Is the Preoperative Level of Procalcitonin a Valid Indicator for Predicting Postoperative Fever After Percutaneous Nephrolithotomy?

Machine learning models to predict systemic inflammatory response syndrome after percutaneous nephrolithotomy

OBJECTIVE

The objective of this study was to develop and evaluate the performance of machine learning models for predicting the possibility of systemic inflammatory response syndrome (SIRS) following percutaneous nephrolithotomy (PCNL).

METHODS

We retrospectively reviewed the clinical data of 337 patients who received PCNL between May 2020 and June 2022. In our study, 80% of the data were used as the training set, and the remaining data were used as the testing set. Separate prediction models based on the six machine learning algorithms were created using the training set. The predictive performance of each machine learning model was determined by the area under the receiver operating characteristic curve (AUC), accuracy, sensitivity and specificity using the testing set. We used coefficients to interpret the contribution of each variable to the predictive performance.

RESULTS

Among the six machine learning algorithms, the support vector machine (SVM) delivered the best performance with accuracy of 0.868, AUC of 0.942 (95% CI 0.890-0.994) in the testing set. Further analysis using the SVM model showed that prealbumin contributed the most to the prediction of the outcome, followed by preoperative urine culture, systemic immune-inflammation (SII), neutrophil to lymphocyte ratio (NLR), staghorn stones, fibrinogen, operation time, preoperative urine white blood cell (WBC), preoperative urea nitrogen, hydronephrosis, stone burden, sex and preoperative lymphocyte count.

CONCLUSION

Machine learning-based prediction models can accurately predict the possibility of SIRS after PCNL in advance by learning patient clinical data, and should be used to guide surgeons in clinical decision-making.

Dynamics of Urinary Extracellular DNA in Urosepsis

Extracellular DNA (ecDNA) is a promising candidate marker for the early diagnosis and monitoring of urinary tract infections (UTIs). The aim of our study is to describe the dynamics of ecDNA in the plasma and urine of patients with urosepsis as well as in a mouse model of UTI. Samples of blood and urine were collected from adult patients with UTIs and obstructive uropathy (n = 36) during the first 3 days at the hospital and during a follow-up. Bacterial burden and urinary ecDNA were evaluated in a mouse UTI model (n = 26) at baseline; 24, 48, and 72 h after UTI induction; and 7 days after UTI induction. The plasma ecDNA did not change during urosepsis, but the plasma DNase activity increased significantly at the follow-up. The urinary ecDNA decreased significantly during hospitalization and remained low until the follow-up (90% lower vs. admission). No change was seen in the urinary DNase activity. C-reactive protein (CRP) and procalcitonin are positively correlated with plasma and urinary ecDNA. A UTI caused sepsis in 23% of mice. The urinary ecDNA decreased by three-fold and remained low until day 7 post-infection. Urinary bacterial burden is correlated with urinary ecDNA. Urinary ecDNA is a potential non-invasive marker for monitoring the effects of treatment during urosepsis and is related to UTI progression in the experimental animal model.

Diagnostic and therapeutic value of biomarkers in urosepsis

Urosepsis is sepsis caused by urogenital tract infection and is one of the most common critical illnesses in urology. If urosepsis is not diagnosed early, it can rapidly progress and worsen, leading to increased mortality. In recent years, with the increase of urinary tract surgery, the incidence of urosepsis continues to rise, posing a serious threat to patients. Early diagnosis of urosepsis, timely and effective treatment can greatly reduce the mortality of patients. Biomarkers such as WBC, NLR, PCT, IL-6, CRP, lactate, and LncRNA all play specific roles in the early diagnosis or prognosis of urosepsis. In addition to the abnormal increase of WBC, we should be more alert to the rapid decline of WBC. NLR values were superior to WBC counts alone in predicting infection severity. Compared with several other biomarkers, PCT values can differentiate between bacterial and non-bacterial sepsis. IL-6 always has high sensitivity and specificity for the diagnosis of sepsis, and CRP also has high sensitivity and specificity for the diagnosis of urosepsis. Lactic acid is closely related to the prognosis of patients with urosepsis. LncRNAs may be potential biomarkers of urosepsis. This article summarizes the main biomarkers, hoping to provide a reference for the timely diagnosis and evaluation of urosepsis.

