This retrospective cohort research focused on episodes of cryptococcosis that occurred in patients who had Western-blot confirmed HIV infection and who were treated in six hospitals in Medellín, Colombia. The C. neoformans isolates from these patients were classified as susceptible, susceptible dose-dependent (SDD) or resistant according to their susceptibility to fluconazole. We evaluated the impact of resistance to fluconazole on treatment failure during the consolidation phase, and on relapse of maintenance therapy.

Episodes of cryptococcosis were identified from the database of the Medical Mycology and Experimental Unit of the Corporation for Biological Research (CIB). This unit performed fluconazole susceptibility tests on 250 C. neoformans isolates collected during the period of January 2000–December 2008, 187 (74.8%) of which came from AIDS patients. All of the isolates were identified as C. neoformans var. grubii. Episodes of cryptococcosis were included in the study if the disease had been confirmed by the isolation of C. neoformans from cerebrospinal fluid, blood, lung, skin or lymph nodes, with samples being obtained for microbiological and pathological analyses for each clinically affected organ if there was information on fluconazole susceptibility reported as minimal inhibitory concentration (MIC), and if the information available in the clinical records was complete enough to determine the outcome of treatment.

For purposes of inclusion in the cohort, patients had to have a positive culture for C. neoformans and the presence of at least one sign or symptom compatible with the disease. The patients included presented with their initial episode or with recurrence of a previously diagnosed episode.

Treatment followed mostly the recommendations that appeared in Practice Guidelines for the Management of Cryptococcal Disease published in 2000 by the Infectious Diseases Society of America.21 However, no patient received flucytosine because this drug was not available in Colombia at the time of the study. Briefly, patients with cryptococcal disease received induction therapy with amphotericin B deoxycholate, 0.7–1mg/kg/day, for two weeks, followed by consolidation therapy with fluconazole, 400–800mg/day, for 10–12 weeks. Some patients did not receive induction therapy with amphotericin B deoxycholate, but were managed with fluconazole alone at 400–800mg/day for 10–12 weeks. At the end of the consolidation phase, maintenance treatment consisted of fluconazole 200–400mg/day, until achievement of CD4T cell counts of at least 200cells/mm3 on two measurements made at least 3 months apart. No patient had intracranial pressure measurement or underwent a new lumbar puncture for pressure management.

A therapeutic failure was considered to have occurred when an event was observed during the 12 weeks of the consolidation phase of treatment. Relapse was taken to be an event occurring after initiation of maintenance treatment and up to 36 weeks on treatment. Failure and relapse were considered proven when recurrence or exacerbation of signs and/or symptoms of disease were accompanied by microbiological (isolation on culture) demonstration of C. neoformans. Failure and relapse were considered probable when mycological tests did not demonstrate the organism, but the patient showed recurrence or exacerbation of signs and/or symptoms compatible with cryptococcosis, had no other disease documented, and had response to antifungal therapy.

The outcomes evaluated were those corresponding to survival time free of failure (to evaluate the consolidation phase) and survival time free of failure or relapse (a combined outcome to evaluate the consolidation and maintenance phases).

The mycological procedures were performed at the CIB's Medical Mycology Laboratory. The isolates were identified by assimilation of urea, production of melanin, and assimilation studies done by means of the Api Aux 20C kit (bioMérieux, Marcy. l’Étoile, France). Species were determined by growth in the canavanine-glycine-bromothymol blue (CGB) medium, while the serotype were defined by means of a commercial kit with specific polyclonal antibodies (Crypto-Check, Iatron, Japan).

Susceptibility tests were performed using the M 44A disk diffusion method, using 25μg fluconazole disks (Becton Dickinson, Sparks, MD), according to CLSI.13 Reading was done with the video camera BIOMIC®, which measures inhibition in millimeters, converts them into MICs by a regression curve, and stores the data electronically (Giles Scientific, 1999, Santa Barbara, CA). Quality control strains tested were ATCC C. albicans 90028 and C. parapsilosis 22019. Given that there is no standard cutoff points for resistance specific for C. neoformans and fluconazole, we used the values published by the CLSI and by Barry et al. for Candida spp., which consider susceptible isolates to be those with a MIC≤8μg/ml (≥19mm inhibition), SDD between 16–32μg/ml (15–18mm inhibition) and resistant MIC≥64μg/ml (≤14mm inhibition).4,13,14

Cryptococcal episodes were classified into three groups according to the fluconazole susceptibility pattern exhibited by the particular C. neoformans isolate, as follows: susceptible, MIC≤8μg/ml; SDD, MIC ranging from 16 to 32μg/ml, and resistant, MIC≥64μg/ml. A second classification was based on a MIC cutoff of 16μg/ml, as described by Aller et al.,1 who grouped the isolates as susceptible if the MIC was <16μg/ml and resistant if they had a MIC≥16μg/ml (epidemiological cut points).1

The mycological data, obtained from the records of the Medical and Experimental Mycology Unit of the CIB, and the clinical data extracted from medical records in the participating hospitals, were recorded in a format designed for the purpose and entered into a database created in Microsoft Excel® (Microsoft, Redmond, WA).

