Predictive value of systemic immunity index for sepsis in low-medium risk community-acquired pneumonia

CHAI Dou-dou, WANG Xiao-miao, XING Bo

1. School of Emergency Traumatology, Hainan Medical University, Haikou, 571199, China

2. Emergency department of the Second Affiliated Hospital of Hainan Medical University, Haikou, 570311, China

Keywords:

ABSTRACT Objective: To assess the predictive value of systemic immune inflammation index (SII) for sepsis in low- and medium-risk community-acquired pneumonia.Methods: A total of 589 elderly patients with low- and medium-risk community-acquired pneumonia admitted to the Emergency Department of the Second Affiliated Hospital of Hainan Medical University from January 2020 to January 2023 were included as the research subjects, and the general information and laboratory test results of the patients were collected, and the optimal cutoff value of continuous variables for predicting sepsis in elderly patients with low- and medium-risk community-acquired pneumonia was determined by plotting the receiver work characteristic (ROC) curve, which was converted into dichotomous variables and univariate and multivariate logistic Regression analysis of the influencing factors of sepsis in elderly patients with low- and medium-risk community-acquired pneumonia.Based on this, a nomogram model is constructed to predict the risk of sepsis.The differentiation, consistency and accuracy of the model were verified by calibration curve and subject operating characteristic (ROC)curve, and the clinical utility of the model was determined by decision curve analysis.Results:A total of 589 elderly patients with low- and intermediate-risk community-acquired pneumonia were included in this study, of which 96 (16.30%) developed sepsis.There were significant differences in age, diabetes mellitus and chronic obstructive pulmonary disease, Lac, PCT,SII and other indexes between sepsis and non-sepsis groups (P＜0.05).Logistics regression analysis showed that age, diabetes mellitus and chronic obstructive pulmonary disease, Lac,and SII were independent risk factors for sepsis in elderly patients with low- and mediumrisk community-acquired pneumonia.The nomogram prediction model was used to verify the results, and the AUC was 0.826 (95% CI: 0.780-0.872), and the calibration curve tended to the ideal curve with good accuracy.The decision curve shows that when the threshold of the model is between 0.10～0.78, the model has the advantage of clinical benefit.Conclusion:The nomogram prediction model constructed based on SII to predict sepsis in elderly patients with low- and medium-risk community-acquired pneumonia has good accuracy, which can predict the occurrence of sepsis early, help early identification of high-risk groups and timely intervention, and thus improve the prognosis of patients.

1.Introduction

Community-acquired pneumonia (CAP) is the leading cause of sepsis, and according to statistics, sepsis occurs in about one-third of patients with severe CAP[1].Although recent studies indicate that the incidence of sepsis has decreased with improved early warning and management of sepsis[2].However, for patients with mild CAP who do not require intensive care treatment, due to their atypical early symptoms or lack of clinical experience of the first attending physician, the severity of pneumonia cannot be accurately assessed,resulting in patients missing the optimal intervention time, which may progress from mild to severe and eventually induce sepsis,resulting in poor prognosis.Therefore, the best screening tools for early sepsis are essential.

Systemic Immune-Inflammation (SII) is a new parameter based on lymphocyte, neutrophil and platelet counts, which can more fully and objectively reflect the balance of inflammation and immunosuppression because it combines three cells into one index.Mangalesh et al[3] found that SII performed better independently in predicting sepsis mortality than individual cell counts, platelet-tolymphocyte ratio (PLR), or neutrophil-to-lymphocyte ratio (NLR)alone.SII was first identified as a good indicator of prognosis for hepatocellular carcinoma[4].Subsequently, it has also shown good prognostic value in various malignancies and cardiovascular diseases[5-7].However, little is known about the role of SII in assessing the prognosis of community-acquired pneumonia.Therefore,this study aims to analyze the predictive value of SII in sepsis in elderly patients with low- and medium-risk community-acquired pneumonia, and construct a nomogram model based on this, in order to provide an effective tool for early identification of high-risk groups of this disease.

