The Journal of Infectious Diseases 1996; 174: by The University of Chicago. All rights reserved /96/ $01.

Transcription

1 812 Molecular Subtypes and Antifungal Susceptibilities of Serial Cryptococcus neoformans Isolates in Human Immunodeficiency Virus-Associated Cryptococcosis Mary E. Brandt, Michael A. Pfaller, Rana A. Hajjeh, Edward A. Graviss, Judy Rees, Eric D. Spitzer, Robert W. Pinner, Leonard W. Mayer, and the Cryptococcal Disease Active Surveillance Group* Emerging Bacterial and Mycotic Diseases Branch, Division of Bacterial and Mycotic Diseases, and Office ofthe Director, National Center for Infectious Diseases, Centers for Disease Control and Prevention, Atlanta, Georgia; Department ofpathology, University ofiowa College of Medicine, Iowa City, Iowa; Infectious Diseases Section, VA Medical Center, and Department ofmedicine, Baylor College ofmedicine, Houston, Texas; School ofpublic Health. University of California. Berkeley; Department of Pathology. State University of New York at Stony Brook Serial isolates of Cryptococcus neoformans from 33 human immunodeficiency virus-infected patients with cryptococcosis were analyzed to determine whether persistence might result from reinfection with a new cryptococcal strain or acquisition of antifungal resistance. Isolates were subtyped by multilocus enzyme electrophoresis (MEE), electrophoretic karyotyping (EK), randomamplified polymorphic DNA (RAPD), and the CNRE-l DNA probe. MICs of amphotericin B, fluconazole, and 5-fluorocytosine were determined. No changes in MEE or RAPD subtypes were detected in serial isolates from any patient. Isolates from 8 patients (24%) showed alterations in EK only (mobility change in two or more bands) but not with any other subtyping method. MICs did not change significantly in isolates from 30 patients. In 1 case, the fluconazole MIC increased stepwise over 18 months, suggesting development of resistance. These overall invariant subtyping and MIC results confirm previous studies suggesting that persistent cryptococcal infection is due to relapse rather than reinfection or antifungal drug resistance. Cryptococcal meningitis, caused by the encapsulated yeast Cryptococcus neoformans, is one of the most life-threatening mycotic infections occurring in persons infected with human immunodeficiency virus (HIV) [1-3]. After initial treatment of cryptococcosis, lifelong antifungal maintenance therapy is recommended to prevent recurrence [4, 5]. The epidemiology of recurrent cryptococcosis remains poorly understood. Although most episodes ofrecurrent disease are believed to be due to persistence of the original infecting strain, the proportion that may be caused by reinfection with a new fungal strain is unclear. Better understanding of this issue could affect the design of prevention strategies for this disease. Ifreinfection is a likely event, identifying and avoiding environmental exposures or behaviors that increase the risk of reinfection could decrease the frequency of disease. On the other hand, ifpersistence ofthe original infecting strain causes Received 5 March 1996; revised 3 June Presented in part: 35th Intersciencc Conference on Antimicrobial Agents and Chemotherapy, San Francisco, September 1995 (paper En). Grant support: NIH (AI to E.D.S.). Reprints or correspondence: Dr. Mary E. Brandt, Emerging Bacterial and Mycotic Diseases Branch, Centers for Disease Control and Prevention, 1600 Clifton Rd., Mailstop D-II, Atlanta, GA * Study group members are listed after the text. The Journal of Infectious Diseases 1996; 174: by The University of Chicago. All rights reserved /96/ $01.00 recurrent disease, as has been suggested [6-9], then strategies to improve therapy and management ofthe initial fungal infection should be emphasized. Molecular subtyping is potentially useful in resolving this issue, by demonstrating the degrees of relatedness of initial and relapse isolates from the same patient. Several previous studies have investigated this question. Spitzer et a1. [10] used electrophoretic karyotyping (EK) and restriction fragment length polymorphism analysis with 2 genomic DNA probes to