Memórias do Instituto Oswaldo Cruz, Vol. 104, No. 6, September, 2009, pp. 905-913
CAS MartinsI, III; LMD FariaI; MC SouzaIII; TCF CamelloIII; E VelascoI; R Hirata JrIII; LCS ThulerII; AL Mattos-GuaraldiIII, +
Nacional do Câncer, Ministério da Saúde, Rio de Janeiro,
Financial support: INCA, FAPERJ, CNPq, CAPES, SR-2/UERJ, PRONEX
Received 15 May
Code Number: oc09190
During a five-year period, 932 clinical isolates from cancer patients treated in a Brazilian reference centre were identified as corynebacteria; 86% of the cultures came from patients who had been clinically and microbiologically classified as infected and 77.1% of these patients had been hospitalised (71.1% from surgical wards). The adult solid tumour was the most common underlying malignant disease (66.7%). The univariate and multivariate analyses showed that hospitalised patients had a six-fold greater risk (OR = 5.5, 95% CI = 1.15-26.30 p = 0.033) related to 30-day mortality. The predominant species were Corynebacterium amycolatum (44.7%), Corynebacterium minutissimum (18.3%) and Corynebacterium pseudodiphtheriticum (8.5%). The upper urinary tracts, surgical wounds, lower respiratory tracts, ulcerated tumours and indwelling venous catheters were the most frequent sources of C. amycolatum strains. Corynebacterium jeikeium infection occurred primarily in neutropenic patients who have used venous catheters, while infection caused by C. amycolatum and other species emerged mainly in patients with solid tumours.
Key words: Corynebacterium amycolatum - Corynebacterium infection - cancer
Corynebacterium species have recently been recognised as important pathogens that infect immunocompromised patients. Despite the increase in the number of reports of severe infection caused by non-diphtheria corynebacteria, these organisms are still frequently dismissed as contaminants. Corynebacterium spp have been cited with an increased frequency as pathogens of nosocomial infections associated with septicaemias, endocarditis, infections of surgical wounds, prostheses and infections related to the central venous catheter. Geographical variations in the frequency of isolated species and variations in the natural and acquired antimicrobial resistance have been described (Osterwalder et al. 1986, Rozdzinski et al. 1991, van der Lelie et al. 1995).
Recent advances in Corynebacterium identification have shown that the genus exhibits considerable taxonomic complexity and that the phenotypic markers used in the past for its identification can be ambiguous (Barreau et al. 1993, Brandenburg et al. 1996, Oteo et al. 2001, Camello et al. 2003, Meyer & Reboli 2005, Otsuka et al. 2006). When these infections are linked to multi-resistant species, they are difficult to treat (Bodey 1995, Funke et al. 1996, Camello et al. 2003, Funke & Bernard 2003). Multiple resistance to antimicrobial agents has been described among some species (Lagrou et al. 1998, Zalas et al. 2004); for example, mutations in the gyrase genes are determinants of the resistance to quinolones in Corynebacterium striatum and Corynebacterium amycolatum strains (Sierra et al. 2005).
In Rio de Janeiro, Brazil, recent studies conducted at a teaching hospital showed that 68% of the clinical isolates of corynebacteria species corresponded to Corynebacterium pseudodiphtheriticum, C. amycolatum, Corynebacterium propinquum and Corynebacterium minutissimum. The urinary tract and the venous access were the sites most commonly affected. In the blood and the respiratory tract, there was a predominance of C. pseudodiphtheriticum and C. propinquum, while Corynebacterium xerosis and C. amycolatum were the most commonly observed species in central venous catheters. C. propinquum and C. minutissimum were the most frequently observed species in surgical wounds. Data indicated the occurrence of multi-resistant phenotypes and the possibility of severe infections due to C. pseudodiphtheriticum, a pathogen usually overlooked in emerging countries (Camello et al. 2003, 2009).
