Performance of Cryptococcus Antigen Latex Agglutination Kits on

Transcription

1 JOURNAL OF CLINICAL MICROBIOLOGY, Feb. 1991, p /91/2333-7$2./ Copyright C) 1991, American Society for Microbiology Vol. 29, No. 2 Performance of Cryptococcus Antigen Latex Agglutination Kits on Serum and Cerebrospinal Fluid Specimens of AIDS Patients before and after Pronase Treatment JOHN R. HAMILTON,'* ANITA NOBLE,' DAVID W. DENNING,12 3'4 AND DAVID A. STEVENS"2'3'4 Microbiology Section, Department of Pathology,' and Division of Infectious Diseases, Department of Medicine,2 Santa Clara Valley Medical Center, 751 South Bascom Avenue, and California Institute for Medical Research,4 San Jose, California 95128, and Division of Infectious Diseases, Department of Medicine, Stanford University School of Medicine, Stanford, California Received 8 June 199/Accepted 23 October 199 Cryptococcal antigen titers in 97 serum and 42 cerebrospinal fluid (CSF) specimens from 37 AIDS patients with culture-proven cryptococcal infection were determined with the Meridian kit (Meridian Diagnostics Inc., Cincinnati, Ohio) before and after treatment with pronase. The geometric mean titers before and after pronase treatment were 1:45 and 1:588 in serum and 1:97 and 1:79 in CSF, respectively. Only on serum (but not CSF) specimens after pronase treatment were (i) titers increased by 2 to 13 dilutions on 57% of the specimens, all of which had titers of -<1:128 before pronase treatment, (ii) false-negative reactions on 27% of specimens before pronase treatment eliminated, all of which had titers from 1:4 to 1:4,96, (iii) prozone-like reactions (titer, <1:256) on 9% of the specimens before pronase treatment eliminated, and (iv) agglutination reactions on all specimens stronger and easier to interpret. Antifungal agents added to serum as well as freeze-thaw cycles did not change antigen titers in serum. After two separate tests, the same titers were obtained on 94% of 35 serum specimens that were treated with pronase and on 96% of 53 CSF specimens that were not treated with pronase. A total of 26 serum specimens and 28 CSF specimens from patients with no cryptococcal disease were negative before and after pronase treatment. The IBL kit (International Biological Labs Inc., Cranbury, N.J.) was compared with the Meridian kit on 41 serum specimens and 14 CSF specimens. Results from the two kits agreed on 54 and 68% of serum specimens and 86 and 93% of CSF specimens before and after pronase treatment, respectively. The IBL kit generally produced higher titers on specimens in disagreement and produced no prozone-like reactions. Routine pronase treatment of serum is recommended with the Meridian kit in order to eliminate false-negative and unclear agglutination reactions by producing a consistent interpretation of agglutination reactions. CSF specimens do not require pronase treatment. Titer results produced by the kits from the two different manufacturers varied considerably: the kits should not be used interchangeably for determining antigen titers in serum specimens. The detection of capsular antigen from Cryptococcus neoformans in serum and cerebrospinal fluid (CSF) is a sensitive and specific test for the rapid diagnosis of cryptococcosis (2, 3, 5, 9, 12, 16). The sensitivity of the latex agglutination method ranges from 1 to 157 ng of capsular polysaccharide antigen per ml, depending upon the reagents used (3, 12, 16). Titration of antigen in serum and CSF specimens has been used for diagnosis, prognosis, and monitoring of antifungal therapy (1-3, 6, 8, 11, 13). With the increasing number of C. neoformans infections caused by the AIDS epidemic, antigen detection is an important test for the clinical microbiology laboratory. The availability of commercial reagents has greatly contributed to the routine testing capability of laboratories; however, the requirements for optimal performance of the various kits and their relative capabilities are not completely defined. Various specimen treatment methods and reagents have been reported which eliminate the false latex agglutination reactions caused by rheumatoid factor and other unknown factors in serum and CSF and allow detection and titration of antigen (4, 7, 9, 1, 15). A concurrent finding was that some of these reagents increase antigen titers. One of the reagents, pronase, produces increased antigen titers on specimens not * Corresponding author. 333 needing treatment (those with no false agglutination reactions). Increased antigen titers were observed on 8.7% of the serum specimens and on 2% of the CSF specimens (9). False-negative reactions with latex agglutination kits have rarely been reported. Most have been corrected with either dilution of the specimen or pronase treatment. A prozonelike reaction was reported on a CSF specimen (14) and a serum specimen (16) that was resolved by dilution of each specimen: antigen titers were as high as 1:1,6. Other negative reactions have been reported from serum and CSF specimens that were converted to positive after the specimens were treated with pronase (9). Nineteen percent of the serum specimens were falsely negative before pronase treatment; after pronase treatment titers as high as 1:1,24 were obtained. One kit (International Biological Labs [IBLI Inc., Cranbury, N.J.) has been studied more than the other three commercially available kits have. Two of these kits have been compared with the IBL kit in only one study (16). The other kit, which is manufactured by Meridian Diagnostics Inc., Cincinnati, Ohio, was used in conjunction with the IBL kit in a study to monitor therapy (3); no other studies with the Meridian kit have been reported. The purpose of this study was to report our experience with the Meridian kit used on clinical specimens from patients with AIDS and with culture-proven cryptococcal

