Clinical Correlates of Antifungal Macrodilution Susceptibility Test Results for Non-AIDS Patients with Severe CandidaInfections Treated with Fluconazole

RESULTS

The mean age of the 32 patients was 60.8 years (range, 16 to 85 years), with 21 men and 11 women. The most frequent underlying disease was malignancy (13 patients). Other underlying diseases included neutropenia (two patients), end-stage renal disease (five patients), obstructive uropathy with nephrostomy (two patients), common bile duct stones and choledocholithotomy (three patients), perforated peptic ulcer and peritonitis (one patient), necrotizing and posttraumatic pancreatitis (three patients), diabetes mellitus and neurogenic bladder (five patients), cerebrovascular disease (two patients), motor neuron disease (one patient), hypovolemic shock (one patient), and benign prostate hypertrophy after transurethral resection (one patient). There were no rapidly fatal underlying diseases according to McCabe and Jackson's definition (13). Twenty-one patients had candidemia, one had postoperative peritonitis, three had peritonitis due to chronic ambulatory peritoneal dialysis (CAPD), four had pyelonephritis, one had pneumonia, one had pyothorax, and one had infective endocarditis (Table 1). In the case of Candida pneumonia, the patient had lung carcinoma and did not respond to broad-spectrum antimicrobial therapy. Culture of purulent bronchoalveolar lavage fluid produced only C. albicans, with colony counts of >10³/ml. Twenty-four patients had coexisting severe bacterial infections, and pneumonia was the most common. Gram-negative bacilli were the most frequent pathogens in the coexisting bacterial infections, including 13 nonfermentative gram-negative bacillus isolates, 11 enteric bacilli, 1Stenotrophomonas maltophilia isolate, 1 Haemophilus influenzae isolate, 5 Staphylococcus aureus isolates, 2Staphylococcus epidermidis isolates, 7 Enterococcus faecalis isolates, 1 group B Streptococcus isolate, 1Gardnerella vaginalis isolate, 1 Bacteroides fragilis isolate, and 1 atypical mycobacterium.

The duration range of fluconazole therapy was 7 to 28 days, with a mean of 15.3 days. A clinical response to fluconazole was obtained within 7 days of the start of therapy in 25 of 32 patients (78.1%), of whom 23 showed complete clearance and 2 had signs of relapse of clinical infection. Two candidemic patients had no clinical response to fluconazole after 7 days of therapy, with persistence of candidemia. In 4 of the 32 patients, coexisting severe bacterial infections due to multiresistant gram-negative bacilli were not controlled by appropriate antimicrobial therapy, and the patients died during fluconazole therapy. In one patient with infective endocarditis, which was confirmed by vegetations on a cardiac echogram, and persistent candidemia, surgical intervention was recommended, but the patient and her family refused. The patient died during fluconazole therapy. A case of pyelonephritis due to C. albicans in a patient with obstructive uropathy and nephrostomy was treated with 2 weeks of fluconazole therapy. Fever and pyuria subsided after fluconazole therapy. However, 3 days after fluconazole was stopped, mild fever developed again. Both blood culture and urine culture from nephrostomy were repeated, and both produced Candida tropicalis.

Of the 32 clinically significant Candida isolates, 17 wereC. albicans and 15 were non-albicans (see Table3). The 24-h MIC results (range, 0.125 to 64 μg/ml; geometric mean, 4.81 μg/ml) were calculated to be three to four times lower than the 48-h MICs by the macrodilution method. Because the correlation between 48-h MIC results and clinical-response results was significantly superior to that of the 24-h MIC results when NCCLS breakpoints of susceptibility were utilized according to the methods of prior studies (5, 6, 12, 14-17, 22, 23), 48-h MICs obtained by a macrodilution method may be adequate predictors of clinical outcome (Table 2). The 48-h MICs of these 32 isolates are shown in Table 3. C. albicans showed a wide range of MICs, from 0.25 to >64 μg/ml, with a geometric mean of 10.1 μg/ml. The MIC results of the reference strains ATCC 90028 and ATCC 6258 were all within the standard acceptable range of MICs. The MICs of 24 isolates in 24 cases were ≤8 μg/ml, and 19 of these cases were cured clinically. The clinical success rate was 19 of 24 (79%; 95% confidence interval [CI], 56 to 93%) for susceptible isolates interpreted according to the criteria of NCCLS standard M27-A (Table 2). The MICs of another six isolates from six patients were 16 to 32 μg/ml, and four of these cases were cured clinically. The clinical success rate was four of six (66%; 95% CI, 19 to 95%) for dose-dependent isolates. The MICs in the remaining two nonneutropenic evaluable cases were ≥64 μg/ml. The patients with these two candidemia cases, one due to C. albicans and one due to Candida guilliermondii, had underlying esophageal carcinoma after chemotherapy and radiotherapy and were not neutropenic and they failed to respond clinically. The success rate in this category was zero of two (0%; 95% CI, 0 to 84%) for resistant isolates (Table 2). In clinically cured patients, the range of 24-h MICs were 0.125 to 2.0 μg/ml, with a mean of 0.94 μg/ml. In cases of clinical failure, the range of 24-h MICs was 0.15 to 64 μg/ml, with a mean of 14.72 μg/ml.