Infection has been recognised as one of the major obstacles to the successful management of patients with malignant diseases. Few studies have described non-diphtheria corynebacteria species as emerging pathogens in patients with solid tumours, with the exception of those that highlighted the importance of Corynebacterium jeikeium (Rozdzinski et al. 1991, van der Lelie et al. 1995). Vancomycin has been demonstrated to be the primary treatment option for C. jeikeium. The choice of using glycopeptides to empirically treat patients with neutropenic fever and persistent fever has been criticised since the advent of the vancomycin-resistant Enterococcus and the heterogeneous vancomycin-intermediate Staphylococcus aureus (Furtado et al. 2006). The new options for therapy are also not completely safe for the treatment of gram positive life-threatening infections partially due to the emergence of microbial resistance (Boucher et al. 2000, Dobbs et al. 2006, Schoen et al. 2009).
The aim of this descriptive study was to assess the microbiological and clinical aspects as well as factors related to 30-day mortality in cancer patients with corynebacteriosis.
PATIENTS, MATERIALS AND METHODS
Study setting - In this descriptive study, we retrospectively reviewed clinical and microbiological data over a five-year period (from January 2000-December 2004). The patient information originated from one of the units (Cancer Hospital I with 200 beds, Cancer Hospital II with 90 beds and Cancer Hospital III with 60 beds) of the National Cancer Institute (INCA) in Rio de Janeiro, Brazil.
The analysis included bacteriological data from 88,541 cultures and 1,100 irregular gram-positive rods isolated from cancer patients. The operational units included the Laboratory of Microbiology and the Hospital Infection Control Committee (HICC) at the INCA and the Laboratory of Diphtheria and Corynebacteria, State University of Rio de Janeiro (UERJ), Brazil.
During the period of the study, there were no changes in the technical team who processed the material for bacteriological cultures or in the medical team who assisted the patients.
Ethics - This paper was submitted and approved by the Ethical Research Committee at INCA (CEP 008/06) and complies with the Brazilian Government's Ethical Guidelines for research involving human beings (resolution of the National Health Council/Ministry of Health).
Diagnostic measures - The study analysed the clinical characteristics of 315 patients with Corynebacterium isolates who had malignant diseases or underwent bone marrow transplantation in the last two years of the study. Data assessment was performed on the basis of a medical records review of patients.
The hospitalised patients were monitored by at least one member of the HICC as a part of the antimicrobial vigilance routine through daily microbiology laboratory charts. Ambulatory patients were identified by the dressing nurses committee, catheter ambulatory (both for children and adults) and bacteriological charts.
In addition to the physicians' experiences of treating patients with cancer in INCA, the diagnosis of bacterial infection was staged according to the Centers for Disease Control and Prevention (CDC) classification (CDC 1992).
The patients with positive cultures were interpreted as infected when these cultures were derived from a normally sterile site associated with a febrile illness. Alternatively, patients with positive cultures were interpreted as infected when a Corynebacterium spp was isolated from two or more non-sterile sites in which there was a suspicion of infection and the physician considered it clinically significant to immediately start a specific antimicrobial therapy. A positive urinary culture was considered as significant in the presence of local (dysuria, polyuria) or systemic signs of infection; pyuria was only required in non-neutropenic patients. Tracheal secretions in intubated patients, sputum or tracheal aspirations in non-intubated patients were considered as positive if they were associated with clinical or radiological signs that indicated an infection. Protected sampling was not performed in bronchoscopy (Garner et al. 1988, Hughes et al. 1996, Berghmans et al. 2003).
Bacterial strains and identification - We retrospectively reviewed 932 strains identified as Corynebacterium spp that were recovered from representative clinical sites of cancer patients with signs and symptoms of bacterial infection. Corynebacterium-like colonies were selected for further identification when they were grown in any quantity from normally sterile body fluid or when they were isolated in significant numbers or in pure culture from other specimens obtained at clinical sites in which infection was suspected (Funke & Bernard 2003). All clinical samples yielding more than three organisms were regarded as contaminated and discarded (Thomson 2007).
