- Email: [email protected]
Serologic tests in the diagnosis of systemic candidiasis
Serologic Tests in the Diagnosis of Systemic Candidiasis EnhancedDiagnostic Accuracy with Crossed lmmunoelectrophoreses
RICHARD H. GLEW. MD
l
HELEN R. BUCKLEY, Ph.D HARVEY M ROSEN, M.D ROBERT C
MOELLERING, Jr,
M.D
JOSEF E FISCHER, MD. Boston, Massachusetts
From the Departments of Medictne and Surgery, Harvard Medical School and the Department of Microbiology, Harvard School of Public Health, Boston, Massachusetts Preliminary results of this study were presented in part at the Seventy-Fifth Annual Meeting of the Amerrcan Society for MIcrobiology, New York, Aprrl30, 1975, and at the Sixty-First Clinical Congress of the American College of Surgeons, San Francisco, California, October 13, 1975 This study was supported in part by NIAID Training Grant TO1 A 1 0044 1 Requests for reprints should be addressed to. Dr. Richard H Glew, The Memorral Hospital, 119 Belmont Street, Worcester, Massachusetts 01605 Manuscript accepted August 18, 1977 Present address. University of Massachusetts Medical School, The Memorial Hospital, 119 Belmont Street, Worcester, Massachusetts, 01605. l
586
April 1978
In order to assess the diagnostic value of serologic methods in discriminating between colonization and disseminated infection with Candlda species, serum samples from 201 consecutive patients receiving parenteral hyperalimentation were examined for the presence of agglutinating and precipltatlng antibodies to Candida albicans. In these patlents, fungal colonization was detected In 50 per cent, transient fungemia In 6 per cent and dlssemlnated candidlasis in 2.5 per cent. Agglutinating antibodles were of no dlagnostlc value in separating patients with Candida colonization from those with disseminated infection (sensitivity 675 per cent, predictive value 620 per cent). Serum Candlda precipltlns were detected by double lmmunodlffuslon (DID) in all five patients with systemic candldlasls (100 per cent sensitivity); however, since only five of 53 patients with preclpitins detected by DID demonstrated evidence of systemic candidiasis, the predlctlve value of the test was unacceptably low (9 per cent). When the serums which were posltlve by DID were submitted to crossed lmmunoelectrophoresis (XIE), precipitln bands to antigens other than mannan were obtained in only nine patients, Including all five patlents with disseminated candldiasis. Thus, detection by XIE of serum Candida precipltins directed against antigens other than mannan was associated with a predlctlve value of 56 per cent and a sensitivity of 100 per cent. The detection of serum Candlda precipltlns by DID Is a sensitive screening test in patients at high risk for the development of systemic candldlasls; patients with persistently negative preclpltins by DID are unlikely to harbor disseminated candldlasis. If preclpltin bands are detected by XIE to antigens other than mannan, there is a high likelihood that the patient has dlssemlnated or invasive Candida infection. Systemic infections with Candida species have become increasingly more prevalent in the past few decades, in part because advances in medical and surgical expertise and technology have prolonged the survival of compromised patients, and in part because of the side effects or consequences of these supportive medical-surgical measures [l-5]. Since patients ultimately documented to have disseminated candidiasis may fail to have Candida grown on repeated blood cultures, it is often difficult to establish the diagnosis of systemic candidiasis [3,4,6]. On the other hand, the isolation of Candida from various clinical sources, including blood cultures, does not necessarily indicate invasive or disseminated infection [2-41. Many laboratories have attempted to establish serologic tests which would assist in the prompt and reliable diagnosis of deep-seated infections. However,
The American Journal of Medicine
Volume 64
SEROLOGIC TESTS IN SYSTEMIC CANMDIASIS-GLEW
ET AL.
reports differ in their assessments of the sensitivity and specificity of these tests [7-l 11. In order to assess the diagnostic value of Candida serology, we carried out a prospective survey of Candida agglutinins and precipitins in a group of 201 hospitalized patients receiving total parenteral nutrition. This group was selected because it provided a diverse patient population with a wide variety of underlying medical and surgical disorders. In general, the patients shared a common background involving one or more factors known to predispose to fungal colonization and infection: surgery, antibiotic administration, genitourinary catheters, malnutrition, malignancy, thermal burns, old age and nutritional support via protracted intravenous alimentation through indwelling plastic venous cannulas. MATERIALS AND METHODS Patients. A total of 577 serum samples were obtained prospectively from 20 1 consecutive hospitalized patients at the Massachusetts General Hospital who received intravenous hyperalimentation during a nine month period from June 1974 through February 1975. As has been the experience in this hospital in recent years, the mean duration of hyperalimentation in these patients was approximately three weeks [ 121. Of this group, 127 patients were male and 74 female; the ages of the patients ranged from nine to 89 years. Serum samples were collected at least once weekly and were processed within two weeks of collection for determination of agglutinating and precipitating antibodies against Candida albicans. Informed consent for blood sampling was obtained from the patient or a responsible family member in each case. Criteria for Diagnosis Candidiasis. Cultures ofmaterial from various clinical sites were obtained at the discretion of the primary physician caring for each patient. Cultures of the urine were obtained in 96 per cent of the patients, of the blood in 89 per cent, of the sputum in 72 per cent, of material from intravenous catheters in 68 per cent, and of various body fluids and wounds in 58 per cent; no specimens for cultures were obtained from six (3 per cent) of the patients. Of the 201 patients, the diagnosis of systemic candidiasis was made in four, each of whom exhibited two or more of the following: (1) repeated isolation of Candida albicans from blood cultures (three patients) or C. parapsilosis (one patient) over several days (six to 85 days) in patients exhibiting clinical signs of septicemia; (2) funduscopic findings diagnostic of Candida endopthalmitis (two patients) [ 13,141; (3) clinical and postmortem pathologic evidence of Candida endocarditis on a prosthetic value (one patient); (4) signs of peritonitis plus growth of C. albicans from repeated paracenteses of peritoneal fluid (one patient). One additional patient was diagnosed as having invasive candidiasis of the genitourinary tract based on repeated isolation of C. albicans over a nine-month period from the urinary bladder, both ureters and bilaterial nephrostomy sites (the latter on deep cultures of material obtained from sinus tracts). The remaining 196 patients without invasive or systemic candidiasis were divided into three groups. In group I were
1 -_
Figure 1. Crossed immunoelectrophoresis of serum from patient without candidiasis demonstrating antibodies directed only against cell wall (mannan) antigens (see arrow). patients with transient self-limited fungemia: in six patients C. albicans was grown on culture of the blood, in three patients Torulopsis glabrata, in one patient C. tropicalis and in one patient C. parapsilosis; in one patient blood cultures grew C. albicans and T. glabrata on different occasions. Among these 12 patients, adequate clinical follow-up (including negative blood cultures) was obtained for two weeks or longer after the initial episode of fungemia in nine. Three other patients died within two days of their fungemic episodes. Postmortem examination (including fungal cultures) demonstrated no evidence of systemic or invasive candidiasis in two of these patients, whereas no autopsy was performed in the remaining patient. In group II were 83 patients in whom Candida species or T. glabrata were grown from various clinical sites exclusive of blood, but who showed no clinical or pathologic signs of deep invasion and were considered to be simply colonized with these fungi. Finally, in group Ill were 101 patients from whom fungi were never isolated in routine clinical specimens. Serologic Tests. Tests for agglutinating antibodies were carried out with suspensions of whole yeast cells of C. albicans (Group A) and tests for precipitating antibodies were performed on all serums by double immunodiffusion (DID) as previously described [ 151. Those serums which gave a positive precipitin reaction by DID were subjected to crossed immunoelectrophoresis (XIE) according to the method of Axelsen and Svendsen [ 161. In the XIE tests, the Candida antigen was subjected to sequential electrophoresis in two directions, the second time into a gel containing the patient’s serum. With XIE, it is possible to detect two types of precipitin bands. The first is a broad, fuzzy, deeply-staining band (Figure 1) which represents a reaction to cell wall antigen (mannan) and can be identified either by prior absorption of the test serum with mannan or by prior reaction of the antigen preparation wlth concanavalin A, which precipitates out mannans
April 1978
The American Journal of Medicine
Volume 84
587
SERCLCGIC TESTS IN SYSTEMIC CANDIDIASIS-GLEW
ET AL.