The value of procalcitonin for predicting urosepsis after mini-percutaneous nephrolithotomy or flexible ureteroscopy based on different organisms

PURPOSE

To assess the value of procalcitonin (PCT) as an early biomarker for predicting urosepsis caused by Gram-negative (GN) bacteria, Gram-positive (GP) bacteria and fungi following mini-percutaneous nephrolithotomy (mPCNL) and flexible ureteroscopy (FURS).

METHODS

A total number of 356 patients with positive preoperative UC (urine cultures) who underwent mPCNL and FURS between June 2017 and January 2021 were retrospectively analyzed. Univariable analysis and multivariable logistic regression analysis were conducted to compare the predictors for urosepsis caused by different organisms. Furthermore, the nomogram was established as a predicted model for urosepsis.

RESULTS

Among 356 positive UC, 265 (74.4%) were positive for GN bacteria, 77 (21.4%) for GP bacteria and 14 (3.9%) for fungal pathogens. Escherichia coli (48.9%) were the predominant pathogens and Enterococcus (54/77) were the most common GP bacteria. Multivariate logistic regression analysis showed that positive nitrite (OR 3.31, 95% CI 1.20-9.14; P = 0.021), operative time > 90 min (OR 3.10, 95% CI 1.10-8.75, P = 0.033) and postoperative PCT > 0.1 ng/mL (OR 56.18, 95% CI 15.20-207.64, P < 0.001) were associated with postoperative urosepsis originated in GN infections, while urosepsis caused by GP bacteria and fungi was not associated with PCT > 0.1 ng/mL (P = 0.198), only stone burden > 800 mm² (OR 3.69, 95% CI 1.01-13.53, P = 0.049) was an independent risk factor.

CONCLUSIONS

For patients with positive preoperative UC, postoperative PCT > 0.1 ng/mL was an independent risk factor of post-PCNL and post-FURS urosepsis caused by GN bacteria rather than GP bacteria and fungi.

The predictive and diagnostic ability of IL-6 for postoperative urosepsis in patients undergoing percutaneous nephrolithotomy

The aim of this study is to investigate the predictive and diagnostic ability of interleukin-6 (IL-6) for postoperative urosepsis in patients undergoing percutaneous nephrolithotomy (PCNL). 90 patients undergoing PCNL between April 2019 and September 2019 were studied. 16 patients progressed to urosepsis (EXP1 group, n = 16) and 74 patients did not (CON group, n = 74); demographic and perioperative data were compared between these two groups. 25 patients who progressed to postoperative urosepsis without receiving the test of IL-6 between March 2018 and March 2019 were enrolled (EXP2 group, n = 25); demographic and perioperative data were compared between EXP1 group and EXP2 group. Compared with CON group, EXP1 group showed higher serum levels of IL-6 (p < 0.001) and neutrophil (p < 0.001) at postoperative hour two; higher serum levels of IL-6 (p < 0.001), procalcitonin (PCT) (p < 0.05), white blood cell (WBC) (p < 0.05), and neutrophil (p < 0.001) on postoperative day one; higher serum levels of PCT (p < 0.05) and WBC (p < 0.05) on postoperative day three. ROC curves showed IL-6 (AUC = 1.000) at postoperative hour two and PCT (AUC = 0.954) on postoperative day three. Compared with EXP2 group, EXP1 group showed shorter time to intervene (p < 0.001), a shorter postoperative hospital stay (p < 0.001), and a lower incidence rate of severe urosepsis (p < 0.05). There were different diagnostic abilities of IL-6, PCT, WBC, and neutrophil for postoperative urosepsis at different time points, and IL-6 was greatly valuable as a predictive and early diagnosing biomarker for postoperative urosepsis in patients undergoing PCNL at postoperative hour two and on postoperative day one.

