Clinical characteristics of patients

We retrospectively analyzed 66 consecutive patients with hematologic disorders diagnosed with S. maltophilia bloodstream infection (BSI) at a single tertiary care center. The cohort had a median age of 41 years (range: 5–73 years), with a male predominance (66.7%, 44/66). The 30-day all-cause mortality rate was 34.8% (23/66).

The most common underlying hematological diseases were acute myeloid leukemia (AML) (51.5%, 34/66), followed by aplastic anemia (AA) (24.2%, 16/66) and acute lymphoblastic leukemia (ALL) (18.2%, 12/66). A majority of patients (56.1%, 37/66) had relapsed or refractory disease, and 27.3% (18/66) had undergone hematopoietic stem cell transplantation (HSCT), predominantly allogeneic (83.3%, 15/18). Severe neutropenia (absolute neutrophil count < 500 cells/µL) was present in 92.4% (61/66) of patients at the time of BSI diagnosis, with prolonged neutropenia (> 1 month) observed in 42.4% (28/66). Broad-spectrum antibiotic use within the preceding 30 days was nearly universal (93.9%, 62/66), primarily carbapenems (78.3%, 52/66).

The median time for positive blood culture of S. maltophilia was 21 h (range: 4h17min–59h18min), with no significant difference between survivors and non-survivors (p = 0.76). All positive cultures were obtained from aerobic bottles. This is in line with the pathogen’s obligate aerobic metabolism. Among patients with repeated testing following initial positivity (n = 32), 43.8% (14/32) had subsequent positive cultures. The intervals ranged from 1 to 15 days. Conversely, 56.2% (18/32) of follow-up cultures reverted to negative within 1–30 days. Repeated positivity occurred most frequently within 1–3 days (68% of cases), It suggests potential persistent bacteremia or inadequate early antimicrobial coverage.

Comparative analysis was done between survivors (n = 43) and non-survivors (n = 23). It showed significant differences in clinical and treatment-related factors (Table 1). Non-survivors had higher rates of pneumonia (91.3% vs. 44.2%, p = 0.002), shock (52.2% vs. 4.7%, p < 0.0001), and prolonged neutropenia (> 1 month) (78.3% vs. 9.3%, p < 0.0001). Inappropriate initial antibiotic therapy (43.5% vs. 14.0%, p = 0.0143) and ceftazidime resistance (30.4% vs. 2.3%, p = 0.019) were also closely linked to mortality. Notably, resistance to other antibiotics (e.g., TMP-SMX, levofloxacin) did not reach statistical significance, likely due to preserved susceptibility in our cohort. This finding is particularly significant given that ceftazidime is often used empirically in febrile neutropenia [12], and its resistance may delay the initiation of effective therapy, thereby exacerbating outcomes.

The link between prolonged neutropenia and mortality matches its role in harming innate immunity, thereby facilitating invasive infections. Similarly, shock reflects end-organ dysfunction due to uncontrolled systemic inflammation, which is particularly detrimental in immunocompromised hosts. The novel identification of ceftazidime resistance as a risk factor underscores the clinical impact of delayed appropriate therapy, as ceftazidime is commonly used empirically in febrile neutropenia.

Table 1 Univariate analysis of risk factors for mortalityDistribution of pathogens and antimicrobial susceptibility patterns

To further elucidate the source of S. maltophilia BSI and the complexity of infections in this immunocompromised population, we conducted a detailed analysis of the pathogen’s distribution in non-blood sites and identified co-infecting pathogens in other body sites. In 37 of 66 patients, S. maltophilia was detected at sites other than the blood, predominantly in respiratory specimens (94.6%, e.g., oropharyngeal specimen, sputum, or bronchoalveolar lavage fluid), reflecting its propensity to colonize the respiratory tract in patients with prolonged hospitalization or mechanical ventilation. Additionally, wound or tissue samples yielded S. maltophilia in 4 cases, particularly in patients with skin breakdown due to chemotherapy-induced mucositis.