analyze serial isolates from 4 patients with recurrent cryptococcosis. These authors suggested that recurrences resulted from persistence of the initial infecting strain. Varma et a1. [11] obtained similar results with 53 serial isolates from 9 non AIDS patients analyzed using the DNA probe UT-4p. They found that for 8 patients, the restriction fragment length polymorphism pattern was stable over a period of weeks of antifungal therapy. In the remaining patient, isolates with two different patterns were recovered at different intervals during treatment. Barchiesi et a1. [12] analyzed sequential isolates from 3 patients using EK. In 2 patients, the same karyotype persisted in multiple isolates collected during a I-month period. In the third patient, two karyotypes were found in 3 isolates collected during a I-month period. In a contrasting study, Haynes et a1. [13] identified dissimilar DNA patterns in each of 3 pairs of patient isolates. Another important question in the epidemiology of recurrent cryptococcal disease is whether acquisition ofantifungal resistance in C. neoformans might contribute to recurrence. Pro-

2 JID 1996; 174 (October) Subtypes of Serial C. neoformans Isolates 813 longed use offluconazole for treatment or prophylaxis ofmucocutaneous candidiasis has led to replacement of f1uconazolesensitive Candida albicans strains with non-c albicans yeasts resistant to this drug [14]. Whether antifungal resistance can develop in C neoformans through the administration ofchronic antifungal maintenance therapy is not clear. This possibility was investigated by Casadevall et a1. [15], who determined in vitro susceptibilities to amphotericin Band fluconazole in 13 isolates from 5 patients with recurrent cryptococcal meningitis. For each patient, isolates from episodes of recurrence showed no increase in antibiotic resistance compared with the initial isolate. These researchers concluded that recurrent disease may have been associated with poor compliance or deficits in immune function. A later study demonstrated similar results for serial isolates from 3 patients with cryptococcal disease in Italy [12]. On the other hand, emergence of clinically significant resistance to amphotericin B [16] or fluconazole [17] has been reported in individual cases ofrecurrent cryptococcosis. In this report, we evaluated subtype variation and changes in antifungal susceptibilities in 102 serially collected isolates from 33 patients obtained through an active, population-based laboratory surveillance for cryptococcal disease in three US cities [18, 19]. Our findings confirm those of previous smaller studies [10-12, 15] that cryptococcal disease in HIV-infected persons is incurable because antifungal therapy fails to eradicate the infection and that failure to eradicate the infection is apparently not due to antifungal drug resistance. Materials and Methods Study design and demographics. C. neoformans clinical isolates were obtained through the Centers for Disease Control and Prevention (CDC) Fungal Active Surveillance from three US metropolitan areas (San Francisco and Atlanta in ; Houston in ). Medical records of all patients from whom serial isolates were collected were reviewed. Information was collected on clinical characteristics ofthe initial and subsequent episodes of cryptococcal disease, other fungal and opportunistic infections, cryptococcal antigen, CD4 cell count, antifungal therapy, and compliance with maintenance therapy. Patients were considered to be noncompliant with therapy if judged so by their health care providers as documented in medical records. Organisms and growth conditions. On receipt at CDC, isolates were subcultured to Sabouraud dextrose agar slants and frozen in 20% glycerol at -20 C. Frozen isolates were thawed when needed and grown on Sabouraud dextrose agar at 37 C. Identification was confirmed by standard methods [18, 20]. Molecular subtyping. Multilocus enzyme electrophoresis (MEE) with 10 enzymes was done as described [18, 21]. Randomamplified polymorphic DNA (RAPD) subtyping was done with five decamer oligonucleotide primers (OPA 2, OPA 3, OPA 19, OPAO 