candidiasis, and in 60 per cent of the patients with candidiasis. Agglutinins of al/256 were observed in from 1 to 33 per cent of the patients without candidiasis and in 40 per cent of the patients with disseminated or invasive disease. In cases in which more than one serum sample had been obtained, at least a e-tube rise in agglutinin titer was detected in from 20 to 78 per cent (31 per cent over-all) of the patients without candidiasis and in 75 per cent of the patients with candidiasis. In patients without candidiasis, the greatest incidence of elevated agglutinin titers (either 2 l/64 or B l/256) and P-tube rises in agglutinin titer were seen in patients with positive blood cultures (transient fungemia). Moreover, these agglutinin values were similar to those seen in the five patients with systemic candidiasis. Preclpitins (Table II). Precipitating antibodies were found by DID in from 15 to 67 per cent (24 per cent over-all) of the patients without candidiasis and in 100 per cent of the five patients with systemic or invasive candidiasis. As with agglutinins, the highest percentage of positive tests among patients without candidiasis was noted in patients with transient fungemia. However, when serums which were positive by DID were submitted to XIE, the number of patients with positive tests (precipitin lines to antigens other than mannan) was markedly reduced in the groups of patients without disseminated candidiasis. Of the patients without systemic candidiasis, only 1 to 8 (2 per cent over-all) had positive precipitins by XIE, whereas all five (100 per cent) of the patients with systemic or invasive candidiasis had positive reactions by XIE. Limited information is available concerning the chronologic sequence of positive serology in our study. Only in patients in whom fungemia ultimately culminates in definite signs of systemic candidiasis can one with certainty define the timing of the initial antigenic stimulus to antibody production. Only five of our patients were documented to have disseminated infection and were able to be evaluated for the temporal sequence of precipitins. The earliest appearance of positive precipitins by DID and XIE after the onset of candidemia
Figure 2. Crossed immunoelectrophoresis of Serum from a patient with endocarditis due to Candida parapsilosis demonstrating antibodies to cytoplasmic antigens as well as to cell wall (mannan) antigens (see arrow).
[ 171. The second type of bands are fine precipitin arcs which represent the reaction of serum antibodies with nonmannan Candida antigens (Figure 2). Since precipitating antibodies to mannan have been demonstrated to occur in normal subjects and in people superficially colonized with candida [ 181, serums were considered positive by XIE only if precipitin bands were formed to antigens other than mannan (Flgure 2). The antigen preparation employed in the precipitin tests (DID and XIE) was a crude cytoplasmic extract derived from homogenization of yeast phase C. albicans grown in synthetic liquid medium [ 151. RESULTS
Agglutinins (Table I). Agglutinating antibodies were detected at a titer of 2 l/64 in from 31 to 50 per cent (33 per cent over-all) of the patients without systemic
TABLE I
Candlda Agglutinins PatientswtthAgglutlnatlng Antibodies
Total
Patients Systemic or invasive candidiasis Patients without systemic or invasive candidiasis Group It Group IIt Group IlIT
TiterH/64
Ttter H/128 No. %
Ttter M256 No. %
2-T&s Titer Rise No. %
(no.)
No.
%
5
3
60
2
40
2
40
314
75’
196 12 a3 101
65 6 28 31
33 50 34 31
20 5 7 a
10 42 a a
6 4 1 3
4 33 1 3
401128 779 11/56 22763
31 78 20 34
Since not all patients had two or more serum samples obtained, the denominator for the a-tube rise in agglutinin is less than for each group as a whole. t Group I = patients with blood cultures positive for fungi; group II = patients with cultures from sources other than blood which yield fungi; and group Ill = patients with no cultures which grew fungi. l
588
April 1979
The American Journal of Medlctne
Volume 64
SERCLDGIC TESTS IN SYSTEMIC CANDIDIASIS-GLEW
TABLE Ii
ET AL.
Candida Preclpitins PatientswithPrecipHathgAnts+ C Total Patients nc.
Systemic or invasive candidiasis Patients without systemic or invasive candidiasis Group I+ Group IIt Group Ill+
lmmlJnodmurJon (DID) lm*No. % No.