The software SPSS 16.0 (SPSS Inc., Chicago, IL) and EPI DAT 3.1 (Consejería de Sanidad, Junta de Galicia and OPS/OMS) were used for statistical analysis. For descriptive analysis, measures of absolute frequency and percentage for qualitative variables were calculated, while for quantitative variables the median and the inter-quartile range (IQR) were used. The baseline characteristics of the three groups (susceptible, SDD and resistant) were compared in a univariate analysis using the Pearson Chi2 test or Fisher's exact test when applied to compare qualitative variables and the Kruskal–Wallis test for comparison of quantitative variables, accepting a two-tailed p-value ≤0.05 as statistically significant. The Shapiro–Wilk test showed that all quantitative variables did not adopt a normal distribution curve (p≤0.05 for at least one group in each variable). Survival analysis was done using the Kaplan–Meier method to estimate the survival function for each of the outcomes and the log-rank test to compare groups. To identify other possible variables associated with the risk of therapeutic failure or relapse, a Cox's proportional hazards model with the Enter method was obtained, including variables with p-values ≤0.25 in the univariate analysis.

This research was approved by the Ethics Committee of Universidad Pontificia Bolivariana and the Research Committees of the participant Hospitals: Hospital La María, Hospital Pablo Tobón Uribe, Clínica Universitaria Bolivariana and Hospital General de Medellín. The study was subjected to the principles of Declaration of Helsinski. No informed consent was prepared due to the retrospective nature of the study and, also, due to the difficulties for contacting those patients seen in consultation years ago.

Seventy-one episodes of cryptococcosis experienced by 68 patients met the study's inclusion criteria (Fig. 1). Male patients (n=59) accounted for 81% of the episodes. The median age of the study population was 35 years (IQR 29–40), and the median duration of the HIV infection was 121 days (3–159). In 25 (35%) episodes, the patients were receiving antiretroviral therapy at the time they developed cryptococcosis. The median viral load was 236,000copies/ml (93,200–733,000copies/ml), and the median CD4 count was 22.5cells/mm3 (10–51cells/mm3). The use of fluconazole prior to the development of cryptococcosis was reported in 22 (31%) episodes.

Fig. 1.

Flowchart of isolates, episodes and patients included in the study.

(0.49MB).

For 60 (85%) episodes of cryptococcosis it was the first event, and for 11 (15%) it was a relapse; for the latter, the initial episode of cryptococcosis had occurred a mean of 5.2 months before relapse. Only 3 of these 11 patients had received fluconazole for maintenance treatment.

Meningeal involvement was present in 66 (93%) episodes; in the remaining 5 episodes involvement of a different site (fungemia, pulmonary, cutaneous, and/or lymph node) other than central nervous system was documented. In 12 (17%) of the episodes, simultaneous isolation of C. neoformans from more than one site was recorded; sixteen episodes had documented fungemia. In 7 of these episodes, susceptibility tests were run for isolates obtained from cerebrospinal fluids, in another 3 from lung samples, and from an equal number from blood cultures. In a single case, the isolate subjected to susceptibility tests came from a lymph node, and in two patients more these studies were carried out using isolates coming from two different involved sites. Documented concurrent opportunistic infections were primarily tuberculosis and pneumocystosis, which were demonstrated in 20 (28%) episodes.

For those patients who had meningitis, lower glucose concentration was the most relevant finding in cerebrospinal fluid (CSF); the increase in protein concentration and leukocytes was mild in all groups. The qualitative cryptococcal antigen test on CSF was positive in 47 (94%) of the 50 episodes in which this test was performed. The India ink test was positive in 60 (85%) of the 71 episodes.

Induction treatment consisted of amphotericin B deoxycholate in 64 (90%) of the episodes; in the remaining 7 episodes, only fluconazole treatment was given from the time of diagnosis. The total mean dose of amphotericin B administered was 737 (581–1000) mg. In all episodes, but two, the patients received 400mg daily of fluconazole during the consolidation phase; these two patients were given 600 and 800mg daily.

Fluconazole susceptibility testing of the 71 isolates showed that 36 (51%) were susceptible, twenty (28%) were SDD, and the remaining 15 (21%) isolates resistant (MIC≥64μg/ml).

When comparing baseline and clinical characteristics from the patients whose isolates fell into each of the 3 groups defined by the fluconazole susceptibility, no statistically significant differences were observed (Tables 1 and 2). From the 71 episodes evaluated during the consolidation phase, 66 completed such a phase while the remaining 5 (7%) experienced treatment failures. From the 33 episodes followed during the maintenance stage, 4 (12%) had cryptoccocal relapses.

