Screening Bacteriuria Using the Sysmex UF-5000 during Intravesical Bacillus Calmette–Guérin Therapy for Bladder Cancer

Abstract

Intravesical Bacillus Calmette–Guérin (BCG) therapy for bladder cancer requires screening for urinary tract infection during treatment, but conventional urinalysis cannot reliably distinguish bacteriuria from BCG-induced inflammatory cystitis. This study evaluated the ability of the Sysmex UF-5000 urine flow cytometer to detect significant bacteriuria and predict Gram stainability during intravesical BCG. This retrospective study included 145 patients who underwent intravesical BCG therapy after transurethral resection of bladder tumors (TURBT). Urine flow cytometry and cultures were performed at baseline, after TURBT, and weekly during BCG instillation. Diagnostic performance was assessed at the patient level over the BCG induction course, using urine cultures (≥10⁵ CFU/mL) as the reference standard. Among 145 patients (median age 71 years; 88.3% male), 37 (25.5%) developed significant bacteriuria. A UF-5000 cutoff of >100 bacteria/µL identified bacteriuria with a sensitivity of 94.6% (95% CI, 82.3-98.5) and specificity of 68.5% (95% CI, 59.3-76.5), and a binary-classifier AUC of 0.816 (95% CI, 0.727-0.905); >1,000 bacteria/µL yielded 62.2% sensitivity, 96.3% specificity, and AUC 0.792. For Gram-stain prediction, sensitivity/specificity reached 83.3%/92.2% (Gram-negative) and 81.0%/84.2% (Gram-positive). The Sysmex UF-5000 shows promise as a rapid screening tool for bacteriuria and Gram-stain prediction during intravesical BCG therapy. Larger prospective studies are needed to validate optimal cutoff values for routine clinical application.

INTRODUCTION

Bladder cancer is the ninth most common cancer worldwide,1 and approximately 75% of patients present with non–muscle-invasive bladder cancer (NMIBC).2 For patients with NMIBC at high risk of recurrence or progression, intravesical Bacillus Calmette–Guérin (BCG) following transurethral resection of the bladder tumor (TURBT) is recommended and reduces recurrence compared with TURBT alone.2,3

Intravesical BCG is generally initiated at least 2 weeks after TURBT once gross hematuria and significant adverse symptoms have resolved, and it is commonly administered as a 6-week induction course.2,4 Local adverse events are frequent and include urinary frequency, urgency, and dysuria.5 Similar irritative urinary symptoms can also occur after TURBT alone,6 making it clinically challenging to distinguish symptomatic bacterial UTI from treatment-related inflammatory cystitis.

Because a urine culture requires 2-3 days to yield results, decisions regarding the continuation of BCG are often based on clinical judgment and routine urinalysis. Inappropriately interrupting BCG due to misclassification of UTI may adversely affect oncologic outcomes.7 Accordingly, rapid and reliable screening methods for significant bacteriuria are needed to support timely decision-making during BCG instillation.

Fluorescence flow cytometry–based urine particle analyzers, such as the Sysmex UF-1000i (Sysmex, Kobe, Japan), can rapidly quantify bacteria and other formed elements and have been evaluated for UTI screening.8,9 Scattergram-based approaches have also been explored for the early prediction of treatment responses in uncomplicated UTI.10 However, earlier UF platforms have limited accuracy for bacterial classification and do not reliably provide clinically actionable bacterial grouping information, including Gram discrimination.11

The latest-generation urine flow cytometer, the Sysmex UF-5000, incorporates an advanced data analysis algorithm that enables improved bacterial enumeration and automated discrimination between GN and GP bacteria.12 In patients undergoing intravesical BCG therapy—where the distinction between bacterial cystitis and BCG-related inflammatory symptoms is consequential—rapid identification of significant bacteriuria and early estimation of Gram status may facilitate appropriate antimicrobial selection and timely resumption of therapy.

Accordingly, this study evaluated the performance of the Sysmex UF-5000 system in detecting significant bacteriuria and predicting Gram status by analyzing urinary bacterial counts and Gram-flag outputs in patients receiving intravesical BCG treatment.

MATERIALS AND METHODS

1. Study population and design

A retrospective analysis was conducted of 145 patients diagnosed with NMIBC who underwent intravesical Bacillus Calmette–Guérin (BCG) therapy following transurethral resection of a bladder tumor (TURBT). Intravesical BCG treatment was scheduled for weekly administration for six consecutive weeks (induction course) in patients with stage Ta-T1 tumors or carcinoma in situ, as determined after surgery.