%
5
5
100
5
100
196 12 63 101
48 8 25 15
24 67 30 15
4 1 1 2
2 8 1 2
See Materials and Methods for definitions. + See Table I for explanabon of groups. l
in a patient with systemic candidiasis was 18 days. Another patient with candidiasis had negative precipitins eight days after the appearance of candidemia; unfortunately, blood for repeat serology was not obtained until 54 days after the onset of candidemia, at which time precipitins were demonstable by DID and XIE. In a third patient with disseminated infection a repeat sample was not obtained until 37 days after infection, at which time precipitins were positive by both methods. In the other two patients with systemic fungai infection, precipitins were already demonstrable in the first serum samples obtained for the study. Diagnostic Value of Tests (Table ill). The results of these serologic tests were evaluated for sensitivity, specificity and predictive value according to the criteria outlined by Krieg and associates [ 191. Sensitivity (per cent of positive results in patients with the disease being tested for) was 100 per cent with positive precipitins detected by either DID or XIE. Similarly, the specificity (per cent of negative results in people who do no have the disease) was 78 per cent with DID and 98 per cent with XIE. However, because of the high percentage of false positive results with DID, the predictive value (per cent of positive results which are true positive results) or serum Candida precipitins by DID was only 9 per cent. The predictive value of positive precipitins by XIE was 58 per cent. An agglutinin titer of either >/l/64, 2 l/128, or >, l/256 gave sensitivity rates of only 60 per cent, 40 per cent and 40 per cent, respectively, and predictive values of only 4 per cent, 9 per cent and 20 per cent, respectively. If a e-tube rise in agglutinin titer was selected, the sensitivity was 75 per cent, but the predictive value was only 7 per cent. COMMENTS The results of this study demonstrate that the serum Candida precipitin test as usually performed by DID is a sensitive screening test for the detection of systemic or invasive candidiasis. However, only five of our 53 patients in whom precipitins were demonstrated by DID
were considered to have disseminated candidiasis, so that the predictive value of precipitins by DID was only 9 per cent. When the serums which were positive by DID were submitted to XIE, positive tests (precipitin bands to antigens other than mannan) were obtained in only nine patients. Of these positive results, four occurred in patients without evidence of systemic candidiasis, resulting in a predictive value of 56 per cent. Thus, determination of Candida precipitins by XIE is of value in detecting or confirming disseminated candidiasis in patients. On the other hand, agglutinating antibodies were of no diagnostic value: regardless of what criteria were applied (peak agglutinin titer or at least two-tube rise in titer), sensitivity rates were no better than 75 per cent and predictive values were no better than 20 per cent. Several laboratories have examined the diagnostic value of Candida serology in the detection and documentation of significant Candida infections. Staliybrass [7] found precipitins in one patient with “Candida septicemia” and in only one of 403 control patients. Similarly, several reports by Taschdjian and co-workers [8,11,20] have demonstrated the appearance of CanTABLE Ill
Evaluation of Serologic Tests for Candldlasls Sessltlvlty’ (%)
Precipitins DID XIE Agglutinins all64 >/l/128 311256 >/2-tube rise l
Sensitivrty =
+ Specificity =
PrsdktlveValue* 46)
100 100
76 98
9 56
60 40 40 75
67 90 96 69
4 9 20 7
True Positives True Positives + False Negatives True Negatives True Negatives + False Positives
+ Predictive Value =
April 1978
9pscHlcHy+ 1%)
True Positives True Positives -I- False PosItives
The American Journal of Medicine
Volume 64
599
SEROLOGIC TESTS IN SYSTEMIC CANDIDIASIS-GLEW
ET AL
dida precipitins (by DID) in up to 94 per cent of the patients with systemic candidiasis, and in no more than 15 per cent of the patients without candidiasis. Gaines and Remington [6] detected Candida precipitins by DID in 21 of 23 patients with deep candidiasis and in only two of 107 patients without invasive disease. Dee and Rytel [21] found precipitins by DID in all nine patients with disseminated candidiasis and in none of 61 patients without invasion. Everett and co-workers [22] demonstrated precipitins by DID in eight of 12 patients with visceral candidiasis and in none of 26 controls. On the other hand, in two studies reported by Preisler and associates [9, 231, positive precipitins were detected by DID in only four of 23 and six of 14 patients with visceral candidiasis; false positive precipitins occurred in none of 10 and in two of 130 patients without candidiasis and in two of 10 patients with superficial Candida infection. Similarly, Dolan and Streid [24] detected positive precipitins by DID in only three of 10 patients with candidiasis. Although these investigators found no false precipitins in 50 blood bank serums, Everall and associates [ 181 detected positive precipitins by DID in seven of 191 healthy adults, and in 30 of 76 patients with urinary tract infection or colonization due to Candida. Recently, Filice and co-workers [25] noted positive precipitins by DID in only six of 10 patients with systemic candidiasis and in 24 of 33 patients with colonization, local infection or transient candidemia. Several possible explanations have been proposed to reconcile these seemingly discrepant findings. First, many of the serums found to be negative by Preisler and co-workers [9,23] had been examined after storage in a frozen state. In a report by Taschdjian and associates [20] precipitating antibodies to Candida were noted to deteriorate when stored at - 15’C for several months. Second, when screening studies are applied to populations with low prevalence of disease, even the most sensitive and specific screening test will have a low predictive value, whereas application of the test to a population with a high incidence of the disease will result in much improved results [26]. Since the prevalence of systemic candidiasis differed in these several studies, it is not surprising that the evaluation of serologic tests varied from study to study. Third, the falsely negative precipitin reactions reported by Filice et al. [25] generally occurred in patients with low serum levels of immunoglobulins and may have been due to preterminal anergy. Furthermore, approximately onehalf of those patients did not have serial determinations, and positive precipitin reactions might have ultimately developed if they had survived longer. Finally, the high incidence of false positive precipitins detected by DID in some laboratories is likely to be a persistent problem with this screening test, since precipitating antibodies to mannan appear in many patients with Candida col-
590
April 1976
The American Journal of Medicine
Volume 64
onization or overgrowth. Thus, in patients without candidiasis, Chew and Theus [27] found antimannan precipitins in all concentrated serum globulin preparations as well as in 62 per cent of unconcentrated serums using a sensitive tube immunodiffusion test. Murray and associates [ 151 prospectively examined serums from cardiac surgical patients for the presence of precipitins by DID; of 47 patients studied on admission to the hospital, none had precipitins, whereas in 27 of 62 patients without evidence of systemic candidiasis precipitins were detected by DID following surgery. Since all the patients had received perioperative antibiotics, it is possible that mucosal overgrowth of Candida in the gastrointestinal tract (with or without transient seeding to other sites as urinary bladder) may have allowed absorption of mannan into the circulation with subsequent stimulation of antimannan precipitins [28]. Similarly, using a sensitive passive hemmagglutination assay, Weiner and Yount [29] detected antimannan antibodies in 100 per cent of the patients with systemic candidiasis as well as in most serums from a variety of control populations. Furthermore, rises in the antibody titer were noted to occur in patients following transient candidemia as well as during superficial Candida infection. In light of the high incidence in most studies of false-positive precipitin reactions by DID, it is encouraging that serums with precipitin directed only against mannan can be screened out by XIE. In our study, a positive test by XIE, as evidenced by the appearance of precipitating antibodies to antigens other than mannan, was found in all five patients with disseminated or invasive candidiasis and in only four of 196 (2 per cent) patients wihtout candidiasis. This results in a predictive value for the test of 56 per cent in a population with a low prevalence (2.5 per cent) of systemic candidiasis. Accordingly, it is suggested that the detection of Candida precipitins by DID is a sensitive screening test in patients at risk of having systemic or invasive candidiasis; patients who persistently show no precipitins by DID are highly unlikely to have systemic candidiasis. Serums which exhibit precipitins by DID should then be submitted to XIE; if precipitin bands are detected to antigens other than mannan, there is a high likelihood (at least 50 per cent) that the patient has disseminated Candida infection. Two additional points need to be emphasized. First, previous studies have demonstrated that precipitating antibodies do not appear until at least 14 days after antigenic stimulation in rabbits [8], and clinical studies suggest a similar interval of 10 to 14 days in human subjects [ 15,281. In our small series of patients with systemic candidiasis, one patient had a negative precipitin reaction eight days after the onset of systemic
SEROLCGIC TESTS IN SYSTEMIC CANDKHASI-W
candidiasis, and the earliest appearance of precipitins by DID and XIE after onset of systemic candidiasis was 18 days. Therefore, in patients suspected of having systemic candidiasis, serial serum specimens must be examined repeatedly for the presence of Candida precipitins. The second point is that a cell sonicate of C. albicans can react with serum precipitins which arise secondary
5.
