The value of pathogen DNA load in evaluating the efficacy of bloodstream infection: a prospective observational study Authors: Shiheng Wu, Jing Liu, Lili Yan, Xinmei Zhong, Chenchen Shi, Li He, Baogui Wang Published: 2025 ABSTRACT Background and objectiveDroplet digital polymerase chain reaction (ddPCR) is increasingly being applied in clinical practice, with absolute quantification as one of its key advantages. This study aims to explore the significance of pathogen DNA load in evaluating the therapeutic efficacy of bloodstream infection (BSI).MethodsA total of 25 hospitalized patients with bloodstream infection were enrolled as research subjects. These patients were confirmed by ddPCR testing in the Department of Infectious Diseases, Fuyang People's Hospital, and admitted between November 2024 and April 2025. All patients showed clinical improvement after treatment and were discharged. Based on ddPCR results, they were divided into the Escherichia coli (E. coli) group (n=15) and the Klebsiella pneumoniae (K. pneumoniae) group (n=10). Each patient underwent consecutive ddPCR testing three times, and data including general clinical information, laboratory test results, and sequential organ failure assessment (SOFA) scores were collected and analyzed. The study further aimed to investigate the value of dynamic changes in pathogen DNA load for evaluating efficacy of treatment in BSI and other potential clinical implications.ResultsAmong the baseline data of two groups, the alanine aminotransferase(ALT)level in the K. pneumoniae group (52.8 U/L) was higher than that in the E. coli group (21.0U/L), and the medianl DNA load in the K.pneumoniae group(8329.2 copies/mL,[IQR]: 291.8–47722.4 copies/mL) was higher than that in the E. coli group (310.2 copies/mL,[IQR]: 129.0–1838.0 copies/mL). Meanwhile, Pearson correlation analysis showed no strong correlation between the initial DNA load and SOFA score in patients of the two groups. Dynamic observation indicated that the pathogen DNA load of BSI patients in both groups showed a decreasing trend; at the same time, the proportion of patients with normalized body temperature increased in each group. Further verification demonstrated a strong correlation between the changing trend of DNA load and C-reactive protein(CRP) as well as procalcitonin(PCT)(r >0.9).In addition, the rate of reduction in pathogen DNA load in the two groups was comparable to CRP and PCT. Finally, in the K. pneumoniae group, it was found when the DNA load cut-off value was set at 8329.23 copies/mL, it exhibited high sensitivity and high specificity for identifying the formation of secondary abscesses(AUC=1).ConclusionOur study demonstrates that DNA load levels of BSI vary significantly among different pathogens. More importantly, the dynamic changes in pathogen DNA load closely correlate with clinical outcomes, serving as a reliable and objective indicator to directly evaluate therapeutic efficacy in BSI patients. This highlights the practical clinical value of pathogen DNA load monitoring for real-time assessment of treatment response and optimization of clinical management strategies.Clinical trial numberNot applicable. Introduction BSI has become one of the major public health burdens worldwide [1]. The incidence of bloodstream infections ranges from 113 to 220 cases per 100,000 people [2, 3]. BSI is also prone to progressing to sepsis and multiple organ dysfunction syndrome (MODS). Meta-analysis suggests that the average 30-day septic shock mortality was 34.7% (95% CI 32.6–36.9%), and 90-day septic shock mortality was 38.5% (95% CI 35.4–41.5%) [4]. Droplet digital PCR (ddPCR), a cutting-edge technology, offers advantages including high sensitivity, rapidity, multi-target detection, and absolute quantification [5, 6]. Its high sensitivity and rapid turnaround time have been validated in clinical settings [7, 8]. Previous studies exploring the correlation between pathogen load and clinical outcomes have primarily focused on viral infections, such as hepatitis B virus (HBV) and coronavirus disease 2019(COVID-19) [9, 10], whereas research on