Authors: T-touch Anawilkul, Sarunyou Chusri, Siripen Kanchanasuwan, Sorawit Chittrakarn
Categories: Research, Corynebacterium bacteremia, Bloodstream infection, Antimicrobial resistance, Vancomycin, Central venous catheter, Hematologic malignancy
Source: BMC Infectious Diseases
Authors: T-touch Anawilkul, Sarunyou Chusri, Siripen Kanchanasuwan, Sorawit Chittrakarn
Corynebacterium species are often regarded as blood culture contaminants, but they are increasingly recognized as true pathogens. However, their clinical significance, resistance patterns, and prognostic factors remain poorly defined. This study aimed to describe the clinical characteristics, antimicrobial resistance profiles, and factors associated with true bacteremia compared with isolates considered contaminants, as well as all-cause mortality.
We conducted a 15-year retrospective cohort study of adults with at least one Corynebacterium-positive blood culture at a tertiary-care university hospital in southern Thailand (2009–2024). Species identification was performed using MALDI-TOF MS from 2017 onward. True bacteremia was defined by clinical and microbiologic criteria. Multivariable logistic regression was used to identify independent predictors of true bacteremia, and Cox proportional hazards regression was applied to identify predictors of 30-day mortality.
Among the 437 patients, 192 (44%) were classified as having true bacteremia. Independent predictors included hematologic malignancy (aOR 3.75; 95% CI: 1.40–10.73), central venous catheter use (aOR 2.39; 95% CI: 1.49–3.84), non-dialysis dependent chronic kidney disease (aOR 2.83; 95% CI: 1.11–7.66), and hypertension (aOR 1.92; 95% CI: 1.20–3.08). The 30-day all-cause mortality rate was significantly greater in true bacteremia patients (43.8% vs. 23.3%, p < 0.001). In the Cox proportional hazards model, hematologic malignancy (aHR 2.32, 95% CI 1.35–4.00) and central venous catheter use (aHR 2.34, 95% CI 1.46–3.74) were identified as independent predictors of 30-day mortality. C. striatum was the most common species (29.5%), highly resistant to penicillin (96.7%) and clindamycin (98.4%), but universally susceptible to vancomycin
Nearly half of Corynebacterium-positive blood cultures represented true infections with high mortality. Vancomycin remains the most reliable empiric therapy. Species-level identification and susceptibility testing are essential for accurate diagnosis and management
Not applicable.
Corynebacterium species are aerobic, gram-positive, non-spore forming rods [1]. that are increasingly recognized as clinically relevant pathogens capable of causing severe bloodstream infections, particularly in immunocompromised individuals and those with invasive devices such as central venous catheters [2–4]. Although Corynebacterium spp. are traditionally considered contaminants when isolated from blood cultures, recent studies have demonstrated their pathogenic potential, emphasizing the importance of accurately distinguishing true bacteremia from contamination [5–9]. Misclassification of true Corynebacterium bacteremia as contamination can significantly worsen patient outcomes, leading to delayed appropriate treatment, increased morbidity, and elevated mortality risk. Previous studies have reported a 30-day all-cause mortality rate approaching 35% for true infections [10], underscoring the critical need for accurate diagnosis.
Despite the increasing recognition of Corynebacterium spp. as true pathogens, the clinical characteristics, antimicrobial resistance profiles, and prognostic factors associated with Corynebacterium bacteremia remain poorly defined in many healthcare settings. Most existing research, primarily from high-income countries, has identified predictors of true bacteremia, including shorter blood culture positivity times, the presence of central venous catheters, hematologic malignancies, and species such as C.striatum and C.jeikeium [6, 10–12]. However, comprehensive data from diverse healthcare environments are limited, highlighting important knowledge gaps in the global epidemiology and resistance patterns of these organisms.
Understanding local epidemiology and species-specific antimicrobial resistance is essential due to the emergence of multidrug-resistant Corynebacterium strains worldwide [10]. C.striatum, in particular, has demonstrated substantial resistance to several commonly used antibiotics, complicating both empiric therapy and antimicrobial stewardship efforts. Despite these concerns, regionally representative studies remain scarce.
In this study, we aimed to characterize the clinical and microbiological features of Corynebacterium bacteremia in hospitalized adults over a 15-year period. We focused on identifying factors associated with true infection, evaluating outcomes, and describing resistance patterns to inform diagnostic and therapeutic decision-making.
