Change in fluconazole susceptibility patterns and genetic relationship among oral Candida albicans isolates

Abstract

Objective: 

To assess the genetic homogeneity or heterogeneity within each set of Candida albicans isolates colonizing/infecting the oral cavities of HIV-infected patients undergoing azole therapy when changes in susceptibility to fluconazole were detected.

Design: 

Fourteen HIV-positive patients suffering recurrent episodes of oral candidosis were prospectively followed from the first episode to the isolation of strains with decreased susceptibility to fluconazole. The strains of C. albicans isolated either from episodes or controls throughout the prospective study were analysed.

Methods: 

Electrophoretic karyotyping and hybridization with the repeated sequence probe 27A were used to delineate sequential isolates. In vitro susceptibility tests to fluconazole and ketoconazole were also performed. The results obtained by DNA fingerprinting with the probe combined with computer-assisted analysis were used to assess the genetic relationships amongst the strains. In addition, comparison with the genetic relatedness of a group of geographically unrelated strains was made.

Results: 

Isogenic populations of sequential isolates were observed only in two patients; 12 patients harboured heterogenic populations over time, although in 11 patients there was a predominant strain that was isolated more than once, and only one of these patients carried strains with a similarity index less than 80%. With the exception of two patients, each patient carried a major strain that became less susceptible to fluconazole. The similarity index for the unrelated strains was 59%.

Conclusions: 

HIV-infected patients may carry a mixed population of strains, but the strains tend to be related to each other. The strains were maintained throughout the course of infection and at least one developed secondary resistance to fluconazole.

Introduction

Oropharyngeal candidosis (OPC) is the commonest opportunistic infection in patients infected with HIV, occurring in up to 90% of patients. Candida albicans is the species most frequently isolated from patients with AIDS and with oral thrush. The use of fluconazole in this clinical setting is widely accepted, although OPC recurs within 3 months of successful treatment in up to 80% of patients treated with azole therapy. The relapses of OPC observed during progression of AIDS may be attributed to strains of C. albicans identical to or different from the initial isolate. During this recurrent infection the appearance of clinical resistance to fluconazole therapy due to a resistant C. albicans has been clearly documented [^1,2]. However, whether the resistant isolate represents a newly acquired strain or emerged from an originally mixed population is frequently unclear. Several studies [^3–6] using molecular techniques to analyse isolates of C. albicans from oral cavities of HIV-positive patients have shed some light on this matter, but their results are variable. Some authors have found that the fluconazole-susceptible and fluconazole-resistant strains belong to the same DNA type. Others, however, have found that in a significant proportion of cases, susceptible and resistant isolates from the same patient are of different DNA types. Apart from the methodological differences between the studies, the different criteria on what is called ‘the same DNA type’ or ‘different DNA type’ could have influence on these discrepancies. Therefore, a computerized analysis of genetic relationships of DNA types obtained with a given molecular assay seems inevitable in epidemiological studies. In a previous study [⁷], in which genetic diversity among C. albicans oral isolates from HIV-infected patients was assessed by four different molecular techniques, it was stated that no particular DNA type was linked with a geographically delimited area. Similarly, the oral isolates showing fluconazole resistance were not grouped into a common DNA type. This lack of cross-infection amongst the patients studied agreed with the idea of endogenous origin of fluconazole-resistant strains, and refuted the idea of strain replacement. On the other hand, that work highlighted the necessity of using at least two different methods simultaneously to obtain a reliable identification of clones among C. albicans isolates in epidemiological studies.

The objective of this study was to explore the genetic homogeneity or heterogeneity within each set of isolates colonizing/infecting oral cavities of patients when changes in susceptibility to fluconazole were detected. An additional objective was to evaluate the clinical significance of the fluconazole-resistant genotypes isolated in this study as causative agents of episodes non-responding to the treatment with this antifungal agent. Part of this study was presented at the 13th Congress of the International Society for Human and Animal Mycology, Parma, in June 1997 [⁸].