the association between bacterial DNA load and clinical outcomes remains limited. A study using ddPCR technology has observed that 9 out of 10 non-surviving sepsis patients showed an upward trend in bacterial DNA load [11]. Therefore, we propose that the “absolute quantification” feature of ddPCR also has the potential to evaluate the therapeutic efficacy of bacterial bloodstream infection. However, this still needs to be compared with the classic inflammatory markers C-reactive protein (CRP) and procalcitonin(PCT)—and these two markers themselves have certain limitations [12]. This study aimed to evaluate whether pathogen DNA load measured by ddPCR can serve as a biomarker for treatment efficacy in BSI caused by E. coli and K. pneumoniae. Materials and methods Patients and study design From November 2024 to April 2025, 25 hospitalized patients with BSI were recruited from the Department of Infectious Diseases, Fuyang People’s Hospital, including 15 cases of E. coli infection and 10 cases of K. pneumoniae infection(Fig. 1). All patients showed clinical improvement after treatment and were discharged. Fig. 1Flow chart. (a) Brief introduction to the enrolled cohort; (b) Detection time points of key indicators such as DNA load Flow chart. (a) Brief introduction to the enrolled cohort; (b) Detection time points of key indicators such as DNA load Inclusion and exclusion criteria Inclusion criteria ≥ 18years;Community-onset infection with initial symptoms including fever(T ≥ 38℃);Confirmation of E. coli or K. pneumoniae infection by ddPCR, with exclusion of sample contamination and clinical assessment (regardless of blood culture results);Completion of three dynamic ddPCR tests (performed on days 1, 3–4, and 6–7 after admission, respectively). ≥ 18years; ≥ 18years; Community-onset infection with initial symptoms including fever(T ≥ 38℃); Community-onset infection with initial symptoms including fever(T ≥ 38℃); Confirmation of E. coli or K. pneumoniae infection by ddPCR, with exclusion of sample contamination and clinical assessment (regardless of blood culture results); Confirmation of E. coli or K. pneumoniae infection by ddPCR, with exclusion of sample contamination and clinical assessment (regardless of blood culture results); Completion of three dynamic ddPCR tests (performed on days 1, 3–4, and 6–7 after admission, respectively). Completion of three dynamic ddPCR tests (performed on days 1, 3–4, and 6–7 after admission, respectively). Exclusion criteria Hospital-acquired infection (non-community onset);Failure to complete three dynamic ddPCR tests or early discharge;Lack of simultaneous C-reactive protein (CRP) and procalcitonin (PCT) testing;Septic shock,clinical deterioration or death;Incomplete clinical data. Hospital-acquired infection (non-community onset); Hospital-acquired infection (non-community onset); Failure to complete three dynamic ddPCR tests or early discharge; Failure to complete three dynamic ddPCR tests or early discharge; Lack of simultaneous C-reactive protein (CRP) and procalcitonin (PCT) testing; Lack of simultaneous C-reactive protein (CRP) and procalcitonin (PCT) testing; Septic shock,clinical deterioration or death; Septic shock,clinical deterioration or death; Incomplete clinical data. Incomplete clinical data. Methods Grouping Patients were divided into the E. coli group and K. pneumoniae group based on ddPCR results. Data collection General information (gender, age, body mass index [BMI], comorbidities, etc.) was collected. ddPCR, CRP, and PCT were measured simultaneously at three time points(Fig. 1). Additional laboratory tests included bilirubin, albumin, white blood cell count, platelet count, and creatinine. The Sequential Organ Failure Assessment (SOFA) score at admission was calculated based on indicators such as blood pressure. ddPCR technology and reagents ddPCR is a third-generation PCR technology. Its core lies in achieving direct absolute quantification of copy numbers through large-scale single-molecule amplification, which enables direct determination of the number of target nucleic acid molecules and improves