Diagnostic Accuracy and Prognostic Value of Neutrophil-to-Lymphocyte and Platelet-to-Lymphocyte Ratios in Septic Patients outside the Intensive Care Unit

Background and Objectives: The aim of this study was to evaluate the diagnostic accuracy and prognostic value of neutrophil-to-lymphocyte (NLR) and platelet-to-lymphocyte (PLR) ratios and to compare them with other biomarkers and clinical scores of sepsis outside the intensive care unit. Materials and methods: In this retrospective study, 251 patients with sepsis and 126 patients with infection other than sepsis were enrolled. NLR and PLR were calculated as the ratio between absolute values of neutrophils, lymphocytes, and platelets by complete blood counts performed on whole blood by Sysmex XE-9000 (Dasit, Italy) following the manufacturer’s instruction. Results: The best NLR value in diagnosis of sepsis was 7.97 with sensibility, specificity, AUC, PPV, and NPV of 64.26%, 80.16%, 0.74 (p < 0.001), 86.49%, and 53.18%, respectively. The diagnostic role of NLR significantly increases when PLR, C-reactive protein (PCR), procalcitonin (PCT), and mid-regional pro-adrenomedullin (MR-proADM) values, as well as systemic inflammatory re-sponse syndrome (SIRS), sequential organ failure assessment (SOFA), and quick-sequential organ failure assessment (qSOFA) scores, were added to the model. The best value of NLR in predicting 90-day mortality was 9.05 with sensibility, specificity, AUC, PPV, and NPV of 69.57%, 61.44%, 0.66 (p < 0.0001), 28.9%, and 89.9%, respectively. Sensibility, specificity, AUC, PPV, and NPV of NLR increase if PLR, PCR, PCT, MR-proADM, SIRS, qSOFA, and SOFA scores are added to NLR. Conclusions: NLR and PLR represent a widely useful and cheap tool in diagnosis and in predict-ing 90-day mortality in patients with sepsis.


Introduction
Sepsis is a systemic syndrome induced by infection and leading to a widespread inflammation up to septic shock, multi organ failure, and death [1,2]. Patients with bacteriemia, sepsis, and septic shock presented a high mortality rate ranging from 25% to 30% and 40% to 50%, respectively [3,4]. Patients' prognosis and mortality rate, however, are strictly affected by a timely performed clinical and laboratory diagnosis as well as by proper therapeutic management [5][6][7].
The aim of this study was to evaluate the diagnostic accuracy and prognostic value of NLR, PLR, and MLR in patients with sepsis and septic shock outside the intensive care unit (ICU) and to compare them with C-reactive protein (CRP), PCT, MR-proADM, SIRS, qSOFA, and SOFA scores.
Furthermore, we evaluated the role of NLR in aetiological diagnosis of sepsis and on length of stay stratification.

Materials and Methods
The study was approved on 23 July 2016 by the Ethical Committee of the University Hospital Campus Bio-Medico of Rome (28.16 TS Com Et CBM). All methods were performed in accordance with the relevant guidelines and regulations. Informed consent was not required for the retrospective design of the study.

Patients Selection and Study Design
Consecutive patients with clinically suspected sepsis or septic shock admitted to the Diagnostic and Therapeutic Medicine Department and General Surgery of the University Hospital Campus Bio-Medico of Rome were retrospectively enrolled between May 2014 and February 2021.
Exclusion criteria were age < 18 years and pregnancy. The control group included patients with infection, but without sepsis admitted to the Diagnostic and Therapeutic Medicine Department between May 2014 and February 2021.
Diagnosis of sepsis was performed according to the Third Consensus Conference Criteria of 2016 when qSOFA or SOFA scores were ≥2 from the baseline in the presence of an infection.
Bloodstream infection was defined as any positive blood culture for pathogens. Pneumonia was defined based on a positive pathogen respiratory culture and other Infectious Diseases Society of America (IDSA) diagnostic criteria [45]. Patients with positive urine cultures were identified as cases based on the CDC National Healthcare Safety Network (NHSN) UTI case definitions [46].

Laboratory and Microbiological Parameters
Complete blood counts (CBCs) were performed on whole blood by Sysmex XE-9000 (Dasit, Italy) following the manufacturer's instruction. NLR, PLR, and MLR were calculated by the ratio between absolute values of neutrophils, lymphocytes, monocytes, respectively, and that of platelets.
CRP protein was measured by Alinity c (Abbott, diagnostics) following the manufacturer's instruction.
Blood specimens from patients were collected in BACTEC bottles containing anaerobic or aerobic broth and resins. Blood culture bottles were incubated in BACTEC FX instrument (Becton Dickinson, Meylan, France) until they were positive for bacterial growth or for a maximum of 5 days. Positive samples were cultivated in selective agar media. Growing colonies were identified by MALDI-TOF (Brahms) [5,21,25,26]. Selective and non-selective media were used for microbiological cultures.