9, OPAO 15; Operon Technologies, Alameda, CA) as described [18]. EK analysis was done by using a CHEF (contourclamped homogeneous electric field) DRII system (BioRad Laboratories, Richmond, CA). Chromosomal DNA preparations were made by using a modification of existing protocols [12, 22-24]. Isolates were considered identical (same DNA type) if all bands matched exactly. Banding patterns differing by a single band were considered similar and were identified as subtypes (shown as subscripts to EK type in table 2). Isolates with differences of two or more bands were considered different. Southern blotting was done on selected isolates with the CNRE 1 repetitive element DNA probe [10] as described by Spitzer and Spitzer [25], except that the probe was labeled with digoxigenin (Genius 1 kit; Boehringer Mannheim, Indianapolis). Prehybridization, washing, and development were done according to the manufacturer's instructions (Genius 1 kit; Boehringer Mannheim). Antifungal susceptibility testing. Broth microdilution testing was done according to National Committee for Clinical Laboratory Standards proposed standard guidelines with the spectrophotometric method of inoculum preparation, an inoculum concentration of 0.5 X 10 3 to 2.5 X 10 3 cells/ml, and RPMI 1640 medium buffered to ph 7.0 with M morpholinepropanesulfonic acid buffer (Sigma, St. Louis) [26,27]. Antifungal agents included amphotericin B, 5-f1uorocytosine, and fluconazole. The MIC end points were read visually and were defined for 5-fluorocytosine and fluconazole as the lowest concentration in which a prominent decrease in turbidity (MIC-2) was observed and for amphotericin B as the lowest concentration producing complete inhibition of growth [26, 27]. Statistical methods. MEE and RAPD gels were analyzed as described [18, 19, 21]. For MEE, each unique combination of electrophoretic variants was designated as a multilocus genotype [21]. For RAPD, results with each of the five RAPD primers were combined and expressed as a RAPD type assigned to each isolate [18, 19]. Statistical comparisons were tested at the significance level of.05 by Fisher's exact test. Results Patient and Clinical Characteristics Serial isolates were available from 33 patients with persistent cryptococcosis during the 3-year study. Relevant clinical data, extracted from retrospective chart review, are summarized in table 1. Fourteen patients resided in metropolitan Atlanta (8 counties), 17 in metropolitan San Francisco (3 counties), and 1 in Houston (Harris County). One patient from Georgia resided in a county outside the surveillance area. All patients were men, with ages ranging from 20 to 54 years (median, 33). All patients were infected with HIV. On initial presentation, 27 patients were diagnosed with cryptococcal meningitis, and 5 patients were diagnosed with cryptococcal pneumonia and meningitis. One patient (RC26) had cryptococcal pneumonia and fungemia, but cerebrospinal fluid (CSF) was negative for cryptococcal antigen. All patients had normal mental status except 2 who were slightly confused on initial presentation, 1 who presented with seizures, and 1 who had seizures during hospitalization. Nine patients died during the surveillance period, 8 presumed due to cryptococcal disease.

3 814 Brandt et al. JID 1996; 174 (October) Table 1. Selected demographic and clinical features of patients at initial and subsequent episodes of cryptococcosis (n = 33). Characteristic No. % Male sex Black race Initial episode LA test*?c512 (n = 29) Second episode LA test*?c512 (n = 24) Received amphotericin B as initial therapy Died after second episode 7 21 Median Range Age (years) CD4 cell count (IJ1L) Interval (months) between first and second clinical episodes (n=31 t ) 4 I-IS Interval (months) between second and third clinical episodes (n = 17) 4 I-II Total amphotericin B dose of initial therapy (mg) * LA, latex agglutination was done on serum or cerebrospinal fluid. Result is reciprocal of titer. t For 2 patients, date of first clinical episode was not known. Antifungal therapy is outlined in table 1. Twenty-four patients received amphotericin B as