Polymicrobial infections were prevalent (84.8%), underscoring the complexity of managing these patients. The most common co-infecting organisms included Gram-negative bacteria (75.8%, e.g., Escherichia coli [29.3%], Klebsiella pneumoniae), Gram-positive bacteria (28.8%, e.g., Staphylococcus aureus [31.6%], Enterococcus faecalis), and fungi (Candida spp., 2 cases). Notably, 7 cases of carbapenem-resistant Enterobacteriales (CRE), 2 cases of carbapenem-resistant Pseudomonas aeruginosa (CRPA), 2 cases of carbapenem-resistant Acinetobacter baumannii (CRAB), and 2 cases of methicillin-resistant Staphylococcus aureus (MRSA) were identified (Fig. 1).

Antimicrobial susceptibility testing revealed that 75.8% of S. maltophilia isolates were susceptible to trimethoprim-sulfamethoxazole (TMP-SMX), 87.8% to levofloxacin, 92.4% to minocycline, and 85.7% to ceftazidime. Multidrug resistance was observed in 11.1% of isolates. Prior to BSI diagnosis, 91% of patients had received broad-spectrum antibiotics, primarily carbapenems (78.3%), oxazolidinones (18.3%), and β-lactams (18.3%), with 78.8% receiving inappropriate empirical therapy (e.g., carbapenems). Following diagnosis, 16.7% were switched to TMP-SMX, 20.0% to levofloxacin, and only one patient received combination therapy (TMP-SMX plus levofloxacin). Despite adjustments, 24.2% of patients continued to receive inappropriate therapy.

These findings highlight the polymicrobial nature of S. maltophilia BSI and the critical need for susceptibility-guided therapy to address emerging resistance and optimize outcomes.

Fig. 1

The pie chart of co-infecting pathogens. (a) A pie chart of Gram-negative co-infecting pathogens detected in non-blood sites; (b) A pie chart of Gram-positive co-infecting pathogens detected in non-blood sites

Multivariate analysis of risk factors for mortality

To identify factors associated with mortality and infection outcomes in patients with S. maltophilia BSI, we performed Cox regression and logistic regression analyses. In Cox proportional hazards regression analysis, prolonged neutropenia (> 1 month) (HR = 1.57, 95% CI: 1.02–2.41, p = 0.04), ceftazidime resistance (HR = 1.95, 95% CI: 1.06–3.59, p = 0.03), and shock (HR = 2.44, 95% CI: 1.41–4.22, p < 0.01) were independently associated with increased mortality. Conversely, pneumonia (HR = 1.26, p = 0.32) and inappropriate antibiotic therapy (HR = 0.89, p = 0.63) did not reach statistical significance.

Logistic regression analysis yielded similar findings for prolonged neutropenia (OR = 2.20, 95% CI: 1.02–2.41, p = 0.02) and shock (OR = 3.44, 95% CI: 1.41–4.22, p = 0.01). However, ceftazidime resistance showed only a marginal association (OR = 1.76, p = 0.08), while pneumonia (OR = 1.58, p = 0.05) and inappropriate antibiotic therapy (OR = 0.88, p = 0.79) remained non-significant. (Table 2).

Table 2 Multivariate analysis of risk factors for mortalityROC analysis of risk factors for mortality in S. maltophilia BSI

To evaluate the predictive performance of various risk factors for mortality in patients with S. maltophilia BSI, we performed receiver operating characteristic (ROC) curve analysis (Fig. 2). The area under the curve (AUC) for prolonged neutropenia lasting more than one month was 0.8564 (95% CI: 0.7467–0.9661, p < 0.0001), indicating excellent predictive ability (Fig. 2A). Similarly, shock showed a significant predictive value with an AUC of 0.7376 (95% CI: 0.5982–0.8771, p = 0.0016) (Fig. 2B). The AUC for pneumonia was 0.7356 (95% CI: 0.6143–0.8569, p = 0.0017) (Fig. 2C), suggesting moderate predictive performance. In contrast, the predictive ability of inappropriate antibiotic therapy (AUC = 0.6810, 95% CI: 0.5370–0.8250, p = 0.016) and resistance to ceftazidime (AUC = 0.6405, 95% CI: 0.4908–0.7903, p = 0.0614) was relatively lower, with the latter not reaching statistical significance (Fig. 2D, E).

