Lymphocyte Count and Serum PCT L ev els Improve the Sensitivity of CURB-65 in Prediction of 30-day Mortality of Patients with Community-acquired Pneumonia

Background Community-acquired pneumonia (CAP) is an infectious disease with high morbidity and mortality worldwide. The CURB-65 score and other blood biomarkers were used to evaluate prognosis of hospitalized patients with CAP. The aim of our study is t o evaluate the prognostic values of other blood biomarkers and the CURB-65 score in hospitalized patients with community-acquired pneumonia (CAP). Methods A retrospective study of clinical data of in-hospitalized adult CAP patients who fulfilled the CAP criteria and were admitted to the Second Affiliated Hospital of Anhui medical university between January 2015 and December 2018 was conducted. The CURB-65 score was calculated, and other biomarkers including blood lymphocyte countat and serum PCT (a propeptide of calcitonin) level were collected at enrolment. Logistic regression analysis was performed to develop combined models to predict 30-day mortality in overall hospitalized and/or the ICU admission of CAP patients. Then, receiver operating characteristics curve (ROC) analysis was conducted to measure and compare the prognostic values of the CURB-65 score and biomarkers in the combined models. new

day mortality. This new CURB-65LP scoring system is simple, but more accurate for evaluating the severity of CAP with higher sensitivity and specificity than the current CURB-65 scoring system.

Background
Community-acquired pneumonia (CAP) is a common infectious disease with high morbidity and mortality worldwide [1][2][3]. The rates of adult patients hospitalized due to CAP are increasing by 22%-42%. In addition, approximately 5% of CAP patients require admission to an intensive care unit (ICU). In these severe cases, mortality rises to 35% [4].
Multiple biomarkers and several established models have been used to assess the severity of CAP to improve management of CAP patients [5][6][7], and among these models, the CURB-65 score is recommended in the guidelines for the diagnosis and treatment of adult CAP in Europe, America and China. The CURB-65 scoring system was created by the British thoracic society in 2003 and requires only five parameters (including confusion, urea >7 mmol/L, respiratory rate ≥30/min, low systolic < 90 mmHg or diastolic ≤ 60 mmHg blood pressure, age≥65 years) with 1 point for each item and a total maximum score of 5 points. The higher the score, the more serious the disease [8]. The CURB-65 score is convenient and simple, and often used in clinical practice. However, it does not include an immune background of CAP patients and other biomarkers, its clinical value is limited [9]. Therefore, a simple and more reliable scoring system is needed.
Lymphocytes are an important cellular component of the immune system and are produced by lymphoid organs. Lymphopenia is considered to be a total lymphocyte count fewer than 1000 mm³ by the definition proposed in Hematology, 9th ed [10], and usually associated with immune deficiency [11].
PCT is a propeptide of calcitonin without hormone activity. The dynamic changes in its serum levels are closely related to bacterial infection and severity. The serum level of PCT in normal subjects is less than 0.046 ng/ml, and begin to increase after endotoxin injection and reach at the peak 24 hours later [12]. The role of PCT in infectious diseases such as sepsis has been described by Assicot et al. [13]. Thus, it can be used as a prognostic indicator for infectious diseases [7,14].
In the present study, we evaluated the prognostic values of lymphocyte count, serum PCT level and other biomarkers alone or in combination with the CURB-65 scoring in patients with CAP, and aimed to build a new model that can more accurately assess the prognosis of CAP.

Statistical analysis
Differences in the demographic clinical characteristics between patient cohorts or sub-groups were calculated using the medians with interquartile ranges. Categorical variables are presented as numbers and percentages. Data from the vital signs and laboratory results were assessed by t-tests or one-way analysis of variance(One-Way ANOVA). Variables yielding a P value <0.05 on t-tests were further included in the logistic regression analysis. Potential confounding variables were selected by assessing the association between those variables shown in Table 1 with 30-day mortality and ICU admission. Variables with significant differences were tested by a binary logistic analysis together with the CURB-65 score and are presented with an odds ratio (OR) and a 95% confidence interval (CI). The association between lymphocyte and PCT (procalcitonin) levels and the risk of 30-day mortality and ICU admission were evaluated by logistic regression analysis [11]. The ability of lymphocyte counts and procalcitonin (PCT) levels to differentiate survivors from non-survivors and ICU admission from non-ICU were assessed by using area under the receiver operating characteristic curve analysis (AUC). All analysis were performed with IBM SPSS Statistics 24.0.