initial treatment; 14 of these received < 500 mg initially (l0- to 14-day course). Most patients (26/33) were receiving an azole drug as maintenance therapy before the second cryptococcal episode; 24 were receiving fluconazole and 2 received itraconazole. Three patients were receiving amphotericin B as outpatient maintenance therapy. Based on review of medical records, 25 patients were considered noncompliant with outpatient therapy; 6 patients were considered compliant. Compliance could not be assessed for the remaining 2 patients. One hundred two isolates from these patients (2-6 isolates from each subject) were collected during the study period (table 2). In 7 cases, isolate(s) from the incident hospitalization were not available because the initial diagnosis occurred before this study. All isolates were from either blood or CSF, except in 1 case in which cryptococci were cultured from a lung biopsy and 2 urine specimens (table 2). The intervals between consecutive isolates ranged from 1 to 14 months; the total interval from first to last isolate ranged from 2 to 21 months. For 18 patients (55%), the interval between the first two culture-positive episodes was at least 4 months. All isolates were identified as Cryptococcus neoformans var. neoformans. Molecular Subtyping All isolates were subtyped by MEE, RAPD, and EK (table 2). All samples generated clear, readable subtyping patterns with all three methods. ME. All isolates from a given patient displayed the same multilocus genotype (table 2). Furthermore, the genotype distribution in this population resembled that ofan unselected population ofclinical isolates from these same geographic locations [19]. Isolates from 22 patients were enzyme type (ET) 1, the predominant genotype found in a recent surveillance for cryptococcosis in the same geographic regions (67% of cases) [19]. Isolates from the other 11 patients were ETs that were present in low prevalence (2% - 10% ofcases) in the unselected population ofclinical isolates [19]. These rare subtypes included ETs 2 and 8 (2 cases each) and 3, 4, 5, 7, 12, 20, and 25 (l case each). RAPD. RAPD was done with five primers, and the results were combined to generate a RAPD type for each patient. No variation was seen within the RAPD results for isolates from a single subject (figure 1, table 2). The distribution of RAPD types in this persistent cryptococcosis group displayed some differences compared with that of an unselected population from the same geographic areas [19], although these differences were not statistically significant by Fisher's exact test. The proportional recovery ofrapd-5 and RAPD-I1 was increased in this patient group (13% and 14%, respectively) compared with recovery from an unselected population (3% and 2%, respectively). Recovery ofrapd-1 was decreased in this group (4.5%) compared to its proportion in an unselected population (20%). The proportions of RAPD-6 (41% in this group vs. 25%), RAPD-12 (4.5% vs. 2%), and RAPD-14 (23% vs. 11%) were similar in the 2 populations [19]. EK analysis. In nearly all cases, isolates from a given patient could be distinguished from those ofevery other patient by karyotype analysis (figure 2, table 2). In only 1 case did the karyotype of 3 isolates from 1 patient (RC30) match that ofthe initial isolate from another patient (RC23). For 18 (55%) of the 33 cases, all isolates from the same patient showed identical karyotypes. With isolates from another 7 patients (21%), a mobility difference in one band only was identified in the respective sequential isolates. In 2 of these cases (RC2 and RC17), this difference was reversible; the karyotype of subsequent isolates reverted to that ofthe initial isolate. Isolates from 8 additional patients (24%) displayed altered mobility in at least two bands. All isolates from these 8 patients were further analyzed by Southern blotting with the CNRE-1 cryptococcal dispersed repetitive element probe [10, 25]. No differences were seen in SstI restriction fragment po1ymorphisms within isolates from the same patient (figure 3). In 3 of these 8 patients, phenotype differences could be associated with changes in karyotype. One patient (RC24), in whom an increasing 5-fluorocytosine MIC was also demonstrated, is described below. In another (RC16), 3 fungal isolates

