Predictors of mortality in bacteremic cancer patients: retrospective analysis of 64 deaths occurring among 262 bacteremic episodes.

A total of 262 bacteremic episodes were observed in cancer patients in a single cancer institution during the last 7 years, and the recorded outcome was death in 65. The 65 patients who died (24.8% overall mortality) were divided retrospectively into two subgroups: (a) those who died of underlying disease with bacteremia (45 cases, 16.9% crude mortality) and (b) those who died of bacteremia (20 patients, 7.7% attributable mortality). Comparison of several risk factors in subgroups of patients who achieved a cure (197 cases) and of those who died and whose deaths were attributable (20 cases) revealed six risk factors that were associated with attributable mortality: (1) chemotherapy-induced neutropenia (P < 0.03), (2) Acinetobacter/Stenotrophomonas spp. bacteremias (P < 0.001), (3) liver failure (P < 0.001), (4) inappropriate therapy (P < 0.0001), (5) organ complications (P < 0.003) and (6) multiresistant organisms (P < 0.001). Enterococci and Pseudomonas aeruginosa, surprisingly, were found more frequently in those who died of an underlying disease with bacteremia than among patients who were cured (17.6% vs 7.6%, P < 0.05 and 29.1% vs 13.8%, P < 0.02). Those who died of infection had higher numbers of positive blood cultures, with 2.05 per episode, than did those who died of underlying disease with bacteremia (1.82) or those who were cured (1.51). Other risk factors, such as underlying disease, type of chemotherapy, origin of bacteremia, age, and catheters did not predict either overall or attributable mortality within the study group.

Nosocomial breakthrough fungaemia during antifungal prophylaxis or empirical antifungal therapy in 41 cancer patients receiving antineoplastic chemotherapy: analysis of aetiology risk factors and outcome.

Forty-one episodes of breakthrough fungaemia occurring over a 7.5 year period in the National and St Elizabeth's Cancer Institutes in Bratislava, Slovakia, were analysed. Five of them occurred during prophylaxis with fluconazole (one Torulopsis glabrata, one Hansenula anomala, two Candida krusei and one Candida parapsilosis), ten with itraconazole (three Trichosporon pullulans, one Trichosporon beigelii, one Cryptococcus laurentii, three Candida albicans and two T. glabrata), 11 during prophylaxis with ketoconazole (one Candida norvegenesis, one C. parapsilosis, one C. krusei, one Candida tropicalis, five C. albicans, one Candida stellatoidea and one C. laurentii and 15 during empirical therapy with amphotericin B (ten C. albicans, two T. beigelii and three Candida lusitaniae). The most frequent risk factors for breakthrough fungaemia were neutropenia, previous therapy with multiple antibiotics and recent catheter insertion. Comparing these episodes with 38 non-breakthrough fungaemias (appearing at the same institute in the same period) differences in certain risk factors were noted: breakthrough fungaemias were more frequently observed in patients with acute leukaemia (39.0% vs 5.2%, P < 0.001), mucositis (34.2% vs 13.1%, P < 0.05), prophylaxis with quinolones (58.5% vs 15.8%, P < 0.0001) and catheter-associated infections (29.3% vs 2.6%, P < 0.003). In this subgroup overall mortality (36.6% vs 28.8%) or early attributable mortality (22.0% vs 23.6%) were not significantly different.

Eight-year surveillance of non-albicans Candida spp. in an oncology department prior to and after fluconazole had been introduced into antifungal prophylaxis.