Material and methods

Patient population

Fourteen HIV-infected patients with a first episode of OPC due to C. albicans were included in this study. The follow-up period ranged from 6 to 41 months (median, 16.4 months). The diagnoses of OPC episodes were based on the presence of a sore mouth or tongue, dysphagia/odynophagia in the presence of oral thrush, and an oropharyngeal culture that yielded C. albicans. Following completion of therapy the clinical response was categorized as follows: cure (complete resolution of symptoms), improvement (partial resolution of symptoms and oral lesions), and failure (either the absence of improvement or an exacerbation of symptoms and oral lesions).

Culture procedure and isolate identification

Samples were taken with sterile swabs either from oral lesions when an episode was active, or from whole oral mucous when controls were performed between active episodes. The specimens were plated onto Sabouraud dextrose agar (Oxoid, Madrid, Spain). Plates were incubated for at least 7 days at 28°C.

The identification of strains was performed by germ tube test, carbohydrate assimilation, fermentation and morphology on corn meal agar, as previously recommended [⁹]. The clones identified as C. albicans were included in the study, whereas the other yeasts isolated (C. glabrata, C. krusei) were not considered as they seem to be not clinically relevant in episodes of mixed OPC in HIV-infected patients (the non-albicans yeast found in our study were always associated with C. albicans in polymicrobial cultures). We found, as have other investigators [^10,11], that C. albicans was isolated from all episodes of OPC.

Strains

Overall, 80 isolates of C. albicans were recovered from the 14 patients over the observation period. Each strain was named with the number of patient from whom it was isolated followed by the sequential number of the strain (i.e., P1.1 relates to the first strain isolated from patient P1). An additional group of 20 geographically unrelated strains was used in the study of genetic relationship within the C. albicans species: 10 were obtained from 10 HIV-positive patients with OPC, and 10 were from 10 healthy women with vaginitis. Each of these 20 strains were from different geographical locations throughout Spain.

Susceptibility testing procedure

Two antifungal agents were used in the study: fluconazole (Pfizer, Madrid, Spain) and ketoconazole (Janssen Farmacéutica, Madrid, Spain). The antifungal susceptibility testing was performed by a microdilution adaptation of the reference method [¹²], which has been exhaustively described in previous reports [^13,14]. Only minor modifications are described here: the inoculum size was 10⁵ colony-forming units/ml and the readings were performed with a spectrophotometer iEMS Reader (Labsystems, Barcelona, Spain) set at 540 nm. The minimal inhibitory concentration (MIC) endpoint was defined as the lowest drug concentration giving rise to an inhibition of growth equal or greater than 50% of that in the drug-free control.

Electrophoretic karyotypes

The preparation of plugs containing yeast chromosomal DNA was similar to a method described previously [¹⁵]. Briefly, yeast cells were cultured overnight in 1% yeast extract–2% glucose–2% peptone medium at 35°C. After several washes, 10⁹ cells were collected by centrifugation, resuspended in 50 mmol/l EDTA (pH 8.0) and mixed with lyticase (Sigma-Aldrich, Alcobendas, Spain). Cells and lytic enzyme were incubated at 37°C for 20 min. Low melting point agarose (BioRad, Alcobendas, Spain) was added to a final concentration of 0.7%, and the mixture was dispensed in moulds and chilled at 4°C to solidify. Two sequential overnight incubations were performed: (i) at 37°C in buffer containing 7.5% β-mercaptoethanol, and (ii) at 50°C in buffer containing proteinase K (0.5 mg/ml) and 1% lauroylsarcosine. Several washes with 0.5 mol/l EDTA (pH 8.0) were performed before storing the plugs at 4°C.

Pulsed field electrophoresis conditions were as previously described [⁷]: the CHEF-XA mapper (BioRad) was set at 3 V/cm for 72 h, with pulse times ramped from 3 to 17 min, and reorientation angle of 106°. The Tris-borate–EDTA running buffer (1×; 90 mmol/l Tris-borate, 2 mmol/l EDTA) was maintained at 10°C. Chromosomal preparation of Hansenula wingei (BioRad) was included in each gel as a standard. The gel was stained with ethidium bromide and DNA visualized by ultraviolet (UV) transillumination.