These results highlight the strong predictive value of prolonged neutropenia and shock for mortality in patients with S. maltophilia BSI, while pneumonia and inappropriate antibiotic therapy showed moderate predictive ability. The limited predictive performance of ceftazidime resistance may reflect its complex role in influencing outcomes.

Fig. 2

ROC analysis of risk factors for mortality in S. maltophilia BSI in pantients with hematologic disorders. (a) ROC analysis of prolonged neutropenia (> 1 month) for mortality prediction; (b) ROC analysis of shock for mortality prediction; (c) ROC analysis of pneumonia for mortality prediction (d) ROC analysis of inappropriate antibiotic therapy for mortality prediction (e) ROC analysis of ceftazidime resistance for mortality prediction

Univariate analysis of risk factors for progression from colonization to BSI

Comparative analysis between the colonization group and the colonization to BSI group showed significant differences in clinical and laboratory parameters. Patients who progressed to BSI exhibited higher levels of inflammatory markers, including CRP (median: 124.96 vs. 24.5 mg/L, p < 0.001) and PCT (median: 0.45 vs. 0.09 ng/mL, p = 0.002). Neutropenia (94.6% vs. 48.8%, OR = 7.0, 95% CI: 2.8–17.5, p < 0.001), chemotherapy (83.8% vs. 43.9%, OR = 3.1, 95% CI: 1.4–7.0, p = 0.003), and immunosuppressor use (94.6% vs. 53.7%, OR = 6.8, 95% CI: 2.5–18.4, p < 0.001) were closely related to progression to BSI. Inappropriate antibiotic use was also related to progression to BSI (OR = 2.98, 95% CI: 1.05–8.45, p = 0.039) (Table 3).

Table 3 Univariate analysis of risk factors for progression to BSIMultivariate analysis of risk factors for progression from colonization to BSI

Multivariate logistic regression confirmed that neutropenia (OR = 6.85, 95% CI: 2.71–17.31, p < 0.001), immunosuppressor use (OR = 7.20, 95% CI: 2.62–19.80, p < 0.001), and elevated CRP (> 100 mg/L, OR = 4.11, 95% CI: 1.83–9.23, p = 0.001) were independent predictors of progression to BSI. COX regression analysis, despite limited time-to-event data, suggested a higher hazard ratio for inappropriate antibiotic use (HR = 2.45, 95% CI: 1.12–5.38, p = 0.025) (Table 4).

Table 4 Multivariate analysis of risk factors for progression to BSIROC analysis of risk factors for progression from colonization to BSI

ROC curve analysis was performed to evaluate the predictive value of various clinical factors for bloodstream infection. Neutropenia demonstrated a significant discriminative ability (AUC = 0.729, 95% CI: 0.616–0.842, p = 0.005), followed by chemotherapy history (AUC = 0.699, 95% CI: 0.582–0.817, p = 0.003) and immunosuppressor use (AUC = 0.705, 95% CI: 0.588–0.821, p = 0.002). Among inflammatory markers, CRP > 100 mg/L showed the highest predictive performance (AUC = 0.732, 95% CI: 0.618–0.846, p < 0.001), whereas PCT > 0.5 ng/mL had a modest but statistically significant association (AUC = 0.632, 95% CI: 0.508–0.756, p = 0.045). These findings suggest that both immune suppression-related factors (neutropenia, chemotherapy, and immunosuppressor use) and elevated CRP levels may serve as valuable indicators for identifying patients at higher risk of bloodstream infection (Fig. 3).

Fig. 3

ROC Analysis of Risk Factors for Progression from Colonization to S. maltophilia BSI in pantients with hematologic disorders. (a) ROC analysis of neutropenia for predicting progression to BSI; (b) ROC analysis of chemotherapy history for predicting progression to BSI; (c) ROC analysis of immunosuppressor use for predicting progression to BSI; (d) ROC analysis of CRP > 100 mg/L for predicting progression to BSI; (e) ROC analysis of PCT > 0.5 ng/mL for predicting progression to BSI