4 JID 1996; 174 (October) Subtypes of Serial C. neoformans Isolates 815 Table 2. Molecular subtypes and antifungal susceptibility of C. neoformans from patients with persistent cryptococcosis. MIC (jlg/ml) Collection Patient Site date ET RT EK Amphotericin B Fluconazole 5-fluorocytosine RCI Blood 7/20/92 I RI A CSF 3/30/93 1 Rl B RC2 Blood 1/26/93 1 R6 C CSF 6/15/93 1 R6 C I CSF 11/10/93 1 R6 C RC3 CSF 2/6/93 I Rll D CSF 3/1/93 1 Rll D CSF 9/3/93 1 Rll D RC4 Blood 7/13/93 1 Rl1 E CSF 7/15/93 1 Rll E Blood 10/02/93 1 Rll E CSF 10/4/93 1 Rll E RC5 Blood 3/25/ F Blood 7/10/ F CSF 7/11/ F I RC6 CSF 4/1/92 1 R14 G Blood 8/27/92 1 R14 G Blood 1/13/93 1 RI4 G RC7 CSF 5/1/ H CSF 7/3/ H CSF 11/30/ H CSF 12/20/ H CSF 1/4/ H RC8 CSF 517/ I Blood 1017/ I RC9 CSF 6/17/92 1 R6 J CSF 7/3/92 1 R6 J CSF 9/29/93 1 R6 K CSF 1/3/94 I R6 K I CSF 2/20/94 1 R6 K RCI0 Blood 7/24/92 1 R6 L CSF 8/17/92 1 R6 L Blood 9/11/92 I R6 L CSF 9/11/92 1 R6 L RCll CSF 10/10/92 1 R12 M CSF 6/11/93 1 R12 M I CSF 7/13/93 1 R12 M RC12 Blood 7/25/93 1 R5 N Blood 12/04/93 1 R5 N RCB CSF 7/28/93 1 Rll Blood 12/18/93 1 Rll CSF 2/20/94 1 Rll RC14 Blood 9/29/ P CSF 12/02/ P RCI5 CSF 11/05/ Q CSF 1/31/ Q RCI6 CSF 3/19/92 1 R5 NN CSF 11/22/92 (1) 1 R5 R CSF 11/22/92 (2) 1 R5 S RC17 CSF 3/13/92 1 R6 T CSF 7/20/92 1 R6 T CSF 12/28/92 1 R6 T I CSF 9/15/93 I R6 T RC18 CSF 5/22/92 1 R14 V CSF 2/11/93 I R14 W

5 816 Brandt et al. Jll) 1996; 174 (October) Table 2. (Continued) MIC (pg/ml) Collection Patient Site date ET RT EK Amphotericin B Fluconazole 5-fluorocytosine RCI9 CSF 5119/ X Blood 7/13/ X Blood 7/29/ XI RC20 CSF 7119/ y CSF 8/18/ y CSF 11124/ y CSF 2/25/ y CSF 3/26/ y CSF 7/13/ yy RC21 CSF 7/16/92 I R5 Z Blood 7/17/92 I R5 Z Blood 11/22/92 1 R5 Z RC22 CSF 11/10/ AA CSF 3/17/ AA RC23 Blood 11/13/92 I R4 BB CSF 2/02/93 1 R4 CC CSF 2/12/93 1 R4 CC RC24 CSF 7/09/92 I R6 U CSF 10/04/92 1 R6 DO CSF 11/27/92 I R >256 RC25 CSF 1/22/93 I R6 EE CSF 2/03/93 I R6 EE Blood 5118/93 1 R6 EE RC26 Blood 4/28/93 I R6 EEl Blood 10/06/93 1 R6 EEl RC27 CSF 7/27/93 1 R6 GG Blood 7/27/93 1 R6 GG CSF 9/27/93 1 R6 GG Blood 9/27/93 1 R6 GG Blood 1/20/94 1 R6 GG RC28 CSF 8/14/ HH CSF 1/10/ HH RC29 Urine 1/20/92 I RI4 1I Urine 3/08/92 I RI4 II Lung biopsy 4/21/92 I RI4 JJ RC30 Blood 6/08/92 1 RI4 BB CSF 6/13/92 1 RI4 BB CSF 3/1 7/93 I RI4 BB RC31 CSF 7/18/ PP CSF 5/25/ PP CSF 10/04/ PP RC32 CSF 4/06/ LL CSF 6/29/ LL RC33 CSF 5/02/93 I R6 MM CSF 7/03/93 I R6 MM CSF 1/02/94 I R6 MM I NOTE. ET, enzyme type (multilocus genotype); RT, random-amplified polymorphic DNA (RAPD) type (RAPD types ofet 1 isolates are labeled with prefix "R"); EK, electrophoretic karyotype (subscripts indicate subtypes); CSF, cerebrospinal fluid. Dates are given as month/day/year. CSF sample. In a third case (RC20), 5 isolates received from July 1992 to March 1993 displayed similar colonial morphol- ogy (large colony type) and the same karyotype, but the sixth isolate (July 1993) displayed both a small colony morphology and an altered karyotype. In all cases, karyotypically distinct isolates displayed identical RAPD and CNRE-l profiles. were recovered from 2 CSF samples collected 8 months apart. The second CSF sample generated two morphologically distinct (large and small) colony types, each of which showed a different karyotype. Both karyotypes were distinct from that ofthe initial isolate, although the initial isolate morphologically resembled the small colony type recovered from the second