Urinalysis using the Sysmex UF-5000 system, urinary nitrite testing, urine culture, and symptom questionnaires were performed at baseline (before TURBT), 7-14 days after TURBT, and before each of the six BCG instillations. Prophylactic antibiotics were administered at the clinician’s discretion in patients considered at high risk for UTI, such as those with a history of recurrent UTIs. Intravesical BCG therapy was discontinued in cases of persistent severe lower urinary tract symptoms, severe hematuria, or suspected systemic BCG-related infection accompanied by high fever (≥38℃).

Midstream urine samples were collected in sterile disposable urine containers. Samples were transferred to centrifuge tubes and delivered to the laboratory for chemical analysis and laser flow cytometry. All analyses were performed within 30 minutes of sample collection to ensure sample integrity.

2. Sysmex UF-5000 analysis

Standard urine sediment analysis was performed using the Sysmex UF-5000 system in conjunction with a chemistry analyzer. Results were expressed as counts per microliter (µL) and per high-power field (HPF), in accordance with the manufacturer’s instructions. Urine samples were aspirated directly by the instrument without additional preparation.

When bacterial counts exceeded 100 bacteria/µL, the UF-5000 system classified bacteria as Gram-negative (GN), Gram-positive (GP), or mixed GP/GN. Samples were flagged as “unclassifiable” when bacterial differentiation was not possible. Pyuria was defined as ≥10 white blood cells (WBCs) per HPF. Significant bacteriuria was defined as the growth of a pathogenic organism at ≥10⁵ colony-forming units per milliliter (CFU/mL) on urine culture.

3. Unit of analysis and handling of repeated urine tests

The primary unit of analysis was the patient. Because the clinical question was whether the UF-5000 could support screening and management decisions during the BCG induction course, the diagnostic-performance analysis was conducted at the patient level rather than as a specimen-level repeated-measures analysis. Serial urinalysis, UF-5000 testing, and urine culture were performed before scheduled BCG instillations as part of routine clinical monitoring; however, repeated specimens from the same patient were not pooled as independent observations, and each patient contributed one analytic outcome. A patient was classified as bacteriuria-positive if at least one urine culture during the BCG induction course showed significant bacteriuria (≥10⁵ CFU/mL), and as bacteriuria-negative when available urine culture results during the induction course showed no significant bacteriuria. All 145 patients had at least one evaluable urine culture during the induction course; therefore, no patient was classified as culture-negative in the absence of culture data. UF-5000 screen positivity was defined at the patient level as at least one pre-instillation bacterial count exceeding the specified cutoff during the induction course. In patients with more than one culture-positive episode, the first culture-confirmed episode was used for the organism-level and Gram-flag analyses. Because no repeated specimens were analyzed as independent observations, adjustment for within-patient correlation was not required.

4. Statistical analysis

Baseline demographic characteristics were summarized using descriptive statistics. Bacterial counts and GN/GP flag outputs obtained from the UF-5000 system were compared with urine culture results. The diagnostic performance of the UF-5000 system for detecting significant bacteriuria was evaluated by calculating sensitivity, specificity, positive predictive value, and negative predictive value.

Sensitivity, specificity, positive predictive value, and negative predictive value were calculated with Wilson score 95% confidence intervals. The UF-5000 bacterial count was evaluated using two predefined clinically relevant thresholds, >100 and >1,000 bacteria/µL. Because the analytic dataset handled UF-5000 bacterial count according to these predefined categories rather than as raw continuous values, an unrestricted continuous ROC cutoff search was not performed. Instead, the two predefined dichotomized UF-5000 criteria were compared, and the primary screening cutoff was selected by comparing the Youden’s index, defined as sensitivity+specificity-1. For each dichotomous diagnostic criterion (pyuria, nitrite, and the two UF-5000 thresholds), the binary-classifier area under the receiver operating characteristic (ROC) curve, equivalent to (sensitivity+specificity)/2, was reported with its 95% CI calculated using the Hanley–McNeil method. All statistical analyses were performed using IBM SPSS Statistics for Windows, version 23.0 (IBM Corp., Armonk, NY, USA). A two-sided p-value <0.05 was considered statistically significant.

5. Ethics statement

The study protocol was reviewed and approved by the Institutional Review Board of Chonnam National University Hwasun Hospital (IRB No. CNUHH-2018-007). The study was conducted in accordance with the Declaration of Helsinki and the Ethical Guidelines for Clinical Studies.