2 334 HAMILTON ET AL. infection, the advantages and uses of pronase treatment of specimens, and comparison of the IBL and Meridian kits. MATERIALS AND METHODS Specimens. A total of 42 CSF specimens and 97 serum specimens from 37 AIDS patients with culture-proven cryptococcal infection were tested before and after pronase treatment from late 1987 through September Serum and CSF specimens were tested with the Meridian kit before pronase treatment and prior to storage at -2 C or concurrently with and without pronase treatment or were stored at -2 C and then tested concurrently with and without pronase treatment. Thirteen different lot numbers of the Meridian kit were used. Eight lot numbers were used on the 61 specimens that produced different titers after pronase treatment. On 56 of 61 (92%) specimens, the same lot number of the kit (one of the eight lots) was used to test the specimen before and after pronase treatment. The differences in titer on these 56 specimens spanned the entire range of titer increases and decreases after pronase treatment. Therefore, the difference in titer between specimens that were treated and not treated with pronase was related to the action of pronase and not the different lots of reagents. Specimens tested with the IBL kit were stored at -2 C and then tested concurrently with and without pronase treatment. Four different lot numbers of the kit were used. Fifty-three CSF specimens from 2 patients were repeat tested with the Meridian kit without pronase treatment. Of the 53 CSF specimens, 21 were selected from the previous group of 42 CSF specimens from 37 patients, an additional 28 CSF specimens were selected from those 37 patients, and 4 other CSF specimens were selected from 2 additional patients. Specimens were stored and tested in borosilicate glass tubes (13 by 1 mm; dispo Tubes; American Scientific Products, McGaw Park, Ill.). The tubes were sealed with plastic caps (Tainertop; Fisher Scientific, Pittsburgh, Pa.) for storage. Agglutination titers were considered different when there was a discrepancy of.2 dilutions. CSF and serum specimens from patients with no cryptococcal disease. A total of 28 CSF specimens and 26 serum specimens were collected from 44 patients without cryptococcal disease. Samples were stored at -2 C and later tested with the Meridian kit concurrently with and without pronase treatment. Meridian kit. The Meridian Diagnostics cryptococcal antigen latex agglutination kit, CALAS, contains anticryptococcal globulin reagent (latex particles coated with rabbit anticryptococcal globulin), normal globulin reagent (latex particles coated with rabbit normal globulin), antiglobulin control (goat anti-rabbit serum), negative control (normal human serum), cryptococcal antigen control (purified polysaccharide antigen from cultures of C. neoformans), and glycine-buffered diluent with albumin at ph 8.4. The kit was used according to the instructions provided by the manufacturer. Briefly, CSF specimens were placed in a boiling water bath for 3 min, and serum specimens were heated at 56 C for 3 min. Twenty-five-microliter samples of CSF or serum were mixed with an equal volume of anticryptococcal reagent or normal globulin reagent. Positive and negative controls were tested daily. The slide was placed on a rotating platform at 12 rpm for 5 min, and the reaction was read within 15 to 3 s without moving the slide. The agglutination reaction was graded negative or 1+ to 4+. Pictures of each reaction were provided in the kit, which helped to produce a consistent interpretation of the reactions. A negative reaction was interpreted as either a homogeneous suspension with no visible clumps or a fine granulation against a milky background (1+ reaction). The agglutinations of 2+ to 4+ were graded by the increasing size of the clumps and the degree of background clearing; these were interpreted as positive. Specimens were diluted in the glycine buffer by using a twofold dilution series to determine the antigen titer. IBL kit. CSF and serum specimens were tested according to the procedures supplied with the Crypto-LA Kit (IBL Laboratories) and defined previously (15). Results were read immediately after mechanical rotation with minimal hand rotation. Reactions were interpreted as defined in the kit. A negative reaction was interpreted as a fine granular background or milky suspension with the absence of agglutination. A positive reaction was interpreted as distinct large clumps against a clear or slightly milky background, or small but definite