6.
7 6 9.
IO
11. 12
13.
14.
15.
Louria DB, Stiff DP, Bennett B: Disseminated moniliasis in the adult. Medicine (Baltimore) 41: 303, 1962. Ellis CA, Spivack ML: The significance of candidemia. Ann Intern Med 67: 511, 1967. Gaines JD, Remington JS: Disseminated candidiasis in the surgical patient. Surgery 72: 730, 1972. Goldstein E, Hoeprich PD: Problem in the diagnosis and treatment of systemic candidiasis J Infect Dis 125: 190. 1972. Young RC, Bennett JE, Geelhoed GW, et al.: Fungemia with compromised host resistance. A study of 70 cases. Ann Intern Med 60: 605, 1974 Gaines JD, Remington JS: Diagnosis of deep infection with Candida. A study of Candida precipitins. Arch Intern Med 132: 699, 1973. Stallybrass FC: Candida precipitins. J Pathol Bact 67. 69, 1964. Taschdjian CL, Kozinn PJ, Oakes A., et al.: Serodiagnosis of systemic candldiasis. J Infect Dis 117: 160, 1967. Preisler HD, Hasenclever HF, Henderson ES: Anti-candida antibodies In patients with acute leukemia. A prospective study Am J Med 61: 352, 1971. Remington JS, Gaines JD, Gllmer MA: Demonstration of Candida preclpitins in human sera by counterimmunoelectrophoresis. Lancet 1: 413, 1972 Kozlnn PJ, Galen RS, Taschdjian CL, et al.: The precipitin test in systemic candldiasis. JAMA 235: 626, 1976. Ryan JA, Abel RM, Abbott WM, et al.: Catheter complications in total parenteral nutrition. A prospective study of 200 consecutive patients. N Engl J Med 290: 757, 1974. Meyers BR, Lieberman TW, Ferry AP: Candida endophthalmitis complicating candidemia Ann Intern Med 79. 647, 1973. Weinstein AJ, Johnson EH, Moellering RC Jr: Candida endophthalmltls: a complication of candidemia Arch Intern Med 132: 749, 1973. Murray IG, Buckley HR. Turner GC: Serological evidence of Candida infectlon after open heart surgery. J Med Microbial 2: 463, 1969.
ET AL.
to colonization or infection with a variety of Candida species [ 8,151. In our study, using a cytoplasmic extract of C. albicans, we detected precipitins by DID in patients colonized with C. albicans, C. tropicalis, C. parapsilosis as well as with Torulopsis glabrata. In addition, one of our patients in whom precipitins were detected by XIE (using the same C. albicans antigen preparation) had endocarditis due to C. parapsilosis (see Figure 2).
16
17
16.
19.
20. 21
22.
23.
24. 25.
26.
27. 26. 29.
April 1976
Axelsen NH, Svendsen PJ: Candida precipitins characterized by a modified antigen-antibody crossed electrophoresis Protides of the Biological Fluids (Peeters H, ed), New York, Pergamon Press, 1972. Buckley HR, Lapa EW, Hipp SS: Immunological significance of some antigens of Candida albicans. 5th International Congress of Infectious Diseases, Vienna, 4: 21, 1970. Everall PH, Morris CA, Morris DF: Antibodies to Candida albicans in hospital patients with and without spinal injury and in normal men and women. J Clin Patho127: 722, 1974. Krieg AF, Gamblno R, Galen RS: Why are clinical laboratory tests performed? When are they valid? JAMA 233: 76, 1975. Taschdjian CL, Philip MS, Kozlnn PJ, et al : Serodiagnosis of candidal infections. J Clin Pathol 57: 195, 1972. Dee TH, Rytel MW: Clinical application of counter-immunoelectrophoresis in detection of candida serum precipltlns. J Lab Clin Med 65: 161, 1975. Everett ED, LaForce FM, Eickhoff TC: Serologic studies in suspected visceral candidiasis Arch Intern Med 135 1075, 1975. Preisler HD, Hasenclever HF, Levitan AA, et al.: Serologic diagnosis of disseminated carxkdtasis in patients with acute leukemia. Ann Intern Med 70: 19, 1969 Dolan CT, Shied RP: Serologic diagnosis of yeast infections Am J Clin Pathol 59: 49, 1973. Filice G, Ya B, Armstrong 0: lmmunodiffusion and agglutination tests for Candida in patients with neoplastic disease. J Infect Dis 135: 349, 1977. Katz MA: A probability graph describing the predictive value of a highly sensitive diagnostic test. N Engl J Med 291: 1115,1974. Chew WH, Theus TLI Candida precipltins. J lmmunol96 220, 1967. Evans EGV, Forster RA: Antibodies to Candii after operations on the heart. J Gen Microbial 9: 303, 1976. Weiner MH, Yount WJ: Mannan antigenemia in the diagnosis of invasive Candida infections. J Clln Invest 56: 1045, 1976.