This retrospective cohort study was conducted at Songklanagarind Hospital, a 900-bed university-affiliated tertiary care center in southern Thailand. The hospital serves as the main referral center for southern Thailand and houses a comprehensive cancer center and hematopoietic stem cell transplantation unit. Approximately 20–25% of inpatient admissions are oncology or hematology patients. Adult patients (aged ≥ 18 years) with at least one blood culture was positive for Corynebacterium spp. between January 1, 2009, and December 31, 2024, were eligible for inclusion.
We included only the first episode of Corynebacterium bacteremia per patient. Patients were excluded if they had incomplete medical records, had only one blood culture obtained, or were transferred to other hospitals before outcome data could be collected. For patients who were readmitted with the same infection episode, only the initial admission was included. This approach was chosen to avoid duplication of baseline risk factors and clinical variables, which could otherwise bias the regression analysis.
Demographic, clinical, and microbiological data were extracted from the hospital information system and microbiology laboratory records. From 2017 onward, species identification was performed via matrix-assisted laser desorption ionization-time of flight mass spectrometry (MALDI-TOF MS; Bruker Microflex LT, Bruker Daltonics, Bremen, Germany). For each isolation, the highest score generated by the Bruker Biotyper software was used for reporting. Prior to MALDI-TOF MS implementation, isolates were identified as Corynebacterium spp. by colony morphology, Gram stain (Gram-positive pleomorphic rods with coryneform arrangement), catalase positivity, and selected biochemical tests such as urease production, carbohydrate utilization, and nitrate reduction.
True bacteremia was defined by any of the (1) growth in both bottles of a single blood culture set. In our hospital, a single set of blood cultures typically consists of two aerobic bottles collected simultaneously from different venipuncture sites; (2) one positive blood culture plus culture from another sterile site; (3) a positive blood culture with systemic signs of infection, for which antimicrobial therapy was initiated by the attending physician (defined as the responsible staff physician directly caring for the patient, such as an internist, intensivist, or infectious disease specialist, depending on the clinical setting); or (4) persistently positive cultures drawn >24 h apart. Isolates that did not meet these criteria were considered contaminants.
Independent variables included demographic data (age, sex), comorbidities (diabetes mellitus, chronic kidney disease, hematologic malignancies, immunosuppression, etc.), the presence of a central venous catheter, the Pitt bacteremia score, the National Early Warning Score (NEWS), and prior antibiotic usage. The dependent variables included the classification of bacteremia (true vs. contamination), 14-day all-cause mortality, 30-day all-cause mortality, intrahospital all-cause mortality, and length of hospital stay.
Antibiotic susceptibility was tested by disk diffusion. The routine panel included penicillin, ciprofloxacin, imipenem, vancomycin, and clindamycin. Susceptibility interpretation followed EUCAST guidelines. Because EUCAST does not provide Corynebacterium-specific breakpoints for imipenem, our laboratory standard-operating procedure interprets imipenem results using the EUCAST Bacillus spp. (non-anthracis) breakpoints, reflecting our local practice of testing imipenem for Gram-positive bacteria.
Descriptive statistics were used to summarize the baseline characteristics. Categorical variables were compared chi-square test or Fisher’s exact, and continuous variables were compared via the Wilcoxon rank-sum test. Multivariable logistic regression was used to identify independent predictors of true bacteremia. Variables with p < 0.10 in univariable analysis and those considered clinically relevant were initially included in the models, with final model selection guided by the Akaike information criterion (AIC). Kaplan–Meier curves were constructed to illustrate mortality over time. For inferential survival analysis, Cox proportional hazards regression was applied to identify independent predictors of 30-day mortality. Proportional hazards assumptions for the Cox model were tested using Schoenfeld residuals. All analyses were performed using RStudio (version 2024.12.1 + 563).
A total of 437 patients with positive blood cultures for Corynebacterium species were included. Among these patients, 192 (44%) were classified as having true bacteremia (Table 1). According to our predefined criteria, 83 patients fulfilled criterion 1 (growth in both bottles of a single blood culture set), 21 patients fulfilled criterion 2 (one positive blood culture plus growth from another sterile site), 88 patients fulfilled criterion 3 (one of two bottles positive with systemic signs of infection and antibiotic therapy initiated), and none fulfilled criterion 4 (persistently positive cultures > 24 h apart).