clumps against a milky background. Four different lot numbers were used. Pronase preparation and use. The pronase reagent (Pronase protease; 5372; Calbiochem, San Diego, Calif.) was prepared as described previously (9), but was stored at -2 C and not lyophilized. J. CLIN. MICROBIOL. The glycine-buffered saline (ph 8.2) supplied with the Meridian kit was used to suspend the pronase. The reagent was divided into.5-ml portions in glass tubes, sealed with plastic caps, and frozen at -2 C for later use. Prior to pronase treatment, all specimens were prepared according to the instructions supplied with each latex agglutination kit. The pronase reagent was thawed at room temperature, added to an equal volume (.2 ml) of CSF or serum specimen in glass tubes, incubated for 15 min at 56 C, and then heated in a boiling water bath for 5 min to inactivate the pronase enzyme. Pronase-treated specimens were not saved beyond the initial day of testing. If repeat testing was performed on a later day, a portion of the original specimen was treated with pronase and tested. Since the specimen is diluted with an equal volume of the pronase reagent, this dilution factor was incorporated into all calculations of antigen titer. The reagent was held at room temperature for the day of use and then discarded. Studies conducted over a 45-day period showed that the reagent was stable after reconstitution when held at -2 or 2 to 8 C. RESULTS No cryptococcal disease. The results from 28 CSF specimens and 26 serum specimens from 44 patients with no cryptococcal disease were negative when tested before and after pronase treatment by using the Meridian kit. Serum tested with Meridian kit. Results from 97 serum specimens from 35 patients with culture-proven cryptococcal disease tested with the Meridian kit are shown in Fig. 1. Before pronase treatment of specimens, most agglutination reactions were 1 to 3+ in reactivity, producing titers that ranged from negative to 1:65,536 with a median and geometric mean of 1:16 and 1:45, respectively. After pronase treatment agglutination reactions were consistently 3+ to 4+ and produced approximately the same range of titers, but the median and geometric mean increased to 1:512 and 1:588, respectively. On 41 of 97 (43%) specimens, results were within 1 dilution of each other before and after pronase treatment. Titers ranged 1:16 to 1:65,536, with most (39 of 41; 95%) being at a titer of.1:256. On 55 of 97 (57%) specimens, titers were higher after pronase treatment compared with titers before pronase

3 VOL. 29, 1991 it PRONASE AND CRYPTOCOCCAL ANTIGEN TESTING , me 8192Ubm * ~496 $ 248_. a cc 512 * w L 256 *o-oo _ 128 in3*_ = - 64 * * * *- o 32 mm. 16 ~~~~~~~ 8 * 4 E E 2 undili neg ) U1) 'a c c Ct D CD CM lt CD CM t co CD c') CM cl cm ) It c- CM CCM t CJ "it m co ) co CD Co) T- C') ooco r1- C-J c LO LO to Reciprocal of bter BEFORE pronase treatment FIG. 1. Comparison of cryptococcal antigen titers on 97 serum specimens (U) and 42 CSF specimens () tested with the Meridian kit before and after pronase treatment. treatment by 2, 3, 4, 5, 6, 7, 8, 9, 1, 11, 12, and 13 dilutions on 4, 6, 7, 13, 5, 1, 5, 2, 5, 3, 3, and 1 specimen, respectively. Titers on all 55 specimens before pronase treatment were s1:128. After pronase treatment, titers on these 55 specimens ranged from 1:4 to 1:4,96. Titers on 26 of 55 specimens (from 1 patients) before pronase treatment were interpreted as negative. On many of these specimens, a 1+ agglutination (negative interpretation) was observed that produced an endpoint titer equal to that observed after pronase treatment. After pronase treatment, titers on these specimens, which were negative before pronase treatment, ranged from 1:4 to 1:4,96. On 1 of 97 (.8%) specimens, pronase treatment produced a fourfold lower titer (1:2,48) compared with the titer (1:8,192) before pronase treatment. Randomly selected serum specimens (35 specimens from 16 patients) with titers that ranged from 1:8 to 1:32,768 were repeat tested to determine the reproducibility of titer results after storage at -2 C and then pronase treatment. Titers on 33 of 35 specimens were within 1 dilution of the previously obtained titer. On 2 of 35 specimens, each from a different patient, the repeat titer differed from the previous one by 2 dilutions; one was greater (initial titer, 1:4,96; repeat titer, 1:16,384); the other was less (initial titer, 1:32,768; repeat titer, 1:8,192). Effect of multiple freeze-thaw cycles on antigen titer. In order to determine whether multiple freeze-thaw cycles on a specimen could cause changes in titer, two patient specimens (titers, 1:1,24 and 1:32,768) were thawed, and a portion was removed for testing and then refrozen. The cycle was repeated weekly for 4 weeks. Each weekly sample was pronase treated and then tested with the Meridian kit. No changes in titer were observed at any of the testing times. All titers were within 1 dilution of the initial titer. CSF tested with Meridian kit. Results on 42 CSF specimens from 19 patients with culture-proven cryptococcal meningitis tested with the Meridian kit before and after pronase treatment are shown in Fig. 1. The range of titers before