The American
Journal of Medlclne
Volume 64
591
RICHARD H. GLEW. MD
l
HELEN R. BUCKLEY, Ph.D HARVEY M ROSEN, M.D ROBERT C
MOELLERING, Jr,
M.D
JOSEF E FISCHER, MD. Boston, Massachusetts
From the Departments of Medictne and Surgery, Harvard Medical School and the Department of Microbiology, Harvard School of Public Health, Boston, Massachusetts Preliminary results of this study were presented in part at the Seventy-Fifth Annual Meeting of the Amerrcan Society for MIcrobiology, New York, Aprrl30, 1975, and at the Sixty-First Clinical Congress of the American College of Surgeons, San Francisco, California, October 13, 1975 This study was supported in part by NIAID Training Grant TO1 A 1 0044 1 Requests for reprints should be addressed to. Dr. Richard H Glew, The Memorral Hospital, 119 Belmont Street, Worcester, Massachusetts 01605 Manuscript accepted August 18, 1977 Present address. University of Massachusetts Medical School, The Memorial Hospital, 119 Belmont Street, Worcester, Massachusetts, 01605. l
586
April 1978
In order to assess the diagnostic value of serologic methods in discriminating between colonization and disseminated infection with Candlda species, serum samples from 201 consecutive patients receiving parenteral hyperalimentation were examined for the presence of agglutinating and precipltatlng antibodies to Candida albicans. In these patlents, fungal colonization was detected In 50 per cent, transient fungemia In 6 per cent and dlssemlnated candidlasis in 2.5 per cent. Agglutinating antibodles were of no dlagnostlc value in separating patients with Candida colonization from those with disseminated infection (sensitivity 675 per cent, predictive value 620 per cent). Serum Candlda precipltlns were detected by double lmmunodlffuslon (DID) in all five patients with systemic candldlasls (100 per cent sensitivity); however, since only five of 53 patients with preclpitins detected by DID demonstrated evidence of systemic candidiasis, the predlctlve value of the test was unacceptably low (9 per cent). When the serums which were posltlve by DID were submitted to crossed lmmunoelectrophoresis (XIE), precipitln bands to antigens other than mannan were obtained in only nine patients, Including all five patlents with disseminated candldiasis. Thus, detection by XIE of serum Candida precipltins directed against antigens other than mannan was associated with a predlctlve value of 56 per cent and a sensitivity of 100 per cent. The detection of serum Candlda precipltlns by DID Is a sensitive screening test in patients at high risk for the development of systemic candldlasls; patients with persistently negative preclpltins by DID are unlikely to harbor disseminated candldlasis. If preclpltin bands are detected by XIE to antigens other than mannan, there is a high likelihood that the patient has dlssemlnated or invasive Candida infection. Systemic infections with Candida species have become increasingly more prevalent in the past few decades, in part because advances in medical and surgical expertise and technology have prolonged the survival of compromised patients, and in part because of the side effects or consequences of these supportive medical-surgical measures [l-5]. Since patients ultimately documented to have disseminated candidiasis may fail to have Candida grown on repeated blood cultures, it is often difficult to establish the diagnosis of systemic candidiasis [3,4,6]. On the other hand, the isolation of Candida from various clinical sources, including blood cultures, does not necessarily indicate invasive or disseminated infection [2-41. Many laboratories have attempted to establish serologic tests which would assist in the prompt and reliable diagnosis of deep-seated infections. However,
The American Journal of Medicine
Volume 64
SEROLOGIC TESTS IN SYSTEMIC CANMDIASIS-GLEW
ET AL.
reports differ in their assessments of the sensitivity and specificity of these tests [7-l 11. In order to assess the diagnostic value of Candida serology, we carried out a prospective survey of Candida agglutinins and precipitins in a group of 201 hospitalized patients receiving total parenteral nutrition. This group was selected because it provided a diverse patient population with a wide variety of underlying medical and surgical disorders. In general, the patients shared a common background involving one or more factors known to predispose to fungal colonization and infection: surgery, antibiotic administration, genitourinary catheters, malnutrition, malignancy, thermal burns, old age and nutritional support via protracted intravenous alimentation through indwelling plastic venous cannulas. MATERIALS AND METHODS Patients. A total of 577 serum samples were obtained prospectively from 20 1 consecutive hospitalized patients at the Massachusetts General Hospital who received intravenous hyperalimentation during a nine month period from June 1974 through February 1975. As has been the experience in this hospital in recent years, the mean duration of hyperalimentation in these patients was approximately three weeks [ 121. Of this group, 127 patients were male and 74 female; the ages of the patients ranged from nine to 89 years. Serum samples were collected at least once weekly and were processed within two weeks of collection for determination of agglutinating and precipitating antibodies against Candida albicans. Informed consent for blood sampling was obtained from the patient or a responsible family member in each case. Criteria for Diagnosis Candidiasis. Cultures ofmaterial from various clinical sites were obtained at the discretion of the primary physician caring for each patient. Cultures of the urine were obtained in 96 per cent of the patients, of the blood in 89 per cent, of the sputum in 72 per cent, of material from intravenous catheters in 68 per cent, and of various body fluids and wounds in 58 per cent; no specimens for cultures were obtained from six (3 per cent) of the patients. Of the 201 patients, the diagnosis of systemic candidiasis was made in four, each of whom exhibited two or more of the following: (1) repeated isolation of Candida albicans from blood cultures (three patients) or C. parapsilosis (one patient) over several days (six to 85 days) in patients exhibiting clinical signs of septicemia; (2) funduscopic findings diagnostic of Candida endopthalmitis (two patients) [ 13,141; (3) clinical and postmortem pathologic evidence of Candida endocarditis on a prosthetic value (one patient); (4) signs of peritonitis plus growth of C. albicans from repeated paracenteses of peritoneal fluid (one patient). One additional patient was diagnosed as having invasive candidiasis of the genitourinary tract based on repeated isolation of C. albicans over a nine-month period from the urinary bladder, both ureters and bilaterial nephrostomy sites (the latter on deep cultures of material obtained from sinus tracts). The remaining 196 patients without invasive or systemic candidiasis were divided into three groups. In group I were
1 -_
Figure 1. Crossed immunoelectrophoresis of serum from patient without candidiasis demonstrating antibodies directed only against cell wall (mannan) antigens (see arrow). patients with transient self-limited fungemia: in six patients C. albicans was grown on culture of the blood, in three patients Torulopsis glabrata, in one patient C. tropicalis and in one patient C. parapsilosis; in one patient blood cultures grew C. albicans and T. glabrata on different occasions. Among these 12 patients, adequate clinical follow-up (including negative blood cultures) was obtained for two weeks or longer after the initial episode of fungemia in nine. Three other patients died within two days of their fungemic episodes. Postmortem examination (including fungal cultures) demonstrated no evidence of systemic or invasive candidiasis in two of these patients, whereas no autopsy was performed in the remaining patient. In group II were 83 patients in whom Candida species or T. glabrata were grown from various clinical sites exclusive of blood, but who showed no clinical or pathologic signs of deep invasion and were considered to be simply colonized with these fungi. Finally, in group Ill were 101 patients from whom fungi were never isolated in routine clinical specimens. Serologic Tests. Tests for agglutinating antibodies were carried out with suspensions of whole yeast cells of C. albicans (Group A) and tests for precipitating antibodies were performed on all serums by double immunodiffusion (DID) as previously described [ 151. Those serums which gave a positive precipitin reaction by DID were subjected to crossed immunoelectrophoresis (XIE) according to the method of Axelsen and Svendsen [ 161. In the XIE tests, the Candida antigen was subjected to sequential electrophoresis in two directions, the second time into a gel containing the patient’s serum. With XIE, it is possible to detect two types of precipitin bands. The first is a broad, fuzzy, deeply-staining band (Figure 1) which represents a reaction to cell wall antigen (mannan) and can be identified either by prior absorption of the test serum with mannan or by prior reaction of the antigen preparation wlth concanavalin A, which precipitates out mannans
April 1978
The American Journal of Medicine
Volume 84
587
SERCLCGIC TESTS IN SYSTEMIC CANDIDIASIS-GLEW
ET AL.
