIG. 1. Sensitivity of Pastorex Candida (0) and Cand-Tec (0) antigen detection tests in relation to the number of serum samples from each patient tested; the number of patients entering each category was defined as the minimal number of serum samples tested.
samples provided by the Pastorex-positive patients. When patients who provided only one or two serum samples were omitted from the analysis, the sensitivity of Pastorex rose to 76.9%. This correlation between sensitivity and number of serum samples could not be observed with Cand-Tec, but it has been described by Burnie and Williams (7). In their study, the average number of samples per patient was 3.1 for Cand-Tec-negative patients and 9.7 for Cand-Tec-positive ones. In the same context, the time interval between samples is important. It is noteworthy that the average interval between the time that serum samples were obtained from four of five patients who were negative by both tests was 10 days or more. We observed discordant results between the two tests in nine patients: five patients were negative by Cand-Tec and positive by Pastorex, while four patients were negative by Pastorex and positive by Cand-Tec. Only 12.5% of the serum samples were positive when they were tested by using both tests. This observation can be explained by the fact that the two latex agglutination tests detect different antigens. The rat monoclonal antibody EB-CA1, which is used in the Pastorex test, recognizes mannan, a major cell wall polysaccharide, from C. albicans, C. tropicalis, C. glabrata, C. guilliermondii, and to a lesser extent, C. pseudotropicalis and C. parapsilosis (14). The antigen detected by Cand-Tec, however, has not been identified, but it is certainly different from mannan (12, 30). The high degree of discord between the two tests suggests that the detection of more than one antigen may improve the sensitivity. The prognosis of the patient improves when a diagnosis can be established in an early phase of the infection (31). It is therefore important to see whether the antigen tests would have allowed an earlier diagnosis. In the nine positive patients, the Cand-Tec test turned positive in five patients before cultures became positive or antibody levels considered as significative of candidosis were detected (29). In the 10 Pastorex-positive patients, Pastorex Candida was the first indication of invasive disease in six of them. Both latex
200 100 200 100
1 2 3 4 5 6 7 8 9 10 2
1 1 0 0 0 1 2 2 2 0 3 3 2 0 0 0
2-3 4-6 7-8 9-15
0
.1 (19)
1
Allergic bronchopulmonary aspergillosis (2) Malaria (13)C Fascioliasis (1) Schistosomiasis (1) Filariasis (1) Echinococcosis (1) Amebiasis (1)
-
2 (NS) -
100
-
100 200
-
400 200 100
-
4 -
-
-
200
-
a 0, absence of precipitin line; 1, nonspecific precipitin line; 2, weak cospecific precipitin line; 3, intense cospecific precipitin line; 4, very intense cospecific precipitin line. b Reciprocal titer. NS, not significant. -, negative. No antigen was detected by Pastorex. c Six of these patients presented with a Cand-Tec titer of 2.
agglutination tests were preceded by positive cultures in three cases and by positive serological tests in one case. We considered antigenemia as positive before culture if the antigen test was positive on day 1 at the latest, since mycological cultures take, on average, 72 h of incubation before they can be interpreted. Our results confirm that circulating antigens appear early in the course of the infection, and therefore, they often provide the clinician with information that allows for the earlier initiation of treatment. Both tests yielded comparable results in this respect. Although the number of patients examined in this study was too small to allow firm conclusions to be made, Pastorex seemed to be a good test for use on immunocompromised patients, because it was positive for all seven patients with a weak serological response, while Cand-Tec was positive for four patients (Table 1). Several other investigators also failed to obtain very good results by Cand-Tec in patients with leukemia (5, 10, 25, 28). On the other hand, Pastorex seemed to be more sensitive to the presence of specific antibodies in the patient's serum (17) (Table 2). This is a surprising observation, because unlike Cand-Tec, the Pastorex Candida protocol includes a serum treatment to dissociate immune complexes. In the four patients whose sera were
ANTIGENEMIA IN CANDIDOSIS
VOL. 30, 1992
positive only by Cand-Tec, the antibody levels were high. Moreover, the majority of Pastorex-positive samples belonged to patients who failed to produce an antibody response to C. albicans (group B). Conversely, in the antibody responder group (group A), only three patients were positive. Among those three patients, only one exhibited high antibody levels at the date of positive antigenemia. This raises the possibility that the patients who respond to the Candida infections immunologically are able to clear mannan from their blood more efficiently (18), leading to a lower incidence of detectable antigenemia. However, since the average number of serum samples per patient was lower in the antibody responder group (2.8 versus 5.7), it cannot be excluded that the lower sensitivity of Pastorex Candida in this group was due to this lower number of serum samples. In this study, both tests were very specific; all 60 control samples were negative by Pastorex Candida (100% specificity), and only 1 control sample was positive by Cand-Tec (98.3% specificity) (Table 3). This positive sample, which was from a patient with aspergillosis, may have been a true-positive result, since serological results were consistent with immunologically confirmed candidosis. For Pastorex Candida, the high specificity is in agreement with earlier findings (23). The specificity of Cand-Tec reported in other studies varied considerably (6, 8, 12, 16, 25, 26, 28). Results of our study suggest that the laboratory diagnosis of invasive candidosis benefits from an integrated, kinetic approach in which patients suspected of having candidosis or, preferably, patients at risk of candidosis are submitted to mycological surveillance, serological tests, and antigen detection at regular time intervals. This approach allows physicians to follow kinetics of antibody fluctuations (29) which complements or relates with (28) results from antigen testing. Depending on the antigen detected, immunoassays for the detection of circulating fungal components may constitute specific markers for invasive candidosis (15, 21, 22, 24, 33). Provided that testing is not restricted to isolated serum samples and that patients are regularly screened, antigen testing is also a sensitive method and is especially useful in immunocompromised patients for whom antibody detection tests remain negative. Moreover, a positive antigen test is often the earliest indication of invasive infection. Although the results of this retrospective study suggest that Pastorex Candida is a useful new tool for the diagnosis of invasive candidosis, more studies are necessary to confirm these results. Further investigations, including prospective studies, are needed to firmly establish the clinical value of this test and to assess its reproducibility with different production batches.