and after pronase treatment was approximately the same, undiluted to 1:8,192. The median and geometric mean titers before and after pronase treatment were 1:128 and 1:97 and 1:128 and 1:79, respectively. On 37 of 42 (88%) specimens, titers after pronase treatment were within 1 dilution of the titers obtained before pronase treatment. Pronase treatment did not enhance reactivity or increase antigen titers. On 5 of 42 specimens, each from a different patient, results were 2 to 3 dilutions lower after pronase treatment. Fifty-three randomly selected CSF specimens from 2 patients were repeat tested without pronase treatment after storage at -2 C. Titers ranged from undiluted to 1:4,96. Titers from 51 of 53 (96%) specimens after storage were within 1 dilution of the previously obtained titer. Two of 53 specimens, each from a different patient, differed by 2 dilutions; one was lower (initial titer, 1:32; repeat titer, 1:8); the other was higher (initial titer, 1:256; repeat titer, 1:1,24). Prozone-like reactions. Only 9 of 97 (13%) serum specimens (from seven patients) and none of 42 CSF specimens produced a prozone-like reaction (no agglutination in low dilutions of the specimen) before pronase treatment with the Meridian kit. All specimens were from patients who had meningitis. The results are shown in Table 1. Six of the nine results occurred on all specimens or on multiple specimens submitted from four patients. The other three of nine prozone-like results occurred on only one of the multiple specimens from three patients. The antigen titers on the specimens with prozone-like reactions ranged from 1:1,24 to 1:32,768 before pronase treatment. The titer of the prozone-like reaction ranged from undiluted to 1:256. There was no correlation between the antigen titer and the presence or titer of the prozone. On seven of nine specimens, the agglutination reactions on the dilutions in the prozone were not clearly negative; a borderline (1+) reactivity was observed. On the other two of nine specimens, each from different patients, a clear negative

4 336 HAMILTON ET AL. J. CLIN. MICROBIOL. TABLE 1. Prozone-like reactions on nine serum specimens tested with the Meridian and IBL kits before and after pronase treatment Meridian kit IBL kit Patient Specimens' Titer before Titer after Titer before Titer after Prozoneb Antigen Prozone Antigen Prozone Antigen Prozone Antigen 1 1/ ,48 No 2,48 No 1,24 No 4,96 2 1/1 1 8,192 No 2,48 No 4,96 No 4,96 3 2/ ,192 No 8,192 No 8,192 No 4, ,192 No 4,96 No 4,96 No 4,96 4 2/3 32 8,192 No 16,384 No 8,192 No 16, ,24 No 1,24 NTC NT NT NT 5 1/ ,48 No 4,96 No 8,192 No 4,96 6 1/5 64 4,96 No 8,192 NT NT NT NT 7 1/5 8 32,768 No 16,384 No 32,768 No 32,768 "Number of specimens with prozone reaction/total number of patient specimens tested. b The endpoint dilution of the prozone-like reaction in which no agglutination was observed. NT, Specimens were not tested. agglutination reaction was observed only on the undiluted specimen; the other dilutions in the prozone showed 1+ reactivity. Pronase treatment eliminated the prozone-like reaction on all dilutions, produced a 3+ to 4+ reactive agglutination to the endpoint dilution, but did not increase antigen titer. Effect of antifungal agents on antigen testing. Antifungal agents and sera from a patient treated with itraconazole were mixed with an antigen-positive specimen and tested with the Meridian kit in order to determine whether these agents caused a direct reduction in antigen titer. Equal portions of an antigen-positive (1:1,24) serum specimen were mixed with (i) serum from a patient with Coccidioides immitis infection treated with itraconazole, (ii) normal human serum supplemented with itraconazole (1,ug/ml), and (iii) normal human serum supplemented with amphotericin B (2,ug/ml) and flucytosine (1 jig/ml). Specimens were tested before and after treatment with pronase. Titers before and after pronase treatment in all test conditions equaled the beginning titer after the dilution factor was calculated. Serum specimens tested with the IBL and Meridian kits. Forty-one serum specimens from 29 patients were tested with both kits. The serum specimens that represented four groups of test results obtained with the Meridian kit before and after pronase treatment were selected. These specimens were tested with the IBL kit before and after pronase treatment. Overall titers produced with the two kits were the same or differed by only 1 dilution on 22 (54%) and 28 (68%) of serum specimens before and after pronase treatment, respectively. On serum specimens in which the two kits produced discrepant titer results, the IBL kit produced higher titers than the Meridian kit did by 2 to >5 dilutions on 18 of 19 specimens before pronase treatment and by 2 to 4 dilutions on 11 of 13 specimens after pronase treatment. A group of eight serum specimens (from eight patients) that were negative before pronase treatment but positive after pronase treatment (tested with the Meridian kit) were selected; titers ranged from 1:8 to 1:1,24. The IBL kit produced titers that ranged from negative to 1:4,96 before pronase treatment and from negative to 1:16,384 after pronase treatment. The