detection sensitivity and accuracy (Fig. 2). The ddPCR platform(relying on the fully automated nucleic acid reaction system construction system AP10 and the D3200 droplet digital PCR system)was provided by Pilot Gene Technology Company (Hangzhou, China). Reagents included 4 pathogen panels (Panels 1–4, covering 18 detection targets in total) and drug-resistance panel (Panel NY, including blaKPC, Oxa-48, blaNDM + IMP, mecA, VanA + VanM). Panel 1, required for the second and third ddPCR tests of each patient, was provided free of charge by Pilot Gene Technology Company, and all tests were performed by the hospital’s laboratory department. The probe design site for E. coli is yccT and the probe design site for K. pneumoniae is khe. Fig. 2Main technical principles of ddPCR Main technical principles of ddPCR Statistical methods IBM SPSS Statistics 26.0 was used for statistical analysis. Categorical variables were described as frequencies and percentages, with intergroup differences analyzed using the chi-square test or Fisher’s exact test. For continuous variables, those following a normal distribution were expressed as mean ± standard deviation (Mean ± SD), while non-normally distributed variables were presented as median and interquartile range (IQR). Intergroup comparisons were performed using the independent samples t-test or Mann-Whitney U test, and correlation analyses were conducted using the Pearson correlation coefficient (r). All statistical tests were two-tailed, with a significance level set at p < 0.05. A p-value < 0.05 was considered statistically significant for intergroup differences or correlations. Prior to plotting the trend chart of pathogen DNA changes, a logarithmic transformation was performed. Raw data were used for all other data and analysis procedures. Receiver operating characteristic (ROC) curve analysis was used to evaluate the efficacy of pathogen load in predicting secondary abscess formation in K. pneumoniae-induced BSI, with the area under the curve (AUC), sensitivity, and specificity calculated accordingly. Results Patient characteristics A total of 25 BSI patients were included, comprising 15 with E. coli infection and 10 with K. pneumoniae infection. There were no significant differences between the two groups in baseline characteristics, including gender, age, BMI, comorbidities, and laboratory indicators such as white blood cell count (WBC)、platelet count (PLT)、total bilirubin (TBil)、creatinine (CREA)、and albumin (ALB) (P > 0.05). However, the alanine transaminase (ALT) level in the K. pneumoniae group (52.8 U/L) was significantly higher than that in the E. coli group (21.0 U/L) (P < 0.05). Additionally, the initial DNA load in the K. pneumoniae group (8329.2 copies/mL) was numerically higher than in the E. coli group (310.2 copies/mL), though this difference did not reach statistical significance (P > 0.05) (Table 1). Table 1Characteristics of the patientsClinical characteristicsE. coli(n = 15)K. pneumoniae(n = 10)P-value Male [no.(%)]5 (33.3)6 (60)0.366 Age (years)61.0 (57.0–75.5)68.5 (58.0–82.0)0.338 BMI (Kg/m2)24.2 (22.8–27.6)24.8 (21.8–27.6)0.935Complication [no.(%)]13 (86.7)8 (80)1Blood laboratory examination WBC × 109/L14.0 ± 6.214.5 ± 11.40.901 PLT × 109/L138.0 (113.0–217.5)109.0 (58.0–168.0)0.144 TBil (umol/L)16.6 (12.6–21.2)27.9 (15.1–44.6)0.196 ALT (U/L)21.0 (17.9–47.3)52.8 (30.6–243.0)0.041 CREA (umol/L)79.6 (72.5 -127.2)87.3 (76.3 -104.6)0.723 ALB(g/L)36.2 ± 5.933.0 ± 4.70.161DNA load (copies/mL)310.2(129.0–1838.0)8329.2(291.8- 47722.4)0.160 Table 1Characteristics of the patientsClinical characteristicsE. coli(n = 15)K. pneumoniae(n = 10)P-value Male [no.(%)]5 (33.3)6 (60)0.366 Age (years)61.0 (57.0–75.5)68.5 (58.0–82.0)0.338 BMI (Kg/m2)24.2 (22.8–27.6)24.8 (21.8–27.6)0.935Complication [no.