ACKNOWLEDGMENTS We thank Laurence Richard, Nadine Francois, and Gilbert Lepage for valuable technical assistance. REFERENCES 1. Armstrong, D. 1989. Problems in management of opportunistic fungal diseases. Rev. Infect. Dis. 7(Suppl. II):1591-1599. 2. Bennet, J. 1987. Rapid diagnosis of candidiasis and aspergillosis. Rev. Infect. Dis. 9:398-402. 3. Biguet, J., P. Tran Van Ky, S. Andrieu, and J. Fruit. 1964. Analyse immunoelectrophoretique d'extraits cellulaires et de milieux de culture d'Aspergillus fumigatus par des immuns6rums experimentaux et des serums de malades atteints d'aspergillome bronchopulmonaire. Ann. Inst. Pasteur 107:7297. 4. Bodey, G. P., and V. Fainstein. 1985. Systemic candidiasis, p. 135-169. In G. P. Bodey and V. Fainstein (ed.), Candidiasis.
2163
Raven Press, New York. 5. Bougnoux, M. E., C. Hill, D. Moissenet, M. Feuilhade de Chauvin, M. Bonnay, I. Vincens-Sprauel, F. Pietri, M. McNeil, L. Kaufman, J. Dupouy-Camet, C. Bohuon, and A. Andremont. 1990. Comparison of antibody, antigen, and metabolite assays for hospitalized patients with disseminated candidiasis. J. Clin. Microbiol. 28:905-909. 6. Burnie, J. P. 1985. A reverse passive latex agglutination test for the diagnosis of systemic candidiasis. J. Immunol. Methods 82:267-280. 7. Burnie, J. P., and J. Williams. 1985. Evaluation of the Ramco latex agglutination test in the early diagnosis of systematic candidiasis. Eur. J. Clin. Microbiol. 4:98-101. 8. Cabezudo, I., M. Pfaller, T. Gerarden, F. Koontz, R. Wendel, R. Gingrich, K. Heckman, and C. Burns. 1989. Value of the Cand-Tec Candida antigen assay in the diagnosis and therapy of systemic candidiasis in high-risk patients. Eur. J. Clin. Microbiol. Infect. Dis. 8:770-777. 9. de Repentigny, L. 1989. Serological techniques for diagnosis of fungal infection. Eur. J. Clin. Microbiol. 8:362-375. 10. Escuro, R. S., M. Jacobs, S. L. Gerson, A. R. Machicao, and H. M. Lazarus. 1989. Prospective evaluation of a Candida antigen detection test for invasive candidiasis in immunocompromised adult patients with cancer. Am. J. Med. 87:621-626. 11. Fung, J., S. Donta, and R. Tilton. 1986. Candida detection system (Cand Tec) to differentiate between Candida albicans colonization and disease. J. Clin. Microbiol. 24:542-547. 12. Gentry, L. O., I. D. Wilkinson, A. S. Lea, and M. F. Price. 1983. Latex agglutination test for detection of Candida antigen in patients with disseminated disease. Eur. J. Clin. Microbiol. 2:122-128. 13. Georges, E., M. L. Garrigues, J. L. Poirot, J. P. Masini, and M. C. Meyohas. 1991. Diagnostic des candidoses sytemiques par un test au latex. R6sultats d'une annee de suivi serologique chez des patients a risque. J. Mycol. Med. 1:25-28. 14. Georges, E., M. L. Garrigues, D. Stynen, and J. L. Poirot. 1991. Specificite in-vitro d'un anticorps monoclonal (AcM) anti-mannane et du latex sensibilise par cet AcM au cours de la candidose dissemin6e experimentale. J. Mycol. Med. 1:21-25. 15. Jones, J. M. 1990. Laboratory diagnosis of invasive candidiasis. Clin. Microbiol. Rev. 3:32-45. 16. Kahn, F. W., and J. M. Jones. 1986. Latex agglutination tests for detection of Candida antigens in sera of patients with invasive candidiasis. J. Infect. Dis. 153:579-585. 17. Kappe, R. 1988. Coexistence of free antigens, free antibodies and immune complexes in sera from patients with suspected deep-seated candidosis. Mycoses 32:24-32. 18. Kappe, R., and J. Muller. 1991. Rapid clearance of Candida albicans mannan antigens by liver and spleen in contrast to prolonged circulation of Cryptococcus neoformans antigens. J. Clin. Microbiol. 29:1665-1669. 19. Lemieux, C., G. St-Germain, J. Vincelette, L. Kaufnan, and L. de Repentigny. 1990. Collaborative evaluation of antigen detec-
tion by commercial latex agglutination test and enzyme immunoassay in the diagnosis of invasive candidiasis. J. Clin. Microbiol. 28:249-253. 20. Mackenzie, D. W. R. 1962. Serum tube identification of Candida albicans. J. Clin. Pathol. 15:563-565. 21. Matthews, R., and J. Burnie. 1988. Diagnosis of systemic candidiasis by an enzyme-linked dot immunobinding assay for a circulating immunodominant 47-kilodalton antigen. J. Clin. Microbiol. 26:459-463. 22. Matthews, R. C., J. P. Burnie, and S. Tabaqchali. 1987. Isolation of immunodominant antigens from sera of patients with systemic candidiasis and characterization of serological response to Candida albicans. J. Clin. Microbiol. 25:230-237. 23. Meulemans, L., A. Andremont, F. Meunier, A. M. Ceuppens, M. L. Garrigues, and D. Stynen. 1991. Pastorex Candida, a new latex agglutination test for mannan detection in serum of patients with invasive candidosis, p. 233-236. In E. Tumbay, H. P. R. Seeliger, and 0 Ang (ed.), Candida and candidamycosis. Plenum Press, New York. 24. Nakamura, A., N. Ishikawa, and H. Suzuki. 1991. Diagnosis of
2164
25. 26.
27. 28.
29.
