Evaluation of urinary inflammatory index in rapid screening of urinary tract infection

The objective of this study was to assess the diagnosis value of urinary inflammatory index (UII) and systemic immune-inflammation index (SII) for UTI. Nine inflammatory indexes including neutrophil-to-lymphocyte ratio, platelet-to-lymphocyte ratio, SII and six UIIs were calculated for Receiver operating characteristic curve analysis to select which one is suitable for the screening of UTIs or distinguishing the types of bacteria. UII3, which calculated from leucocyte esterase (LE), nitrite, white blood cells and bacteria, was preferentially used as an indicator for the diagnosis of UTI when the threshold was set at 0.53. UII2 was more suitable for the distinction between groups when the cutoff is set to 0.94. Appropriate urinary inflammation index calculated by rapid urinalysis of urine dipstick and urine sediment can help us to predict urinary tract infection and bacterial type, and reduce the workload and costs of urine culture.

For the agents involved in the positive urine cultures, Escherichia coli is followed in prevalence by Enterococcus, fungi, Klebsiella, Streptococcus, Proteus, Pseudomonas, Staphylococcus, Enterobacter and Acinetobacter enter the top ten, and there was no significant difference in the infection rate among four seasons (Fig. 1). Then we analyzed the non-negative rate of each item of urinalysis among these bacteria, WBCs and LE were higher in each group, other indicators such as nitrite only had a higher positive rate in Escherichia, Klebsiella, Pseudomonas and Enterobacter, but almost no value for fungi and Streptococcus. This is in accordance with the characteristics of the reduction of nitrate to nitrite by Enterobacteriaceae ( Table 1). The WBCs of fungi, Klebsiella and Pseudomonas were significantly higher than those of other genera (P < 0.001), there was no significant difference in WBCs between Gram-negative and Gram-positive bacteria (P = 0.6476) ( Supplementary Fig. 2).
Diagnostic performance of UII and SII for UTIs. The inflammation indexes for SII, NLR, PLR and different UIIs were calculated based on the data in 2019. The ROC curves and AUC for theses indexes are shown in Fig. 2A,B and Table 2. The diagnostic value of NLR, PLR and SII will not be considered in the following calculation due to their smaller AUC. Compared with the normal controls, the negative controls had lower AUC for UII1 to UII6. UII2 has the highest sensitivity of 88.4% at the cutoff of 0.44 while UII4 has the highest specificity of 99.3% at the cutoff of 0.14. UII3, with the highest AUC of 0.927 while the sensitivity and specificity were 84.3% and 95% at the cutoff of 0.53, become the preference diagnostic index for UTI screening. Meanwhile, the performance of individual items including LE, nitrite, WBCs, bacteria, WBCC and ECs were evaluated by the ROC curve. It is not difficult to see from Fig. 2C,D that LE and WBCs have higher AUC, keeping pace with the high non-negative rate in Table 1.
The value of UII in the diagnosis of various bacteria is different. Escherichia, fungi, Klebsiella, Pseudomonas and Enterobacter have higher UII especially for UII2, UII3 and UII4, while Streptococcus, Staphylococcus and Acinetobacter have lower UII (Fig. 3A). In the same way, we found that the UIIs of Gram-negative bacteria were significantly higher than those of Gram-positive bacteria, and both of them were higher than those of the negative control and the normal control ( Fig. 3B) (P < 0.001). Therefore, we may predict the bacterial type of UTI by UII values and clinical symptoms. Then, we chose UII2, UII3 and UII4 to distinguish the bacterial groups which group A contains Escherichia, Enterococcus, fungi, Klebsiella, Pseudomonas and Enterobacter while group B consists of Streptococcus, Staphylococcus and Acinetobacter. According to ROC curve analysis, AUCs of UII2, UII3 and UII4 were 0.623, 0.647 and 0.642 respectively (P < 0.001). Youden index analysis shows that when we set the cutoff of UII2, UII3 and UII4 at 0.94, 1.81 and 1.52 respectively, the sensitivity and specificity were 81.7% and 62.3% for UII2, 56.1% and 33.6% for UII3, 57.4% and 37.4% for UII4 (P < 0.001). Therefore, UII2 is more suitable for the distinction between groups when the cutoff is set to 0.94 ( Supplementary Fig. 3). www.nature.com/scientificreports/