candidiasis, and in 60 per cent of the patients with candidiasis. Agglutinins of al/256 were observed in from 1 to 33 per cent of the patients without candidiasis and in 40 per cent of the patients with disseminated or invasive disease. In cases in which more than one serum sample had been obtained, at least a e-tube rise in agglutinin titer was detected in from 20 to 78 per cent (31 per cent over-all) of the patients without candidiasis and in 75 per cent of the patients with candidiasis. In patients without candidiasis, the greatest incidence of elevated agglutinin titers (either 2 l/64 or B l/256) and P-tube rises in agglutinin titer were seen in patients with positive blood cultures (transient fungemia). Moreover, these agglutinin values were similar to those seen in the five patients with systemic candidiasis. Preclpitins (Table II). Precipitating antibodies were found by DID in from 15 to 67 per cent (24 per cent over-all) of the patients without candidiasis and in 100 per cent of the five patients with systemic or invasive candidiasis. As with agglutinins, the highest percentage of positive tests among patients without candidiasis was noted in patients with transient fungemia. However, when serums which were positive by DID were submitted to XIE, the number of patients with positive tests (precipitin lines to antigens other than mannan) was markedly reduced in the groups of patients without disseminated candidiasis. Of the patients without systemic candidiasis, only 1 to 8 (2 per cent over-all) had positive precipitins by XIE, whereas all five (100 per cent) of the patients with systemic or invasive candidiasis had positive reactions by XIE. Limited information is available concerning the chronologic sequence of positive serology in our study. Only in patients in whom fungemia ultimately culminates in definite signs of systemic candidiasis can one with certainty define the timing of the initial antigenic stimulus to antibody production. Only five of our patients were documented to have disseminated infection and were able to be evaluated for the temporal sequence of precipitins. The earliest appearance of positive precipitins by DID and XIE after the onset of candidemia
Figure 2. Crossed immunoelectrophoresis of Serum from a patient with endocarditis due to Candida parapsilosis demonstrating antibodies to cytoplasmic antigens as well as to cell wall (mannan) antigens (see arrow).
[ 171. The second type of bands are fine precipitin arcs which represent the reaction of serum antibodies with nonmannan Candida antigens (Figure 2). Since precipitating antibodies to mannan have been demonstrated to occur in normal subjects and in people superficially colonized with candida [ 181, serums were considered positive by XIE only if precipitin bands were formed to antigens other than mannan (Flgure 2). The antigen preparation employed in the precipitin tests (DID and XIE) was a crude cytoplasmic extract derived from homogenization of yeast phase C. albicans grown in synthetic liquid medium [ 151. RESULTS
Agglutinins (Table I). Agglutinating antibodies were detected at a titer of 2 l/64 in from 31 to 50 per cent (33 per cent over-all) of the patients without systemic
TABLE I
Candlda Agglutinins PatientswtthAgglutlnatlng Antibodies
Total
Patients Systemic or invasive candidiasis Patients without systemic or invasive candidiasis Group It Group IIt Group IlIT
TiterH/64
Ttter H/128 No. %
Ttter M256 No. %
2-T&s Titer Rise No. %
(no.)
No.
%
5
3
60
2
40
2
40
314
75’
196 12 a3 101
65 6 28 31
33 50 34 31
20 5 7 a
10 42 a a
6 4 1 3
4 33 1 3
401128 779 11/56 22763
31 78 20 34
Since not all patients had two or more serum samples obtained, the denominator for the a-tube rise in agglutinin is less than for each group as a whole. t Group I = patients with blood cultures positive for fungi; group II = patients with cultures from sources other than blood which yield fungi; and group Ill = patients with no cultures which grew fungi. l
588
April 1979
The American Journal of Medlctne
Volume 64
SERCLDGIC TESTS IN SYSTEMIC CANDIDIASIS-GLEW
TABLE Ii
ET AL.
Candida Preclpitins PatientswithPrecipHathgAnts+ C Total Patients nc.
Systemic or invasive candidiasis Patients without systemic or invasive candidiasis Group I+ Group IIt Group Ill+
lmmlJnodmurJon (DID) lm*No. % No.