J. CLIN. MICROBIOL.
HERENT ET AL.
invasive candidiasis by detection of mannan antigen by using the avidin-biotin enzyme immunoassay. J. Clin. Microbiol. 29:2363-2367. Ness, M. J., W. P. Vaughan, and G. L. Woods. 1989. Candida antigen latex test for detection of invasive candidiasis in immunocompromised patients. J. Infect. Dis. 159:495-503. Phillips, P., A. Dowd, P. Jewesson, G. Radigan, M. G. Tweeddale, A. Clarke, I. Geere, and M. Kelly. 1990. Nonvalue of antigen detection immunoassays for diagnosis of candidemia. J. Clin. Microbiol. 28:2320-2326. Platenkamp, G. J. 1988. Application of serological tests in the diagnosis of invasive candidiasis. Mycoses. 31(Suppl. 2):27-33. Poulain, D., A. Ayadi, and J. Fruit. 1988. Detection d'un antigene temoin d'infection systemique a Candida a l'aide d'un anticorps monoclonal coupl6 a l'or colloidal. Ann. Biol. Clin. 45:565-573. Poulain, D., J. Fruit, L. Fournier, E. Dei Cas, and A. Vernes. 1986. Diagnosis of systemic candidiasis: application of co-
30. 31. 32. 33.
34.
counterimmunoelectrophoresis. Eur. J. Clin. Microbiol. 5:427434. Ruchel, R. 1989. Identification of certain false positive results in the Cand-Tec test for candidal antigen. Mycoses 32:627-630. Saral, R. 1991. Candida and Aspergillus infections in immunocompromised patients: an overview. Rev. Infect. Dis. 13:487492. Taschdjian, C. 1957. Routine of identification of C. albicans, current methods and a new medium. Mycologia 49:332-338. Walsh, T., W. Hathorn, J. Sobel, W. Herz, V. Sanchez, M. Maret, H. Buckley, H. Pfaller, R. Schausele, C. Silva, E. Navaro, J. Lecciones, P. Chandrasekar, J. Lee, and P. A. Pizzo. 1991. Detection of circulating Candida enolase by immunoassay in cancer patients with invasive candidiasis. N. Engl. J. Med. 15:1026-1031. Walsh, T. J., and P. A. Pizzo. 1988. Nosocomial fungal infections: a classification for hospital acquired fungal infections and mycosis arising from endogenous flora or reactivation. Annu. Rev. Microbiol. 42:517-545.
Vol. 30, No. 8
0095-1137/92/082158-07$02.00/0 Copyright © 1992, American Society for Microbiology
Retrospective Evaluation of Two Latex Agglutination Tests for Detection of Circulating Antigens during Invasive Candidosis PHILIPPE
HERENT,1 DIRK STYNEN,2 FERNANDO HERNANDO,3 JEANINE FRUIT,1
AND DANIEL POULAINl* Institut National de la Sante et de la Recherche Medicale, Unite 42, 369 rue Jules Guesdes, F-59650 Villeneuve d'Ascq, and Laboratoire de Parasitologie et Mycologie, Faculte de Me6decine, Place de Verdun, F-59000 Lille, France'; Sanofi Diagnostics Pasteur, B-3600 Genk, Belgium2; and Departamento de Microbiologia e Inmunologia, Facultad de Ciencias, Universidad del Pais Vasco, E-48080 Bilbao, Spain3
Received 3 February 1992/Accepted 4 May 1992
Two latex agglutination tests for the detection of Candida antigens, Pastorex Candida (Sanofi Diagnostics Pasteur, Marnes-la-Coquette, France) and Cand-Tec (Ramco Laboratories, Inc., Houston, Tex.), were applied to 79 serum samples from 19 patients who were retrospectively selected on the basis of mycological and clinical evidence of C. albicans infection and the availability of serial serum samples taken near the date of a positive culture. The specificity in 60 control individuals was 100% for Pastorex and 98.3% for Cand-Tec. The tests scored positive for 10 (52.6%) and 9 (47.4%) patients, respectively. Pastorex detected antigen in only 3 of 12 patients (25%) with positive antibody detection tests, but was positive for all 7 patients (100%Y) who produced no or a low antibody response, suggesting that the test performs better in the absence of antibodies. However, the sensitivity of Pastorex also increased with the number of samples available per patient, which was lower for high-antibody-responder patients (2.8 versus 5.7). If the patients who provided only one or two serum samples were eliminated, the sensitivity of Pastorex rose to 76.9%. For the Cand-Tec, the sensitivity was not related to the presence of antibodies, nor was it related to the number of samples per patient. The observed antigenemia was transient with both Pastorex and Cand-Tec. Only 12.5% of the positive reactions occurred on the same serum sample, confirming that the two tests react with different antigens. A positive antigen test preceded other diagnostic indications for 6 of 10 Pastorex-positive patients and 5 of 9 Cand-Tec-positive patients. In the past 15 years, the incidence of hospital infections caused by yeasts of the genus Candida has shown a steady increase (1, 31, 34). Since the clinical symptoms of deep candidosis are nonspecific (4), the mycological laboratory can play an essential role by providing assistance to the clinician with the establishment of a diagnosis. Furthermore, this biological diagnosis must be made early, since timely initiation of antifungal therapy is essential for a favorable prognosis (1, 31). To reach these goals, three types of biological procedures are available, namely, mycological examination and tests for detection of antibodies and antigen; each of these has its limitations. The subject has been extensively discussed in the literature (2, 9, 15, 27). It is generally agreed that these three types of tests may complement each other and that mycological and serological surveys of high-risk patients, defined according to clinical data, are necessary. The aim of this study was to evaluate the diagnostic value of two commercially available latex agglutination tests, i.e., Cand-Tec (Ramco Laboratories, Inc., Houston, Tex.) which has been available for several years (7, 11, 12), and Pastorex Candida (Sanofi Diagnostics Pasteur, Marnes-la-Coquette, France), which was introduced more recently (13, 23). This evaluation was made in the context of a routine mycological laboratory to which mycological and serological samples are sent at the initiative of clinicians. It involved 19 patients who were retrospectively selected on the basis of mycological and clinical evidence of deep C. albicans infection and the availability of several *
serial serum samples taken near the date of a positive culture of samples from normally sterile sites. Our routine serological procedure involves the detection of antibodies by cocounterimmunoelectrophoresis (Co-CIE) and an indirect immunofluorescence assay (IFA) (29) plus antigen detection by the Cand-Tec test in cases of nonsignificant antibody levels (12). The objective of the present analysis was therefore to assess the advantages, if any, of Pastorex Candida compared with those of Cand-Tec.