The Maki's semi-quantitative method was used to distinguish infection (> 15 colonies) from contamination of catheter-tips (Maki et al. 1977). For quantitative BAL fluid cultures, a colony count > 103 colony-forming units (CFU) mL-1 of potential pathogens was considered positive. Isolation of two or three species of microorganisms was regarded as a polymicrobial infection for catheter tips and the lower respiratory tract, respectively. Microorganisms were identified from the urine cultures in cystine lactose electrolytes deficient agar (CLED; Merck, Darmstadt, Germany) and were considered to be potential pathogens when the growth was > 104 CFU mL-1 as the only isolate or > 105 CFU mL-1 as the predominant isolate; > 103 CFU mL-1, in cases of nephropathies, was also considered a potential pathogen. Blood cultures were always obtained in pairs, wherein at least one of the samples was collected through the central venous catheter, if present. Blood specimens were inoculated in Bactec Plus anaerobicD aerobic vials and processed in a Bactec 9240 continuous-monitoring system (Becton-Dickinson Microbiology System, Cockeysville, MD, USA). Other clinical specimens were inoculated onto a Columbia agar base with the addition of 5% sheep's blood and incubated at 37ºC in 3-5% CO2 atmosphere and monitored for 72 h.
In addition to Gram staining, colonial morphology, pigmentation and haemolysis, Corynebacterium-like colonies were characterised using the API-Coryne System (BioMérrieux, Lyon, France) (Freney et al. 1991). Microorganisms were also submitted to the following conventional biochemical assays: catalase, pyrazinamidase, lypophilic activities, motility, nitrate reduction, hydrolysis of urea and esculin, acidification of glucose, maltose, sucrose, mannitol and xylose, as well as oxidation-fermentation and CAMP reaction tests (Camello et al. 2003, Funke & Bernard 2003).
Susceptibility testing - Antimicrobial susceptibility testing was performed by the Kirby Bauer's disk diffusion method, using an inoculum of 0.5 McFarland standard (150 x 106 CFU mL-1 by direct colonial suspension), adjusted for optical density at λ = 550 nm (Vitek colorimeter Durham, NC, USA). The plates were incubated at 35ºC in ambient air for 24 h and reconfirmed at 48 h in a cation-adjusted Mueller-Hinton Agar with the addition of 5% sheep's blood (Funke et al. 1997). As there is not yet a defined standard by the Clinical and Laboratory Standards Institute for interpreting the results of disk diffusion tests (CLSI 2007), we used the breakpoint for penicillin suggested for Staphylococcus.
For the other antimicrobial agents, we used the breakpoints for other microorganisms but not Haemophilus spp or Neisseria gonorrhoeae, which had been validated by previous studies (Martinez-Martinez et al. 1995, Weiss et al. 1996, Zinkernagel et al. 1996, Funke et al. 1997).
The intermediate results were included as resistant. Microorganisms were tested against 15 antimicrobial agents, according to the clinical criteria for empirical therapy in patients with underlying malignancy and infection. The in vitro antimicrobial susceptibility test associated with the analysis of clinical relevance used the interpretation criteria suggested by the Sanford Guide (Gilbert et al. 2006): (+) usually clinically effective or > 60% susceptible, (±) clinical trials lacking or 30-60% susceptible and (0) not clinically effective or < 30% susceptible.
Statistical analysis - The variables were compared using a Pearson Chi square or Fisher exact test. For statistical significance, the value was established at p < 0.05. A logistic regression model was developed to identify the variables independently associated with the 30-day mortality. This model included every variable that showed a statistically significant association (p < 0.05) in the univariate analysis. The softwares used for the statistical analysis were Epi-Info 2000 for Windows, version 3.3.2 and SPSS, version 14.
During a five-year period (from January 2000-December 2004), the clinical microbiology laboratory at INCA conducted 88,541 cultures, of which 25,173 (28.4%) were positive (n = 36,199). Gram-positive rods represented 1,436 (4%) of the clinical isolates. The genus Corynebacterium corresponded with 932 (84.7%) of the total (1,100 strains) number of irregular gram-positive rods (Table I).