Table 1Demographic data and outcomes of patients with Corynebacterium spp. isolated from blood culturesCharacteristicTrue Bacteremia(n = 192)Contamination (n = 245)p-value Demographics Age (years), median (IQR)66.4 (55.5–77.7)65.4 (52.0–78.1)0.809Male sex, n (%)104 (54.2)144 (58.8)0.385 Comorbidities Hypertension89 (46.4)84 (34.3)0.014Diabetes mellitus49 (25.5)50 (20.4)0.249Dyslipidemia57 (29.7)46 (18.8)0.011Cardiovascular disease^a^44 (22.9)49 (20.0)0.534Chronic pulmonary disease^b^18 (9.4)20 (8.2)0.783Liver cirrhosis15 (7.8)8 (3.3)0.058Chronic Kidney Disease (CKD)Non dialysis dependent CKD16 (8.3)8 (3.3)0.018Dialysis-dependent CKD21 (10.9)17 (6.9)0.072Hemodialysis2012Peritoneal dialysis15Bone marrow failure disorders15 (7.8)6 (2.4)0.017- Myelodysplastic syndrome6 (3.1)6 (2.4)0.893- Aplastic anemia9 (4.7)0 (0.0)0.001Hematologic malignancy47 (24.5)12 (4.9)< 0.001- Lymphoma18 (9.4)1 (0.4)< 0.001- Leukemia31 (16.1)9 (3.7)< 0.001- Multiple myeloma1 (0.5)2 (0.8)1HIV infection2 (1.0)1 (0.4)0.585Autoimmune diseases13 (6.8)16 (6.5)1Solid cancers52 (27.1)65 (26.5)0.984Implanted devices (prosthetic joints/cardiac devices)6 (3.1)3 (1.2)0.190 Clinical conditions NEWS, median (IQR)8.00 (6.00, 10.00)3.00 (2.00, 5.00)< 0.001Pitt score, median (IQR)2.00 (0.00, 4.00)0.00 (0.00, 0.00)< 0.001Chemotherapy7 (3.6)7 (2.9)0.849Systemic corticosteroid use73 (38.0)45 (18.4)< 0.001Neutropenia24 (14.2)8 (3.4)< 0.001Central venous catheter use87 (45.3)54 (22.0)< 0.001Duration of central venous catheter (days), median (IQR)3.0 (1.0–8.5)4.0 (1.0–11.8)0.356Hospital days before blood culture positive, median (IQR)9.5 (0.0–20.3)1.0 (0.0–13.0)< 0.001Prior antibiotic use155 (80.7)154 (62.9)< 0.001Co-infection ^c^32 (16.7)35 (14.3)0.581 Outcome Length of hospital stay (days), median (IQR)25.5 (12.0–42.3)17.0 (7.0–33.0)< 0.00114-day mortality63 (32.8)41 (16.7)< 0.001In-hospital mortality82 (42.7)54 (22.0)< 0.00130-day mortality84 (43.8)57 (23.3)< 0.001Abbreviations: NEWS, National Early Warning Score; CVC, Central venous catheter^a^ Cardiovascular disease includes ischemic heart disease and heart failure^b^ Chronic pulmonary disease includes COPD and asthma^c^ Co-infection was defined as concurrent growth of other pathogens in the same blood culture setData are presented as the median (IQR) or number (%). Comparisons were made using the Wilcoxon rank-sum test for continuous variables and the chi-square test or Fisher’s exact test for categorical variables
The sources of Corynebacterium isolates among patients with true bacteremia are shown in Table 2. Nearly half of the patients (n = 84, 43.8%) had no identifiable infectious focus (primary bacteremia). Catheter-related bloodstream infection was the most common source (n = 71, 37.0%), followed by urinary tract infection (n = 16, 8.3%), pus specimens (n = 8, 4.2%), and surgical site infections (n = 4, 2.1%). Sterile body fluids accounted for 9 cases (4.7%), including ascitic fluid (n = 2), pleural effusions (n = 3), JP drain fluid (n = 2), hematoma collections (n = 1), and joint fluid (n = 1).