differences in antigen titer between the two kits are shown in Fig. 2A. The titer results from the two kits were in agreement on 2 (25%) of the specimens before pronase treatment and on 5 (63%) of the specimens after pronase treatment. Before pronase treatment, two of eight specimens were also negative when tested with the IBL kit; on the other six specimens, the IBL kit had positive titers at c eni a ) 2} j E i-t ~ -D 1 E z 5 4; B r II >5 Dilutions FIG. 2. Cryptococcal antigen titer differences (lower [-], higher [+]) between the IBL and Meridian kits tested before (U) and after (O) pronase treatment on specimens selected on the basis of test results obtained with the Meridian kit. (A) Eight serum specimens that were negative before pronase treatment and positive after pronase treatment; (B) 9 serum specimens that had titers that were.2 dilutions higher after pronase treatment; (C) 17 serum specimens that had the same titers before and after pronase treatment; (D) 14 randomly selected CSF specimens. n I

5 VOL. 29, 1991 PRONASE AND CRYPTOCOCCAL ANTIGEN TESTING 337 2, 4, and >5 dilutions on 1, 2, and 3 specimens, respectively. After pronase treatment, seven of eight specimens were positive with the IBL kit; however only five of seven specimens had equal titers with both kits. On two of seven specimens, the IBL kit produced higher titers than the Meridian kit did by 2 and 4 dilutions. On the one of eight specimens that was negative before and after pronase treatment and tested with the IBL kit, the Meridian kit produced a titer that was 5 dilutions higher after pronase treatment than that before pronase treatment. A group of nine serum specimens (from nine patients) that produced increased titers by -2 dilutions after pronase treatment (tested with the Meridian kit) were selected. Titers ranged from 1:1 to 1:64 before pronase treatment and from 1:32 to 1:2,48 after pronase treatment. The IBL kit produced titers that ranged from 1:4 to 1:2,48 before pronase treatment and from 1:64 to 1:4,96 after pronase treatment. Five of the nine specimens tested with the IBL kit had increased titers as a result of pronase treatment. The differences in antigen titer between the two kits are shown in Fig. 2B. The titer results from the two kits were in agreement on two (22%) of the specimens before pronase treatment and on three (33%) of the specimens after pronase treatment. Before pronase treatment seven of nine specimens had higher titers with the IBL kit than they did with the Meridian kit by 3, 4, and 5 dilutions on three, one, and three specimens, respectively. After pronase treatment the IBL kit produced higher titers compared with those produced by the Meridian kit on six of nine specimens; five of the six titer increases were by 2 dilutions; the other was a 3-dilution increase. A group of 17 serum specimens (from 12 patients) that produced the same titer before and after pronase treatment (tested with the Meridian kit) were selected; titers ranged from 1:16 to 1:16,384. The IBL kit produced equal titers before and after pronase treatment on 15 of 17 specimens; titers ranged from 1:64 to 1:65,536 before and after pronase treatment. The differences in antigen titer between both kits are shown in Fig. 2C. The titer results from the two kits were in agreement on 12 (71%) of the specimens before pronase treatment and on 13 (77%) of the specimens after pronase treatment. Before pronase treatment, titers were higher with the IBL kit than they were with the Meridian kit on 4 of 17 specimens by 2 dilutions, but were lower than those obtained with the Meridian kit on 1 of 17 specimens by 3 dilutions. After pronase treatment titers were higher with the IBL kit than they were with the Meridian kit on 3 of 17 specimens by 2 to 3 dilutions but were lower on 1 specimen by 3 dilutions. A group of seven serum specimens (from six patients) that produced a prozone-like reaction before pronase treatment (tested with the Meridian kit) were selected. Titers ranged from 1:2,48 to 1:32,768 with the Meridian kit and did not increase after pronase treatment. Pronase treatment eliminated the prozone-like reaction on all dilutions. The IBL kit produced equal titers before and after pronase treatment on six of seven specimens; titers ranged from 1:1,24 to 1:32,768 before pronase treatment and from 1:4,96 to 1:32,768 after pronase treatment, as shown in Table 1. The titer results from the two kits were in agreement on six (86%) of the specimens before pronase treatment and on seven (1%) of the specimens after pronase treatment. Antigen titers between the two kits agreed on six of seven specimens before pronase treatment and on seven of seven specimens after pronase treatment. In contrast to the Meridian kit, no prozone-like reactions were observed with the IBL kit. On all specimens that had equal titers before and after pronase treatment (tested with the Meridian kit) (the preceding two groups), results from the two kits agreed on 18 of 24 (75%) specimens before pronase treatment and 2 of 24 (83%) specimens after pronase treatment. CSF specimens tested with the IBL and Meridian kits. Fourteen CSF specimens from 13 patients were tested with both