The retrospective study of perioperative application of dexamethasone and furosemide for postoperative anti-inflammation in patients undergoing percutaneous nephrolithotomy

Objectives

To investigate whether the perioperatively combined application of dexamethasone and furosemide could alleviate the inflammation in patients undergoing percutaneous nephrolithotomy (PCNL).

Patients and methods

147 patients undergoing PCNL between November 2018 and October 2019 were enrolled in the study. 77 patients accepted a single dose of dexamethasone and furosemide administration (EXP group, n = 77), and 70 patients did not (CON group, n = 70). Demographic and perioperative data, inflammatory markers including interleukin-6 (IL-6) and procalcitonin (PCT), and clinical outcomes were compared between the two groups.

Results

Compared with the CON group, the incidence rate of urosepsis of the EXP group were significantly lower (11.69% vs. 24.29%, p = 0.046). 3 patients developed severe urosepsis in the EXP group, while 5 patients developed severe urosepsis in the CON group. Compared with those in the CON group, the patients with postoperative urosepsis in the EXP group showed lower serum levels of IL-6 at postoperative hour two (p = 0.045) and at postoperative day one (p = 0.031) and lower serum levels of PCT at postoperative day one (p = 0.015). There was a better clinical outcome of a shorter postoperative hospital stay (p = 0.015) in patients with postoperative urosepsis in the EXP group than in those in the CON group.

Conclusion

The perioperatively combined application of dexamethasone and furosemide was beneficial for alleviating postoperative inflammatory reaction and caused a better clinical outcome of a shorter postoperative hospital stay.

Risk factors for the fever after percutaneous nephrolithotomy: a retrospective analysis

Background

It’s very common to see the onset of fever after percutaneous nephrolithotomy (PCNL), it’s necessary to analyze the risk factors for the fever following PCNL, and to provide evidence for infection prevention after PCNL.

Methods

A total of 546 adult PCNL patients were included as study subjects and retrospective studies were performed. We collected clinical data of patients using a prospectively designed database. Univariate and multivariate logistic regression analyses were performed to identify the potential risk factors for the fever after PCNL.

Results

Of the included 546 PCNL patients, there were 82 fever patients and 464 no-fever patients following PCNL. Escherichia coli and Proteus mirabilis are the two most common infectious bacteria. Preoperative urinary tract infection (OR =4.38, 95% CI: 1.15–9.53), multiple access (OR =5.31, 95% CI: 1.23–10.75), diabetes (OR =4.97, 95% CI: 1.37–9.86), length of operation ≥60 min (OR =5.67, 95% CI: 2.24–13.42), estimated blood loss in PCNL ≥500 mL (OR=2.78, 95% CI: 2.32–3.61) were the independent risk factors associated with postoperative infection.

Conclusions

Effective control of urinary tract infection, reduction of access number, strict control of blood glucose, length of operation control, reduction of intraoperative bleeding should be considered as measures to prevent postoperative fever for patients with PCNL.

Construction of a Risk Prediction Model for Fever After Painless Bronchoscopy

Background

The aim of this study was to construct a risk prediction model for fever after painless bronchoscopy.

Material/Methods

A total of 188 patients were included, and a self-designed data collection form was used. By collecting relevant clinical data of patients before, during, and after the painless bronchoscopy, the influencing factors were analyzed through univariate analysis, and multiple logistic regression analysis was performed to construct the prediction equation, which was tested by ROC curve analysis.

Results

Of the 188 patients undergoing painless bronchoscopy, 49 had postoperative fever, and the incidence rate was 26.0%. The prediction probability model was: P=e^x/1+e^x, where e is the natural logarithm, X=−4.337+0.020×(CRP)+1.014 (whether the examination time was greater than 30 minutes)+1.912×(whether remifentanil was used during anesthesia)+1.514×(whether nausea or vomiting occurred during surgery or during recovery). The prediction sensitivity and specificity were 78.26% 76.72%, respectively.

Conclusions

Use of this risk prediction model of fever after painless bronchoscopy can improve the recognition of people at high risk of postoperative fever, and it has good ability to guide clinical nursing observation and early screening of fever after painless bronchoscopy.