When is surgical intervention needed in oral and maxillofacial space infection patients? A retrospective case control study in 46 patients

Abstract

Objective

Patients with mild oral and maxillofacial space infection (OMSI) usually need only antimicrobial therapy. However, surgical intervention is eventually needed after using antibiotics for a period. The objective of this study was to explore the risk factors for drug therapy failure in OMSI.

Subjects and methods

A retrospective case‒control study was designed. From August 2020 to September 2022, patients at Shanghai Jiao Tong University Affiliated Ninth People’s Hospital who were diagnosed with OMSI were retrospectively reviewed. The outcome variable was surgical intervention after the use of antibiotics. We collected common biological factors, including demographic characteristics, routine blood test results, C-reactive protein (CRP) levels and composite indicators, such as neutrophil to lymphocyte ratio (NLR) and platelet to lymphocyte ratio (PLR). The χ2 test and binary logistic regression were used to examine the association between biological factors and the outcome variable.

Results

Forty-six patients were included in this study. Further surgical intervention was needed in 20 patients (43.5%). The NLR showed a significant association with further surgical drainage (p = 0.01). A binary logistic regression equation was found by using stepwise regression based on the Akaike information criterion (R2 = 0.443), which was associated with sex (odds ratio [OR], 0.216; p = 0.092), NLR (OR, 1.258; p = 0.045), red blood cell (RBC) count (OR, 4.372; p = 0.103) and monocyte (MONO) count (OR, 9.528, p = 0.023). Receiver operating characteristic analysis produced an area under the curve for NLR of 0.725 (p = 0.01) and for the binary logistic regression model of 0.8365 (p < 0.001).

Conclusion

Surgical interventions are needed in some mild OMSI patients when antimicrobial therapy fails to stop the formation of abscesses. The binary logistic regression model shows that NLR can be used as an ideal prognostic factor to predict the outcome of antimicrobial therapy and the possibility of requiring surgical intervention.

Statement of clinical relevance

Using simple, inexpensive, and easily achieved biological parameters (such as routine blood test results) and composite indicators calculated by them (such as NLR) to predict whether surgical intervention is needed in the future provides a reference for clinical doctors and enables more cost-effective and efficient diagnosis and treatment.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12903-024-04737-1.

Introduction

Oral and maxillofacial space infection (OMSI) primarily originates from odontogenic infection and can also occur as a result of adenogenic, haematogenous or latrogenic factors. The infection can spread widely through cellular-adipose tissue and fascial planes in the maxillofacial region. This allows the infection to penetrate deeper into the tissues and can result in the development of OMSI cellulitis or the formation of abscess [1]. When an abscess has formed with purulent collection during infection evolution, surgical intervention is needed to expel pus and necrotic tissue from the body. It is necessary and effective in reducing local pain and swelling, as well as preventing apnoea. It also guards against infection diffusion to the craniocerebral region or blood circulation, causing serious complications, including brain abscess, Ludwig’s angina and descending necrotizing mediastinitis [2]. These complications can have detrimental effects on the patient’s health and may require additional interventions for management.

If a contrast CT scan does not reveal obvious signs of abscess, the common practice is to administer antimicrobial therapy. After a few days of treatment, medical staff need to monitor the patient closely to assess if the cellulitis is shrinking or if the infection is worsening to the extent of developing one or multiple abscesses. Once a contrast CT scan confirms the presence of purulent collection after antibiotherapy, surgical intervention is required to achieve the therapeutic goals mentioned above.

Patients who require surgery often experience more severe clinical symptoms, accompanied by a higher risk of complications. For some of them, the infection may become severe enough to warrant hospitalization. This not only adds to the physical and emotional burden experienced by patients but also leads to increased financial expenses. Moreover, the prolonged absence from work or school due to hospitalization further adds to the financial burden and can have negative consequences on the patients’ professional or academic life. In addition, intraoral surgical incisions for drainage tend to be less concerning aesthetically for patients; however, extraoral incisions may result in aesthetic sequelae that could impact the patient’s self-esteem and quality of life. Furthermore, more medical resource inputs are placed in these patients since they require specialized care, including frequent monitoring and wound dressing [3]. Consequently, the surgical intervention should be well indicated and justified. However, to our knowledge, few studies have explored risk factors associated with surgical intervention in patients with mild infection. Further studies in this area can improve patient care and outcomes.