6 lid 1996; 174 (October) Subtypes of Serial C. neoformans Isolates 817 A. Figure 1. Random-amplified polymorphic DNA subtypes of C. neoformam,' DNA from patients with persistent cryptococcosis, using selected decamer primers. A, Primer OPA 3; B, primer OPAO 15. DNA Ladder (Life Technologies GIBCO BRL, Gaithersburg, MD) was used as molecular size B. standard. Patient no. and isolate collection date are listed above each lane. BP = base pairs. - g RC24 RC16 RCS RC1 RC14 RC28 RC33 I I I I I I I..- "- C\J C\J C\J C\J C\J ~ ~ ~ C\J C\l C\l M CO') CO') ('t) 'l:t C\J 0'> m 0'> 0'> 0'> ~ s ~ ~ 0'> ('t) ~ M 'l:t en f' ~ en m ~ ~ ~..- ~ ~ LO..- 0'> 0 or- or- C\l ('t) C\l or-..- C\] CO') t--..- or-..- 0r- (\') t-- f'-.. 0'> or-..- l() t-- or- ~ -..- M r:::: - M - CO II II II II BP...r Antifungal Drug Susceptibility Amphotericin B, 5-fluorocytosine, and fluconazole MICs were measured (table 2). In 30 cases, no change in MIC over time was noted. With 1 patient (RC24), an increased 5-fluorocytosine MIC was demonstrated in 3 CSF isolates, from 4 j1g/ ml (July 1992) to 8.0 j1g/ml (October 1992) to >256 j1g/ml (November 1992). Interestingly, each of these isolates displayed a different karyotype. This patient had received amphotericin Band 5-fluorocytosine (1500 mg three times daily) for 5 days only during his first (July 1992) and third (October 1992) hospitalizations. He continued to receive both drugs from October 1992 until 2 weeks before his fourth hospitalization in November In a second case (RC 17), a stepwise increase in fluconazole MIC was detected, from 4 j1g/ml on 13 March 1992 to 64 j1g/ ml on 15 September This patient was initially (March 1992) treated with amphotericin B (total 316 mg), then received fluconazole (400 mg daily) for 14 weeks, then 200 mg daily. During his second hospitalization (July 1992), he again received amphotericin B (total 540 mg) and then fluconazole (400 mg daily) on discharge. The same pattern was repeated during his third hospitalization in December With isolates from a third patient (RC22), a modest but significant increase in amphotericin B MIC was noted, from 0.25 j1g/ml (November 1992) to 1.0 j1g/ml (March 1993). Isolates from the incident hospitalization (November 1991) were not available on this patient. He was treated with amphotericin Band 5-fluorocytosine during hospitalization for cryptococcal meningitis in November He continued to receive these drugs as an outpatient until his next hospitalization in March No change in 5-fluorocytosine M1C was noted. In 2 additional cases (RC29 and RC31), the fluconazole MIC never varied, but initial and subsequent MICs were all 64 /.log/ ml. One ofthese patients (RC31) was given itraconazole (400 mg/day) maintenance therapy after his initial hospitalization in July 1992 but relapsed in May 1993 and again in October Discussion For 76% of patients in this study, persistent cryptococcal disease was associated with continued recovery of a fungal strain identical or similar to that causing the initial infection. Serial isolates from the same patient were defined as identical if no differences could be detected within MEE, EK, and RAPD subtype patterns. Strains that displayed variation in the mobility of one DNA band in the EK were considered to be similar. No evidence for strain replacement was found in these cases, as has been documented with Candida species [14]; rather, individual profiles characteristic of a particular MEE, RAPD, or EK subtype remained stable despite intervals between fungal

7 818 Brandt et al. lid 1996; 174 (October) C") L() co I' ex) CJ) Figure 2. Karyotypcs of C. neoformans DNA from patients with persistent cryptococcosis. Patient no. and isolate collection date are listed above each lane. Saccharomyces cerevisiae chromosomal DNA molecular weight markers (FMC Bioproducts, Rockland, ME) were included in each gel as standards (lanes S). isolations ranging from 1 to 14 months. These molecular epidemiologic results are consistent with clinical data suggesting that these patients were never cured of the initial cryptococcal infection. Serial isolates from 8 patients (24%) revealed karyotype differences as alterations in the mobilities of at least two DNA bands in the EK. No isolates from the same patient showed any differences in MEE, RAPD, or CNRE-I profile. These karyotype polymorphisms were not related to the time interval between positive cultures, since we could also identify patients with similar duration between positive cultures in whom karyotype alterations were not found (table 2). Results similar to ours have been reported in an earlier study of recurrent cryptococcosis [10], in which I