DISCUSSION

In nine studies of oropharyngeal candidiasis in patients with AIDS, clinical correlation of the results of the NCCLS antifungal susceptibility test M27-A was high, ranging from 73 to 98% (5, 6, 13, 14-17, 22, 23). However, MIC data in deep-seated or hematogenous Candida infection is still infrequently reported (12, 15), except in animal studies (2, 4, 11, 21). Rex et al. recently reported an inverse correlation of MICs in patients with nonneutropenic candidemia, which is inconsistent with prior studies (18, 19). However, a majority of our cases of candidemia and deep-seated Candida infections in non-AIDS patients did not reveal inverse clinical correlation of MICs obtained by the same method. In the Rex study, 36 (from 19 patients) of the 100 isolates from 64 fully evaluable fluconazole-treated patients for which the fluconazole MICs were ≤16 μg/ml were associated with failure (18, 19). Four isolates for which the fluconazole MICs were ≥32 μg/ml were obtained from four patients who responded to initial fluconazole therapy (18, 19). The inverse clinical correlation of MICs in the Rex report may be due to some pitfalls in evaluation. Clinical failure in cases with MICs of ≤16 μg/ml may be due to failure to control coexisting bacterial infections, which are common and which are often not discovered in immunocompromised or debilitated patients with candidemia. Seventy-five percent of our patients had severe coexisting bacterial infections. Coexisting bacterial infections should be controlled with appropriate antibiotics before or during fluconazole therapy. In four of our patients, coexisting severe bacterial infections due to multiresistant gram-negative bacilli were not controlled by appropriate antibiotics, and the patients died during fluconazole therapy. These four cases were evaluated as clinical failures according to the criteria of prior studies. An inadequate duration of fluconazole therapy might also result in clinical failure, even with MICs of ≤16 μg/ml. Two weeks of fluconazole therapy for candidemia might not be adequate, even with MICs in the susceptible range, if the candidemia arises from peritonitis or intraperitoneal abscess secondary to intestinal perforation. Of the 21 patients with candidemia in our study, 2 had MICs of ≤8 μg/ml which arose from peritonitis and intraperitoneal abscess due to intestinal perforation. After 2 weeks of fluconazole therapy and appropriate antibiotics for coexisting mixed bacterial infections, fever subsided, repeated blood cultures became negative, and intraperitoneal abscess size diminished. Repeated cultures of peritoneal discharge from the intra-abdominal abscesses were still positive for the same Candida species. Thus, we believed this was due to inadequate fluconazole therapy and continued fluconazole for about two more weeks combined with appropriate antibiotics. Repeated cultures of discharge drained from the abscess site after another 2 weeks of fluconazole therapy became negative forCandida species. Repeated computerized tomograms of the abdomen later revealed complete resolution of the intraperitoneal abscesses. Two to 4 weeks of fluconazole therapy, depending on the origin of the candidemia, would be more reasonable for candidemia and would result in good correlation with the MICs.

Superinfection or reinfection by Candida species other than the initial Candida species is different from clinical failure for the initial Candida infection. In our study, we had a case of pyelonephritis due to C. albicans in a patient with obstructive uropathy and nephrostomy that was treated with 2 weeks of fluconazole therapy, and the patient developed reinfection due toC. tropicalis 3 days after fluconazole was stopped. This case should be considered a clinical cure for the original infection due to C. albicans, although it was complicated by reinfection by C. tropicalis.

Candidemia due to susceptible Candida isolates can be complicated with infective endocarditis requiring surgical intervention, but it may not be detected before death. In our study, we had a patient with peritonitis due to intestinal perforation with candidemia due to Candida glabrata secondary to central venous catheter infection. The catheter was removed, and the fever subsided before the culture result was known, and an antifungal agent was not given. However, 4 weeks later, fever and candidemia recurred due to infection by C. glabrata with a fluconazole MIC of ≤0.125 μg/ml. Candidemia persisted, and a cardiac echogram revealed prominent vegetations diagnostic of infective endocarditis. Intravenous fluconazole was given, and surgical intervention was recommended. However, the patient and her family refused surgery. The patient finally died during fluconazole therapy. This case was regarded as clinical failure according to the criteria of a prior study (19).

Although the incidence of coexisting bacterial infections is high in severe Candida infections, coexisting bacterial infection should not exclude clinical evaluation of antifungal agents, because increasingly potent antibiotics have become available in recent years which can control many severe bacterial infections. The clinical correlation of dose-dependent sensitive MICs of 16 to 32 μg/ml in our study was four of six (66.6%). In one case of Candida famata candidemia with a dose-dependent sensitive MIC (16 μg/ml), fever subsided after 3 days of parenteral fluconazole therapy, and 400 mg of fluconazole a day was given for a total of 20 days. However, 10 days later fever recurred, and the blood culture remained positive for C. famata. Thus, this case was evaluated as clinical failure according to the criteria of a prior study (19).

Of the 32 cases, two isolates from two nonneutropenic cases with MICs of ≥64 μg/ml failed to respond clinically and microbiologically. The clinical success rate of resistant isolates with MICs of ≥64 μg/ml was zero of two (0%; 95% CI, 0 to 84%). Although there was a lack of high-MIC isolates, the results were consistent and supportive of the NCCLS interpretive breakpoints. In summary, our data suggest that the 48-h MICs obtained by the macrodilution reference method of the NCCLS correlate well with the clinical response in candidemia and deep-seatedCandida infections treated with 400 mg of fluconazole per day.