The neutrophil to lymphocyte ratio (NLR) is a systemic inflammatory indicator derived from routine haematological parameters, which currently has been reported as an independent predictor in inflammation-associated diseases such as malignant tumours, cardiovascular diseases, and neurological diseases [4–7]. NLR has attracted considerable attention for its ease and low cost in prediction. Research has reported that during inflammatory stress, neutrophils increase, and lymphocytes undergo apoptosis, resulting in an elevated NLR [8]. In terms of OMSI, NLR has been mentioned as an emerging predicator. Research has reported that the NLR contributes to the severity of infection in terms of the involved spaces and complications in severe and extremely severe oral and maxillofacial space infection patients [9]. It is also regarded as a prognostic marker of deep neck space infections second to odontogenic infection [10, 11]. There is evidence that NLR significantly decreases after surgical drainage in odontogenic cervicofacial phlegmon patients, while it remains at a high level on admission to the hospital [12]. Higher NLR values may lead to unfavourable prognoses in sepsis patients [13].

Some routine haematological parameters are associated with severity in OMSI, such as white blood cell (WBC) count, neutrophil (NEU) count and C-reactive protein (CRP) level. Utilizing these haematological parameters, previous studies have attempted to predict clinical outcomes such as the length of hospital stay and reoperation rate [14–17].

The purpose of the study was to explore the diagnostic value of common haematological parameters and composite parameters, such as NLR, in predicting the possibility that surgical intervention may eventually be required in mild OMSI patients to assist clinicians in making accurate early diagnoses.

Method

Patients

From August 2020 to September 2022, 46 patients at Shanghai Jiao Tong University Affiliated Ninth People’s Hospital diagnosed with OMSI were included in a retrospective study (Fig. 1).

Fig. 1.

Flow chart of the retrospective study

The inclusion criteria for patients with OMSI were as follows: (1) clinical manifestations of typical inflammation, including local redness, swelling, pain, increased skin temperature, and local dysfunction; (2) inflammatory indicators at a high level supporting the diagnosis of inflammation; (3) a contrast CT scan showing no obvious evidence for abscess formation at the first visit; and (4) treatment with antimicrobial therapy received from hospital for no less than 3 days.

The exclusion criteria were as follows: (1) malignant tumour-related infection; (2) previous surgical intervention before the first visit at our hospital; (3) pregnancy; and (4) self-medication with antibiotics before or after the first visit.

All the procedures of the study were in accordance with the Declaration of Helsinki and approved by the Ethics Committee of Shanghai Jiao Tong University School of Medicine Affiliated Ninth People’s Hospital (SH9H-2021-T400-2).

Variables

The outcome variable was surgical intervention after antimicrobial therapy failure.

Surgical intervention was performed when there was presence of abscess formation, which was detected by a contrast CT. The potential predictor variables consisted of demographic characteristics and laboratory examination. Demographic characteristics included patient age at first visit to the hospital and gender (male or female). Laboratory examination comprised both routine full and differential blood cell counts (lymphocytes, leukocytes, neutrophils, eosinophils, basophils, and monocytes) (categorized as normal or abnormal), CRP levels (categorized as normal or abnormal), and composite indicators, including NLR and platelet to lymphocyte ratio (PLR). The clinical laboratory of Shanghai Jiao Tong University School of Medicine Affiliated Ninth People’s Hospital conducted thorough examinations to determine whether the laboratory test results fell within the normal range or exhibited abnormalities.

Statistical analysis

The study variables were acquired from each patient’s medical records. Multiple imputation was used to fill 4 missing data points. Continuous variables are reported as the mean and standard deviation. The Mann‒Whitney U test was used to test the association between continuous predictor variables and the outcome variable. Categorical variables are reported as percentages and were tested by the χ2 test. p < 0.05 was considered statistically significant. All variables with p < 0.2 were included in the multicollinearity test, and then the selected variables were used to build a binary logistic regression model by stepwise selection based on the Akaike information criterion (AIC).

Receiver operating characteristic (ROC) curve analysis was used to evaluate the predictive effect and diagnostic value of different biological variables and the combination of multiple factors on surgical intervention. We used a calibration curve to assess calibration accuracy, accompanied by the Hosmer‒Lemeshow test. We also used forest plots and nomograms to visualize the comprehensive results of the model.

Analyses were performed using R statistical software.