2.Information and methodology

2.1 Research object

From January 2020 to January 2023, 589 elderly patients with community-acquired pneumonia were admitted to the Emergency Department of the Second Affiliated Hospital of Hainan Medical University as the study subjects, including 323 males and 266 females, aged 65～88 years old, with an average age of (73.02±6.96)years.Inclusion Criteria: (1) Age 65 years; (2) Meet the diagnostic criteria for community-acquired pneumonia[8] and the Pneumonia Severity Index (PSI) score is 130[9] (intermediate risk is 91～130 points, low risk 71～90, then 130 is low and intermediate risk patients); (3) No serious infection in other parts.Exclusion Criteria:(1) Age＜ 65 years; (2) Tuberculosis, lung cancer and non-infectious interstitial lung disease; (3) Previous blood system diseases,autoimmune diseases and other related diseases that affect the function of the immune system; (4) The length of hospital stay is less than 24 h; (5) Receiving immunosuppressant therapy, radiotherapy and chemotherapy, etc.; (6) Incomplete clinical data.According to whether elderly patients with low- to medium-risk communityacquired pneumonia were complicated by sepsis, they were divided into sepsis group (n=96 cases) and non-sepsis group (n=493 cases).This study is retrospective and was approved by the Ethics Committee of the Second Affiliated Hospital of Hainan Medical College (approval number: LW2023154) before data collection.

2.2 Research methods

2.2.1 Diagnostic criteria

It meets the diagnostic criteria for community-acquired pneumonia[8], including detailed medical history, detailed physical examination, improvement of relevant laboratory tests and imaging,and further exclusion of pulmonary tuberculosis and lung tumors.Then, according to the PSI score[8], relevant risk factors, including vital signs, laboratory tests, underlying diseases, imaging, etc., were collected and the score was calculated and risk stratification was performed.Among them, it is divided into 5 levels according to different scores, which are grade I (＜ 50 years old, no underlying diseases); grade II ( 70 points); grade III (71～90 points); grade IV.(91～130 points); grade V (＞ 130 points); Risk stratification is divided into low, medium and high risk, among which low risk includes grade I (＜ 50 years old, no underlying disease), grade II( 70 points), grade III (71～90 points), intermediate risk: grade IV.(91～130 points); high risk: grade V (＞ 130 points).Finally, patients with PSI score I.～IV.were included, that is, patients with low- to medium-risk community-acquired pneumonia.

2.2.2 Data Collection

Data were collected on risk factors for sepsis in patients with community-acquired pneumonia based on previous studies[3, 9,10].Including: (1) basic information: including age, gender, body mass index (BMI), underlying diseases (hypertension, coronary heart disease, diabetes and chronic obstructive pulmonary disease),smoking and drinking history; (2) Laboratory indicators: the included patients were tested for blood at the time of admission, and the test items were routine blood cells, including: neutrophil count(NeuC), lymphocyte count (LymC), peripheral platelet count (PLT);Blood lactate (Lac) in arterial blood gas analysis; liver function indicator albumin (ALB); Procalcitonin (PCT), C-reactive protein(CRP).Among them, systemic immune inflammatory trophic index(SII) = NeuC/LymC×PLT (×109/L).

2.2.3 Statistical processing

The collected data is analyzed using SPSS 26.0 software.Measurements that conform to the normal distribution are expressed as mean ± standard deviation (±s), and independent sample t-tests are used for comparison between groups.The counting data are expressed in frequency or example (%), and the χ2test is used for comparison between groups.By plotting the receiver working characteristic (ROC) curve, the cut-off value of continuous variables for predicting sepsis in elderly patients with low- and medium-risk community-acquired pneumonia was determined, the cut-off value was used to convert the continuous variable into a dichotomous variable, and multivariate logistic regression analysis was used to explore the independent risk factors for predicting sepsis in elderly patients with low- to medium-risk community-acquired pneumonia.Based on this, a nomogram prediction model of sepsis risk is constructed through the rms package in R-studio 4.2.2 software.The internal validation of the model uses the bootstrap repeated sampling method.The receiver operating characteristic (ROC)curve, calibration curve and decision curve were used to verify the discrimination, consistency and clinical utility of the model.The difference inP＜0.05 was statistically significant.