%
5
5
100
5
100
196 12 63 101
48 8 25 15
24 67 30 15
4 1 1 2
2 8 1 2
See Materials and Methods for definitions. + See Table I for explanabon of groups. l
in a patient with systemic candidiasis was 18 days. Another patient with candidiasis had negative precipitins eight days after the appearance of candidemia; unfortunately, blood for repeat serology was not obtained until 54 days after the onset of candidemia, at which time precipitins were demonstable by DID and XIE. In a third patient with disseminated infection a repeat sample was not obtained until 37 days after infection, at which time precipitins were positive by both methods. In the other two patients with systemic fungai infection, precipitins were already demonstrable in the first serum samples obtained for the study. Diagnostic Value of Tests (Table ill). The results of these serologic tests were evaluated for sensitivity, specificity and predictive value according to the criteria outlined by Krieg and associates [ 191. Sensitivity (per cent of positive results in patients with the disease being tested for) was 100 per cent with positive precipitins detected by either DID or XIE. Similarly, the specificity (per cent of negative results in people who do no have the disease) was 78 per cent with DID and 98 per cent with XIE. However, because of the high percentage of false positive results with DID, the predictive value (per cent of positive results which are true positive results) or serum Candida precipitins by DID was only 9 per cent. The predictive value of positive precipitins by XIE was 58 per cent. An agglutinin titer of either >/l/64, 2 l/128, or >, l/256 gave sensitivity rates of only 60 per cent, 40 per cent and 40 per cent, respectively, and predictive values of only 4 per cent, 9 per cent and 20 per cent, respectively. If a e-tube rise in agglutinin titer was selected, the sensitivity was 75 per cent, but the predictive value was only 7 per cent. COMMENTS The results of this study demonstrate that the serum Candida precipitin test as usually performed by DID is a sensitive screening test for the detection of systemic or invasive candidiasis. However, only five of our 53 patients in whom precipitins were demonstrated by DID
were considered to have disseminated candidiasis, so that the predictive value of precipitins by DID was only 9 per cent. When the serums which were positive by DID were submitted to XIE, positive tests (precipitin bands to antigens other than mannan) were obtained in only nine patients. Of these positive results, four occurred in patients without evidence of systemic candidiasis, resulting in a predictive value of 56 per cent. Thus, determination of Candida precipitins by XIE is of value in detecting or confirming disseminated candidiasis in patients. On the other hand, agglutinating antibodies were of no diagnostic value: regardless of what criteria were applied (peak agglutinin titer or at least two-tube rise in titer), sensitivity rates were no better than 75 per cent and predictive values were no better than 20 per cent. Several laboratories have examined the diagnostic value of Candida serology in the detection and documentation of significant Candida infections. Staliybrass [7] found precipitins in one patient with “Candida septicemia” and in only one of 403 control patients. Similarly, several reports by Taschdjian and co-workers [8,11,20] have demonstrated the appearance of CanTABLE Ill
Evaluation of Serologic Tests for Candldlasls Sessltlvlty’ (%)
Precipitins DID XIE Agglutinins all64 >/l/128 311256 >/2-tube rise l
Sensitivrty =
+ Specificity =
PrsdktlveValue* 46)
100 100
76 98
9 56
60 40 40 75
67 90 96 69
4 9 20 7
True Positives True Positives + False Negatives True Negatives True Negatives + False Positives
+ Predictive Value =
April 1978
9pscHlcHy+ 1%)
True Positives True Positives -I- False PosItives
The American Journal of Medicine
Volume 64
599
SEROLOGIC TESTS IN SYSTEMIC CANDIDIASIS-GLEW
ET AL
dida precipitins (by DID) in up to 94 per cent of the patients with systemic candidiasis, and in no more than 15 per cent of the patients without candidiasis. Gaines and Remington [6] detected Candida precipitins by DID in 21 of 23 patients with deep candidiasis and in only two of 107 patients without invasive disease. Dee and Rytel [21] found precipitins by DID in all nine patients with disseminated candidiasis and in none of 61 patients without invasion. Everett and co-workers [22] demonstrated precipitins by DID in eight of 12 patients with visceral candidiasis and in none of 26 controls. On the other hand, in two studies reported by Preisler and associates [9, 231, positive precipitins were detected by DID in only four of 23 and six of 14 patients with visceral candidiasis; false positive precipitins occurred in none of 10 and in two of 130 patients without candidiasis and in two of 10 patients with superficial Candida infection. Similarly, Dolan and Streid [24] detected positive precipitins by DID in only three of 10 patients with candidiasis. Although these investigators found no false precipitins in 50 blood bank serums, Everall and associates [ 181 detected positive precipitins by DID in seven of 191 healthy adults, and in 30 of 76 patients with urinary tract infection or colonization due to Candida. Recently, Filice and co-workers [25] noted positive precipitins by DID in only six of 10 patients with systemic candidiasis and in 24 of 33 patients with colonization, local infection or transient candidemia. Several possible explanations have been proposed to reconcile these seemingly discrepant findings. First, many of the serums found to be negative by Preisler and co-workers [9,23] had been examined after storage in a frozen state. In a report by Taschdjian and associates [20] precipitating antibodies to Candida were noted to deteriorate when stored at - 15’C for several months. Second, when screening studies are applied to populations with low prevalence of disease, even the most sensitive and specific screening test will have a low predictive value, whereas application of the test to a population with a high incidence of the disease will result in much improved results [26]. Since the prevalence of systemic candidiasis differed in these several studies, it is not surprising that the evaluation of serologic tests varied from study to study. Third, the falsely negative precipitin reactions reported by Filice et al. [25] generally occurred in patients with low serum levels of immunoglobulins and may have been due to preterminal anergy. Furthermore, approximately onehalf of those patients did not have serial determinations, and positive precipitin reactions might have ultimately developed if they had survived longer. Finally, the high incidence of false positive precipitins detected by DID in some laboratories is likely to be a persistent problem with this screening test, since precipitating antibodies to mannan appear in many patients with Candida col-
590
April 1976
The American Journal of Medicine
Volume 64
onization or overgrowth. Thus, in patients without candidiasis, Chew and Theus [27] found antimannan precipitins in all concentrated serum globulin preparations as well as in 62 per cent of unconcentrated serums using a sensitive tube immunodiffusion test. Murray and associates [ 151 prospectively examined serums from cardiac surgical patients for the presence of precipitins by DID; of 47 patients studied on admission to the hospital, none had precipitins, whereas in 27 of 62 patients without evidence of systemic candidiasis precipitins were detected by DID following surgery. Since all the patients had received perioperative antibiotics, it is possible that mucosal overgrowth of Candida in the gastrointestinal tract (with or without transient seeding to other sites as urinary bladder) may have allowed absorption of mannan into the circulation with subsequent stimulation of antimannan precipitins [28]. Similarly, using a sensitive passive hemmagglutination assay, Weiner and Yount [29] detected antimannan antibodies in 100 per cent of the patients with systemic candidiasis as well as in most serums from a variety of control populations. Furthermore, rises in the antibody titer were noted to occur in patients following transient candidemia as well as during superficial Candida infection. In light of the high incidence in most studies of false-positive precipitin reactions by DID, it is encouraging that serums with precipitin directed only against mannan can be screened out by XIE. In our study, a positive test by XIE, as evidenced by the appearance of precipitating antibodies to antigens other than mannan, was found in all five patients with disseminated or invasive candidiasis and in only four of 196 (2 per cent) patients wihtout candidiasis. This results in a predictive value for the test of 56 per cent in a population with a low prevalence (2.5 per cent) of systemic candidiasis. Accordingly, it is suggested that the detection of Candida precipitins by DID is a sensitive screening test in patients at risk of having systemic or invasive candidiasis; patients who persistently show no precipitins by DID are highly unlikely to have systemic candidiasis. Serums which exhibit precipitins by DID should then be submitted to XIE; if precipitin bands are detected to antigens other than mannan, there is a high likelihood (at least 50 per cent) that the patient has disseminated Candida infection. Two additional points need to be emphasized. First, previous studies have demonstrated that precipitating antibodies do not appear until at least 14 days after antigenic stimulation in rabbits [8], and clinical studies suggest a similar interval of 10 to 14 days in human subjects [ 15,281. In our small series of patients with systemic candidiasis, one patient had a negative precipitin reaction eight days after the onset of systemic
SEROLCGIC TESTS IN SYSTEMIC CANDKHASI-W
candidiasis, and the earliest appearance of precipitins by DID and XIE after onset of systemic candidiasis was 18 days. Therefore, in patients suspected of having systemic candidiasis, serial serum specimens must be examined repeatedly for the presence of Candida precipitins. The second point is that a cell sonicate of C. albicans can react with serum precipitins which arise secondary
5.