Restriction endonuclease digestion of DNA and hybridization with probe 27A

For DNA fingerprinting by restriction fragment-length polymorphism (RFLP) analysis, the plugs were processed further. A slice of agarose block was rinsed twice with 1 ml Tris–EDTA buffer (10 mmol/l Tris-HCl, pH 7.5) containing 1 mmol/l phenylmethylsulfonylfluoride (Sigma-Aldrich). Two dialysis steps were performed with Tris-EDTA buffer (10 mmol/l Tris-HCl, 1 mmol/l EDTA, pH 7.5) before placing the plugs in 2 ml flat-bottomed microfuge tubes with 0.5 ml restriction buffer recommended by the endonuclease manufacturer and keeping on ice for 30 min. The buffer was replaced by 0.2 ml fresh buffer with 20 U EcoRI (Pharmacia-Biotech, Alcobendas, Spain) and tubes were incubated at 37°C overnight.

After digestion, excess buffer was removed and the agarose blocks were melted by incubation in a bath at 65°C for 5 min. Aliquots of 15–20 μl of the molten agarose were mixed with 6× gel-loading buffer and pipetted into wells of a 0.8% (weight/vol) agarose gel. The samples were electrophoresed in 1× Trisacetate–EDTA (40 mmol/l Tris–acetate, 1 mmol/l EDTA) buffer at 4 V/cm for 4 h. The gel was stained with ethidium bromide and DNA visualized by UV transillumination. An aliquot of bacteriophage lambda DNA digested with HindIII and fluorescein-labelled (Amersham Ibérica, Madrid, Spain) was sampled in all the gels performed as the molecular marker. In addition, EcoRI-digested DNA from C. albicans ATCC 64548 strain was included in all the gels as the reference to compare the banding patterns of strains included in different gels.

DNA fragments were transferred to a nylon membrane (Hybond-N+, Amersham Ibérica) by means of a vacuum blotting system (VacuGene XL, Pharmacia-Biotech) according to the instructions supplied by the manufacturers. The probe used (probe 27A) was an EcoRI fragment of C. albicans DNA cloned into the plasmid pUC18, which was kindly supplied by S. Scherer (University of Minnesota, Minneapolis, Minnesota, USA) [¹⁶]. Labelling of the probe with fluorescein (electrochemiluminescence random prime labelling system), hybridization, stringency washes and detection by chemiluminiscence (electrochemiluminescence detection system) were performed according to the manufacturer's instructions (Amersham Ibérica). The reproducibility of this hybridization method was assessed by including 10 samples of digested DNA from C. albicans ATCC 64548 in 10 separate gels.

Definition of DNA types

CHEF patterns were interpreted visually, and isolates were considered to have different karyotypes if they differed in the number/mobility of one or more chromosome-sized DNA bands. The hybridization profiles were digitized into the database of the Lane Manager software program (TDI, Madrid, Spain) using a Scanjet scanner with the transparency option (Hewlett-Packard, Madrid, Spain). Patterns were unwarped and normalized against the marker for cross comparisons. Similarity coefficients (S_AB) were computed for each pair of lanes (A and B) based on band positions alone according to the Dice's formula S_AB = 2E/2E(a + b), where E is the number of bands in patterns A and B sharing the same positions, a is the number of bands in pattern A with no correlates in pattern B, and b is the number of bands in pattern B with no correlates in pattern A. Dendrograms based on S_AB values were generated using the unweighted pair group method with averages to visualize the relationships amongst the isolates. On the basis of the results of the reproducibility assay performed with hybridization profiles of C. albicans ATCC 64548 (see Results), two isolates were considered identical when they were clustered at a S_AB ≥ 95%.