Concurrent opportunistic infections were present in 12 (33%) of the 36 episodes with susceptible isolates, in 6 (30%) of the episodes with SDD isolates, and 2 (13%) of the episodes with resistant isolates. In the group with susceptible isolates the concurrent opportunistic infections were tuberculosis (9) and Pneumocystis jirovecii pneumonia (3). In the SDD group, such concurrent infections were P. jirovecii pneumonia (3), Salmonella spp. bacteremia (1), disseminated Mycobacterium avium complex (1) and cytomegalovirus (1). In the group with resistant isolates, Salmonella spp. bacteremia and Kaposi's sarcoma occurred in one patient each. No relapse or death events occurred in patients receiving simultaneous anti-tuberculosis treatment.

There were 5 (7%) treatment failures, all mycologically confirmed; two were in the group of episodes whose isolates were susceptible to fluconazole, two in the SDD group, and one was in the group of patients whose isolates were fluconazole resistant. The latter patient died of cryptococcosis at week 4 after diagnosis. Two patients in the group whose isolates were susceptible to fluconazole died within one week of switching from amphotericin B to fluconazole therapy, and their deaths were considered to be attributable to cryptococcosis. Two patients in the SDD group died in the consolidation phase from causes unrelated to cryptococcosis. All negative outcomes occurred in patients with meningitis; one of these patients had, additionally, fungemia and skin involvement.

Fig. 2 shows MICs grouped by results of therapy with fluconazole. For episodes in which the cryptococcal infection was controlled, the mean MIC was 14 (2–48)μg/ml, compared with a mean of 48 (4.4–113)μg/ml in the group that failed during the consolidation phase, and a mean of 4 (1.8–16)μg/ml in the group that relapsed during maintenance treatment (p=0.51).

Fig. 2.

Distribution of minimal inhibitory concentrations (MIC) according to the outcome of treatment with fluconazole in 71 episodes of cryptococcosis.

(0.05MB).

The unadjusted comparison of time of survival free of therapeutic failure and time of survival free of failure or relapse showed no statistically significant differences among the three groups when using the log-rank test (p=0.65 and p=0.38, respectively) (Fig. 3).

Fig. 3.

Kaplan–Meier estimates of the probability of (A) remaining free of therapeutic failure according to the level of resistance to fluconazole of (B) remaining free of therapeutic failure or relapse according to the level of resistance to fluconazole (susceptible, SDD and resistant) (N=71).

(0.15MB).

For 33 of the 71 episodes (46%), follow-up was recorded during the maintenance treatment phase. A total of 4 (12%) relapses occurred, 3 in the group of susceptible isolates and 1 in the SDD group. Of these 4 relapse episodes, 3 were confirmed and one was probable. No differences were found between the 3 groups in regard to relapses (p=0.42).

A Cox's model, including as covariates age, use of antiretroviral treatment, and CD4 counts, was built for each outcome as these variables had a p-value ≤0.25 in the bivariate analysis. The time of survival free of failure was not significantly affected by age (hazards ratio (HR) per year of change in age, 1.13; CI 95% 0.97–1.32; p=0.11), CD4 count (HR per change in cell/mm3, 1.00; CI 95% 0.97–1.03; p=0.88), or the use of antiretroviral drugs (HR 2.66; CI 95% 0.22–31.82; p=0.44). The time of survival free of failure and relapse was not affected by age (HR per year of change in age, 1.05; CI 95% 0.95–1.17; p=0.35), CD4 count (HR per change in cell/mm3, 0.99; CI 95% 0.97–1.02; p=0.60), or by the use of antiretroviral drugs (HR 2.55; CI 95% 0.39–16.59; p=0.33).

A comparison between isolates with MIC <16μg/ml (susceptible) versus those with MIC≥16μg/ml (resistant) yielded similar non-significant results. The unadjusted comparison of time of survival free of therapeutic failure and time of survival free of failure or relapse showed no statistically significant differences among the two groups when using the log-rank test (p=0.49 and p=0.86, respectively) (Fig. 4).

Fig. 4.

Kaplan–Meier estimates of the probability of (A) remaining free of therapeutic failure according to the level of resistance to fluconazole or (B) remaining free of therapeutic failure or relapse according to the level of resistance to fluconazole (MIC<16μg/ml vs MIC≥16μg/ml) (N=71).

(0.15MB).

The Cox's model for a breakpoint of 16μg/ml also included as covariates age, CD4 T-cells and the use of antiretroviral drugs. The time of survival free of therapeutic failure was not significantly influenced by age (HR per year of change in age, 1.13; CI 95% 0.97–1.32; p=0.11), CD4 T-cells count (HR per change in cell/mm3, 1.0; CI 95% 0.97–1.03; p=0.95) or by the use of antiretroviral drugs (HR 2.4, CI 95% 0.23–24.62; p=0.46). The time of survival free of failure or relapse was not significantly influenced by age (HR per year of change in age, 1.06; CI 95% 0.95–1.18; p=0.30), CD4 T-cells count (HR per change in cell/mm3, 0.99; CI 95% 0.97–1.02; p=0.57), or by the use of antiretroviral drugs (HR 0.97; CI 95% 0.39–10.70; p=0.40).