6.
7 6 9.
IO
11. 12
13.
14.
15.
Louria DB, Stiff DP, Bennett B: Disseminated moniliasis in the adult. Medicine (Baltimore) 41: 303, 1962. Ellis CA, Spivack ML: The significance of candidemia. Ann Intern Med 67: 511, 1967. Gaines JD, Remington JS: Disseminated candidiasis in the surgical patient. Surgery 72: 730, 1972. Goldstein E, Hoeprich PD: Problem in the diagnosis and treatment of systemic candidiasis J Infect Dis 125: 190. 1972. Young RC, Bennett JE, Geelhoed GW, et al.: Fungemia with compromised host resistance. A study of 70 cases. Ann Intern Med 60: 605, 1974 Gaines JD, Remington JS: Diagnosis of deep infection with Candida. A study of Candida precipitins. Arch Intern Med 132: 699, 1973. Stallybrass FC: Candida precipitins. J Pathol Bact 67. 69, 1964. Taschdjian CL, Kozinn PJ, Oakes A., et al.: Serodiagnosis of systemic candldiasis. J Infect Dis 117: 160, 1967. Preisler HD, Hasenclever HF, Henderson ES: Anti-candida antibodies In patients with acute leukemia. A prospective study Am J Med 61: 352, 1971. Remington JS, Gaines JD, Gllmer MA: Demonstration of Candida preclpitins in human sera by counterimmunoelectrophoresis. Lancet 1: 413, 1972 Kozlnn PJ, Galen RS, Taschdjian CL, et al.: The precipitin test in systemic candldiasis. JAMA 235: 626, 1976. Ryan JA, Abel RM, Abbott WM, et al.: Catheter complications in total parenteral nutrition. A prospective study of 200 consecutive patients. N Engl J Med 290: 757, 1974. Meyers BR, Lieberman TW, Ferry AP: Candida endophthalmitis complicating candidemia Ann Intern Med 79. 647, 1973. Weinstein AJ, Johnson EH, Moellering RC Jr: Candida endophthalmltls: a complication of candidemia Arch Intern Med 132: 749, 1973. Murray IG, Buckley HR. Turner GC: Serological evidence of Candida infectlon after open heart surgery. J Med Microbial 2: 463, 1969.
ET AL.
to colonization or infection with a variety of Candida species [ 8,151. In our study, using a cytoplasmic extract of C. albicans, we detected precipitins by DID in patients colonized with C. albicans, C. tropicalis, C. parapsilosis as well as with Torulopsis glabrata. In addition, one of our patients in whom precipitins were detected by XIE (using the same C. albicans antigen preparation) had endocarditis due to C. parapsilosis (see Figure 2).
16
17
16.
19.
20. 21
22.
23.
24. 25.
26.
27. 26. 29.
April 1976
Axelsen NH, Svendsen PJ: Candida precipitins characterized by a modified antigen-antibody crossed electrophoresis Protides of the Biological Fluids (Peeters H, ed), New York, Pergamon Press, 1972. Buckley HR, Lapa EW, Hipp SS: Immunological significance of some antigens of Candida albicans. 5th International Congress of Infectious Diseases, Vienna, 4: 21, 1970. Everall PH, Morris CA, Morris DF: Antibodies to Candida albicans in hospital patients with and without spinal injury and in normal men and women. J Clin Patho127: 722, 1974. Krieg AF, Gamblno R, Galen RS: Why are clinical laboratory tests performed? When are they valid? JAMA 233: 76, 1975. Taschdjian CL, Philip MS, Kozlnn PJ, et al : Serodiagnosis of candidal infections. J Clin Pathol 57: 195, 1972. Dee TH, Rytel MW: Clinical application of counter-immunoelectrophoresis in detection of candida serum precipltlns. J Lab Clin Med 65: 161, 1975. Everett ED, LaForce FM, Eickhoff TC: Serologic studies in suspected visceral candidiasis Arch Intern Med 135 1075, 1975. Preisler HD, Hasenclever HF, Levitan AA, et al.: Serologic diagnosis of disseminated carxkdtasis in patients with acute leukemia. Ann Intern Med 70: 19, 1969 Dolan CT, Shied RP: Serologic diagnosis of yeast infections Am J Clin Pathol 59: 49, 1973. Filice G, Ya B, Armstrong 0: lmmunodiffusion and agglutination tests for Candida in patients with neoplastic disease. J Infect Dis 135: 349, 1977. Katz MA: A probability graph describing the predictive value of a highly sensitive diagnostic test. N Engl J Med 291: 1115,1974. Chew WH, Theus TLI Candida precipltins. J lmmunol96 220, 1967. Evans EGV, Forster RA: Antibodies to Candii after operations on the heart. J Gen Microbial 9: 303, 1976. Weiner MH, Yount WJ: Mannan antigenemia in the diagnosis of invasive Candida infections. J Clln Invest 56: 1045, 1976.
The American
Journal of Medlclne
Volume 64
591