MATERLILS AND METHODS Human sera. (i) Sera from patients with C. albicans infection. The following criteria were applied to laboratory and clinical files as retrospective selection rules: (i) selection of patients having presented evidence of the presence of C. albicans, which was isolated at least once in large quantities from normally sterile sites (blood, catheter, drain, wound, peritoneal cavity, gastric or lung biopsy specimens, bile, pancreatic cyst, pericardial fluid); (ii) availability of serial serum samples for the corresponding patients taken near the date of positive cultures; and (iii) subsequent analysis of the selected patients' clinical files (together with the physicians in charge of the patients), leading to consideration of candidosis as highly probable once it was integrated into the underlying conditions (e.g., fever resistant to antibacterial agents and response to antifungal therapy). Application of this procedure for the period between February 1988 and May 1989 led to the selection of 73 serum samples from 18 patients from the surgery, intensive care, and hematology wards of the Lille University Hospital.
Corresponding author. 2158
VOL. 30, 1992
Details concerning patients with C. albicans infection are given in Tables 1 and 2. In Tables 1 and 2, the data are for 19 patients, not 18 patients, since 1 patient who underwent two episodes of invasive candidosis in a 1-year interval (separated by a total recovery) was treated as two different patients (patients A9 and B5). (ii) Control sera. Four groups of control individuals (a total of 60 individuals) were tested in order to determine the specificity of the methods. The first group consisted of 10 members of the laboratory staff, the second group included 20 surgery patients without any sign of clinical candidosis, the third group encompassed 10 patients with immunologically confirmed aspergillosis and 2 patients with allergic bronchopulmonary aspergillosis (3), and the fourth group included 18 patients in acute phases of parasitic diseases, as follows: 13 patients with malaria and 1 patient each with fascioliasis, schistosomiasis, filariasis, echinococciasis, and amebiasis. Mycological examinations. Cultures were grown on Sabouraud glucose agar containing kanamycin (800 ,g/ml). The isolates were identified as C. albicans on the basis of germ tube (20) and chlamydosporulation (32) tests. Antibody detection. Two methods for antibody detection were used. (i) Co-CIE. Co-CIE (bio Merieux, Marcy l'Etoile, France) has been described in detail before (29). It essentially consists of a counterimmunoelectrophoresis on cellulose acetate membranes in which three 20-,ul serum deposits, placed at the anode, are run against a line of antigen consisting of 15 ,ul of a 25-mg/ml solution of somatic antigen from C. albicans VW32 placed at the cathode. Patient sera are run side by side with a rabbit anti-VW32 germ tube antiserum to detect precipitating antibodies that present a line of identity with anti-germ tube antiserum. This line of identity is named the cospecific precipitin line (CSPL). Interpretation of Co-CIE results refers to the presence or absence of CSPL and quantitative evolution, namely, migration toward the cathode and variation in intensity for successive sera from the same patient run side by side with the anti-germ tube antiserum. (ii) IFA. IFA was performed against unfixed blastoconidia of a C. albicans VW32 serotype A cloned strain (bioMerieux,) incubated with doubling dilutions (initial dilution, 1/100) of patient sera in phosphate-buffered saline (PBS). Antibody binding was revealed by incubation with a fluorescent goat anti-human immunoglobulin antiserum (1/100 in PBS containing Evans blue at a final dilution of 1/50,000); Sanofi Diagnostics Pasteur). Results were expressed as the reciprocal titer, i.e., the inverse of the highest serum dilution that gave strong fluorescence of all blastoconidia. Standardization of this test was achieved by the use of a pool of sera from patients with candidosis (bio-Merieux). A titer equal to or greater than 400 was considered indicative of candidosis. Relying on the results obtained with Co-CIE and IFA, the following conclusions were assigned to the serum samples: N, no immunological suspicion of candidosis (no CSPL and IFA < 400); S, immunologically suspected candidosis (presence of CSPL and/or IFA > 400); C, immunologically confirmed candidosis (strong CSPL and IFA > 400). Detection of circulating antigens. Both antigen detection kits were used according to the manufacturer's instructions. The Cand-Tec test (Ramco) uses latex beads coated with a polyclonal rabbit anti-Candida serum against an unknown antigen (12). The test is positive when serum, which is diluted 1/4, agglutinates the latex beads. Serial dilutions of serum were tested to determine the antigen titer, which is
ANTIGENEMIA IN CANDIDOSIS
2159