Results
During the study period, 201 adult patients were admitted to The Second Affiliated Hospital of Anhui Medical University. The baseline characteristics of these patients were shown in Table S1. The median age of these patients was 66 (51-77) years old, and 64.7% were male. The top three main complications were chronic heart failure (31.0%),chronic obstructive pulmonary disease (COPD) (25.4%) and diabetes mellitus (19.4%). The overall 30-day mortality rate was 12.9%, and the proportion of  (Table 1).
PCT level and lymphocyte count as good predictors for 30-day mortality and ICU admission Biomarkers in the blood including lymphocyte count, serum PCT level, serum albumin level and activity of LDH were analysed. As shown in the Table 2, PCT levels in the non-survival and in the ICU admission groups were 6.31(4.67-8.41) and 4.25(3.23-6.25) ng/mL, significantly higher than 1.87(0.63-3.36) and 1.56(0.39-3.11) ng/mL in the survival and in the non-ICU admission groups (p<0.001 both), while lymphocyte counts in the non-survival and in the ICU admission groups were significantly lower than those in the survival and in the non-ICU admission groups  (Table 2). These results suggested that these biomarkers were potential predictors for 30-day mortality and ICU admission.
For prediction of 30-day mortality and ICU admission, these biomarkers alone were analysed on the basis of the ROC curves. As shown in the Table 3, PCT level or lymphocyte count alone had better performance, with the sensitivity and specificity of 81% and 90%, 71% and 77%, respectively, in the prediction of 30-day mortality; and 94% and 60%, 65% and 68%, respectively, in the prediction of ICU admission. Incorporation of lymphocyte count, or PCT level or both also increased the sensitivity and specificity of the CURB-65 scoring system. As shown in Table 4 and Fig 1, a CURB-65 score ≥ 2 points presented with sensitivity at 65% and specificity at 73% for predicting 30-day mortality, while a CURB-65L score ≥ 3 points, CURB-65P score ≥ 3 points and CURB-65LP score ≥ 4 points presented sensitivity at 65%, 62% and 81%, and specificity at 98%, 85% and 68%, respectively, for predicting 30day mortality. Similarly, a CURB-65 score ≥ 2 points presented with sensitivity at 91% and specificity at 54% for predicting ICU admission, while a CURB-65L score ≥ 3 points, CURB-65P score ≥ 3 points and CURB-65LP score ≥ 4 points revealed sensitivity at 83%, 82% and 79%, and specificity at 66% 71% and 77% for predicting ICU admission.

Score performance of new models
These results indicate that both CURB-65L, CURB-65P and CURB-65LP models were more sensitive than CURB-65 scoring, with the highest sensitivity in the CURB-65LP model, in predicting 30-day mortality.