of4 isolates from a patient displayed karyotype polymorphisms despite identical CNRE-I profiles. We cannot rule out the possibility that some recurrences could have resulted from acquisition of new strains in these 8 patients. However, we are reluctant to suggest that karyotype alteration alone, in the absence of any supporting change in RAPD or CNRE profiles, defines a new strain or provides compelling evidence that reinfection or superinfection has occurred in these patients. It is equally possible that the extent of in vivo karyotype instability may be greater than previously appreciated. This effect may have been obscured by low sample sizes of both patients and isolates in earlier studies, such that statistically significant conclusions could not reliably have been drawn. The absence of karyotype differences among isolates with less common multilocus genotypes (ETs 2, 3, 4, 5, 7, 8, 12, 20, and 25) also argues against the reinfection/superinfection hypothesis. If reinfection with new strains were likely, patients initially infected with a less prevalent ET would more likely be reinfected with a common ET, primarily ET I [19]. The observation that 7 of 8 chromosomal alterations were seen in ET I strains (the eighth occurred in ET 20, which is closely related to ET I [18, 19]) raises the intriguing possibility that karyotype instability is associated with particular genetic backgrounds. Karyotype instability could also result in potentially significant phenotypic differences. In 3 patients, karyotype polymorphisms coincided with phenotype changes in colony morphology (RCI6 and RC20) or 5-fluorocytosine MIC (RC24). At present, the genetic basis ofchromosomal polymorphisms is not known. Resolution of this problem will require physical mapping of the genomes of clinical isolates. In these patients, development of in vitro antifungal resistance to amphotericin B, 5-fluorocytosine, or fluconazole did not appear to contribute significantly to relapse. In I patient, a stepwise increase in fluconazole MIC (from 4 to 64 p,g/ml over 18-months) was identified; no subtype changes were found in these isolates. The importance of this finding is not clear

8 HD 1996; 174 (October) Subtypes of Serial C. neotormans Isolates 819 Kbp Figure 3. Southern blot of C. neoformans DNA from patients with persistent cryptococcosis probed with digoxigenin-labeled CNRE-l repetitive DNA element. Patient no. and isolate collection date are listed above each lane. Digoxigenin-Iabeled markers (Boehringer Mannheim) were used as molecular size standard. Kbp = kilobase pairs.! and will require determination ofthe prevalence offluconazole resistance in cryptococcal populations. Our results confirm those of earlier studies [10-12, 15] suggesting that relapse occurs because the initial infection is either not adequately treated or not appropriately controlled. The findings of this study may serve as a framework for future investigations directed at determining risk factors for relapse and improving management of the initial fungal infection in the prevention of persistent cryptococcal disease. Study Group Members Members ofthe CDC Cryptococcal Disease Active Surveillance Group participating in this study are David Stephens, Monica Farley, David Rimland, Wendy Baughman, Chris Lao, Jodie DUe, and Christopher Harvey (Atlanta); Richard Hamill (Houston); and Arthur L. Reingold, Gretchen Rothrock, Bharat Pattni, and Pam Daily (San Francisco). Acknowledgments We gratefully acknowledge the many participants in the CDC Fungal Active Surveillance, who collected isolates for this study. We thank Bradley Perkins and Jay Wengerfor reviewingthe manuscript. References 1. Panther LA, Sande MA. Cryptococcal meningitis in the acquired immunodeficiency syndrome. Semin Respir Infect 1990;5: Sugar AM. Overview: cryptococcosis in the patient with AIDS. Mycopathologia 1991; 14: Grant IH, Armstrong D. Fungal infections in AIDS: cryptococcosis. Infect Dis Clin North Am 1988; 2:457 ~ Pinner RW, Hajjeh RA, Powderly WG. Prospects for the prevention of cryptococcosis in persons infected with human immunodeficiency virus. Clin Infect Dis 