(%)]13 (86.7)8 (80)1Blood laboratory examination WBC × 109/L14.0 ± 6.214.5 ± 11.40.901 PLT × 109/L138.0 (113.0–217.5)109.0 (58.0–168.0)0.144 TBil (umol/L)16.6 (12.6–21.2)27.9 (15.1–44.6)0.196 ALT (U/L)21.0 (17.9–47.3)52.8 (30.6–243.0)0.041 CREA (umol/L)79.6 (72.5 -127.2)87.3 (76.3 -104.6)0.723 ALB(g/L)36.2 ± 5.933.0 ± 4.70.161DNA load (copies/mL)310.2(129.0–1838.0)8329.2(291.8- 47722.4)0.160 Characteristics of the patients Correlation between initial DNA load and SOFA score For both the E.coli group and the K.pneumoniae group, the Pearson correlation coefficient between the initial DNA load and SOFA score was low, with no significant statistical difference observed (Table 2). Table 2Pearson correlation analysis between initial DNA load and SOFA scoreGroupr-valveP-value E. coli 0.1430.611 K. pneumoniae 0.1410.697 Table 2Pearson correlation analysis between initial DNA load and SOFA scoreGroupr-valveP-value E. coli 0.1430.611 K. pneumoniae 0.1410.697 Pearson correlation analysis between initial DNA load and SOFA score Dynamic changes in pathogen DNA load and clinical symptoms Due to the wide range of pathogen load values, data were converted to logarithmic indices. Dynamic monitoring revealed a marked decrease in DNA load over time in parallel with therapeutic improvement, observed in both the E. coli and K. pneumoniae groups (Fig. 3). To further verify the hypothesis that DNA load can reflect therapeutic efficacy, we observed the proportion of febrile patients in the two groups at different time points. It was found that the proportion of febrile patients in both groups showed an overall decreasing trend. In the E. coli group, the difference was more significant (P1, P2 < 0.05), while there was also a certain difference in the K. pneumoniae group (P1 > 0.05, P2 < 0.05) (Fig. 4). Fig. 3(A) Changes in DNA load of E.coli group; (B) Changes in DNA load of K.pneumoniae group (A) Changes in DNA load of E.coli group; (B) Changes in DNA load of K.pneumoniae group Fig. 4(C) The proportion of febrile patients in the E.coli groups; (D) The proportion of febrile patients in the E. coil groups. (P1 means 1st vs. 2nd, P2 means 1st vs. 3rd) (C) The proportion of febrile patients in the E.coli groups; (D) The proportion of febrile patients in the E. coil groups. (P1 means 1st vs. 2nd, P2 means 1st vs. 3rd) Correlation between bacterial DNA load and classic inflammatory markers In the E. coli group, Pearson correlation analysis showed that DNA load was positively correlated with CRP and PCT, but without statistical significance (P > 0.05). In the K. pneumoniae group, Pearson correlation analysis indicated that DNA load was significantly positively correlated with CRP, with statistical significance (P < 0.05); while it was still positively correlated with PCT, but without statistical significance (P > 0.05) (Fig. 5). Fig. 5(E) The linear relationship between DNA load and CRP, PCT in the E. coli group; (F) The linear relationship between DNA load and CRP, PCT in the K. pneumoniae group; (G) The results of Pearson correlation (E) The linear relationship between DNA load and CRP, PCT in the E. coli group; (F) The linear relationship between DNA load and CRP, PCT in the K. pneumoniae group; (G) The results of Pearson correlation Comparison of reduction rates of bacterial DNA load and classic inflammatory markers In the E. coil group, compared with CRP, the reduction rate of DNA load at the 1–2nd time point was higher (72.28 ± 37.09% vs. 55.94 ± 19.58%), though not statistically significant (P > 0.05); At the 1–3rd time point, the DNA load reduction rate was significantly higher than that of CRP (95.64 ± 14.15% vs. 87.88 ± 11.17%),with statistical significance (P < 0.05);Compared with PCT, the reduction rate of DNA was slightly higher at both the 1–2nd and 1-3rd time points (72.28 ± 37.09% vs. 62.33 ± 48.26%) and (95.64 ± 14.15% vs. 91.83 ± 10.11%),but there was no statistical significance (P > 0.05)(Fig. 6). In the K. pneumoniae group, compared with CRP, the DNA load reduction rate at the 1–2nd time point was lower (14.95 ± 100.25% vs. 35.66 ± 27.80%),with no statistically significant differences (P > 0.05); At the 1–3rd time point, the DNA load reduction rate was higher (80.27 ± 45.35% vs. 62.68 ± 25.43%), though not statistically significant (P > 0.05).Compared with PCT, the DNA load reduction rate was lower at both the 1–2nd (14.95 ± 100.25% vs. 68.26 ± 15.70%) and 1–3rd (80.27 ± 45.35% vs. 