defined as the reciprocal of the highest serum dilution that still causes agglutination. In the Pastorex Candida test (Sanofi Diagnostics Pasteur), the latex beads were sensitized with a monoclonal antimannan antibody (23). Serum samples were incubated with a serum treatment solution at 100°C. After centrifugation, the supematants were incubated with the latex beads. A serum sample scored positive when undiluted supernatant agglutinated the latex beads. Serial dilutions of the supernatant were tested so that we could obtain semiquantitative results. RESULTS The patients were divided into two groups according to their antibody responses as analyzed by Co-CIE and IFA. Group A consisted of 12 patients with immunologically confirmed candidosis, as established by these routine procedures. The average number of serum samples per patient in this group was 2.8 + 1.3 (Table 1). Group B included seven patients who failed to produce an antibody response, which was considered indicative of infection, but they were judged to suffer from an invasive candidosis on clinical and mycological grounds. The average number of serum samples per patient in this group was 5.7 ± 3.1 (Table 2). For group A patients (Table 1), Cand-Tec was positive at a significant titer for 5 of 12 patients, whereas Pastorex Candida was positive for 3 patients. For group B patients (Table 2), four of seven patients were positive by Cand-Tec; Pastorex Candida yielded positive results for all seven patients. When the results for both patient groups were combined, 9 of 19 cases scored positive by Cand-Tec, as opposed to 10 of 19 by Pastorex Candida. Patients were often positive by only one test (Tables 1 and 2). Five patients (patients A6, All, B3, B4, and B6) were negative by Cand-Tec but positive by Pastorex Candida, while four patients (patients A3, A4, A9, and A12) remained negative by Pastorex Candida but were positive by CandTec. Both tests failed to detect antigen in five patients (patients Al, A2, A5, A8, and A10). A remarkable feature shared by both tests in both groups of patients was that when more than one serum sample was available per patient, antigenemia was always transient. For the serum samples tested and sent at the initiative of clinicians, the minimum delay observed for the appearance of detectable antigen by Cand-Tec and Pastorex were, respectively, 4 days (patient B7) and 1 day (patient A6); conversely, it was possible to observe the disappearance of detectable antigens in both tests within 48 h (patient B5, Cand-Tec; patient B7, Pastorex). The longest antigenemia period detected without any intervening negative test was of 47 days and concerned the Pastorex Candida in patient B5. Figure 1 shows the sensitivities of both tests as a function of the number of serum samples tested per patient. The sensitivity of the Pastorex Candida increased when there were increasing numbers of serum samples available for each patient. In the entire patient population, the sensitivity of Pastorex was 52.6%. When data for the patients who provided only one or two samples were omitted, the sensitivity increased to 76.9%, eventually attaining 100% in patients who provided five or more serum samples. The Cand-Tec test did not show such a correlation. Both tests were negative for samples from the healthy controls and from surgery patients (Table 3). In patients with different fungal and parasite infections, one sample was positive by Cand-Tec. This sample came from a patient with aspergillosis. For this patient, the Candida serology tests,
2160
HERENT ET AL.
J. CLIN. MICROBIOL.
TABLE 1. Patient information, culture data, and results of antibody and antigen testing in patients with an antibody response to C. albicans (group A) Patient
Al A2
Sex'
M M
Age (yr)
63
Underlying condition Pancreas carcinoma
86
Site
ofn'
Bile Catheter
sera were Co-CIEd Conclusionf yFo IFCCand-Tec collectedc
-2 8
4 3
1,600 400
C C
2 (NS)
-10 15
0 3
800
N C
2 (NS) -
M
27
Ileostomy Jejunostomy
Blood (day 0) Drain (day 6)
1 -7
4 3
1,600 400
C C
4
A4
M
51
Oesophagal neoplasm
Drain
1 12 18
3 4 4
400 400 800
C C C
4
AS
M
62
Duodenal perforation
Peritoneal cavity
-1 13
0 4
100 1,600
N C
2 (NS)
A6
M
18
Liver transplant
Gastric biopsy
-5 -4 16
2 3 4
200 200 800
S S C
2 (NS)
-18 -12 -4 4
0 2 3 3
200 400 400 400
N S C C
8 2 (NS) 4 2 (NS)
0 7
2 3
400 800
S
-
C
-
-17 -1 4
2 2 3
100 400 400
10 21 34 >57
2 2 2 4
200 400 400 1,600
-5 0 3 5 14
2 3 3 3 3
400 800 800 800 800
0
2
400
M
43
Non-Hodgkin's malignant lymphoma Pneumopathy
Lung biopsy
A8
M
49
Chronic pancreatitis
Pancreatic cyst
A9
A10
All
A12
F
M
M
M
69
64
46
70
Liver cancer
Sternal resynthesis
Bronchopneumopathy
Duodenostomy Gastrostomy
Blood
Pericardial fluid (day 0) Median wound (day 27) Blood
Peritoneal cavity
S
-
-
1
2 (NS)
C
2 (NS) 2 (NS) 4
S
2 (NS)
S
-
_
A3
A7
Pastorex
2
S S C
C
-
C
-
1
C
-
-
C
-
-
C
-
-
C
M, male; F, female. Unless otherwise indicated, yeasts were isolated on day 0. Indicated as the days before (negative numbers) or after the first isolation of C. albicans. d 0, absence of a precipitin line; 1, nonspecific precipitin line; 2, cospecific precipitin line; 3, intense cospecific precipitin line; 4, precipitin line. e Reciprocal titer. NS, not significant. -, negative. f N, no immunological suspicion of candidosis; S, immunologically suspected candidosis; C, immunologically confirmed candidosis.