Results

A total of 46 patients diagnosed with OMSI were enrolled in this study. None of the patients had formed an abscess by their first visit. Table 1 presents the study variables, encompassing the population characteristics and haematological parameters of the patients. The study population consisted of 27 males (58.7%) and 19 females (41.3%), resulting in an almost equal distribution in terms of gender. Their ages ranged from 8 to 94 years, with a mean age of 53.6 (± 19.0) years. Abnormal haematological results accounted for 73.9% (34) of WBC counts, 80.4% (37) of neutrophil cell (NEU) counts, and 82.6% (38) of CRP levels. Additional details can be found in Table 1. The specific data of the haematological examination results are provided in Table S1 in the Supplementary Materials. In Table S2 in the Supplementary Materials, abnormal value markers for all haematological indicators are provided, indicating whether the abnormal values are higher or lower relative to the normal values. Except for HCT, EOS count, and MONO percentage, the abnormal values for all other indicators uniformly demonstrate a consistent trend of being either higher or lower in comparison to the normal values. The mean NLR was 7.7 ± 4.9.

Table 1.

Study variables

Study Variable	Data(N = 46)
Age, yr	53.6 $\pm$ 19.6
Male gender, n(%)	27(58.7)
Abnormal WBC count, n(%)	34(73.9)
Abnormal NEU count, n(%)	37(80.4)
Abnormal NEU percentage, n(%)	28(60.9)
Abnormal LYM count, n(%)	11(23.9)
Abnormal LYM percentage, n(%)	36(78.3)
Abnormal Hb, n(%)	12(26.1)
Abnormal RBC count, n(%)	17(37.0)
Abnormal HCT, n(%)	18(39.1)
Abnormal BASO count, n(%)	0(0)
Abnormal BASO percentage, n(%)	0(0)
Abnormal EOS count, n(%)	17(37.0)
Abnormal EOS percentage, n(%)	23(50)
Abnormal MONO count, n(%)	32(69.6)
Abnormal MONO percentage, n(%)	8(17.4)
Abnormal PLT count, n(%)	7(15.2)
Abnormal CRP, n(%)	38(82.6)
NLR	7.7 $\pm$ 7.1
PLR	189.3 $\pm$ 89.0

Abbreviations: WBC white blood cell, NEU neutrophil cell, LYM lymphocyte cell, Hb haemoglobin, RBC red blood cell, HCT haematocrit, BASO basophil granulocyte, EOS eosinophil granulocyte, MONO monocyte, PLT platelet, CRP C-reactive protein, NLR neutrophil to lymphocyte ratio, PLR platelet to lymphocyte ratio

Twenty patients underwent surgical drainage, resulting in a surgery rate of 43.5% in the study population. The results of the association between the study variables and the outcome variable are presented in Table 2. Four variables (WBC count, NEU count, MONO count and NLR) exhibited a significant correlation with surgical intervention (p < 0.05). Additionally, there was a tendency for RBC count, EOS percentage (0.05 < p < 0.1), sex, NEU count and LYM count (0.1 < p < 0.2) to be correlated with surgical intervention. The abnormal values of the above haematological variables all show the same trend of being either higher or lower relative to the normal values.

Table 2.

Analyses of the association between the study variables and outcome of surgical intervention

Study Variable	No incision	Incision	P value
Sample, n	26	20
Age, yr	55.6 $\pm$ 18.9	51.0 $\pm$ 20.6	0.393
Gender, n(%)			0.172*
Male	13(48.1)	14(51.9)
Female	13(68.4)	6(31.6)
WBC count, n(%)			0.029***
Normal	10(83.3)	2(16.7)
Abnormal	16(47.1)	18(52.9)
NEU count, n(%)			0.151*
Normal	7(77.8)	2(22.2)
Abnormal	19(51.4)	28(48.6)
NEU percentage, n(%)			0.020***
Normal	14(77.8)	4(22.2)
Abnormal	12(42.9)	16(57.1)
LYM count, n(%)			0.122*
Normal	22(62.9)	13(37.1)
Abnormal	4(36.4)	7(63.6)
LYM percentage, n(%)			0.331
Normal	7(70.0)	3(30.0)
Abnormal	19(52.3)	17(47.2)
Hb, n(%)			0.227
Normal	21(61.8)	13(38.2)
Abnormal	5(41.7)	7(58.3)
RBC count, n(%)			0.060**
Normal	19(65.6)	10(34.4)
Abnormal	6(35.7)	11(64.3)
HCT, n(%)			0.474
Normal	17(60.7)	11(39.3)
Abnormal	9(50.0)	9(50.0)
BASO count, n(%)			/
Normal	26(56.5)	20(43.5)
Abnormal	/	/
BASO percentage, n(%)			/
Normal	26(56.5)	20(43.5)
Abnormal	/	/
EOS count, n(%)			0.322
Normal	18(62.1)	11(37.9)
Abnormal	8(47.1)	9(52.9)
EOS percentage, n(%)			0.074**
Normal	16(69.6)	7(30.4)
Abnormal	10(43.5)	13(56.5)
MONO count, n(%)			0.008***
Normal	12(85.7)	2(14.3)
Abnormal	14(43.8)	18(56.2)
MONO percentage, n(%)			0.246
Normal	20(52.6)	18(47.4)
Abnormal	6(75.0)	2(25.0)
PLT count, n(%)			0.971
Normal	22(56.4)	17(43.6)
Abnormal	4(57.1)	3(42.9)
CRP, n(%)			0.707
Normal	5(62.5)	3(37.5)
Abnormal	21(55.3)	17(44.7)
NLR	4.6(2.5,7.3)	7.6(5.6,11.2)	0.010***
PLR	145.6(118.3,204.6)	179.2(131.6,277.9)	0.240