Results

The results of fluconazole MIC, ketoconazole MIC and DNA types are summarized in Table 1.

Table 1: Fluconazole and ketoconazole susceptibilities, DNA types, and genetic relationship within sets of Candida albicans from 14 patients with recurrent oral candidosis.

Table 1: (continued)

Susceptibility test

In vitro susceptibility testing of the 80 isolates revealed that MIC of fluconazole ranged from ≤ 0.12 to 32 μg/ml, and the MIC of ketoconazole ranged from ≤ 0.0002 to 0.25 μg/ml. The increase in fluconazole MIC of sequential isolates from single patients paralleled an increase in ketoconazole MIC in all the cases studied. The fluconazole breakpoints for MIC determined by the method used in this study have been defined in a previous report [¹⁷]: isolates for which MIC are ≤ 2 μg/ml are susceptible to fluconazole, whereas those for which MIC are ≥ 16 μg/ml are resistant. Isolates for which the MIC of fluconazole is 4–8 μg/ml are considered susceptible dependent upon dose.

OPC episodes

The group of 14 patients suffered a total of 53 episodes of OPC, and 53 C. albicans strains were isolated during these active episodes. The remaining 27 strains included in this study were isolated from controls performed after antifungal therapy or during intervals between episodes. Forty-two episodes were treated with fluconazole, and the results of fluconazole MIC of strains isolated from these episodes and clinical outcomes are summarized in Table 2. One of the responding episodes was due to a resistant strain (P5.9; fluconazole MIC, 32 μg/ml), which was isolated from an active episode of patient P5, which occurred while he was undergoing fluconazole prophylaxis; this episode responded to a 200 mg daily dose of fluconazole. This patient suffered a further episode from which a resistant isolate (P5.10; fluconazole MIC, 32 μg/ml) was isolated; therapy with itraconazole was started but the clinical response could not be evaluated because of patient's compliance.

Table 2: Clinical outcome of episodes treated with fluconazole.

The non-responding episode that was due to a susceptible strain (strain P14.8; fluconazole MIC, 0.5 μg/ml) was initially treated with 100 mg daily; this episode was further treated with fluconazole at 200 mg daily dose and responded. Ten episodes were treated with ketoconazole, eight episodes were caused by strains with ketoconazole MIC ≤ 0.0002 μg/ml and responded to therapy with 400 mg daily ketoconazole. The fluconazole MIC of these strains was ≤ 2 μg/ml. The remaining two episodes were caused by strains P2.6 and P13.3, whose ketoconazole MIC were 0.06 and 0.0019 μg/ml, respectively, and responded to therapy with ketoconazole solution. The fluconazole MIC of these two strains were 8 and 32 μg/ml, respectively.

Comparison of karyotyping versus 27A fingerprinting in strain delineation

The CHEF method placed the 77 strains analysed in 16 different electrophoretic karyotypes, and the hybridization method placed the 78 strains analysed in 32 DNA types (Table 1). Analysis of electrophoretic karyotypes of sets of isolates from individual patients demonstrated homogeneous electrophoretic karyotypes within each set of strains from 10 test individuals. Sets of strains from four patients (P3, P7, P12 and P13) presented more than one electrophoretic karyotype. An example of karyotypes obtained for the set of strains from patient P3 is shown in Fig. 1: strain P3.2 (lane 2) showed the karyotype EK3 and the other strains from patient P3 showed the karyotype EK1. We considered that isolate P3.2 bore a different karyotype on the basis of the double band of about 1 Mb that can be seen in lane 2 of Fig. 1; in the remaining karyotypes of strains from patient P3 this band is single. On the other hand, in the lanes corresponding to the isolates P3.1, P3.2, P3.3, P3.4 and P3.6, a more or less fuzzy band between the two largest bands (3.13 and 2.35 Mb) could be seen. However, according to Magee et al. [¹⁸] the fuzzy band that appeared in this size range for some karyotypes corresponded to one of the homologues of rDNA and it was considered too highly variable to be considered as the basis to separate strains into different types.