3.Results

3.1 Determination of cut-off values for SII, Lac, CRP, PCT

The cut-off value for sepsis in elderly patients with low- and intermediate-risk community-acquired pneumonia was determined by the participant operating characteristics (ROC) curve (Jordon index = sensitivity + specificity - 1).P-value less than 0.05 is considered statistically significant.The area under the curve (AUC)of SII, Lac, CRP and PCT were 0.751 (95%CI: 0.699～0.804),0.585 (95%CI: 0.524～0.647), 0.535 (95%CI: 0.475～0.596) and 0.564 (95%CI: 0.501～0.628), respectively, and the AUC of SII was significantly better than other variables.The cut-off value of SII is 2363.03, its sensitivity is 0.802, and the specificity is 0.649; the cutoff value of Lac is 2.16, its sensitivity is 0.667, the specificity is 0.538; the cut-off value of CRP is 87.81, its sensitivity is 0.406, the specificity is 0.706; the cut-off value of PCT is 1.13, its sensitivity is 0.448, and the specificity is 0.680.According to the cut-off value,the patients were divided into SII ＜2363.03 group and SII 2363.03 group, Lac＜2.16 group, Lac 2.16 group, CRP ＜ 87.81 group,CRP 87.81 group, PCT ＜ 1.13 group and PCT 1.13 group.Among them, the P value of CRP ＞ 0.05, which was not statistically significant.See table 1.

Tab 1 Cut-Off value of SII, Lac, CRP, PCT

3.2 Data from the Sepsis group compared with the Non-sepsis group

A total of 589 elderly patients with low- and intermediate-risk community-acquired pneumonia were included, with a sepsis incidence of 16.30% (96/589).There were no significant differences in body mass index (BMI), gender, history of coronary heart disease,history of cerebrovascular disease, history of hypertension, and history of smoking and drinking between the two groups (P＞0.05).The proportion of patients aged 75 years, diabetes mellitus and chronic obstructive pulmonary disease, Lac 2.16, PCT 1.13 and SII 2363.03 was statistically significant (P＜0.05).See table 2.

3.3 Logistic regression analysis of influencing factors

Taking whether sepsis occurs in elderly patients with low- and medium-risk community-acquired pneumonia as the dependent variable (assignment: yes=1, no=0), and the statistically significant index (P＜0.05) comparing the two groups in Table 2 as the independent variable (assignment: yes=1, no=0), including age(assignment: 75=1, ＜75=0), Lac (assignment: 2.16=1, ＜2.16=0),PCT (assignment: 1.13=1, ＜1.13=0,), SII (assignment: 2 363.03=1, ＜2 363.03=0), DM (assignment: yes=1, no=0), COPD(assignment: yes=1, no=0).The results showed that PCT had no significant effect on sepsis in elderly patients with low- and medium-risk community-acquired pneumonia within 24 hours of admission (P＞0.05), and age 75 years, DM, COPD, Lac 2.16,SII 2363.03 were independent risk factors for sepsis in elderly patients with low- and medium-risk community-acquired pneumonia(P＜0.05).See table 3.

3.4 Establishment of nomogram prediction model

Based on the results of multivariate logistic regression analysis, age 75 years,DM, COPD, Lac 2.16, SII 2363.03 were independent risk factors for sepsis in elderly patients with low- and mediumrisk community-acquired pneumonia, and a nomogram prediction model of sepsis risk in elderly patients with low- and medium-risk community-acquired pneumonia was established by R software.According to the drawn nomogram prediction model, by drawing a straight line perpendicular to the axis of the variable upward, the point on the scoring scale of each variable can be corresponded, that is, the score of the variable, and the total score of the corresponding point of the independent variable is obtained by adding, and then the point corresponding to the risk axis is drawn vertically downward,that is, the probability of sepsis in elderly patients with low- and medium-risk community-acquired pneumonia, see Figure 1.