-
4
a
b
c
however, also yielded results compatible with immunologically confirmed candidosis. The other serum samples were all negative by the Pastorex Candida and Cand-Tec tests. DISCUSSION The presence of Candida antigens in serum appears to be specific indicator of invasive disease (2, 9, 15, 22, 24, 28, 33). However, the low and varying antigen concentrations, which often fall below the detection limits of the available a
very
intense cospecific
methods (5, 26, 33), also lead to controversy about the clinical usefulness of antigen detection tests. In this study, we compared the results of culture and of antibody and antigen detection in a group of 19 patients with or without an antibody response. Special emphasis was placed on the role of antigen detection and, more particularly, on the comparison of two commercially available test kits. We found an overall sensitivity of 47.5% for Cand-Tec and 52.6% for Pastorex Candida (Tables 1 and 2). The Cand-Tec test has been the subject of numerous studies which have yielded
2161
ANTIGENEMIA IN CANDIDOSIS
VOL. 30, 1992
TABLE 2. Patient information, culture data, and results of antibody and antigen testing in patients with an antibody response to C albicans (group B) Patient
Sex°
Age (yr)
Underlying condition
Site of yeast
CoyCIE sera were
IF
Conclusio
collected IlAl
Bi
F
23
Acute
lymphoblastic --ia ---
Blood
leukemia
B2
B3
B4
B5
B6
B7
F
M
M
F
M
F
46
21
65
69
43
50
Gastrectomy
Acute lymphobastic leukemia
Ileostomy
Liver cancer
Acute myelomonocytic leukemia
Catheter
Blood
Blood (day 0) Blood (day 1)
Bile
Blood
Drain
-14 -11 -10 3
1 2 1 1
100 100 100
-1 5 14
0 2 2
100 200 200
-1 15 >30
0 0 2
-21 -16 -13 -8 -5 -2 -1 0 8 11 12 28
N N N N
Cand-Tec
Pastorex
-
-
4
4
N S S
4
4
100 200 200
N N
2 (NS)
1 0 1 2 2 2 2 2 2 2 1 0
200 200 200 200 200 200 200 200 200 200 200 100
N N N S S S S S S S S N
-3 2 4 10 48 51 56
1 2 0 1 2 1 0
100 200 200 100 200 200 100
N S N N S S N
-30 -13 1 8 14 >30
0
100 100 100 100
N N N N N N
-9 -7 8 10 14
1 1 1 0 0
400 400 200 200 200
S S
1 1 1 0 0
2 (NS)
S
-
-
-
2 (NS) 2 (NS) 2 (NS) 2 (NS) 2 (NS) 2 (NS)
2 (NS)
1
2 (NS) 4 4 2 (NS) 2 (NS)
1 4 1 2
4
2 (NS) 2 (NS)
2
2 (NS)
1 4
S
N N
1
-
4
F, female; M, male. Unless otherwise indicated, yeasts were isolated on day 0. Indicated as the days before (negative numbers) or after the first isolation of C. albicans. d 0, absence of precipitin line; 1, nonspecific precipitin line; 2, cospecific precipitin line; 3, intense cospecific precipitin line; 4: very intense cospecific precipitin line. I Reciprocal titer. NS, not significant. -, negative. f N, no immunological suspicion of candidosis; S, immunologically suspected candidosis; C, immunologically confirmed candidosis. I
b
I
variable results; the sensitivity of the test has ranged from 19 to 91% (7, 8, 10-12, 16, 28). In an early study of Pastorex Candida, Meulemans et al. (23) found 5 positive patients of 10 autopsy-proven cases, but the number of serum samples per patient tested in that study was only one to three. The number of serum samples per patient is very important, since our data showed that both tests are liable to become positive or negative within 24 to 48 h. In animal models, mannan clearance from the blood has been proven to be very rapid. The antigen accumulated in the liver and
very
the spleen and was also excreted into the urine (18). Because of the efficiency of the clearance mechanism, mannan concentrations remain low and may drop below the detection limit of a reverse passive latex agglutination test. Patients at risk or patients suspected of being at risk should therefore be tested as frequently as possible. Our results with Pastorex Candida confirm that the sensitivity of the test increases with the number of serum samples available per patient. The average number of serum samples provided by the Pastorexnegative patients was 2.3 versus an average of 5.2 serum
2162
HERENT ET AL. 100
J. CLIN. MICROBIOL.
TABLE 3. Result of antibody and antigen testing in control populations
T
80 +
Population (no. of subjects)
Patient no.
Co-Ciea
Serology by IFAb
Antigen detection by Cand-
Tec"
60
+
40
+
Healthy individuals (10)
1
0
200
-
2 3
0 0 0 2
100 100
-
200
-
4-10 Surgery patients (20)
20 +
.4 .5 .3 (5) (13) (9) n samples/patient (n patients/category)
.2 (17)
.6 (3)
1 2-13 1 1 1 1 1
0 0 0 0 0 0 0
200
16-20 Aspergillosis (10)
FIG. 1. Sensitivity of Pastorex Candida (0) and Cand-Tec (0) antigen detection tests in relation to the number of serum samples from each patient tested; the number of patients entering each category was defined as the minimal number of serum samples tested.
samples provided by the Pastorex-positive patients. When patients who provided only one or two serum samples were omitted from the analysis, the sensitivity of Pastorex rose to 76.9%. This correlation between sensitivity and number of serum samples could not be observed with Cand-Tec, but it has been described by Burnie and Williams (7). In their study, the average number of samples per patient was 3.1 for Cand-Tec-negative patients and 9.7 for Cand-Tec-positive ones. In the same context, the time interval between samples is important. It is noteworthy that the average interval between the time that serum samples were obtained from four of five patients who were negative by both tests was 10 days or more. We observed discordant results between the two tests in nine patients: five patients were negative by Cand-Tec and positive by Pastorex, while four patients were negative by Pastorex and positive by Cand-Tec. Only 12.5% of the serum samples were positive when they were tested by using both tests. This observation can be explained by the fact that the two latex agglutination tests detect different antigens. The rat monoclonal antibody EB-CA1, which is used in the Pastorex test, recognizes mannan, a major cell wall polysaccharide, from C. albicans, C. tropicalis, C. glabrata, C. guilliermondii, and to a lesser extent, C. pseudotropicalis and C. parapsilosis (14). The antigen detected by Cand-Tec, however, has not been identified, but it is certainly different from mannan (12, 30). The high degree of discord between the two tests suggests that the detection of more than one antigen may improve the sensitivity. The prognosis of the patient improves when a diagnosis can be established in an early phase of the infection (31). It is therefore important to see whether the antigen tests would have allowed an earlier diagnosis. In the nine positive patients, the Cand-Tec test turned positive in five patients before cultures became positive or antibody levels considered as significative of candidosis were detected (29). In the 10 Pastorex-positive patients, Pastorex Candida was the first indication of invasive disease in six of them. Both latex
200 100 200 100
1 2 3 4 5 6 7 8 9 10 2
1 1 0 0 0 1 2 2 2 0 3 3 2 0 0 0
2-3 4-6 7-8 9-15
0
.1 (19)
1
Allergic bronchopulmonary aspergillosis (2) Malaria (13)C Fascioliasis (1) Schistosomiasis (1) Filariasis (1) Echinococcosis (1) Amebiasis (1)
-
2 (NS) -
100
-
100 200
-
400 200 100
-
4 -
-
-
200
-
a 0, absence of precipitin line; 1, nonspecific precipitin line; 2, weak cospecific precipitin line; 3, intense cospecific precipitin line; 4, very intense cospecific precipitin line. b Reciprocal titer. NS, not significant. -, negative. No antigen was detected by Pastorex. c Six of these patients presented with a Cand-Tec titer of 2.