RESULTS

Of the 145 enrolled patients, 128 (88.3%) were male, and the median age was 71 years (interquartile range [IQR], 63-77 years). During intravesical BCG treatment, 13 patients (9.0%) received prophylactic antibiotics, whereas 40 patients (27.6%) received therapeutic antibiotics. Intravesical BCG instillation was discontinued in 27 patients (18.6%) (Table 1).

TABLE 1. Baseline characteristics of patients with intravesical Bacillus Calmette-Guérin treatment.

BMI: body mass index, BCG: Bacillus Calmette-Guérin.

Bacteriuria was detected in 6 patients (4.1%) at baseline and in 6 patients (4.1%) during the postoperative period. During the BCG induction course (assessed before instillation), urinalysis revealed pyuria in 111 patients (76.6%) and nitrite positivity in 18 patients (12.4%) (Fig. 1).

FIG. 1. Urinalysis and urine culture findings during the perioperative period.

In the patient-level BCG induction-course analysis, 37 of 145 patients developed culture-confirmed significant bacteriuria. A UF-5000 bacterial count cutoff of >100 bacteria/µL identified these patients with a sensitivity of 94.6% (95% CI, 82.3-98.5) and a specificity of 68.5% (95% CI, 59.3-76.5). At a cutoff of >1,000 bacteria/µL, sensitivity was 62.2% (95% CI, 46.1-75.9) and specificity was 96.3% (95% CI, 90.9-98.6) (Table 2). The sensitivity and specificity remained consistent regardless of whether the patients receiving prophylactic antibiotics were included or excluded (Table 3).

TABLE 2. Sensitivity, specificity, and binary-classifier area under the curve of urine microscopy, nitrite test, and urine flow cytometry for screening bacteriuria in urine culture.

Each patient contributed one outcome to the BCG induction-course analysis. UF-5000 positivity was defined as at least one pre-instillation bacterial count exceeding the threshold during the induction course. TP: true positive, TN: true negative, FP: false positive, FN: false negative, PPV: positive predictive value, NPV: negative predictive value, AUC: area under the curve, CI: confidence interval.

TABLE 3. Sensitivity and specificity of bacterial counts in urine flow cytometry without prophylactic antibiotics.

Each patient was counted once. TP: true positive, TN: true negative, FP: false positive, FN: false negative, PPV: positive predictive value, NPV: negative predictive value, CI: confidence interval.

Urinalysis for pyuria demonstrated a low specificity (28.7%; 95% CI, 21.0-37.9), whereas the urinary nitrite test showed high specificity (97.2%; 95% CI, 92.2-99.1) but low sensitivity (40.5%; 95% CI, 26.4-56.5). Among the four evaluated dichotomous diagnostic criteria, the UF-5000 bacterial count >100 bacteria/µL showed the highest Youden index and binary-classifier AUC (Youden index, 0.631; AUC, 0.816 [95% CI, 0.727-0.905]). The >1,000 bacteria/µL threshold showed a slightly lower binary-classifier AUC (0.792 [95% CI, 0.699-0.885]) but higher specificity, and was therefore retained as a high-specificity threshold (Table 2, Fig. 2).

FIG. 2. Receiver operating characteristic curve for the UF-5000 bacterial-count screening criterion of >100 bacteria/µL in diagnosing significant bacteriuria (urine culture ≥10⁵ CFU/mL). The binary-classifier AUC was 0.816 (95% CI, 0.727-0.905).

Among patients who received therapeutic antibiotics, 14 patients (35.0%) showed no bacteriuria on urine culture; within this culture-negative subgroup, 28.6% had bacterial counts ≤100 bacteria/µL on urine flow cytometry (Table 4).

TABLE 4. Pyuria, nitrite test, and urine flow cytometry outcomes relative to bacteriuria in patients receiving therapeutic antibiotics.

The UF-5000 bacterial-count categories are cumulative: the>1,000 bacteria/µL group is a subset of the >100 bacteria/µL group. One bacteriuria patient had no bacteria detected by UF-5000 (count 0) and is included in the ≤100 bacteria/µL category. Pyuria and nitrite are independent urinalysis results.

During intravesical BCG treatment, 37 patients (25.5%) developed bacteriuria, including 12 cases of GN bacteria, 21 cases of GP bacteria, and 4 cases of mixed bacterial growth (Table 5). The Sysmex UF-5000 system demonstrated better discrimination for GN bacteria than for GP bacteria. The sensitivity and specificity for GN bacteria detection were 83.3% (95% CI, 55.2-95.3) and 92.2% (95% CI, 86.3-95.7), respectively, whereas those for GP bacteria were 81.0% (95% CI, 60.0-92.3) and 84.2% (95% CI, 76.6-89.6), respectively (Table 6).