Statistical Analysis
Data were analysed using Med-Calc 11.6.1.0 statistical package (MedCalc Software, Mariakerke, Belgium). Receiver operating characteristic (ROC) analysis was performed among independent variables associated with sepsis to define the cutoff point for NLR, PLR, plasma PCR, PCT, MR-proADM, SIRS, SOFA, and qSOFA score values. ROC curves and areas under the curve (AUCs) were calculated for all markers and compared in patients with sepsis or septic shock versus control patients.
χ2 for proportions test was used to compare the relative percentage of patients with positivity and/or negativity to SIRS criteria, SOFA score, qSOFA score, and other demographic characteristics of septic patients and control patients.
Positive predictive value (PPV) and negative predictive value (NPV) were calculated for each variable, based on sensitivity, specificity, and disease prevalence. Younden Index was used for cut-off selection.
The multivariate logistic regression model is performed to evaluate the association between all evaluable laboratory markers and 90-day mortality.
Mann-Whitney test was used for median values' comparison. p-value < 0.05 was considered significant.

Baseline Patients' Characteristics
Demographic and clinical characteristics of patients with sepsis (251 patients) and the control group (126 patients) are reported in Table 1.
Patients with sepsis were younger than the control group (73 vs. 80, p = 0.001), while roughly half of the patients in both groups were male (52.6 vs. 50.4%, p = 0.771). The vast majority of baseline patients' characteristics were similar between septic patients and control group (Table 1), except for the presence of presence of cancer and chronic lung disease that was more (36.7% vs. 23.8%, p = 0.016) and less frequent (23.1 vs. 34.1, p = 0.031), respectively, in the former.
The median length of stay was higher in septic patients than the control group (15 days (IQR, 11 to 26) vs. 10 days (IQR 7 to 13), p ≤ 0.001) and a significantly higher proportion of patients with sepsis died during 90-day follow-up (27.5% vs. 0.8%, p < 0.001).

Diagnostic Role of NLR
For the diagnosis of sepsis, the best value of NLR was 7.97 with sensibility of 64.26%, specificity of 80.16%, AUC of 0.74 (p < 0.001), PPV of 86.49%, and NPV of 53.18%. The ROC curve is reported in Figure 1A. In Table 2, the diagnostic role of NLR is compared with that of PLR, PCR, PCT, and MR-proADM, as well as with that of SIRS, q-SOFA, and SOFA scores. MLR did not reach a significant role in the diagnosis of sepsis.
curve is reported in Figure 1A. In Table 2, the diagnostic role of NLR is compared with that of PLR, PCR, PCT, and MR-proADM, as well as with that of SIRS, q-SOFA, and SOFA scores. MLR did not reach a significant role in the diagnosis of sepsis.  Area under the curve (AUC); positive predictive value (PPV); negative predictive value (NPV). NLR, neutrophil-to-lymphocyte; PLR, platelet-to-lymphocyte; PCR, C-reactive protein; PCT, procalcitonin; MR-proADM, mid-regional pro-adrenomedullin; SIRS, systemic inflammatory response syndrome; SOFA, sequential organ failure assessment.
The diagnostic role of NLR significantly increases when PCR, PCT, and MR-proADM values, as well as SIRS, qSOFA, and SOFA scores, were added to the model (Table 3).
When just PCT and MR-proADM are considered in the diagnosis of sepsis, the model reached a PPV of 96% and a NPV of 69%. PPV and NPV for SIRS ≥2, qSOFA ≥2, and SOFA ≥2 were 96.77% and 43.3%, 99.23 and 50.21%, and 82.94% and 54.21%, respectively.
The best values of PPV and NPV are reached when NLR, PLR, and SIRS scores (99.7% and 94%, respectively), or NLR, PLR, and qSOFA scores (99.9% and 95.6%, respectively), are included in the model. NLR, MLR, and PLR did not show a significant role in aetiological diagnosis of sepsis. Conversely, our results confirm the role of PCT in aetiological diagnosis of sepsis with higher values in Gram-negative versus Gram-positive bacteria (p = 0.0022). Furthermore, MR-proADM values are significantly higher in Gram-negative (p = 0.037) and polymicrobial (p = 0.037) than Gram-positive sepsis.
The prognostic role of NLR in comparison with that of PLR, PCR, PCT, and MR-proADM values, as well as with that of SIRS, qSOFA, and SOFA scores, is listed in Table 4. Area under the curve (AUC); positive predictive value (PPV); negative predictive value (NPV). NLR, neutrophilto-lymphocyte; PCR, C-reactive protein; PCT, procalcitonin; MR-proADM, mid-regional pro-adrenomedullin; SIRS, systemic inflammatory response syndrome; SOFA, sequential organ failure assessment.
MLR was not statistically significant in the 90-day mortality prediction. Multivariate logistic regression model including all evaluable laboratory markers showed as just MR-proADM is significantly associated with 90-day mortality ( Table 6).

Discussion
The results of this study showed that NLR values of 7.97 had a good diagnostic accuracy, whereas a value of 9.05 allowed a prognostic stratification of patients with sepsis that is increased by the association with PLR values of 370.59. Conversely to PCT and MR-proADM, NLR did not help identify the type of bacterial pathogen responsible for sepsis. MLR evaluation did not yield significant results.
Patients with sepsis presented a higher 90-day mortality (27.5%) and need for ICU transfer (18.7%) than the control group. However, these proportions of patients resulted lower than data available from previous studies, where mortality and ICU transfer reached values as high as 37.5% and 80.8%, respectively [47].
Performing a complete blood count and calculating NLR and PLR in a clinical suspicion of sepsis may, therefore, help the clinician in diagnostic evaluation and prognostic stratification of patients with significant values of sensibility, specificity, PPV, NPV, and AUC (p < 0.0001). These latter values were similar to the values of PCT >0.41 and MR-proADM >1.83 and are increased by the association with PLR values (PPV of 96% and NPV of 83%) and clinical score of sepsis such as SIRS (PPV of 99.7% and NPV of 94.0%), qSOFA (PPV of 99.9% and NPV of 95.6%), and SOFA (PPV 98.3% and NPV on 91.6%). In our study, the association between NLR and SIRS or qSOFA reached higher diagnostic power than the association between NLR and SOFA. This may be related to the clinical setting; our patients, indeed, were hospitalized in a medical ward and outside the ICU.
As for sepsis diagnosis, NLR values of 9.05 showed a good role in prognostic stratification in terms of 90-day mortality. This is increased by its combination with both MR-proADM (PPV of 52% and NPV of 50%) and clinical scores of sepsis such as SIRS (PPV of 95% and NPV of 86%), q-SOFA (PPV of 96% and NPV of 88%), and SOFA (PPV of 94.6% and NPV of 55.7%).
Knowing that, the shorter the time between clinical presentation and diagnosis, the better the patients' prognosis, NLR may ameliorate septic patients' management. This, latter, further increases when the clinical score such as SIRS and qSOFA is used in association with NLR.
The results of our study certainly showed that a prompt and accurate diagnosis of sepsis may be achieved by the use of rapid, cheap, and widely performed biomarkers, as well as in those clinical setting where the use of other biomarkers may be not available or too expensive. Outside the ICU, adding information derived by these biomarkers to clinical score such as SIRS or qSOFA reached a diagnostic accuracy of about 100%.
A limitation of the study is the monocentric enrollment of patients, which should be expanded in the future to be multicentric, thus increasing the number of patients, which is limited to 251 in this first study.

Conclusions
NLR is a good, rapid, cheap, and widely performed biomarker useful in diagnosis and prognostic stratification of patients with sepsis. The association of NLR with other biomarkers and clinical scores further increases these characteristics. Only the association between clinical signs and several biomarkers may help increase the diagnostic sensibility of sepsis and predict disease severity and mortality. Biomarkers must be performed in supporting a clinical diagnosis. We hope that the use of NLR may improve the management and ameliorate the prognosis of patients with sepsis.  Informed Consent Statement: Informed consent was not required for the retrospective design of the study.

Data Availability Statement:
The data presented in this study are available on request from the corresponding author. The data are not publicly available due to their containing information that could compromise the privacy of research participants.

Conflicts of Interest:
The authors declare no conflict of interest.

AUCs
Areas under the curve CBC Complete blood counts CRP C-reactive protein ICU Intensive care unit MR-proADM Mid-regional pro-adrenomedullin PCR Polymerase chain reaction PCT Procalcitonin PPV Positive predictive value NPV Negative predictive value ROC Receiver operating characteristic SIRS Systemic inflammatory response syndrome SOFA Sequential sepsis-related organ failure assessment WBC White blood cell