Table 2Sources of Corynebacterium isolates among patients with true bacteremiaSource of isolaten (% of true bacteremia, n = 192)Clinical DiagnosesNo obvious focus84 (43.8%)Primary bacteremiaCentral venous catheter71 (37.0%)Catheter-related bloodstream infectionUrine16 (8.3%)Urinary tract infectionPus8 (4.2%)Wound/soft tissue abscessTissue4 (2.1%)Surgical site infection / Soft tissue infectionSterile body fluids9 (4.7%) • Ascitic fluid2Peritonitis • Pleural fluid3Infected pleural effusion • JP drain fluid2Infected intra-abdominal collection • Hematoma collection1Infected hematoma • Joint fluid1Septic arthritis
The median age was similar in the true bacteremia and contamination groups (66.4 vs. 65.4 years, p = 0.809). Compared with patients in the contamination group, those with true bacteremia had significantly higher rates of hypertension (46.4% vs. 34.3%, p = 0.014), dyslipidemia (29.7% vs. 18.8%, p = 0.011) and nondialysis-dependent chronic kidney disease (8.3% vs. 3.3%, p = 0.018). Bone marrow failure disorders were also more common in the true bacteremia group (7.8% vs. 2.4%; p = 0.017), particularly aplastic anemia (4.7% vs. 0.0%; p = 0.001). Hematologic malignancies, including lymphoma (9.4% vs. 0.4%) and leukemia (16.1% vs. 3.7%), were significantly associated with true bacteremia (p < 0.001 for both).
Patients with true bacteremia had significantly higher National Early Warning Scores (median 8 vs. 3, p < 0.001) and Pitt bacteremia scores (median 2 vs. 0, p < 0.001). They were also more likely to have received systemic corticosteroids (38.0% vs. 18.4%, p < 0.001), experience neutropenia (14.2% vs. 3.4%, p < 0.001), have prior antibiotic exposure (80.7% vs. 62.9%, p < 0.001), and have central venous catheter use (45.3% vs. 22.0%, p < 0.001). The duration of hospitalization prior to positive blood culture was also longer in this group (median 9.5 vs. 1 day; p < 0.001).
No significant differences were observed in the prevalence of diabetes mellitus, liver disease, HIV infection, autoimmune disease, or solid tumors. Implanted devices, including cardiac devices and orthopedic prostheses, were uncommon and did not differ significantly between groups (3.1% vs. 1.2%, p = 0.190).
Among 192 patients with true Corynebacterium bacteremia, 184 (95.8%) received empirical antibiotic treatment, while 8 (4.2%) did not. The most frequently prescribed empirical agents were meropenem (46.4%), piperacillin–tazobactam (44.8%), and vancomycin (39.1%). Other commonly used antibiotics included ceftriaxone (19.8%), trimethoprim–sulfamethoxazole (17.2%), imipenem–cilastatin (8.3%), ceftazidime (7.8%), ciprofloxacin (7.3%), and clindamycin (6.8%). Smaller numbers of patients received other antibiotics, such as amoxicillin–clavulanate, ampicillin–sulbactam, colistin, or linezolid (< 5% each).
Multivariable logistic regression analysis (Table 3) revealed significant independent predictors of true Corynebacterium hypertension (adjusted OR 1.92, 95% CI: 1.20–3.08, p = 0.007), CKD non-dialysis dependent (adjusted OR 2.83, 95% CI: 1.11–7.66, p = 0.033), hematologic malignancy (adjusted OR 3.75, 95% CI: 1.40–10.73, p = 0.010), and central venous catheter presence (adjusted OR 2.39, 95% CI: 1.49–3.84, p < 0.001). Variables with p < 0.10 in univariable analysis and those considered clinically relevant were initially included; however, the final model was selected using the Akaike Information Criterion (AIC). As a result, some variables (e.g., prior antibiotic exposure, Pitt bacteremia score, NEWS) were excluded during model selection.
Table 3Multivariate logistic regression identifying independent predictors of true Corynebacterium bacteremiaVariableAdjusted OR95% CIp-valueHypertension1.921.20–3.08 0.007 CKD, non dialysis dependent2.831.11–7.66 0.033 CKD, dialysis-dependent1.560.73–3.320.246Hematologic malignancy3.751.40–10.73 0.01 Liver fibrosis/cirrhosis2.450.97–6.600.064Central venous catheter presence2.391.49–3.84 < 0.001 Systemic corticosteroid use1.70.99–2.910.053Neutropenia2.170.76–6.320.145Abbreviations: CKD, chronic kidney disease; OR, odds ratio; CI, confidence interval
Among the 437 Corynebacterium-positive blood cultures, species-level identification was available for 241 cases (55.1%). The most frequently identified species was Corynebacterium striatum (129 isolates, 29.5% of all), with a high proportion of true bacteremia (73.6%). Other commonly identified species included C. afermentans (8.0%), C.amycolatum (1.4%) and C.jeikeium (1.4%). Notably, C.jeikeium, although rare, had a true bacteremia rate of 50.0%, while C.amycolatum accounted for a high proportion of true bacteremia at 83.3%.