82.90 ± 26.94%) time points, with no statistically significant differences (P > 0.05)(Fig. 6). Fig. 6(H) Bar chart of the reduction rate in the E. coil group; (I) Bar chart of the reduction rate in the K. pneumoniae group (H) Bar chart of the reduction rate in the E. coil group; (I) Bar chart of the reduction rate in the K. pneumoniae group Predictive Value of High K. pneumoniae DNA load for abscess formation in BSI In this study, five patients in the K. pneumoniae group developed abscesses, suggesting that high DNA load levels are associated with an increased likelihood of abscess formation. ROC analysis was performed using abscess formation as the outcome and initial DNA load as the diagnostic indicator. The optimal cut-off value was 8329.23 copies/mL, which identified abscesses with 100% sensitivity (AUC = 1)(Fig. 7). Fig. 7ROC curve ROC curve Discussion BSI caused by gram-negative bacteria such as E. coli and K. pneumoniae remain a critical clinical challenge due to their high mortality and association with sepsis and MODS [13, 14]. This study explored the clinical value of pathogen DNA load, measured by ddPCR, in evaluating therapeutic efficacy and predicting complications in BSI caused by E. coli and K. pneumoniae, yielding several key findings that warrant further discussion. Differences in DNA load between E. coli and K. pneumoniae BSI Our results showed that K. pneumoniae had a higher median DNA load (8329.2 copies/mL) than E. coli (310.2 copies/mL) in BSI patients, though the difference was not statistically significant (P > 0.05). This may reflect inherent biological differences: K. pneumoniae produces virulence factors (e.g., capsular polysaccharides, biofilm-forming capabilities) that enhance immune evasion and bloodstream persistence [15], while E. coli—a common BSI pathogen—may be more readily cleared by innate immunity, leading to lower circulating loads. Numerically, the groups differed by nearly an order of magnitude; the lack of significance likely stems from a large standard deviation and small sample size. Notably, K. pneumoniae patients had significantly higher ALT levels (52.8 U/L vs. 21.0 U/L in E. coli group, P<0.05), potentially due to differing pathogen tropism. Classical K. pneumoniae strains are highly virulent and often cause pyogenic liver or extrahepatic abscesses even without biliary disease [16]. Few prior studies have focused on bacterial DNA load in BSI, with most research on viral infections [9, 10]. Our findings suggest pathogen-specific DNA load variations may exist in bacterial BSI, offering insights into pathogenic mechanisms. Larger studies are needed to confirm these differences’ statistical and clinical significance, which could guide pathogen-targeted management. Dynamic changes in pathogen DNA load as a marker of therapeutic efficacy This study found that pathogen DNA load decreased in both groups with effective treatment, accompanied by a gradual reduction in febrile patients. Combined with DNA load line charts and febrile patient bar charts, DNA load shows potential as a marker for evaluating BSI therapeutic efficacy. Hu et al. prospectively observed critically ill ICU patients with BSI and reported decreasing DNA loads in improvers and increasing loads in non-survivors [8], consistent with our findings, supporting the feasibility of using DNA load trends to predict clinical outcomes. While line charts provide visual insights, further validation is needed to confirm the exact significance of DNA load changes in efficacy assessment. We further observed strong correlations (r > 0.9) between pathogen DNA load changes and CRP/PCT in both groups, with only the DNA load-CRP correlation in the K. pneumoniae group reaching statistical significance. As classic infectious disease efficacy indicators [12, 17], this partially supports the hypothesis that DNA load could serve as a novel evaluation tool. In the E. coli group, DNA load reduction rates at 1–2nd and 1–3rd time points were higher than CRP/PCT (P > 0.5), with a significant difference between DNA load and CRP at 1–3rd time points (95.64 ± 14.15% vs. 