^***: p < 0.05 **: 0.05 < p < 0.1 *: 0.1 < p < 0.2

After including the correlated variables (p < 0.2) into the binary logistic regression analysis as mentioned earlier, we found that four variables, including sex, NLR, RBC count and MONO count, remained significant in the final model (Table 3).

Table 3.

Binary logistic regression analysis for outcome (Surgical intervention)

Study Variable	Regression coefficient	Odds ratio(95% CI)	P value
Gender	-1.53	0.21(0.03–1.28)	0.092
NLR	0.23	1.26(1.01–1.57)	0.045
RBC count	1.48	4.37(0.74–25.78)	0.103
MONO count	2.25	9.53(1.37–66.17)	0.023
Intercept	-3.50	0.03	0.008

An increase in NLR and abnormalities in RBC count and MONO count were found to increase the likelihood of undergoing surgical intervention; however, being female was associated with a decreased likelihood of undergoing surgical intervention.

The ROC curve was used to analyse the performance of the regression model and independent variables in the model in predicting surgical intervention (Fig. 2).

Fig. 2.

ROC curve

The ROC curve indicated that NLR was able to effectively distinguish the rate of surgical intervention after antimicrobial therapy failure. NLR demonstrated a 72.5% accuracy (AUC 0.725 [95% CI 0.578–0.872], p = 0.01), with a cut-off value of 5.50 predicting surgical intervention with 80.0% sensitivity and 61.5% specificity.

Moreover, the ROC curve revealed that the predictive ability of multiple variables was superior to that of a single factor. The regression model had 83.7% accuracy (AUC 0.837 [95% CI 0.720–0.953], p < 0.001).

We also used forest plots and nomograms to visually present the comprehensive results of the regression model. In the forest plot, the influence of various variables is graphically depicted, providing a comprehensive overview of the effect size for each individual variable as well as the collective effect size. It becomes evident that an elevation in the count of MONO, RBC, and NLR is indicative of a risk factor. Conversely, being female is associated with a protective effect. Additionally, NLR, as an influential factor within the multifactorial model, exhibits a relatively narrow confidence interval and a statistically significant p-value (Fig. 3). Utilizing logistic regression, a scoring criterion is established within the nomogram, based on the regression coefficients of the independent variables. The ‘risk’ indicated on the final line, derived from the ‘total points’, serves to represent the model’s predicted risk level, thereby enhancing the comprehensibility and interpretability of the predictive outcomes (Fig. 4).

Fig. 3.

Forest plot. *The dashed line (x = 1) represents an invalid line. The solid horizontal line represents the results of each study variable. Lines on the left side of the valid line indicate protective factors, while lines on the right side indicate risk factors. The size of the diamond represents the weight of each variable in the model, and the colour indicates the p value. The length of the solid lines represents the 95% CI

Fig. 4.

Nomogram. *To determine the risk of undergoing surgical intervention predicted by the logistic regression model, these steps were followed: a vertical line was made from the axis of each parameter, and the corresponding value on the line labelled ‘Points’ was located. The points of all the parameters were added together. Then, another vertical line was drawn from the axis labelled ‘Total Points’, and the corresponding number on the ‘Risk’ axes was found. The number on the line labelled ‘Risk’ represents the predicted risk of undergoing surgical intervention

Next, we utilized the calibration curve to demonstrate a strong agreement between the prediction and observation in the study cohort (Fig. 5). Additionally, the Hosmer‒Lemeshow test yielded a nonsignificant statistic (p = 0.693), indicating no deviation from a perfect fit.