agglutination tests were preceded by positive cultures in three cases and by positive serological tests in one case. We considered antigenemia as positive before culture if the antigen test was positive on day 1 at the latest, since mycological cultures take, on average, 72 h of incubation before they can be interpreted. Our results confirm that circulating antigens appear early in the course of the infection, and therefore, they often provide the clinician with information that allows for the earlier initiation of treatment. Both tests yielded comparable results in this respect. Although the number of patients examined in this study was too small to allow firm conclusions to be made, Pastorex seemed to be a good test for use on immunocompromised patients, because it was positive for all seven patients with a weak serological response, while Cand-Tec was positive for four patients (Table 1). Several other investigators also failed to obtain very good results by Cand-Tec in patients with leukemia (5, 10, 25, 28). On the other hand, Pastorex seemed to be more sensitive to the presence of specific antibodies in the patient's serum (17) (Table 2). This is a surprising observation, because unlike Cand-Tec, the Pastorex Candida protocol includes a serum treatment to dissociate immune complexes. In the four patients whose sera were
ANTIGENEMIA IN CANDIDOSIS
VOL. 30, 1992
positive only by Cand-Tec, the antibody levels were high. Moreover, the majority of Pastorex-positive samples belonged to patients who failed to produce an antibody response to C. albicans (group B). Conversely, in the antibody responder group (group A), only three patients were positive. Among those three patients, only one exhibited high antibody levels at the date of positive antigenemia. This raises the possibility that the patients who respond to the Candida infections immunologically are able to clear mannan from their blood more efficiently (18), leading to a lower incidence of detectable antigenemia. However, since the average number of serum samples per patient was lower in the antibody responder group (2.8 versus 5.7), it cannot be excluded that the lower sensitivity of Pastorex Candida in this group was due to this lower number of serum samples. In this study, both tests were very specific; all 60 control samples were negative by Pastorex Candida (100% specificity), and only 1 control sample was positive by Cand-Tec (98.3% specificity) (Table 3). This positive sample, which was from a patient with aspergillosis, may have been a true-positive result, since serological results were consistent with immunologically confirmed candidosis. For Pastorex Candida, the high specificity is in agreement with earlier findings (23). The specificity of Cand-Tec reported in other studies varied considerably (6, 8, 12, 16, 25, 26, 28). Results of our study suggest that the laboratory diagnosis of invasive candidosis benefits from an integrated, kinetic approach in which patients suspected of having candidosis or, preferably, patients at risk of candidosis are submitted to mycological surveillance, serological tests, and antigen detection at regular time intervals. This approach allows physicians to follow kinetics of antibody fluctuations (29) which complements or relates with (28) results from antigen testing. Depending on the antigen detected, immunoassays for the detection of circulating fungal components may constitute specific markers for invasive candidosis (15, 21, 22, 24, 33). Provided that testing is not restricted to isolated serum samples and that patients are regularly screened, antigen testing is also a sensitive method and is especially useful in immunocompromised patients for whom antibody detection tests remain negative. Moreover, a positive antigen test is often the earliest indication of invasive infection. Although the results of this retrospective study suggest that Pastorex Candida is a useful new tool for the diagnosis of invasive candidosis, more studies are necessary to confirm these results. Further investigations, including prospective studies, are needed to firmly establish the clinical value of this test and to assess its reproducibility with different production batches.
ACKNOWLEDGMENTS We thank Laurence Richard, Nadine Francois, and Gilbert Lepage for valuable technical assistance. REFERENCES 1. Armstrong, D. 1989. Problems in management of opportunistic fungal diseases. Rev. Infect. Dis. 7(Suppl. II):1591-1599. 2. Bennet, J. 1987. Rapid diagnosis of candidiasis and aspergillosis. Rev. Infect. Dis. 9:398-402. 3. Biguet, J., P. Tran Van Ky, S. Andrieu, and J. Fruit. 1964. Analyse immunoelectrophoretique d'extraits cellulaires et de milieux de culture d'Aspergillus fumigatus par des immuns6rums experimentaux et des serums de malades atteints d'aspergillome bronchopulmonaire. Ann. Inst. Pasteur 107:7297. 4. Bodey, G. P., and V. Fainstein. 1985. Systemic candidiasis, p. 135-169. In G. P. Bodey and V. Fainstein (ed.), Candidiasis.