The Corynebacterium species isolated from cancer patients presenting signs and symptoms of infection are displayed in Table II. Most of the isolates (44.7%) were recognised as C. amycolatum, followed by C. minutissimum (18.3%) and C. pseudodiphtheriticum (8.5%). C. jeikeium was the sixth in frequency among the isolates (4.7%). The species Corynebacterium ulcerans and Corynebacterium pseudotuberculosis, which are usually associated with zoonoses, were absent or observed in a very low incidence in our population, respectively.
Taking into account the retrospective analysis of our study from 2000-2004, the isolation of C. amycolatum/xerosis increased from 10.4-72.8%. On the other hand, C. minutissimum decreased from 31.3-1.3%; the percentage of Corynebacterium spp (unidentified species) went from 11.5-0% and C. striatum from 12.5-0.8%. C. xerosis comprised 20 isolates in 2000 and six in 2001 and was absent from new isolates in the subsequent years.
During the second phase of the study (from January 2003-December 2004), 10 Corynebacterium species were observed in 425 clinical isolates obtained from 315 cancer patients. C. amycolatum was the most frequent clinical isolate, followed by C. pseudodiphtheriticum, C. jeikeium and C. minutissimum (Table III). When only bloodstream infections were considered, catheter-related infections were frequently observed. C. amycolatum was widely distributed in all topographies. The upper urinary tract, surgical wounds, lower respiratory tract, ulcerated tumours and indwelling venous catheters were the most frequent sources of C. amycolatum strains. C. pseudodiphtheriticum strains were mainly isolated from the lower and upper respiratory tract and C. jeikeium from the intravenous sites and skin lesions. The antimicrobial spectra of microorganisms are exhibited in Table IV.
Clinical data of the 315 cancer patients with fever and/or other signs of infection were retrieved. The epidemiological characteristics, the topographies that were involved and the factors that predisposed cancer patients to corynebacteria infection are depicted in Table V. The findings were interpreted as infection in 86% of patients and specific treatments were initiated. Corynebacterium strains were observed as the only microorganism in 46% of the cultures. Bacterial colonisation was confirmed in 44 cases (14%), 35 of which originated in the respiratory tract (25 rhinosinusal, 3 lower respiratory and 7 upper respiratory), four cases were in central venous catheters (Maki method negative), two cases were in skin/tumours and one case was in a surgical wound; two cases were isolated from the digestive tract.
The main underlying malignant diseases were 66.7% adult solid tumours, followed by tumours of the central nervous system and paediatric solid tumours (10.8% and 9.2%, respectively). The patients from surgical wards (71.1%) were previously submitted to head and neck (25.1%), gynaecological (13.3%) and abdominal-pelvic (9.5%) surgeries. Surgical wound infections comprised 32.7% of these patients. The patients from clinical wards were from pediatrics (28.9%), oncology (12.1%) and haematology (7.3%). The majority of patients had been hospitalised (77.1%) and exposed to a hospital environment for a long term (median = 13 days).
The clinical-epidemiological characteristics of 315 patients with 30-day mortality involved a uni and multivariate analysis of risk factors for corynebacteria-associated infection (Table IV).
Identification of non-diphtheria bacteria at the species level is often problematic. Recent advances in identification have shown that the genus exhibits considerable taxonomic complexity and the phenotypic markers that have been used in the past for its identification can be ambiguous. Even when sent to a reference laboratory, 30-50% of coryneform bacteria isolates cannot be reliably identified at the species level. Consequently, there is a low rate of identification from clinical isolates (Schiff et al. 2004).
Variations in the occurrence of Corynebacterium species during the course of the study were probably due to improvements in the taxonomy and laboratory diagnosis of the genus Corynebacterium. The substantial increase in the number of C. amycolatum isolates was partially due to the disappearance of C. xerosis and the significant reduction of C. striatum and C. minutissimum, which reflects the progress in the diagnosis of Corynebacterium species and consequently, the variability in antimicrobial sensitivity patterns. The former Corynebacterium CDC F-2 and CDC I-2 groups have been reclassified as C. amycolatum.