kits before and after pronase treatment. Titers obtained with the Meridian and IBL kits ranged from 1:8 to 1:4,96 and 1:16 to 1:16,384 before pronase treatment and from 1:8 to 1:8,192 and 1:16 to 1:16,384 after pronase treatment, respectively. The differences in titers obtained with the two kits are shown in Fig. 2D. Antigen titers obtained with the two kits were the same on 12 (86%) specimens before pronase treatment and on 13 (93%) specimens after pronase treatment. Before pronase treatment the Meridian kit produced higher titers than the IBL kit did on 2 of 14 specimens by 2 dilutions. After pronase treatment, the IBL kit produced higher titers than the Meridian kit did on 1 of 14 specimens by 3 dilutions. On this specimen, both kits produced equal titers (1:8,192) before pronase treatment, but after pronase treatment the Meridian and IBL kits produced titers of 1:2,48 and 1:16,384, respectively. DISCUSSION This is the first extensive study of cryptococcal antigen testing in the AIDS era. It may have uncovered a diagnostic problem that is unique to these patients and their disease process, as demonstrated with the Meridian and the IBL kits. Pronase treatment of serum specimens enhanced the detection and titration of cryptococcal antigen by uncovering and resolving false-negative results, eliminating potentially false-negative results from a prozone-like reaction, and eliminating the variable endpoint determinations for titration of antigen. No identifiable factors were found to explain the effects of pronase treatment. The Meridian and IBL kits were shown to produce different results even after pronase treatment of specimens. Pronase treatment of specimens, as originally described, was used to eliminate false-positive reactions caused by rheumatoid and other factors. Only one study (9) has shown the effects of pronase treatment of all specimens on antigen detection and titration. The IBL kit was used to test 57 serum specimens and 7 CSF specimens before and after pronase treatment. More of the serum specimens (81%) in that study had increased titers (.2 dilutions) after pronase treatment compared with those (57%) in the present study tested with the Meridian kit. It is interesting in both studies that the range of titers before pronase treatment were the same. On CSF specimens tested after pronase treatment in the previous study (9), titers were increased on 2% of the specimens, none had decreased titers, but one specimen that was negative before pronase treatment was positive. These findings are in contrast to the results of the present study, in which titers decreased after pronase treatment on 12% of the specimens tested with the Meridian kit. The greater effect of pronase treatment shown in the previous study may reflect variables with individual patients and, perhaps, not differences in the two kits. It is interesting that in the present and previous studies, pronase treatment of CSF changed titers regardless of the degree of titer before pronase treatment. There is no explanation for the opposite effect of pronase treatment on CSF specimens tested with the Meridian kit in the present study. Results from the previous study (9) recommend the use of pronase treatment of CSF specimens. Results from this study show that on CSF specimens tested

6 338 HAMILTON ET AL. with the Meridian kit, routine pronase treatment is not needed. In the previous study (IBL kit) (9), in which all specimens were treated with pronase, specimens that were negative before pronase treatment but positive after pronase treatment were found on 19% of serum specimens and 1.4% of CSF specimens. Very similar results were obtained in the present study on serum specimens tested with the Meridian kit; false-negative results were uncovered with the use of pronase treatment on 19% of serum specimens but not on CSF specimens. These results demonstrate that routine treatment of serum specimens with pronase is essential in order to eliminate false-negative results, regardless of the kit used. Rarely, other studies have reported false-negative cryptococcal antigen test results. In two studies (14, 16) the false-negative result was resolved by dilution of the specimen. The specimens (one serum specimen and one CSF specimen) were positive to titers of.1:1,24 when tested after dilution. This prozone-like phenomenon that occurred in the serum specimen was observed with three different kits (16). In the other study (14), all CSF specimens from a single patient tested with commercially unavailable reagents produced the prozone-like phenomenon. The titers of the prozone-like reactions in those two studies were not determined. In the present study the prozone-like reactions occurred on 9% of the serum specimens, a higher rate than that reported previously. It is interesting in the present study that no CSF specimens produced a prozone-like reaction; however, all serum specimens with a prozone-like reaction were from patients who had cryptococcal meningitis. In addition, prozone-like reactions did not occur on all specimens from the same patient, as reported in the one study (14), but could occur randomly among multiple specimens. A new finding in the