Fig. 5.

Calibration curve. *The calibration curve illustrates the agreement between the predicted risks of surgical intervention and the observed actual risks of surgical intervention. The x-axis represents the predicted probability of surgical intervention, while the y-axis represents the observed probability of surgical intervention. The diagonal dotted line represents the ideal predicted results of a perfect model. On the other hand, the solid line represents the predictions made by our bias-corrected regression model. The closer the solid line is to the diagonal dotted line, the more accurate the calibration of the model

Discussion

The aim of the study was to explore risk factors associated with surgical intervention in OMSI patients. Although we found that several variables showed a correlation tendency with the outcome in the single-factor screening, binary logistic regression ultimately identified 4 key variables: male sex showed negative associations, increased NLR showed positive associations, and abnormal MONO count and RBC count showed positive associations. Among all the factors, NLR, as a composite index, was found to have a significant association with surgical intervention. When used as an independent predictor, the NLR showed good discrimination between outcomes. Additionally, when included as a component in the multivariable regression model, it improved the prediction performance, making the model’s ability to predict superior to that of a single factor.

The proportion of male and female patients in the study was relatively balanced, and males comprised 58.7% of the study population. The average age was 53.6 years, which was also close to the population characteristics in other studies on OMSI [18, 19].

Gender differences are a controversial risk factor related to OMSI. Samuel et al. analysed 1002 hospitalized patients diagnosed with OMSI and found that the probability of males needing immediate airway management was significantly higher than that of females (male 7% vs. female 2%, p = 0.001), the WBC count in males was significantly higher than that in females (male 12.4*10⁹/L vs. female 11.1*10⁹/L, p = 0.000), and C-reactive protein was also significantly higher than that in females (male 78 mg/L vs. female 59 mg/L, p = 0.001) [20]. Multiple studies have indicated that male patients face a higher risk of hospitalization following surgeries and a greater likelihood of requiring intensive care unit (ICU) treatment [14, 21]. However, there were also studies showing little correlation between gender and OMSI severity in terms of length and cost of hospitalization, severe complications and reoperation risk [3, 10, 17]. The final regression model of our study retained gender as a risk factor associated with the outcome. The OR value of sex was 0.21 < 1, indicating that female sex was a protective factor, which was consistent with previous studies. The final regression model of our study found that gender was not statistically significant in relation to the outcome (p = 0.092). However, importantly, insignificance does not necessarily mean that the variable should be removed from the model, as a small sample size could also contribute to this result. The use of backwards stepwise selection based on AIC made the result acceptable.

Serum laboratory tests are relatively simple and inexpensive. They are easy to perform and widely accepted by most individuals. We selected routine blood tests and CRP levels as the parameters to be examined, which can provide results quickly. Since these parameters are routinely tested in almost every infection patient who visits the hospital, the prediction can be made at no additional cost. In summary, the prediction can be done quickly and does not impose any extra financial or time burden on the patients.

The NLR has been widely used in prognostic prediction in inflammation-associated diseases such as malignant tumours. In the context of OMSI, it is commonly used as a predictive factor for severe complications and the length of hospitalization [9, 10, 22]. However, we have not come across any research related to NLR in relation to mild OMSI. Our study demonstrated that the NLR significantly influenced surgical intervention and acted as a risk factor (p = 0.045, OR = 1.26 [95% CI 1.01–1.57]). The ROC curve analysis revealed that NLR had higher accuracy than other risk factors (AUC = 0.725, p = 0.01), indicating its ability to correctly identify patients in need of surgical intervention. The determined cut-off value was 5.50, which may provide a reference for clinical work. Furthermore, the final regression model showed even better discrimination (AUC = 0.837, p < 0.001), suggesting that multiple variables have a stronger impact on the outcome and yield higher discrimination in the results. This provides a reliable basis for determining further surgery and can be considered a criterion for judgement in subsequent clinical decisions. The forest plot and nomogram were used to visualize the results of the regression model. The nomogram, in particular, makes it easy and quick to use the model. These tools provide reliable references for clinical doctors and enable a higher level of nurse care for patients. Additionally, they can help decrease unnecessary consumption of medical resources.