Discussion
In this study, the 30-day mortality rate of CAP patients was 12.9%, higher than that reported previously [16]. This may be ascribed to better economic conditions, better medical conditions and strong health awareness in developed countries where previous study was conducted [17]. In addition, our hospital is a third-grade hospital in Anhui Province that frequently admits older patients; thus, the patients are generally in a worse condition than those admitted to community hospitals. A hospitalization delay often happens, also contributing to the poor prognosis.
The higher mortality in this study is associated not only with the above factors, but also with the exclusion criteria used in this study. Advanced age, a history of COPD and cerebrovascular disease, and ventilator use are independent risk factors for 30day mortality. The correlation between sex and mortality was consistent with the meta-analysis of Fine et al. [18].
Early identification of CAP patients at risk of death is an important step in the diagnosis and treatment process [19]. Current accepted CAP scoring systems include CURB-65 and PSI. PSI has high sensitivity and specificity, but it needs to be combined with 20 clinical parameters (including age, sex, nursing staff, tumour, etc.) and thus is complex and extremely inconvenient to apply in the clinic. The CURB-65 is more convenient and simpler, and often used in clinical practice.
However, blood pressure and heart rate are easily affected by the environment, leading to interference with the assessment of the severity of CAP [20]. Our study indicate that the CURB-65 scoring system is an independent predictor of CAP patients' 30-day mortality, but the AUC was only 0.73 (0.63-0.83) and the area under the curve was lower than that from Chalmers et al. [21], suggesting that the CURB-65 scoring system has a limited predictive value for 30-day mortality. Among biomarkers analysed in this study, lymphocyte counts were lower in the nonsurvivors and ICU admissions than in the survivors and non-ICU patients. In the research by Marrie et al [22], a lymphocyte count < 1000 cell/mm³ is associated with early but not late mortality. The most important difference is that patients with lymphocyte counts < 1000 cells/mm³ and those with critical illnesses are excluded in their study. In addition, a hospital mortality is analysed in their study, but not 30day mortality as analysed in this study. The value of PCT in predicting 30-day mortality was superior to that of CRP, which is consistent with a report by Kruger et al [23]. LDH and albumin are also independently associated with 30-day mortality in CAP patients [24,25].
In the present study, we established new models, in which lymphocyte count and PCT alone improve the predictive efficacy of the CURB-65 scoring system. The addition of both PCT and lymphocyte count to form the CURB-65LP scoring system can further improve the predictive value, especially in the prediction of 30-day mortality. The area under the ROC curve of the CURB-65LP scoring system was 0.84 (0.77-0.91), much higher than that of CURB-65 (0.73 (0.63-0.83). The new model can more comprehensively reflect the severity of a CAP patient's condition from multiple aspects such as autoimmune status and degree of inflammatory response, Lymphopenia could be a cause or a consequence of CAP. Impaired production or increased apoptosis of lymphocytes leads to critical illness [26]. However, the mechanisms still needs further study.
Although our new models show higher predictive value than the CURB-65 scoring, limitations exist in our study. First, it was a single-centre secondary analysis and the sample size was relatively small; thus, multi-centre prospective studies are needed to validate the clinical value of lymphocyte counts and PCT. Second, the criteria for admission to the ICU vary widely in different hospitals. Third, we did not explore the influence of antibiotic treatment on these serum biomarkers because of insufficient data. ROLE IN LUNG INFECTIONS. IT IS GENERALLY BELIEVED THAT ELEVATED LYMPHOCYTE  INNATE IMMUNE RESPONSE MECHANISMS OF HUMAN INFLAMMATION. AS CAN BE   SEEN IN OUR STUDY, SOME OF THE PATIENTS WITH BACTERIAL INFECTION, THE   LYMPHOCYTE COUNT IS DECREASED, ACCOMPANIED BY AN INCREASE IN   PROCALCITONIN LEVEL, AND PROCALCITONIN IS AN INDICATOR OF BACTERIAL   INFECTION, WHICH MAY FURTHER EXPLAIN THAT IF THE  Corresponding author Correspondence to zhaodahai@ahmu.edu.cn or xujiegou@ahmu.edu.cn.

Ethics approval and consent to participate
The study was approved by the Ethics Committee of the Second Affiliated Hospital of Anhui Medical University (approval number is PJ-YX2019-036), and written informed consent was received from all of the participants or their relatives after explaining the objectives and details of the study to them.

Consent for publication
Not applicable.

Competing of interests
All the authors declare that they have no competing interests.  Variables did not conform to the normal distribution and were described by the median (25 th percentile, 75 th percentile). The T test was used for comparisons between the two groups.  Supplemental materials.docx