Rare species were also identified. C.confusum was isolated from a 70-year-old man with progressive head-and-neck cancer and tracheostomy who presented with sepsis and an infected tumor. C.saperdae was isolated from a 62-year-old woman with tonsillar carcinoma on chemoradiation and a ventriculoperitoneal shunt, who presented with febrile illness and improved after empiric piperacillin–tazobactam.
Antimicrobial susceptibility testing revealed universal susceptibility to vancomycin (100%) across all the tested species. However, resistance to penicillin and clindamycin was widespread, especially among C. striatum, with only 3.3% and 1.6% of the isolates being susceptible, respectively. C. amycolatum and C.jeikeium showed moderate susceptibility to penicillin (66.7%), but clindamycin resistance remained high. Imipenem and ciprofloxacin demonstrated variable activities. C. striatum exhibited low susceptibility to both agents (4.8% for imipenem and 1.6% for ciprofloxacin), whereas other species, such as C. amycolatum and C. jeikeium, retained higher susceptibility.
The remaining 196 isolates (44.9%) were identified only at the genus level (Corynebacterium spp.), with a lower proportion of true bacteremia (24.5%), reflecting potential contamination or a lack of recognition as pathogens prior to MALDI-TOF implementation (Table 4).
Table 4Distribution of Corynebacterium species, proportion of true bacteremia cases and susceptibility profileSpeciesn (% of total)% True BacteremiaCiprofloxacin(% susceptible)Imipenem(% susceptible)Penicillin(% susceptible)Vancomycin(% susceptible)Clindamycin(% susceptible) C. striatum 129 (29.5%)73.60%1.6% (2/124)4.8% (6/124)3.3% (4/122)100% (126/126)1.6% (2/126) C. afermentans 35 (8.0%)57.10%6.1% (2/33)39.4% (13/33)9.7% (3/31)100% (33/33)0% (0/32) C. amycolatum 6 (1.4%)83.30%16.7% (1/6)100% (6/6)66.7% (4/6)100% (6/6)60% (3/5) C. jeikeium 6 (1.4%)50.00%0% (0/6)16.7% (1/6)0% (0/6)100% (6/6)0% (0/6) C. aurimucosum 5 (1.1%)20.00%40% (2/5)100% (5/5)80% (4/5)100% (5/5)20% (1/5) C. mucifaciens 4 (0.9%)50.00%50% (2/4)100% (4/4)75% (3/4)100% (4/4)50% (2/4) C. coyleae 3 (0.7%)0.00%66.7% (2/3)100% (3/3)66.7% (2/3)100% (3/3)33.3% (1/3) C. minutissimum 2 (0.5%)50.00%50% (1/2)100% (2/2)100% (2/2)100% (2/2)0% (0/2) C. riegelii 2 (0.5%)100%0% (0/2)100% (2/2)100% (2/2)100% (2/2)50% (1/2) C. confusum 1 (0.2%)100%100% (1/1)100% (1/1)100% (1/1)100% (1/1)0% (0/1) C. simulans 1 (0.2%)100%100% (1/1)100% (1/1)100% (1/1)100% (1/1)0% (0/1) C. saperdae 1 (0.2%)100%100% (1/1)0% (0/1)0% (0/1)100% (1/1)0% (0/1) C. propinquum 1 (0.2%)100%100% (1/1)100% (1/1)100% (1/1)100% (1/1)100% (1/1)Corynebacterium spp.241 (55.1%)24.50%–––––Note: Values represent the percentage of susceptible isolates (number susceptible/total tested) Imipenem susceptibility interpreted using EUCAST Bacillus spp. (non-anthracis) breakpoints
Patients with true Corynebacterium bacteremia had significantly worse clinical outcomes those with contaminant isolates. The median length of hospital stay was longer in the true bacteremia group (25.5 days; IQR, 12–42.3) than in the contamination group (17 days; IQR, 7–33; p < 0.001). Similarly, the 30-day all-cause mortality rate was significantly greater among patients with true bacteremia (43.8%) than among those with contamination (23.3%; p < 0.001).