1995; I(suppl 1):S Powderly WG. Therapy for cryptococcal meningitis in patients with AIDS. Clin Infect Dis 1992; 14(suppl l):s Larsen RA, Bozzette S, McCutchan A. Persistent Cryptococcus neoformans infection of the prostate after successful treatment of meningitis. Ann Intern Med 1989;111: Bozzette SA, Larsen RA, Chiu J, et a!. Fluconazole treatment of persistent Cryptococcus neoformans prostatic infection in AIDS. Ann Intern Med 1991; 115: Bailly MP, Boibieux A, Biron F. Persistence of Cryptococcus neotormans in the prostate: failure of fluconazole despite high doses. J Infect Dis 1991; 164: Ndimbie OK, Dekker A, Martinez AJ, Dixon B. Prostatic sequestration of Cryptococcus neotormans in immunocompromised persons treated for cryptococcal meningoencephalitis. Histol Histopathol 1994; 9: Spitzer ED, Spitzer SG, Freundlich LF, Casadevall A. Persistence ofinitial infection in recurrent Cryptococcus neoformans meningitis. Lancet 1993; 341:595~ Varma A, Swinne D, Staib F, et al. Diversity of DNA fingerprints in Cryptococcus neoformans. J Clin MicrobioI199S;33:

9 820 Brandt et al. JID 1996; 174 (October) 12. Barchiesi F, Hollis Rf, Messer SA, et al. Electrophoretic karyotype and in vitro antifungal susceptibility of Cryptococcus neoformans isolates from AIDS patients. Diagn Microbiol Infect Dis 1995;23: Haynes KA, Sullivan 01, Coleman DC, et al. Involvement of multiple Cryptococcus neoformans strains in a single episode of cryptococcosis and reinfection with novel strains in recurrent infection demonstrated by random amplification ofpolymorphic DNA and fingerprinting. 1 Clin Microbiol 1995; 33: Wingard lr, Merz WG, Rinaldi MG, et al. Increase in Candida krusei among patients with bone marrow transplantation and neutropenia treated prophylactically with fluconazole. N Engl J Med 1991;325: IS. Casadevall A, Spitzer ED, Webb D. Susceptibilities ofserial Cryptococcus neoformans isolates from patients with recurrent cryptococcal meningitis to amphotericin B and fluconazole. Antimicrob Agents Chemother 1993; 37: Powderly WG, Keath EJ, Sokol-Anderson M, et al. Amphotericin B resistant Cryptococcus neoformans in a patient with AIDS. Infect Dis Clin Pract 1993; 1: Paugam A, Dupouy-Camet 1, Blanche P, et al. Increased fluconazole resistance of Cryptococcus neoformans isolated from a patient with AIDS and recurrent meningitis. Clin Infect Dis 1994; 19: Brandt ME, Hutwagner LC, Kuykendall RJ, et al. Comparison of multilocus enzyme electrophoresis and random amplified polymorphic DNA analysis for molecular subtyping of Cryptococcus neoformans. 1 Clin Microbiol 1995;33: Brandt ME, Hutwagner LC, Klug LA, et al. Molecular subtype distribution of Cryptococcus neoformans in four areas of the United States. 1 Clin MicrobioI1996;34: Kwon-Chung Kl, Polacheck 1, Bennett le. Improved diagnostic medium for separation of Cryptococcus neoformans var. neoformans (serotypes A and D) and Cryptococcus neoformans val gattii (serotypes Band C). 1 Clin Microbiol1982; 15: Brandt ME, Bragg SL, Pinner RW. Multilocus enzyme typing of Cryptococcus neoformans. 1 Clin MicrobioI1993;31: Perfect JR, Magee BB, Magee PT. Separation of chromosomes of Cryptococcus neoformans by pulsed field gel electrophoresis. Infect Immun 1989; 57: Varma A, Kwon-Chung Kl. Rapid method to extract DNA from Cryptococcus neoformans. 1 Clin Microbiol 1991;29: Pfaller MA, Hollis Rl, Sader HS. Chromosomal restriction fragment analysis by pulsed-field gel electrophoresis. In: Isenberg HD, ed. Clinical microbiology procedures handbook. Suppl I. Washington, DC: American Society for Microbiology, 1994: 10.5.C.l-10.5.C Spitzer ED, Spitzer SG. Use of a dispersed repetitive DNA element to distinguish clinical isolates of Cryptococcus neoformans. 1 Clin MicrobioI 1992; 30: National Committee for Clinical Laboratory Standards. Reference method for broth dilution antifungal susceptibility testing for yeasts. Proposed standard. Document M27-P. Villanova, PA: NeCLS, Pfaller MA, Grant C, Morthland V. Comparative evaluation of alternative methods for broth dilution susceptibility testing of fluconazole against Candida albicans. 1 Clin Microbiol 1994; 32:506-9.