TABLE 5. Discrimination of Gram-negative and Gram-positive bacteria by the UF-5000 compared to urine culture results after intravesical Bacillus Calmette-Guérin treatment.

Counts are per patient at the first culture-confirmed bacteriuria episode. GP: Gram-positive bacteria, GN: Gram-negative bacteria.

TABLE 6. Predictive value for Gram stainability of bacteriuria by the UF-5000.

The 141 patients comprised 12 with Gram-negative-only bacteriuria, 21 with Gram-positive-only bacteriuria, and 108 without culture-confirmed bacteriuria during the BCG induction course; four mixed-culture cases were excluded. For patients with more than one culture-positive episode, the first culture-confirmed episode was used. GP: Gram-positive bacteria, GN: Gram-negative bacteria, TP: true positive, TN: true negative, FP: false positive, FN: false negative, PPV: positive predictive value, NPV: negative predictive value, CI: confidence interval.

Table 7 summarizes the patient-level distribution of UF-5000 flags by cultured organism at the first culture-confirmed monomicrobial episode. The most frequent organisms were Enterococcus spp. (n=11), Staphylococcus spp. (n=8), and Escherichia coli (n=6).

TABLE 7. Discrimination of Gram-negative and Gram-positive bacteria by UF-5000 according to cultured bacteria, excluding polymicrobial samples.

Counts are per patient at the first culture-confirmed monomicrobial episode. Enterococcus spp. includes E. faecalis and E. faecium. GP: Gram-positive bacteria, GN: Gram-negative bacteria.

DISCUSSION

Rapid assessment of bacteriuria is essential for safe and timely clinical decision-making during intravesical BCG therapy. Automated urine flow cytometry has been proposed as a screening strategy to triage urine cultures and expedite clinical decisions in suspected UTI.12

Our group has previously evaluated urine flow cytometry during intravesical BCG therapy using the earlier-generation Sysmex UF-1000i system, which provides bacterial enumeration but not Gram discrimination.13,14 Drawing on the same institutional BCG registry, the present study extends this work by evaluating the newer UF-5000 platform; its automated Gram-negative/Gram-positive flag output—the principal focus of this study—has not been reported previously in BCG-treated patients (Supplementary Table 1). The bacterial-count findings are consistent with our earlier results and serve to benchmark the UF-5000 in this setting.

In the present study, pyuria demonstrated high sensitivity (91.9%) but very low specificity (28.7%) for diagnosing significant bacteriuria. Notably, during the postoperative assessment and throughout the BCG induction course, the incidence of pyuria exceeded 70%, whereas the actual rate of bacteriuria was substantially lower (4%-25%). Conversely, nitrite positivity showed high specificity (97.2%) but low sensitivity (40.5%), indicating that false-negative nitrite results should be carefully considered when evaluating symptomatic patients.

Flow cytometry–based urine particle analysis enables high-throughput quantification of bacteria and leukocytes; the UF-1000i includes a dedicated bacterial channel and has been evaluated in routine diagnostic workflows.9 However, attempts to infer bacterial morphology or classification using earlier UF platforms have demonstrated insufficient accuracy to allow for stand-alone clinical decision-making.11 Although direct pathogen identification approaches combining urine screening with MALDI-TOF mass spectrometry have been described, these strategies require additional instrumentation and workflow complexity, limiting their applicability as a rapid, stand-alone screening solution in time-sensitive outpatient settings.15

By contrast, the UF-5000 incorporates an updated analysis algorithm that improves bacterial enumeration and provides automated GN/GP discrimination.12 In our cohort—characterized by postoperative and treatment-related urinary inflammation—a bacterial count cutoff of >100 bacteria/µL achieved high sensitivity (94.6%) but modest specificity (68.5%) for detecting culture-confirmed bacteriuria (≥10⁵ CFU/mL), whereas a higher cutoff (>1,000 bacteria/µL) improved specificity (96.3%) at the expense of sensitivity (62.2%). These findings may support a symptom-contextual approach in which lower thresholds are favored when prioritizing sensitivity in symptomatic patients, while higher thresholds are used when the primary goal is to reduce unnecessary cultures and antibiotics.

Regarding Gram status prediction, the UF-5000 demonstrated better discrimination for GN bacteria than for GP bacteria (Table 6). Lu et al.16 described a mechanistic basis for improved GN specificity on the UF-5000 platform, reporting that GN organisms generate characteristic scattergram patterns attributable to greater nucleic-acid stainability associated with thinner cell walls. This observation is concordant with our finding of high GN specificity (92.2%) and supports the potential utility of early GN/GP discrimination as an adjunct to empiric decision-making while awaiting culture confirmation.