Fig. 1.: Electrophoretic karyotypes of sequential isolates of Candida albicans from patient P3. Lane M, Hansenula wingei chromosomes with sizes denoted in Mb (on the left). Lanes 1–6, isolates P3.1 to P3.6 from patient P3; lane c2, C. albicans ATCC 64548.

On the basis of hybridization results, only two sets of strains (P8 and P10) maintained the same hybridization pattern (H pattern) over the study period. However, 12 sets of strains (P1, P2, P3, P4, P5, P6, P7, P9, P11, P12, P13 and P14) each contained at least two different H patterns. An example of hybridization profiles obtained for strains from set P3 is shown in Fig. 2: different H patterns were showed by four strains from this patient, whereas the isolates P3.1 and P3.3 shared the pattern H6. A dendrogram showing the genetic distance among the six isolates from patient P3 calculated by the software is included in Fig. 2.

Fig. 2.: (a) Hybridization profiles of isolates from patient P3. Lane M, fluorescein-labelled HindIII digest of bacteriophage lambda, with sizes denoted in kb (on the left). Lanes 1–6, isolates P3.1 to P3.6 from patient P3. (b) Dendrogram of the Candida albicans isolates from patient P3.

In eight patients (P1, P2, P3, P4, P6, P9, P11 and P14) analysis of hybridization with probe 27A demonstrated variations in genetic background of strains that were not detected by the CHEF technique, revealing that sequential populations of these patients were not as genetically homogeneous as described by the CHEF technique. An example of this situation (strains of set P3) is shown in Figs 1 and 2.

Genetic relationship within each set of strains from single patients versus a population of unrelated strains

Having established that hybridization with probe 27A had greater potential to differentiate between C. albicans strains than karyotyping, the genetic relatedness within each set of strains was assessed by the software package on the basis of data from hybridization images. Of the 14 sets of three or more C. albicans strains, two consisted of isolates with an identical H pattern, whereas 12 consisted of isolates with two or more H patterns. In 11 out of the 12 heterogenic sets of isolates, one or two patterns were isolated more than once, whereas in one set (P13), no repetitions in the H patterns exhibited by its three successive isolates were found. In each patient, repetitive patterns were isolated either from lesions or from episode-free intervals; therefore, no association could be established between a given DNA type and the presence of clinical symptoms.

The similarity indices automatically computed for strains within each set are shown in Table 1. Sets P8 and P10 each had an S_AB of 100%, and 11 out of 12 sets containing heterogeneous H profiles showed an S_AB ≥ 80%. A set with 10 strains (P5) had a median S_AB of 72.4%. However, the results derived from the reproducibility assay for hybridization patterns showed that 10 replicas of the control strain ATCC 64548 were identified in the dendrogram generated by the software at an S_AB of 95%; this point was defined as the threshold above which two patterns were considered identical. The genetic relationship obtained for each sets of strains was compared with the genetic relationship obtained for a population of 20 unrelated strains and a combined dendrogram was constructed for the whole related and unrelated strains of this study (Fig. 3). The branch point at which the 100 isolates of C. albicans were connected was an S_AB of 59.5%, and the sets of strains from each patient remained clustered at S_AB ≥ 80% in all but two of the patients studied (P5 and P13). Strains from P13, which were clustered at an S_AB of 81.2% when tested separately, appeared randomly distributed in the combined dendrogram.

Fig. 3.: Combined dendrogram of the entire collection of isolates from the 14 patients of this study and 20 geographically unrelated strains. The sequential isolates from single patients are named ‘P’ plus patient number plus isolate number. The unrelated strains are named ‘CL’ plus isolate number. The vaginal isolates are marked with ‘V'. The mean S_AB is presented on the left of the rule shown below the dendrogram. The threshold for identity (S_AB = 95%) is denoted by a vertical bold line. The isolates from a single patient showing identical patterns are shown in parentheses.