From 1989 until 1996, during the last 8 years, the proportion of Candida (C.) krusei, and other non-albicans Candida spp. isolated from surveillance cultures and from sterile body sites, was analyzed among 13,758 admissions in a National Cancer Institute. During these admissions a total of 9,042 isolates were prospectively collected from surveillance cultures, and 126 from blood cultures. The proportion of C. krusei among all organisms was 12.7% to 16.5% in 1989 through 1991, i.e., before fluconazole was introduced into prophylactic protocols. After the introduction of fluconazole into prophylaxis in acute leukemia in 1992 the incidence of C. krusei was 7.9% to 8.6% during 1994 to 1996. After 5 years of using this drug for prophylaxis, the incidence of C. krusei was lower than before this drug was introduced in our institute. Among yeasts, the most frequently isolated pathogen was still Candida albicans (72.2% of all isolated fungal organisms). Among molds, Aspergillus spp. was the most frequently isolated agent. Analyzing the etiology of proven fungal infections (fungemias) confirmed by positive blood cultures, C. albicans was the most common causative organism in 53.8% of cases. The incidence of fungemia due to Torulopsis (C.) glabrata and C. krusei before and after fluconazole introduction did not change. Of 126 organisms isolated from blood cultures, there was no increase in T. (C.) glabrata or C. krusei after introduction of fluconazole for prophylaxis and therapy, and the quoted 6.4% of fungemic episodes remained stable with an incidence of 1 fungemia/year since 1991. The proportion of C. krusei and C. glabrata among Candida spp. was decreasing in our center between 1989 and 1996. Also, the proportion of non-albicans Candida spp. among isolates decreased from 25.7% in 1990 to 11.9% in 1996.

Analysis of 553 episodes of monomicrobial bacteraemia in cancer patients: any association between risk factors and outcome to particular pathogen?

Relationships between aetiology, various risk factors (such as neutropenia, catheter insertion, endoscopy, therapy with corticosteroids, therapeutic use of antimicrobials, antibiotic prophylaxis, source of infection), symptomatology and outcome were studied in 553 monomicrobial bacteraemic episodes in cancer patients observed within 7 years at the National Cancer Institute of the Slovak Republic. The ratio of gram-positive to gram-negative bacteraemia was 1:1 (43.5% vs 43.8%), and yeasts caused 7.2% of monomicrobial episodes. The highest mortality was associated with Pseudomonas aeruginosa (19.2%), non-albicans Candida yeasts (25%) and Bacteroides fragilis (22.6%). Independent risk factors for particular pathogens were investigated by a computerized logistic regression model. The only independent risk factor for staphylococcal and enterococcal bacteraemia was vascular catheter insertion (OR = 1.95 and 2.05, CI = 95%, P = 0.035 and 0.044, respectively). However, there were no independent specific risk significant factors for viridans streptococcal bacteraemia and bacteraemia due to Enterobacteriaceae or Ps. aeruginosa. Neutropenia was found to be an independent predictor for development of Acinetobacter spp. bacteraemia (OR = 3.84, CI = 95%, P = 0.044). Prior therapy with third-generation cephalosporines was a predictive, independent risk factor for the development of fungaemia (OR = 1.99, CI = 95%, P = 0.028) but not of enterococcal bacteraemia. We also did not observe any association between prior therapy with imipenem and Stenotrophomonas maltophilia bacteraemias. Multivariate analysis confirmed that fungaemia may be independently associated with higher mortality than bacteraemia caused by Enterobacteriaceae and staphylococci. However, the mortality of fungaemia was statistically no different from that of Ps. aeruginosa, Stenotrophomonas spp. and viridans streptococci bacteraemias.

Bacteremia in cancer patients with solid tumors undergoing chemotherapy versus surgery: risk factors, etiology and outcome in 276 patients.

Etiology, risk factors, outcome and complications of bacteremia in 276 patients with solid tumors were analyzed. A group of 78 patients with solid tumors and surgical therapy only was compared with 172 patients with solid tumors who were treated with chemotherapy only. The most frequently observed risk factors of bacteremia in patients after surgery was urinary catheter insertion, wound as source of bacteremia, age > 60, staphylococci, enterococci and Enterobacteriaceae as etiologic agents. In comparison, viridans streptococci and Pseudomonas aeruginosa as etiologic agents as well as vascular catheters were significantly more frequently found in those treated with chemotherapy only. Patients with bacteremia after surgery only had a lower incidence of septic shock (6.4 vs. 16.9%, P < 0.03) and also lower mortality (5.6 vs. 14.9%, P < 0.04) attributable to shock than patients being treated for solid tumors with chemotherapy only.