BCG-related irritative lower urinary tract symptoms are common and can mimic bacterial cystitis. In the EORTC randomized phase III study evaluating BCG dosages and maintenance durations, approximately 62% of patients experienced local side effects and 23% developed bacterial cystitis.17 In our cohort, 25.5% of patients developed bacteriuria during BCG therapy, which is comparable to trial-based estimates.17 Notably, adverse events do not reliably predict BCG efficacy, emphasizing the importance of maintaining the therapy when clinically safe.18 BCG-related symptomatology largely reflects bladder wall inflammation and immune activation after intravesical instillation,19,20 and long-term follow-up data suggest that symptom burden and bladder health remain clinically meaningful considerations for patients treated with BCG.21

Accurate differentiation between bacterial cystitis and BCG-induced inflammatory cystitis is therefore critical. Pyuria, microscopic hematuria, or asymptomatic bacteriuria alone are not contraindications to intravesical BCG instillation, and routine prophylactic antibiotics are not recommended in these settings.2,22,23 In contrast, symptomatic UTI warrants postponement of instillation and appropriate antimicrobial therapy, as infection is considered a risk factor for BCG-related complications and a contraindication to BCG instillation.2

In this study, therapeutic antibiotics were administered to 27.6% of patients and BCG instillation was discontinued in 18.6%; no serious infectious complications were observed. However, the proportion receiving antibiotics exceeded the proportion with culture-confirmed bacteriuria, and only 65% of patients treated with therapeutic antibiotics had bacteriuria on culture. Collectively, these findings suggest potential overtreatment driven by nonspecific urinalysis findings and symptom overlap. Rapid bacteriuria screening tools may help reduce unnecessary interruption of BCG therapy and inappropriate antibiotic exposure.

The GN/GP discrimination provided by the UF-5000 may further support empiric antibiotic selection in symptomatic patients. Although antibiotic susceptibility cannot be inferred from flow cytometry, early GN detection can help prioritize urine culture and guide initial empiric choices based on local resistance patterns and patient history, pending definitive culture and susceptibility testing.

Beyond initial screening, urine flow cytometry may offer opportunities for early therapeutic monitoring. Taguchi et al.24 reported that the UF-5000 can detect antibiotic-associated bacterial morphologic changes within six hours, suggesting potential utility for early assessment of empiric antibiotic effect. If validated in BCG-treated populations, such functionality could help clinicians reassess empirical therapy earlier and avoid prolonged unnecessary antibiotic exposure in patients with overlapping inflammatory and infectious symptoms.

This study has several limitations. First, the retrospective design and relatively small, single-center sample limit generalizability. Second, the present analysis should be interpreted as a patient-level BCG induction-course analysis rather than a per-visit specimen-level diagnostic study. Although serial urine testing was performed in most patients, repeated specimens were not analyzed as independent observations; therefore, the reported diagnostic indices reflect the ability of the UF-5000 to identify patients who developed culture-confirmed bacteriuria during the induction course rather than per-visit specimen-level accuracy. Because culture-negative status was determined from available culture data, incomplete monitoring in a small number of patients may have led to under-detection of bacteriuria. Third, large, multicenter studies of the UF-5000 in bladder cancer patients undergoing intravesical BCG therapy are still lacking. Further research correlating urinary symptoms with bacterial counts is needed to optimize cutoff values and improve differentiation of non-bacterial inflammatory cystitis.

Although the UF-5000 system cannot replace urine culture for definitive bacterial identification or antibiotic susceptibility testing, this study demonstrates that it provides high sensitivity for bacteriuria screening and useful GN/GP discrimination prior to culture, which could support more timely clinical decision-making.

In conclusion, this study demonstrated the potential clinical utility of the Sysmex UF-5000 system as a rapid screening tool for significant bacteriuria and for predicting Gram status in patients with bladder cancer undergoing intravesical BCG therapy. Although conventional urinalysis and urine culture remain essential for definitive clinical decision-making, bacteriuria assessment using the UF-5000 system may facilitate earlier decisions regarding UTI evaluation and empiric antibiotic selection. To establish its role in routine clinical practice, large-scale prospective studies are warranted, and further investigation is required to optimize bacterial count cutoff values for culture-defined significant bacteriuria in bladder cancer patients receiving BCG treatment.

SUPPLEMENTARY MATERIAL

Supplementary Table 1

Comparison of the present study with the authors’ previous related publications