Kaplan–Meier curves illustrated lower survival probabilities in the true bacteremia group (Fig. 1). In the Cox proportional hazards model restricted to patients with true bacteremia, hematologic malignancy (aHR 2.32, 95% CI 1.35–4.00, p = 0.002) and central venous catheter use (aHR 2.34, 95% CI 1.46–3.74, p < 0.001) were identified as independent predictors of 30-day mortality. The proportional hazards assumption was satisfied (global Schoenfeld test p = 0.49) (Table 5).
Fig. 1Kaplan-meier survival curves comparing true Corynebacterium bacteremia and contamination. Kaplan‒Meier survival curve comparing patients with true Corynebacterium bacteremia (red line) versus contamination (blue line) over a 90-day period. Inferential survival analysis was performed using Cox proportional hazards regression
Table 5Cox proportion hazard ratio identifying predictors of 30-day all-cause mortality in true Corynebacterium bacteremia patientsVariableAdjusted HR95% CIp-valueC. afermentans ^a^1.250.50–3.130.6C. striatum ^a^1.350.82–2.240.2Bone marrow failure disorders ^b^1.560.84–2.910.2Hematologic malignancy2.321.35–4.00 0.002 Central venous catheter presence2.341.46–3.74 < 0.001 Systemic corticosteroid use1.470.87–2.480.2Abbreviations: HR, Hazard ratio; CI, confidence interval^a^ Reference category is other Corynebacterium species^b^ Includes aplastic anemia and myelodysplastic syndromeNote Variables with p < 0.05 were considered statistically significant
In this 15-year retrospective study of 437 patients with Corynebacterium-positive blood cultures, nearly half (44%) were classified as having true bacteremia. This proportion is consistent with previous studies, which reported true bacteremia rates ranging from 44% to 48% [6, 11, 12]. These findings emphasize that the isolation of Corynebacterium spp. from blood cultures should prompt clinicians to carefully assess for true infection rather than dismissing the organism as a contaminant. Patients with true infections had significantly higher 30-day all-cause mortality (43.8%) compared to those with contaminant isolates (23.3%). These findings highlight the critical importance of accurately distinguishing true bloodstream infections from contamination in clinical practice, as misclassification can lead to delayed or inappropriate treatment with potentially fatal outcomes [5, 13, 14]. Our mortality rate was higher than the 35% reported previously [10], likely reflecting an older patient population with greater illness severity at presentation, as indicated by higher Pitt bacteremia scores.
Several clinical factors, including hypertension, non-dialysis dependent chronic kidney disease, hematologic malignancy, and the presence of a central venous catheter are independently associated with true bacteremia. These findings align with previous studies from Japan, which have reported similar associations between true Corynebacterium bacteremia, hematologic malignancy, and catheter use [6, 11]. Notably, the effect of hypertension appeared to be modified by CKD in patients with non–dialysis-dependent CKD, hypertension increased the likelihood of true bacteremia, whereas this effect was absent in dialysis-dependent CKD. This suggests that vascular and renal dysfunction may contribute to host susceptibility. Kang et al. further noted that patients with severe underlying diseases are more likely to develop clinically significant C. striatum bacteremia [15], potentially because of increased healthcare exposure, vascular manipulation, or skin barrier defects and innate immune dysfunction.
Although Corynebacterium can cause infections in prosthetic material such as cardiac valves and orthopedic joints [16], implanted devices were uncommon in our cohort and did not differ significantly between true bacteremia and contaminant cases. This likely reflects the oncology- and hematology-dominant patient population at our institution. Moreover, Corynebacterium prosthetic joint infections typically present as chronic infections and only rarely manifest as bacteremia [17].
With respect to microbiology, Corynebacterium striatum was the most frequently identified species in our study and exhibited high resistance rates to several antibiotics, while remaining universally susceptible to vancomycin. This resistance profile aligns with global reports describing the emergence of multidrug-resistant C. striatum as an important nosocomial pathogen [6, 10, 18–21]. Interestingly, the second most common species in our cohort was C. afermentans, which contrasts with prior studies where C. jeikeium was more commonly reported as the second leading isolate and was associated with a greater proportion of true bacteremia [6, 22]. This discrepancy may reflect regional differences in species distribution, patient populations, or underlying host factors such as immunosuppression or device exposure. Although C.confusum and C.saperdae are rarely reported as pathogens, C.confusum has previously been described in human infections, including foot infections and bacteremia [23, 24]. Our findings suggest that such rare species may occasionally play a pathogenic role in immunocompromised hosts.