present study was that pronase treatment eliminated the prozone-like reactions on all specimens. A practice of screening a diluted and undiluted sample of serum in order to eliminate the potential prozone-like reactions and detect the presence of cryptococcal antigen is questionable. In the present study, 9% of the serum specimens tested with the Meridian kit would have required dilution in order to test positive. The optimal dilution would be 1:512 because all of the prozone-like reactions were l1:256 and the antigen titers were.1:1,24. It is interesting that results from other studies of prozone-like reactions support this suggestion (14, 16); however, it is unknown whether all potential combinations of antigen and prozonelike titers could be detected with this dilution. The data from the present study suggest that the treatment of specimens with pronase would be a more reliable method to eliminate false-negative reactions caused by prozone-like phenomenon. The cause for the false-negative results revealed by testing the specimens after dilution or after pronase treatment is not known. The presence of antifungal agents, a resulting metabolic product from these agents, or repeat testing after freezing and thawing did not account for these findings. Antigen excess is probably not an explanation of the prozone-like reaction since the prozone-like reaction occurred over a broad antigen titer range, and in a previous study (14) a prozone was not created with the addition of whole organisms. The fact that pronase treatment eliminated the prozone-like reaction with the Meridian kit and the falsenegative results obtained with both kits suggests the possibility that bound antibody or a nonspecific protein masks the antigen. In addition, the fact that the different kits produced variable results on the same specimens suggests the possibility that different antigenic components are recognized by J. CLIN. MICROBIOL. the different kits. These findings may represent a more universal diagnostic problem with cryptococcal antigen testing regardless of the kit used. Different cryptococcal antigen kits have been compared in only one other study (16). The IBL kit was compared with the Myco-immune kit (American Scientific Products) and the IMMY kit (Immuno-mycologics, Norman, Okla.). One kit produced negative results on 55% of serum specimens and 17% of CSF specimens that were positive when they were tested with the IBL kit. Titers with the IBL kit on these falsely negative specimens ranged from 1:2 to 1:16 on CSF specimens and 1:16 to 1:128 on serum specimens. In serum, titers were as much as 8- to 16-fold higher than those produced with either of the other kits, and in CSF titers were as much as 8-fold higher than those produced with one of the other kits. These results are consistent with the results from the present study. On specimens that were not treated with pronase, the IBL kit produced titers that ranged from 1:2 to 1:4,96 on six of eight serum specimens that were negative when they were tested with the Meridian kit. The IBL kit generally produced higher titers than the Meridian kit did on serum specimens in which discrepant titer results were obtained between the two kits. Titers agreed on only 54% of 41 serum specimens and on 86% of 14 CSF specimens before pronase treatment. Even after pronase treatment, agreement between the kits increased only to 68% on serum specimens and 93% on CSF specimens. The best titer correlation of 83% between the two kits was observed with 24 serum specimens that on previous testing showed no titer change after pronase treatment when tested with the Meridian kit. On specimens that were falsely negative, titers after pronase treatment obtained with the two kits were similar, except for one specimen. That specimen was tested with the IBL kit and was negative before and after pronase treatment. It appears from the results of these studies that the IBL kit more characteristically produces higher titers compared with those produced by other kits, especially when specimens are not treated with pronase before testing. There are uncontrollable and unpredictable differences in the results between the IBL and Meridian kits, even with the use of pronase, to the degree that the different kits cannot be used interchangeably. This is especially important in comparing serial changes in titer, as in following the response to treatment. Comparison of results from one laboratory to another is also problematic when one considers the variables in kit sensitivity, the differences in patient populations and disease conditions that may exist, and the use of pronase or other reagents, as well as the differences produced by subjective interpretation of some agglutination reactions. Agglutination was difficult to interpret without pronase treatment with both kits because of the minimal degree of agglutination that occurred on many specimens. In a previous study (11), the 1+ reactive serum and CSF specimens were shown to correlate only 86 and 33%, respectively, with cryptococcal infection. The results of the present study on serum specimens parallel those findings somewhat, in which the 1+ reactions were indicators of a positive reaction that markedly changed to 3+ to 4+ reactivity after treatment of the specimen with pronase. It was necessary to differentiate between a 1+ and a 2+ reactivity with the Meridian kit. This was difficult to do consistently, and as a result it led to variable endpoint interpretations in which the 1+ reactivity could be interpreted as positive or negative and, at times, the 2+ reactivity could be interpreted as negative. A similar problem occurred with the IBL kit in determining whether a reaction was