87.88 ± 11.17%, P < 0.01). In the K. pneumoniae group, DNA load reduction was lowest at 1–2nd time points and higher than CRP at 1–3rd time points (both P > 0.5). The slower early-to-middle-stage DNA load reduction in K. pneumoniae BSI may relate to: (1) hypervirulent strains in community-acquired infections [16]; (2) high risk of abscess formation (which may persist/progress short-term despite effective antibiotics [18]; 3) smaller sample size in the K. pneumoniae group. Nevertheless, DNA load changes were non-inferior to CRP/PCT for BSI efficacy evaluation. Unlike CRP/PCT—elevated by non-infectious conditions (e.g., tissue injury, autoimmunity) [17, 19]—DNA load directly measures the causative pathogen, offering greater specificity for infection control. These results extend ddPCR’s application from bacterial BSI diagnosis to therapeutic monitoring, complementing existing biomarkers. Its rapidity and sensitivity [noted in prior studies] further support its potential as a point-of-care tool to guide timely antibiotic adjustments. High K. pneumoniae DNA load and the risk of abscess formation A striking finding was the strong association between high K. pneumoniae DNA load and subsequent abscess development. Among K. pneumoniae BSI patients, those with abscesses had higher initial DNA loads, with a cutoff value of 8329.23 copies/mL showing 100% sensitivity in predicting abscess formation (AUC = 1). This is clinically relevant because K. pneumoniae is well-documented to cause metastatic abscesses, particularly in the liver, lungs, or soft tissues, due to its ability to form biofilms and resist phagocytosis [18, 20]. A high initial DNA load may indicate a more aggressive infection with greater pathogen dissemination, increasing the likelihood of localized abscess formation despite systemic antibiotic therapy. This finding suggests that K. pneumoniae DNA load could serve as a prognostic marker, prompting closer clinical and radiological surveillance for abscesses in patients with high initial loads. Early detection of abscesses is critical, as they often require drainage in addition to antibiotics to achieve cure [21]. Lack of correlation between DNA load and SOFA scores No significant correlation was observed between initial DNA load and SOFA scores in either group. Many previous studies have also yielded similar results [8, 22]. The SOFA score reflects the severity of organ dysfunction, which is influenced by multiple factors including host immunity, comorbidities, and the timeliness of treatment [23], whereas DNA load primarily reflects pathogen burden. This dissociation may indicate that organ dysfunction in BSI is not solely determined by the quantity of circulating pathogens but also by host responses (e.g., cytokine storms) or pre-existing organ impairment. This finding underscores the complexity of BSI pathophysiology and suggests that DNA load and SOFA score serve complementary roles: the former monitors pathogen clearance, and the latter assesses organ function. Limitations of the study Several limitations should be acknowledged. Firstly, the sample size was small, which may have reduced the statistical power to detect subtle. Secondly, the study included only E. coli and K. pneumoniae, limiting generalizability to other BSI pathogens (e.g., Staphylococcus aureus or fungi). Thirdly, all patients included improved and were discharged, excluding those who died or deteriorated—this selection bias may have underestimated the association between DNA load and poor outcomes. Finally, the follow-up period was relatively short, precluding assessment of long-term outcomes beyond discharge. Conclusion and future directions This study demonstrates that pathogen DNA load, measured by ddPCR, is a valuable tool in BSI management. Differences in DNA load between pathogens may reflect their pathogenic properties, while dynamic changes in load offer direct insights into therapeutic efficacy, comparable or superior to traditional inflammatory markers. Notably, high K. pneumoniae DNA load strongly predicts subsequent abscess formation, aiding in targeted surveillance.