Changes in susceptibility to fluconazole and DNA types

A significant increase in fluconazole MIC, of four twofold dilutions or greater, was detected over time in the isolates from all 14 patients included in the study. The hybridization results were used in this study to ascertain whether the sensitive and resistant strain belonged to the same DNA type. Two patients maintained a unique DNA type over the study period, regardless of the change in fluconazole MIC of strains (P8 and P10). One fluconazole-susceptible strain and two fluconazole-resistant strains were isolated from P8, and all of them exhibited the pattern H13. In the case of patient P10, three susceptible strains were isolated before the occurrence of the strain with decreased susceptibility to fluconazole and all of them shared the H23 profile. Twelve patients had populations of sequential isolates in which more than one DNA type was observed. Within 10 out of these 12 mixed populations the DNA type of at least one isolate with secondary resistance to fluconazole was identical to one of the prior susceptible strains, although more than one resistant DNA type was observed in four of these patients (P2, P5, P7 and P14). In the other two patients, P3 and P13, the H pattern of the resistant isolate did not match with any of the patterns shown by the previous susceptible strains.

Discussion

In an effort to simplify the methodology of the study, the same DNA sample preparation procedure was used for both karyotyping and hybridization. A simple method is described by which the intact chromosomal DNA, extracted by techniques conventionally used in pulsed field gel electrophoresis analysis, is also used to generate RFLP with the probe 27A. The combined use of these two methods confirmed that pulsed-field gel electrophoresis was less discriminatory than hybridization with a specific probe, as we have reported previously [⁷]. Hybridization with probe 27A achieved the distinction among strains that were previously considered isogenic by the karyotyping method; this was true for strains within eight sets. This result reflects the risk of bias on the results of any epidemiological study that is exclusively based on karyotype delineation of strains. Conversely, one set (P12) contained two strains with identical hybridization profiles and different karyotypes. This is a surprising data because the usual scenario in this study with 80 isolates and in a previous study with 38 isolates [⁷] was that two strains exhibiting distinct karyotypes always bore different hybridization patterns. The major advantage of the CHEF method was the ease of visual interpretation of the patterns obtained, due to the discrete number of bands of karyotypes. However, more complex interpretation of hybridization profiles has now been improved with computer-aided analysis and has the advantage of the possibility of creating a datafile, which is essential in large epidemiological studies.

The aim of this study was to explore the genetic homogeneity or heterogeneity of C. albicans strains sequentially isolated from oral cavities of single patients when changes in susceptibility to fluconazole were detected. In concordance with previously published data [^3,4], heterogenic populations of C. albicans strains were detected in the oral cavities of 12 out of 14 patients even though in the majority of these cases a major DNA type was repetitively obtained among successive strains of one population. However, the two patients of this study who presented respective isogenic populations (S_AB = 100%) of strains were followed for only 6 months and this could have influenced the result. A detailed study of the genetic relationship of the strains within each of the 12 heterogenic sets revealed that oral isolates from the same patient were related each other (S_AB ≥ 80%) in all but one of the sets analysed. This analysis showed that the most common scenario in recurrent OPC was the maintenance of the same strain or the occurrence of strains with minor genetic variations, related to the prior strain. In 12 out of 14 patients the same DNA type showed increasing MIC for fluconazole over time, suggesting that the occurrence of strains with decreased susceptibility to fluconazole in this clinical setting is mainly due to the development of resistance in a previously susceptible strain. The fact that the strains within each of the sets individually analysed clustered at an S_AB ≥ 80% in all but one set, and that they remained clustered when compared with a collection of unrelated strains suggests that both susceptible and resistant isolates sequentially obtained from single patients may have a clonal origin. In the two cases (P3 and P13) in which DNA types of susceptible and resistant isolates did not match, we cannot be sure whether either the resistant strain was newly acquired or it emerged from a previously susceptible strain, because only one colony from each specimen culture was analysed and the chance of missing another strain also present in the specimen existed. However, on the basis of the degree of relationship of each of these two resistant isolates with their respective sets of strains (S_AB ≥ 80% in both cases), the second option is the most probable scenario.