7 VOL. 29, 1991 PRONASE AND CRYPTOCOCCAL ANTIGEN TESTING 339 discernible, since any agglutination was interpreted as positive. These problems in endpoint interpretation may have accounted for some of the higher titers with the IBL kit in the present study. For some of the specimens in which the IBL kit produced titers that were 2 or 3 dilutions higher than those produced by the Meridian kit before pronase treatment, the Meridian kit produced 1+ reactivity to the endpoint of the IBL kit reaction. We also found that results from a titration could vary considerably, especially with a weakly reactive specimen, in which it was tempting to continue to manipulate the glass plate to "better" resolve the questionable reaction. The longer the reaction was observed with manual tilting and rotating, the more difficult it became to resolve an endpoint. The titer could increase tremendously during this interval. As an example of this phenomenon, a specimen of weak reactivity (1+ to 2+) was given to several technologists for titration; titers ranged from 1:1,24 to 1:65,536. Pronase treatment of the specimen eliminated the discrepant readings and produced titers that were the same or that varied by only 1 dilution. It is suspected that one serum specimen with a reported titer of 1:1,, from another laboratory referred to our laboratory for testing was a result of this discrepant reading phenomenon. Our result on that specimen before pronase treatment was difficult to determine because of the weak reactivity (1+ to 2+), and with continued manipulation of the agglutination glass plate, a titer of >1:262,144 was produced. After pronase treatment, a titer of 1:16,384 was produced with a very clear endpoint reading of 3+ to 4+ reactivity. We recommend pronase treatment of serum specimens but not CSF specimens for testing with the Meridian kit. We do not recommend that kits from different manufacturers be used interchangeably for detection or titration of antigen. With the use of pronase treatment of serum specimens, consistent results are obtained by eliminating the subjective interpretations of the agglutination reactions, and the detection of antigen is enhanced by eliminating false-negative results and prozones. These findings have important implications for the diagnosis of cryptococcal meningitis and for monitoring patient responses to therapy. REFERENCES 1. Denning, D. W., R. M. Tucker, L. H. Hanson, J. R. Hamilton, and D. A. Stevens Itraconazole therapy for cryptococcal meningitis and cryptococcosis. Arch. Intern. Med. 149: Diamond, R. D., and J. E. Bennett Prognostic factors in cryptococcal meningitis, a study in 111 cases. Ann. Intern. Med. 8: Eng, R. H. K., E. Bishburg, S. M. Smith, and R. Kapila Cryptococcal infections in patients with acquired immune deficiency syndrome. Am. J. Med. 81: Eng, R. H. K., and A. Person Serum cryptococcal antigen determination in the presence of rheumatoid factor. J. Clin. Microbiol. 14: Fisher, B. D., and D. Armstrong Cryptococcal interstitial pneumonia, value of antigen determination. N. Engl. J. Med. 297: Goodman, J. S., L. Kaufman, and M. G. Koenig Diagnosis of cryptococcal meningitis, value of immunologic detection of cryptococcal antigen. N. Engl. J. Med. 285: Gordon, M. A., and E. W. Lapa Elimination of rheumatoid factor in the latex test for cryptococcosis. Am. J. Clin. Pathol. 61: Gordon, M. A., and D. K. Vedder Serologic tests in diagnosis and prognosis of cryptococcosis. J. Am. Med. Assoc. 197: Gray, L. D., and G. D. Roberts Experience with the use of pronase to eliminate interference factors in the latex agglutination test for cryptococcal antigen. J. Clin. Microbiol. 26: Hay, R. J., and D. W. R. Mackenzie False positive latex tests for cryptococcal antigen in cerebrospinal fluid. J. Clin. Pathol. 35: Kaufman, L., and S. Blumer Cryptococcosis: the awakening giant. Proceedings of the Fourth International Conference on the Mycosis, PAHO, publication 356. Center for Disease Control, Atlanta. 12. Prevost, E., and R. Newell Commercial cryptococcal latex kit: clinical evaluation in a medical center hospital. J. Clin. Microbiol. 8: Snow, R. M., and W. E. Dismukes Cryptococcal meningitis, diagnostic value of cryptococcal antigen in cerebrospinal fluid. Arch. Intern. Med. 135: Stamm, A. M., and S. S. Polt False-negative cryptococcal antigen test. J. Am. Med. Assoc. 244: Stockman, L., and G. L. Roberts Specificity of the latex test for cryptococcal antigen: a rapid, simple method for eliminating interference factors. J. Clin. Microbiol. 17: Wu, T. C., and S. Y. Koo Comparison of three cryptococcal latex kits for detection of cryptococcal antigen. J. Clin. Microbiol. 18: