Platelet-to-Lymphocyte ratio (PLR) is a Novel Independent Risk Factor for Newborn Patients in the Neonatal Intensive Care Unit (NICU)

Objective: To investigate the prognostic signicance of the platelet-to-lymphocyte ratio (PLR) for newborn patients in the neonatal intensive care unit (NICU). Design: A retrospective cohort study. Setting and participants: Data on 5240 newborn patients in the NICU extracted from the Multiparameter Intelligent Monitoring in Intensive Care III (MIMIC III) database. Methods: Spearman correlation was used to analyze the association of PLR with length of hospital and ICU stays. The chi-square test was used to analyze the association of PLR with mortality rate. Multivariable logistic regression was used to determine whether the PLR was an independent prognostic factor of mortality. The area under the receiver operating characteristic (ROC) curve was used to assess the predictive ability of models combining PLR with other variables. Results: PLR was negatively associated with length of hospital stay and ICU stay (hospital stay: Spearman’s rho=-0.416, P<0.0001; ICU stay: Spearman’s rho=-0.442, P<0.0001). PLR was signicantly correlated with hospital mortality (P<0.0001). Lower PLR was associated with higher hospital mortality (OR=0.85, 95% CI=0.75-0.95, P=0.005) and 90-day mortality (OR=0.85, 95% CI=0.76-0.96, P=0.010). The prognostic predictive ability of models combining PLR with other variables for hospital mortality was moderately good (AUC for Model 1=0.804; AUC for Model 2=0.964). Conclusion: PLR is a novel independent risk factor for newborn patients in the NICU. care unit; PLR, platelet-to-lymphocyte ratio; RBC, red blood cell; ROC, receiver operating characteristic curve; SOFA, Sequential Organ Failure Assessment score; SAPS II, Simplied Acute Physiology Score II; SQL, Structured Query Language; WBC, white blood cell.


Introduction
The rst month of life is riskiest time for child survival, accounting for approximately 40% of all childhood mortality [1][2][3] . Each year, 2.6 million neonates die globally, with 75% of neonatal deaths occurring in the rst week of life and 99% of deaths occurring in low-and middle-income countries 3,4 . Premature delivery and birth-related complications (such as birth asphyxia and neonatal sepsis) are considered to be the main causes of neonatal death 5 .
In recent years, the neutrophil-to-lymphocyte ratio (NLR) 6 , lymphocyte-to-monocyte ratio (LMR) 7 and platelet-to-lymphocyte ratio (PLR) 8 have been found to be independent predictors of prognosis in various benign and malignant conditions 9,10 . Moreover, NLR, LMR and PLR were reported to be related to the outcome of intensive care unit (ICU) patients, because of their rapid response to systemic in ammation and stress [11][12][13] . However, little is known about the associations of NLR, LMR and PLR with prognosis in newborn patients in the neonatal intensive care unit (NICU).
The primary purpose of this study was to determine the associations of NLR, LMR and PLR with hospital mortality in newborn patients in the NICU. In the present study, we were the rst to report that PLR can serve as an independent risk factor for newborn patients in the NICU.

Data Source
A retrospective cohort study design was used in this study. Data were obtained from the ICU database, a free accessible critical care database of Medical Information Mart for Intensive Care III (MIMIC-III). The clinical data of patients who stayed in the ICU of Beth Israel Deaconess Medical Center (BIDMC) between 2001 and 2012 were selected 14 . The institutional review boards of both the BIDMC and the Massachusetts Institute of Technology A liates approved the access to the database. No informed consent was required because all of the data were deidenti ed.

Patient Selection
Clinical data of eligible patients in the MIMIC-III database were selected for analysis in this study. The eligibility criteria were (1) newborn patents admitted to the NICU; and (2) patients with routine preoperative blood examinations within 24 hours of admission.

Data Extraction
All of the data were obtained and extracted by using the Structured Query Language (SQL), and pgAdmin4 for PostgreSQL was used as the administrative platform. The extracted data mainly included age, sex, birthweight, heart rate (HR), laboratory parameters (red blood cell count (RBC), peripheral white blood cell count (WBC), platelet count, lymphocyte count, neutrophil count, monocytes count), comorbidities (congestive heart failure, cardiac arrhythmias, valvular disease, pulmonary circulation disorder, hypertension, liver disease and renal failure), the Simpli ed Acute Physiology Score (SAPS) II, the Sequential Organ Failure Assessment (SOFA) score and the model for end-stage liver disease (MELD) score. PLR was calculated by dividing the platelet count by the lymphocyte count. NLR was calculated by dividing the neutrophil count by the lymphocyte count. LMR was calculated by dividing the lymphocyte count by the monocyte count. Given that the proportion of missing data for each variable was < 1.5%, we directly omitted these data in further analyses.

Outcome Variables
The following outcome variables were extracted: hospital mortality, length of ICU stay, length of hospital stay and 90-day mortality (post-ICU admission). Because a patient may have had more than one ICU admission during a single hospitalization, the length of ICU stay was entirely determined by the rst ICU hospitalization.

Statistical Analysis
Continuous variables are presented as the mean ± standard deviation or the median (interquartile range), and were compared via t-test or the Mann-Whitney U test. Categorical data are presented as numbers with proportions and were analyzed via the χ 2 test. Correlation of length of ICU stay and hospital stay with the laboratory parameters were assessed with the nonparametric Spearman's rank correlation test. Logistic regression with the univariate and multivariate analyses was used to identify independent prognostic factors of mortality (hospital mortality and 90-day mortality) for newborn NICU patients. Two different models were designed to adjust for potential confounders. Model 1 was adjusted for NLR, LMR, MELD, SAPS II and liver disease. Moreover, Model 2 was adjusted for NLR, LMR and MELD. Receiver operating characteristic (ROC) curves were constructed, and the area under the curve (AUC), sensitivity and speci city were calculated. P-values of less than 0.05 were considered to indicate statistical signi cance.

Baseline Characteristics of the Study Population
In total, 5240 patients who met the selection criteria were enrolled in our study, among whom 43 patients (0.82%) died in the hospital. The baseline characteristics of the enrolled patients are summarized in the Table 1.  In the present study, the correlation of the in ammatory markers with the mortality in the newborn patients in the NICU was investigated. Quartiles of LMR, NLR and PLR were signi cantly correlated with hospital mortality (all P < 0.0001) ( Table 3). A higher rate of hospital mortality was observed in patients in the fourth LMR quartile than is those in the rst, second and third quartiles. For NLR and PLR, a higher rate of hospital mortality was observed in patients in the rst quartile than in those in other quartiles.  The results of the multivariate analysis are summarized in Table 5, and only PLR was signi cantly associated with hospital mortality ( Meld. The data suggested that PLR could be an independent risk factor for hospital mortality and 90-day mortality in the newborn patients in the NICU.  (Fig. 1).

Discussion
In the present study, we found that LMR was signi cantly positively was associated with length of hospital stay and ICU stay while both NLR and PLR were negatively associated with length of hospital stay and ICU stay. PLR, NLR and LMR were associated with hospital mortality and 90-day mortality, but only PLR was signi cantly associated with hospital mortality and 90-day mortality in the multivariate analyses. The prognostic predictive ability of models combining PLR with other variables for hospital mortality was moderately good. To our knowledge, this is the rst investigation to demonstrate that PLR can serve as an independent risky factor for newborn patients of NICU.
In ammatory cells, including WBCs and their subtypes (such as lymphocytes, monocytes and neutrophils), have been well validated to play an indispensable role in various benign and malignant conditions. Moreover, platelets could play a critical role in the immunomodulatory and in ammatory process 15 , by inducing the release of in ammatory cytokines and interacting with different kinds of immune cells, such as neutrophils, T-lymphocytes, and macrophages (the precursors of macrophages are monocytes), which contribute to the initiation or exacerbation of the in ammatory process 16,17 . Low lymphocyte counts could represent a suppressed immune and in ammatory response, which is related to in ammatory disease 18, 19 . Thus, PLR was proposed to serve as a novel systematic in ammatory indicator 20 .
The association between PLR and outcomes was different in different cohorts. Both high and low PLR were associated with increased mortality, among critically ill patients with acute kidney injury (AKI) 21 . In another study, high PLR was positively associated with increased epicardial adipose tissue deposition in diabetes patients 22 . Wang et al showed that high PLR was independently associated with shorter disease-free days and lower overall survival rates in lung adenosquamous carcinoma 23 . For fetal malnutrition, cord-blood PLR negatively correlated with term fetal malnutrition gestational age neonates 24 . Maternal PLR was negatively correlated with the week of birth and birth weight of the infant 25 . Our data showed that PLR was associated with the prognosis of newborn patients in the NICU. PLR was negatively associated with length of hospital stay and ICU stay. A higher rate of hospital mortality was observed in patients in the rst PLR quartile than in those in the other quartiles. The prognostic predictive ability of models combining PLR with other variables for hospital mortality was moderately good (AUC for Model1 = 0.804; AUC for Model2 = 0.964). These data suggested that PLR can serve as an independent risky factor for newborn patients in the NICU.

Conclusion
In summary, we demonstrated that lower PLR was signi cantly associated with higher hospital mortality. The prognostic predictive ability of models combining PLR with other variables for hospital mortality was moderately good. PLR is a novel independent risk factor for newborn patients in the NICU.
List of Abbreviations AKI, acute kidney injury; AUC, area under the curve; BIDMC, Beth Israel Deaconess Medical Center, HR, heart rate; ICU, intensive care unit; LMR, lymphocyte-to-monocyte ratio; MIT, Massachusettes Institute of Technology; MELD, model for end-stage liver disease score; MIMIC III, Multiparameter Intelligent Monitoring in Intensive Care III database; NLR, neutrophil-to-lymphocyte ratio; NICU, neonatal intensive care unit; PLR, platelet-to-lymphocyte ratio; RBC, red blood cell; ROC, receiver operating characteristic curve; SOFA, Sequential Organ Failure Assessment score; SAPS II, Simpli ed Acute Physiology Score II; SQL, Structured Query Language; WBC, white blood cell.

Declarations
Ethics approval and consent to participate The institutional review boards of the MIT (Cambridge, Massachusetts) and BIDMC (Boston, Massachusetts) reviewed and approved studies involving human participants. According to national laws and institutional requirements, this study does not require written informed consent.