3.5 Validation of the nomogram prediction model

In the model, the nomogram prediction model was internally verified by using BootStrap repeated sampling 1 000 times, and thecalibration curve almost coincided with the ideal curve trend, and Hosmer-Lemesshow tested P＞0.05, indicating that the prediction results are in good agreement with the actual clinical observation results and have a good calibration degree.See Figure 2.The ROC curve analysis of sepsis risk in elderly patients with low- to mediumrisk community-acquired pneumonia showed that the AUC was 0.826 (95% CI: 0.780-0.872), which had good accuracy, as shown in Figure 3.Finally, the decision curve was used to evaluate the clinical practicability of the model, and it was concluded that when the threshold of the model was between 0.10～0.78, the model had the advantage of clinical benefit, as shown in Figure 4.

Tab 2 Comparison of data between sepsis group and non-sepsis group

Fig 1 Nomogram prediction model for sepsis in elderly patients with low- to medium-risk community-acquired pneumonia

Fig 2 Calibration plot of the predictive model

Fig 3 ROC curve of the predictive model

4.Discussion

Sepsis is caused by the inflammatory response to organ tissue damage throughout the body, involving the interaction and signaling of multiple inflammatory factors and immune cells, and has high morbidity and high mortality.In addition, sepsis can be caused bya variety of factors and affect the function of multiple organs, with community-acquired pneumonia being one of the main causes of sepsis[11].The results of this study showed that the incidence of sepsis in elderly patients with low- and medium-risk communityacquired pneumonia was 16.30% (96/589), which was basically consistent with the results reported by Zhou Yunhang, Aliberti.S et al[9, 12].Because sepsis is a time-dependent disease, early recognition and risk stratification to guide timely treatment strategies are essential to reduce sepsis morbidity and mortality.However, a single biomarker is not sufficient to comprehensively assess the immune status of a patient with sepsis, so a combination of different markers may be advantageous over a single biomarker assessment.The predictive value of SII combined NeuC, LymC, and PLT reflects the balance of inflammatory, immunological, and thrombotic pathways in patients, and may be an early screening tool for potential sepsis risk[13].

Tab 3 Logistic regression analysis of factors influencing sepsis in elderly patients with low- to medium-risk community-acquired pneumonia

Fig 4 DCA curve of the predictive model

This study found that the AUC of SII was 0.751 (95% CI:0.699～0.804), and the AUC of Lac, CRP and PCT were 0.585,0.535 and 0.564, respectively, indicating that compared with the three commonly used sepsis indicators of Lac, CRP and PCT, SII has a higher predictive value in predicting the development of sepsis in low- and medium-risk community-acquired pneumonia.This study also showed that when the Cut-off value of SII was 2 363.03, the sensitivity and specificity predicted by this indicator were 0.802 and 0.649, respectively.This means that older patients with low- to intermediate-risk community-acquired pneumonia may progress to sepsis when the SII is greater than 2 363.03,significantly increasing the risk of death.This conclusion is similar to the findings of Ma.k et al[14], which found that as SII values increased, the more severe the condition of sepsis patients, the risk of poor prognosis in this group of patients increased significantly when SII was greater than 2 106.46.Progression to sepsis in lowand medium-risk community-acquired pneumonia is a complex pathophysiologic process.First, pneumonia activates the immune system and induces the release of a large number of inflammatory mediators, thereby causing an increase in NeuC and a systemic inflammatory response, leading to the occurrence of sepsis[15, 16].In addition, various inflammatory cytokines and mediators in the body during periods of high inflammatory response can cause apoptosis of a large number of lymphocytes by inducing immunosuppression,resulting in a decrease in LymC, which further reduces immune defenses, and ultimately leads to immune collapse and sepsis[17].Secondly, inflammatory mediators can promote the transformation of megakaryocytes into platelets in the bone marrow, cause increased PLT in peripheral blood, promote vascular endothelial cell adhesion and leukocyte extravasation at the site of infection, not only activate the coagulation system, produce thrombus in local capillaries, and play a protective mechanism to limit the infection foci, and activated platelets can aggravate systemic inflammation and coagulation abnormalities by interacting with endothelial cells and inflammatory cells, thereby playing a key role in the pathophysiology of sepsis and in organ damage[18].It can be seen that the increase in SII values in patients with sepsis may be related to elevated NeuC and PLT and decreased LymC, which can fully reflect the acute inflammation and immune defense state of the body, and thus be more reliable than predictors based on a single factor.