The coexistence of diverse strains in the place of colonization/infection could not be stated in this study, due to the strategy employed to isolate strains from clinical samples, in which a single colony from the specimen culture was picked up. This approach probably led us to the isolation of the strain that predominated at the moment of sampling, as shown by the fact that the clinical response to fluconazole or ketoconazole of almost all active episodes registered in the study was determined by the fluconazole or ketoconazole MIC exhibited by the strain isolated during the episode. To assess the simultaneous carriage of multiple strains in HIV-infected OPC patients would require a more complex sampling procedure than employed here, which was beyond the scope of this study. Some other studies have been recently published that demonstrated this issue [^19–22]. All these studies were based on analysing several colonies (usually 5–10 colonies) from the same sampling. The genetic relationship among several strains from the same culture is analysed only in the work of Boerlin et al. [²²] and this fact may have influenced the different conclusions reached by these and other investigators. One of these studies [²¹] stated that in some cases different DNA subtypes and fluconazole MIC were observed amongst isolates obtained from a single culture and in other cases they emerged in subsequent cultures. These authors explained this phenomenon as a consequence of multiple exposures to different subtypes in an increasingly immunocompromised host, although they did not analyse the genetic distance between the strains classified as ‘different sub-types'. Boerlin et al. [²²] then conducted an study with seven patients, and using a sensitive DNA fingerprinting method demonstrated that the simultaneous strains isolated from the same sampling were highly related or clonal, as consequence of minor genetic changes that may take place within a single strain. They supported their result by the fact that only one C. albicans clone per patient was found in 20 out of 21 patients with sequential oral isolates. This second alternative seems to agree with our finding of minor genetic variation or no variation among sequential isolates obtained from single patients. The microevolution and shuffling processes that have been described between subgenotypes in sequential vaginal infections [²³] may account for genotypic variation in sequential isolates obtained from the patients in our study.

Regarding the susceptibility results, the vast majority of strains (63.7%) were susceptible to fluconazole and clinical response of these isolates to fluconazole treatment was usually positive, although there were a few exceptions. Only 21.2% of the strains were susceptible dependent upon dose and an even smaller percentage (15%) were fluconazole-resistant. Six of the seven episodes caused by a C. albicans strain with fluconazole MIC ≥ 16 μg/ml did not respond to treatment with this azole. A value of fluconazole MIC ≥ 16 μg/ml seems to predict quite accurately the failure of the treatment with standard dosage of fluconazole. Regarding cross-resistance, in all of the patients studied an increase in fluconazole MIC of four twofold dilutions or greater was detected among their sequential isolates, which was paralleled by a similar rise in ketoconazole MIC. Furthermore, of the 13 strains that were resistant to fluconazole (MIC ≥ 16 μg/ml), five isolates had ketoconazole MIC ≥ 0.06 μg/ml (defined as the breakpoint for resistant strains [²⁴]), six isolates had ketoconazole MIC that fell in the range of indeterminate susceptibility to ketoconazole (0.003–0.03 μg/ml), and the remaining two isolates had ketoconazole MIC of 0.0019 μg/ml (defined as the breakpoint for susceptible strains). The breakpoints of ketoconazole mentioned here were defined in a previous study [²⁴] and despite the fact that they are tentative, they may be useful until more detailed studies of the correlation between clinical response to ketoconazole and the MIC for this azole are performed. These data reflect the fact that different azole resistance patterns may exist, including resistance to all azole derivatives and resistance to only fluconazole [¹³].

In conclusion, mixed genotypes of C. albicans exist in the oral cavities of HIV-infected patients. However, they may well be related to each other, and the presence of a major genotype predominating over time together with minor genetic variants is not uncommon. We demonstrated that at least one of the previously susceptible strains was maintained over time and became less susceptible to azoles due to the selective pressure of prolonged antifungal treatment. This increase in fluconazole MIC was accompanied by an absence of clinical response.