Discussion
It is generally believed that when UTI occurs, it is often accompanied by inflammation, and the number of red blood cells (RBCs), WBCs and bacteria in urine will increase. In addition to UTI, urinary WBCs may also come from female genital tract pollution, while some patients with low immunity may have normal urinary WBCs, and bacterial may also come from specimen contamination with periurethral, epidermal, perianal, and vaginal flora 15 , so the clinical diagnosis of UTI still mainly depends on the quantitative culture of bacteria in urine 16 .
In addition, the discovery of the female urinary microbiota makes the sterile urine paradigm is no longer valid. Unfortunately, Culture-dependent techniques are severely limited because the vast majority of bacteria are not or cannot be cultured by standard clinical laboratory techniques 17 . Therefore, we hope to judge whether there is UTI by the results of routine urinalysis. Rapid urine tests, such as microscopy, for bacteria and WBCs, and dipsticks, for LE and nitrite, are often used to guide early diagnosis and treatment of UTI. Many researchers have tried to compare the correlation between rapid urinalysis and positive urine culture. Generally, LE is more sensitive than the nitrite test in screening for UTI 18 because of the high detection limit of the commercial dipstick. Urinary dipsticks are an effective rapid test for screening for asymptomatic, catheter-associated UTIs in intensive care unit patients 11 . However, traditional indicators such as LE and nitrite have high specificity but low sensitivity in the diagnosis of UTI, so their limited diagnostic value when used alone does not lead to satisfactory results. In this study, the non-negative rates of nitrite, WBCC, bacteria and ECs were all very low, which was not a powerful evidence for UTI screening, as reported by Koeijers et al. 19 . Another study reported the nitrite/nitrate ratio determined by the LC-MS/MS was significantly more sensitive (95%) and exhibited a satisfactory specificity (91%) in the screening of UIT 20 . Although WBCs and LE had high non-negative rates in various bacterial infections, they are not suitable for screening UTI alone. In order to improve the sensitivity and specificity of diagnosis, we combined the indicators related to inflammation in urinalysis to make a comprehensive judgment, so the UII came into being.
Multiple inflammation indexes such as NLR, PLR, SII and UII were calculated in this study for the screening of UTIs, only UII2, UII3, UII4 and UII6 with a higher AUC (> 0.9) while UII2 had the highest sensitivity of 88.2% and UII4 had the highest specificity of 99.3% at the cutoff of 0.44 and 0.18, respectively. NLR, PLR and SII are widely used in various tumors as useful prognostic indicators [21][22][23] . However, they seem to have no value in predicting UTIs, this suggests that UTIs are mostly focal inflammation, except for complicated UTI which is often accompanied by systemic diseases such as renal insufficiency, transplantation, diabetes or immune-deficiency 4 . LE, nitrite, WBCs and bacteria participate in the calculation of UII2, UII3, and UII4. This reminds us that these Furthermore, a culture that shows significant bacterial growth may not reflect an active infection, for example, asymptomatic bacteriuria occurs in about 10% of community-dwelling older women 24 . Pyuria is a nonspecific finding that is frequent in older patients with or without bacteriuria 25 , while the negative predictive value can reach 95% or more to rule out infection if it is absent 26 .
The clinical treatment of UTIs mainly depend on experience and the adjustment of drug use according to the results of antimicrobial susceptibility test. With the spread of antibiotic resistance and its increasing threat to public health caused by UTIs became the second most common diagnosis for empirical antibiotics, national guidelines and antimicrobial management programs have been proposed to meet these challenges 27 . If we can predict the types of pathogens in advance, it will be of great value for the guidance of early experience drug use. In this study, UII was calculated by rapid urine test for the first time and compared with the results of urine culture. The AUC of 0.927 for prominent applications of UII3 which integrated the characteristics of four items and became the preferred prediction index according to Swets 28 . To our knowledge, the current study is the only report to focus on the value evaluation of UII in UTI screening.
UII can be used not only to judge the existence of UTI, but also to roughly identify the types of bacteria. In summary, prediction of UTIs by simple calculation of UII might be helpful in significantly reducing workload and costs of urine culture.

Methods
Urine culture and urinalysis. All patients are required to take midstream urine in sterile containers and submitted for bacterial culture. In the microbial laboratory, the uropathogens were isolated by using 1ul quantitative inoculation ring evenly smeared on the Columbia blood agar plate, cultured at 35 °C for 18-24 h, and continue to culture until 48 h when there is no bacterial growth. Then, they were identified by Vitek-2 compact Complete automated ID/AST platform and MALDI-TOF mass spectrometry (bioMérieux, France), growth of ≥ 10 5 CFU/ml was considered to be a positive result of a urine culture 4 . In the instance of culture of > 1 microorganism, only the predominant microorganism was considered 19 .Urinalysis were performed for urine chemistry and microscopy values by the iRICELL3000 (IRIS diagnostic) automatic urinalysis line which were designed to streamline the UTI testing workflow by generating work lists of possible UTI patient sample values. The indicators measured by iChemVelocity were LE and nitrite, and the iQ200 module assessed were WBCs, white blood cell clot (WBCC), bacteria and epithelial cells (ECs).
Data collection and UII. A total of 7279 positive urine culture results were collected from the laboratory information system between January 1, 2016, and Dec 31, 2019. The 1597 results in total of blood cell analysis, urine dipstick and urine sediment in 2019 were collected simultaneously. Considering the changes of bacterial virulence, we gathered the blood cell analysis, urine dipstick and urine sediment results within 2 days before or after the urine culture 30 . According to the principle of randomization, we selected 500 outpatients or inpatients with negative urine culture results and 500 normal physical examination results as controls, those samples that are tested repeatedly or cultured to be positive again will be removed. Urine and venous blood were obtained and examined in the clinical laboratory Department of Affiliated hospital of Nanjing university of Traditional Chinese Medicine. Data collection was approved by the institutional review board and the Ethics Committee of Table 3. Basic characteristics and result representation of each urinary item. a he results of leucocyte esterase and nitrite were obtained by urine dipstick analysis while other results were obtained by urine sediment test. b To facilitate UII calculation, the report format of all parameters including quantitative and ordinal data are transformed to converted format. Ordinal data such as negative, ± , 1+, 2+, 3+, 4+ represent 0, 0.5, 1, 2, 3, 4, respectively. In addition, the ordinal data of white blood cell clot were converted in the same way. c The rank data represent for within 1, 5, 10, 20, 30, 40, 50 and more than 50 times of the upper limit of the reference interval, respectively.