In this review, we included predictors of sepsis in previous studies[3, 9, 10], including age, underlying conditions such as diabetes mellitus or COPD, and blood Lac, PCT, and SII, and found differences between the above indicators in the sepsis group and the non-sepsis group.Multivariate analysis also showed that age, associated diabetes, chronic obstructive pulmonary disease,Lac, and elevated SII were independent risk factors for sepsis in older patients with low- and medium-risk community-acquired pneumonia.Previous studies[19] have shown that with age, due to different degrees of atrophy and functional decline of body tissues and organs, as well as decline in physiological defense function, it is easy to progress to sepsis after infection under the invasion of a variety of pathogens.Diabetes mellitus is a major comorbidity of sepsis, accounting for 20 to 23 percent of patients with sepsis[20].Because the body of these patients is in a high-glucose environment,the host’s immune response is destroyed, resulting in pathogenic bacteria easily penetrating the body’s defense system and infection,and promoting the intensification of the inflammatory response on the basis of chronic inflammation-induced insulin resistance, thereby inducing sepsis[21].Previous studies[22, 23] have found that COPD is a chronic inflammatory lung disease characterized by persistent airway inflammation and immune dysfunction, and the levels of pro-inflammatory cytokines and oxidative stress in the airway and peripheral blood in these patients are significantly higher than in the normal population, and pneumonia in this persistent inflammatory state tends to induce an inflammatory cascade, thereby increasing the risk of sepsis.In addition, a study by Montull.B et al [24] also identified COPD as a risk factor for severe sepsis due to pneumonia,consistent with this study.Studies[25, 26] have shown that serum Lac levels are strongly associated with pro-inflammatory cytokine levels and disease severity, and higher serum Lac levels are associated with higher sepsis incidence and mortality, similar to the results of this study.

Nomogram is a multivariate analysis model built by integrating multiple predictors, which can personalize and accurately predict the probability of an event, so it is widely used in the medical field [27].In this study, a nomogram model was further constructed and validated based on five independent risk factors: SII, Lac,age, diabetes mellitus and COPD, and the results showed that the AUC of the model was 0.826 (95% CI: 0.780-0.872), and the calibration curve tended to be ideal, with good discrimination and consistency.In addition, the decision curve also shows that when the threshold is between 0.10～0.78, the nomogram model has the advantage of clinical benefit.Therefore, by predicting the risk of sepsis in elderly patients with low- and medium-risk communityacquired pneumonia, a nomogram model with high reliability and high practicability was constructed, which provided a reference for individualized prevention and treatment.

In summary, SII value at admission is a major factor influencing whether sepsis progresses to sepsis in older patients with low- and medium-risk community-acquired pneumonia.In addition, the nomogram prediction model constructed based on SII to predict sepsis in elderly patients with low- and medium-risk communityacquired pneumonia has good accuracy, which can predict the occurrence of sepsis early, help early identification of high-risk groups and timely intervention, and improve the prognosis of patients.Of course, this study was a single-center, retrospective study with limited sample sizes and possible selection bias.In addition, the nomogram prediction model constructed in this study is only internally validated, and the generalization of the model needs external verification.Therefore, the association between SII and sepsis in patients with low- and medium-risk community-acquired pneumonia and the constructed nomogram prediction model need to be further collected and externally verified in order to ensure that it has better clinical applicability.