2163
Raven Press, New York. 5. Bougnoux, M. E., C. Hill, D. Moissenet, M. Feuilhade de Chauvin, M. Bonnay, I. Vincens-Sprauel, F. Pietri, M. McNeil, L. Kaufman, J. Dupouy-Camet, C. Bohuon, and A. Andremont. 1990. Comparison of antibody, antigen, and metabolite assays for hospitalized patients with disseminated candidiasis. J. Clin. Microbiol. 28:905-909. 6. Burnie, J. P. 1985. A reverse passive latex agglutination test for the diagnosis of systemic candidiasis. J. Immunol. Methods 82:267-280. 7. Burnie, J. P., and J. Williams. 1985. Evaluation of the Ramco latex agglutination test in the early diagnosis of systematic candidiasis. Eur. J. Clin. Microbiol. 4:98-101. 8. Cabezudo, I., M. Pfaller, T. Gerarden, F. Koontz, R. Wendel, R. Gingrich, K. Heckman, and C. Burns. 1989. Value of the Cand-Tec Candida antigen assay in the diagnosis and therapy of systemic candidiasis in high-risk patients. Eur. J. Clin. Microbiol. Infect. Dis. 8:770-777. 9. de Repentigny, L. 1989. Serological techniques for diagnosis of fungal infection. Eur. J. Clin. Microbiol. 8:362-375. 10. Escuro, R. S., M. Jacobs, S. L. Gerson, A. R. Machicao, and H. M. Lazarus. 1989. Prospective evaluation of a Candida antigen detection test for invasive candidiasis in immunocompromised adult patients with cancer. Am. J. Med. 87:621-626. 11. Fung, J., S. Donta, and R. Tilton. 1986. Candida detection system (Cand Tec) to differentiate between Candida albicans colonization and disease. J. Clin. Microbiol. 24:542-547. 12. Gentry, L. O., I. D. Wilkinson, A. S. Lea, and M. F. Price. 1983. Latex agglutination test for detection of Candida antigen in patients with disseminated disease. Eur. J. Clin. Microbiol. 2:122-128. 13. Georges, E., M. L. Garrigues, J. L. Poirot, J. P. Masini, and M. C. Meyohas. 1991. Diagnostic des candidoses sytemiques par un test au latex. R6sultats d'une annee de suivi serologique chez des patients a risque. J. Mycol. Med. 1:25-28. 14. Georges, E., M. L. Garrigues, D. Stynen, and J. L. Poirot. 1991. Specificite in-vitro d'un anticorps monoclonal (AcM) anti-mannane et du latex sensibilise par cet AcM au cours de la candidose dissemin6e experimentale. J. Mycol. Med. 1:21-25. 15. Jones, J. M. 1990. Laboratory diagnosis of invasive candidiasis. Clin. Microbiol. Rev. 3:32-45. 16. Kahn, F. W., and J. M. Jones. 1986. Latex agglutination tests for detection of Candida antigens in sera of patients with invasive candidiasis. J. Infect. Dis. 153:579-585. 17. Kappe, R. 1988. Coexistence of free antigens, free antibodies and immune complexes in sera from patients with suspected deep-seated candidosis. Mycoses 32:24-32. 18. Kappe, R., and J. Muller. 1991. Rapid clearance of Candida albicans mannan antigens by liver and spleen in contrast to prolonged circulation of Cryptococcus neoformans antigens. J. Clin. Microbiol. 29:1665-1669. 19. Lemieux, C., G. St-Germain, J. Vincelette, L. Kaufnan, and L. de Repentigny. 1990. Collaborative evaluation of antigen detec-
tion by commercial latex agglutination test and enzyme immunoassay in the diagnosis of invasive candidiasis. J. Clin. Microbiol. 28:249-253. 20. Mackenzie, D. W. R. 1962. Serum tube identification of Candida albicans. J. Clin. Pathol. 15:563-565. 21. Matthews, R., and J. Burnie. 1988. Diagnosis of systemic candidiasis by an enzyme-linked dot immunobinding assay for a circulating immunodominant 47-kilodalton antigen. J. Clin. Microbiol. 26:459-463. 22. Matthews, R. C., J. P. Burnie, and S. Tabaqchali. 1987. Isolation of immunodominant antigens from sera of patients with systemic candidiasis and characterization of serological response to Candida albicans. J. Clin. Microbiol. 25:230-237. 23. Meulemans, L., A. Andremont, F. Meunier, A. M. Ceuppens, M. L. Garrigues, and D. Stynen. 1991. Pastorex Candida, a new latex agglutination test for mannan detection in serum of patients with invasive candidosis, p. 233-236. In E. Tumbay, H. P. R. Seeliger, and 0 Ang (ed.), Candida and candidamycosis. Plenum Press, New York. 24. Nakamura, A., N. Ishikawa, and H. Suzuki. 1991. Diagnosis of
2164
25. 26.
27. 28.
29.
J. CLIN. MICROBIOL.
HERENT ET AL.
invasive candidiasis by detection of mannan antigen by using the avidin-biotin enzyme immunoassay. J. Clin. Microbiol. 29:2363-2367. Ness, M. J., W. P. Vaughan, and G. L. Woods. 1989. Candida antigen latex test for detection of invasive candidiasis in immunocompromised patients. J. Infect. Dis. 159:495-503. Phillips, P., A. Dowd, P. Jewesson, G. Radigan, M. G. Tweeddale, A. Clarke, I. Geere, and M. Kelly. 1990. Nonvalue of antigen detection immunoassays for diagnosis of candidemia. J. Clin. Microbiol. 28:2320-2326. Platenkamp, G. J. 1988. Application of serological tests in the diagnosis of invasive candidiasis. Mycoses. 31(Suppl. 2):27-33. Poulain, D., A. Ayadi, and J. Fruit. 1988. Detection d'un antigene temoin d'infection systemique a Candida a l'aide d'un anticorps monoclonal coupl6 a l'or colloidal. Ann. Biol. Clin. 45:565-573. Poulain, D., J. Fruit, L. Fournier, E. Dei Cas, and A. Vernes. 1986. Diagnosis of systemic candidiasis: application of co-
30. 31. 32. 33.
34.
counterimmunoelectrophoresis. Eur. J. Clin. Microbiol. 5:427434. Ruchel, R. 1989. Identification of certain false positive results in the Cand-Tec test for candidal antigen. Mycoses 32:627-630. Saral, R. 1991. Candida and Aspergillus infections in immunocompromised patients: an overview. Rev. Infect. Dis. 13:487492. Taschdjian, C. 1957. Routine of identification of C. albicans, current methods and a new medium. Mycologia 49:332-338. Walsh, T., W. Hathorn, J. Sobel, W. Herz, V. Sanchez, M. Maret, H. Buckley, H. Pfaller, R. Schausele, C. Silva, E. Navaro, J. Lecciones, P. Chandrasekar, J. Lee, and P. A. Pizzo. 1991. Detection of circulating Candida enolase by immunoassay in cancer patients with invasive candidiasis. N. Engl. J. Med. 15:1026-1031. Walsh, T. J., and P. A. Pizzo. 1988. Nosocomial fungal infections: a classification for hospital acquired fungal infections and mycosis arising from endogenous flora or reactivation. Annu. Rev. Microbiol. 42:517-545.