The empirical antibiotics administered in our cohort frequently included broad-spectrum agents with limited intrinsic activity against Corynebacterium, such as meropenem and piperacillin–tazobactam. This prescribing pattern reflects real-world practice in our setting, where broad Gram-negative coverage is prioritized for febrile, immunocompromised patients with suspected sepsis. Only about 40% of patients received vancomycin empirically, despite its universal activity against our isolates. This underscores the challenge of early recognition of high-risk patients who may benefit from empiric Gram-positive coverage. Cox regression analysis confirmed hematologic malignancy and central venous catheter use as independent predictors of 30-day mortality, highlighting the vulnerability of immunocompromised hosts and the pathogenic potential of Corynebacterium in the presence of intravascular devices.
At our institution, species-level identification using MALDI-TOF mass spectrometry has been available since 2017. Prior to its implementation, the rate of recognizing Corynebacterium as a true pathogen was substantially lower, with only 24.5% of isolates classified as true bacteremia. This trend may be explained by several first, patients today are more severely immunocompromised, with a higher incidence of hematologic malignancies and central venous catheter use; second, there is increased clinical awareness of Corynebacterium as a potential pathogen rather than a contaminant; and third, species-level identification likely contributes to improved diagnostic accuracy and confidence in clinical interpretation.
In healthcare settings where MALDI-TOF MS or advanced microbiological diagnostics are not readily available, the differentiation between true infection and contamination remains particularly challenging. Our findings suggest that certain clinical features such as the presence of hematologic malignancy, central venous catheter use, and high severity score can serve as useful surrogate markers for identifying patients at high risk of true Corynebacterium bacteremia. In such cases, empiric coverage with vancomycin may be justified while awaiting culture and susceptibility results, especially in immunocompromised patients. This pragmatic approach may help guide early and appropriate therapy, reduce mortality risk, and improve clinical outcomes in resource-constrained environments.
Our study has several strengths. To our knowledge, this represents one of the largest single-center cohorts of Corynebacterium bloodstream infections reported to date. The long study period and robust statistical analyses allowed us to examine key factors associated with infection classification and patient outcomes. Since 2017, MALDI-TOF MS has enabled more reliable species identification and antimicrobial susceptibility profiling.
However, this study has several limitations. As a retrospective analysis from a single tertiary-care center, the findings may not be generalizable to other settings with different patient populations or laboratory practices. The absence of blood culture time-to-positivity data limited our ability to assess this useful diagnostic criterion for distinguishing true bacteremia from contamination. In addition, species-level identification was unavailable prior to 2017, which may have led to underestimation of certain trends and restricted species-specific analyses. Importantly, the observed associations between true Corynebacterium bacteremia and adverse outcomes such as prolonged hospitalization and higher mortality should not be interpreted as causative. Patients with hematologic malignancies, neutropenia, or higher severity scores were more likely to be treated and therefore classified as true bacteremia, which could have biased outcomes toward sicker patients independent of infection. Furthermore, no consensus guideline exists for defining true Corynebacterium bacteremia, and our classification relied on predefined criteria adapted from prior studies [5, 6, 10, 12]. The inclusion of patients with one of two positive blood culture bottles plus systemic signs of infection may have introduced further bias, since sicker patients were more likely to receive antibiotics and thus be classified as true bacteremia, even when contamination could not be entirely excluded. Because the decision to initiate therapy rested with the attending physician, variability in clinical judgment may also have influenced classification. Finally, imipenem susceptibility was interpreted using EUCAST Bacillus spp. (non-anthracis) breakpoints, reflecting institutional practice in the absence of Corynebacterium-specific criteria, but these results may not fully reflect true carbapenem susceptibility and should be interpreted cautiously.
In conclusion, nearly half of the Corynebacterium isolates from blood cultures represented true bloodstream infections and were associated with high all-cause mortality, particularly among patients with hematologic malignancies or central venous catheter use. C.striatum was the most frequently identified species and exhibited extensive multidrug resistance, although all the isolates remained susceptible to vancomycin. These findings underscore the importance of timely species-level identification and susceptibility testing, which can guide appropriate therapy and improve patient outcomes. Clinicians should maintain a high index of suspicion for true infection when Corynebacterium is isolated in high-risk hosts rather than presuming contamination.