Resistance pattern of 2816 isolates isolated from 17631 blood cultures and etiology of bacteremia and fungemia in a single cancer institution.

The resistance pattern of 2816 isolates from 17631 blood cultures and the etiology of isolates causing bacteremia and fungemia among 14591 admissions were investigated in an 80-bed single cancer institute during seven years (1990-1996) under the same empiric therapeutic antibiotic policy but with different prophylactic strategies. No change was found in the proportion of Gram-positive versus Gram-negative bacteria isolated from bacteremias (70% vs. 30%) during the past seven years. Furthermore, the proportion of coagulase-negative staphylococci and enterococci was about the same before and after the introduction of ofloxacin in prophylaxis. However, the proportion of Pseudomonas aeruginosa and Stenotrophomonas maltophilia causing bacteremia increased. There was no increase in Candida krusei and Candida glabrata after the introduction of fluconazole into our prophylactic regimen in 1992. Penicillin-resistance in viridans streptococci increased after penicillin was introduced into prophylaxis in acute leukemia in 1993. Until 1995 no quinolone-resistant Enterobacteriaceae were observed. Susceptibility to quinolones did not significantly change within the past seven years in Enterobacteriaceae after their introduction to prophylaxis in 1991, but Pseudomonas aeruginosa decreased from 90 to 58.2%. Glycopeptide resistance in enterococci and staphylococci was minimal in the observed period (0.9-4.3%).

Fungemia in cancer patients undergoing chemotherapy versus surgery: risk factors, etiology and outcome.

26 patients with fungemia and cancer treated with chemotherapy (group A) were compared to 25 patients with fungemia and cancer treated with surgery (group B), to assess differences in etiology, risk factors and outcome. Candida albicans was responsible for 42% of fungemias in group A, and for 92% of fungemias in group B (p < 0.005). Breakthrough fungemia occurring during antifungal prophylaxis appeared in 46.6% of group A vs 12% of group B (p < 0.02). There was significant difference in outcome between the groups: 20% of patients after surgery vs 7.7% of those after chemotherapy died from fungemia (p < 0.04). Most common risk factors recorded in both groups were catheter insertion and previous therapy with broad spectrum antibiotics.

Do vancomycin serum levels predict failures of vancomycin therapy or nephrotoxicity in cancer patients?

The purpose of this study was to determine if patients with high vancomycin (VAN) serum levels experience more toxicity than underdosed patients with lower (VAN) levels, and whether low VAN serum levels cause therapeutic failures in patients with gram-positive bacteremia. In 198 cancer patients trough and peak serum levels of VAN were measured. Acute toxicity (Red Man syndrome) appeared in 3 patients (1.5%). Patients previously or currently treated with other nephrotoxic compounds (134 patients) presented the same incidence of nephrotoxicity as those receiving VAN for the first time in monotherapy (64 patients). VAN did not increase the toxicity when patients were dosed simultaneously or previously with aminoglycosides or amphotericin B. Our second observation, when studying serum levels in our 198 patients was that high VAN trough serum levels (trough > 15 microg/mL) were associated with significantly more nephrotoxicity (33.3% vs. 11.1%, P < 0.03) than low levels in the subgroups of either pretreated patients or unpretreated with other nephrotoxic drugs. None of 198 patients who had trough levels below 15 microg/mL had peak levels exceeding 40 microg/mL. This suggests that only serum monitoring of trough levels may predict nephrotoxicity. A case control study was conducted to compare a group of 22 VAN failures with 22 successfully treated patients matched in underlying disease and neutropenia who were treated in the same period, under the same antibiotic policy, at the same cancer center, for gram-positive bacteremia. Persisting, enterococcal, or mixed enterococcal plus staphylococcal bacteremia were the only statistically significant risk factors which predicted therapy failure in cancer patients. Neither peak nor trough VAN serum levels predicted failure or cure of gram-positive bacteremia in cancer patients.