Due to imprecise diagnosis at the species level, data from the first three years (507 strains) were not included in the analysis of the bacteriological and clinical characteristics of cancer patients with corynebacteriosis.
Many non-diphtheria corynebacteria-caused access infections can be effectively treated by antibiotics and local care. Susceptibility testing of corynebacteria is highly recommended to establish a specific therapy (CLSI 2007). Except for the unvarying activity of vancomycin against corynebacteria, the variability in resistance to other classes of antimicrobial agents emphasises the need for the continuous surveillance of their resistance patterns. Resistance to oxacillin was observed for all corynebacteria species isolated from cancer patients at INCA. The data emphasise the relevance of susceptibility testing of Corynebacterium isolates during the diagnosis and treatment of infections in cancer patients.
A growing number of reports have confirmed the importance of C. amycolatum in the aetiology of a variety of infectious processes. In the reviewed literature, many C. amycolatum clinical isolates were at first identified in different laboratories as C. xerosis (Funke et al. 1996) and C. minutissimum (Zinkernagel et al. 1996). After the mid-1990s, with the advent of the taxonomical characterisation of C. amycolatum, this species has been described as a causal agent of infections with high morbidity and mortality rates. Multi-resistant C. amycolatum strains were isolated from patients with septicaemia (Berner et al. 1997, de Miguel et al. 1999, Camello et al. 2003), septic arthritis after vascular transplants (Clarke et al. 1999), cardiac electrode implantation (Vaneechoutte et al. 1998), infections in vascular prostheses and open fractures (von Graevenitz et al. 1998), peritonitis in patients undergoing peritoneal dialysis (Chiu et al. 2005), infectious endocarditis accompanied by aorto-atrial fistula (Daniels et al. 2003), nosocomial endocarditis (Knox & Holmes 2002) and septicaemia in leukemic patients (de Miguel et al. 1999). These reports drew attention to the clinical significance of C. amycolatum strains in different clinical materials (Esteban et al. 1999).
The clinical characteristics of the patients with favourable outcomes and documented corynebacteriosis could not be compared with those patients who died (controls). Due to the nature of the present research, which was a retrospective study, we could not investigate the pathogenicity of the isolates in relation to 30-day mortality in cancer patients. For the same reason, it was impossible to establish prognostic factors in these patients considering their descriptive characters. Another limitation of this study was the exclusion of isolates from 2000-2002 in the complete analysis due to major changes in taxonomical classification. Nevertheless, the sampling procedures adopted in this study allowed us to detect some patterns and characteristics of the population under investigation.
The univariate and multivariate analyses showed that hospitalised patients had a six-fold greater risk (OR = 5.5, 95% CI = 1.15-26.30 p = 0.033) related to 30-day mortality. Also of statistical significance were patients bedridden for longer than 50% of the day with neoplastic disease in progress or diabetes mellitus.
The study revealed that corynebacteria are increasingly being recognised as a cause of infections in cancer patients. These organisms have been underreported, but they may account for approximately 4% of the cases of infection in patients presenting solid tumours. The data reflects a tendency towards infection or colonisation by Corynebacterium in cancer patients who have undergone surgical procedures and a long subsequent hospitalisation period. From the study data, we also emphasise the need for the rigorous identification of Corynebacterium species isolates from different sites, especially of invasive strains in patients with clinical conditions of persistent fever and no other attributable site of infection. C. jeikeium infection occurs primarily in neutropenic patients who have used venous catheters (Rozdzinski et al. 1991), while C. amycolatum infection appears mainly in patients with solid tumours.
Copyright 2009 - Instituto Oswaldo Cruz - Fiocruz
The following images related to this document are available:
Photo images[oc09190t2.jpg] [oc09190t5.jpg] [oc09190t6.jpg] [oc09190t4.jpg] [oc09190t3.jpg] [oc09190t1.jpg]