During infection, MONOs further differentiate into tissue macrophages and dendritic cells, thereby mediating the immune response. They also possess the ability to be recruited to the site of infection and directly engage in antibacterial activity. Additionally, they participate in the initial inflammatory response by releasing factors such as tumour necrosis factor (TNF) and chemokines [23, 24]. All 32 abnormal MONO counts in our study were higher than the normal range. In our statistical analysis, MONO count was found to be significantly associated with the outcome (p = 0.023), suggesting that a higher MONO count could be regarded as a risk factor (OR = 9.53 [95% CI 1.37–66.17]). However, the confidence interval for this association was wide, possibly due to the use of categorical variables in the statistical analysis or the small sample size and unstable distribution. In further studies, it is recommended to increase the sample size and report MONO count as a continuous variable to better observe its impact on the outcome.

RBCs undergo activation during inflammation triggered by bacterial infection, leading to a sequence of pathological alterations such as erythrocyte deformation and programmed cell death [25]. E. Pretorius et al. reported that the erythrocyte membrane interacts with inflammatory molecules, resulting in erythrocyte deformation and programmed cell death. These processes have an impact on haemorheology and can serve as a parameter for identifying the presence and extent of inflammation [26]. In our study, all 17 abnormal RBC counts were found to be lower than the normal range, which is consistent with previous studies. The final model in our study included RBC count as a risk factor (OR = 4.37 [95% CI 1.37–66.17], p = 0.103]). Similar to the MONO count, the confidence interval was also wide, indicating that further confirmation is needed to determine its specific impact on the outcome.

In addition, we observed a lack of correlation between surgical drainage and CRP levels. This finding is in contrast to previous studies that have demonstrated a relationship between CRP levels and factors such as the length of hospitalization, the number of spaces involved, and the rate of reoperation in OMSI patients [15, 17, 27]. In our study cohort, 82.6% of patients had abnormal CRP levels. This high prevalence can be attributed to the fact that CRP is a highly sensitive and responsive indicator of acute infection [28, 29]. Furthermore, we treated CRP level as a categorical variable in our analysis, so the results we obtained showed no significant difference between the group that underwent a surgical intervention and the group that did not.

The study is an exploratory retrospective case‒control study, which inherently has limitations. The small sample size resulted in some statistical results being unstable. Only population characteristics and simple serum laboratory test results were analysed, while other potential risk factors, such as spaces involved, potential systemic diseases, clinical symptoms at admission (fever, pain, mouth opening, etc.), and basic vital signs (blood pressure, pulse, temperature, etc.), were not included in the analysis.

Conclusion

Within limitations associated to the present study, NLR has been found as an effective parameter for predicting surgical intervention in mild OMSI patients. Additionally, sex, MONO count and RBC count also contributed to the outcome. The application of simple and easily accessible serum tests to predict surgical intervention enables the efficiency and effectiveness of medical resources while also providing more opportunities to deliver higher quality care to patients.

Abbreviations

AIC

Akaike information criterion

AUC

Area under curve

BASO

Basophil granulocyte

CI

Confidence Interval

CRP

C-reactive protein

EOS

Eosinophils

Hb

Haemoglobin

HCT

Haematocrit

ICU

Intensive care unit

LYM

Lymphocyte cell

MONO

Monocyte

NEU

Neutrophil cell

NLR

Neutrophil to lymphocyte ratio

OMSI

Oral and maxillofacial space infection

OR

Odds ratio

PLR

Platelet to lymphocyte ratio

PLT

Platelet

RBC

Red blood cell

ROC

Receiver operating characteristic

TNF

Tumour necrosis factor

WBC

White blood cell

Authors’ contributions

Lingyan Zheng and Huan Shi made substantial contributions to the conception. Yimin Liu and Huan Shi designed the work. Yimin Liu, Hanyi Zhu, Xin Bao and Zhiyuan He acquired the data. Yinyi Qin, Yimin Liu and Hanyi Zhu analyzed and interpreted the data. Yimin Liu, Hanyi Zhu and Xin Bao drafted the work. Lingyan Zheng and Huan Shi substantively revised it. All authors approved the submitted version, agreed both to be personally accountable for the own contributions and to ensure that questions related to the accuracy or integrity of any part of the work. Yimin Liu and Hanyi Zhu contributed equally to this work.

Funding

This study was supported by the National Natural Science Foundation of China (Grant No.82174041, 82302553, 82370976), Shanghai Young Science and Technology Talents Sailing Program (Grant No. 22YF1422300) and the Biological Sample Bank Project of Ninth People’s Hospital Affiliated with Shanghai Jiao Tong University School of Medicine (Grant No.YBKB202107, YBYB202212).

Availability of data and materials

All data generated or analysed during this study are included in this published article and its supplementary information files.

Declarations

Ethics approval and consent to participate

The study was approved by the Ethics Committee of Shanghai Jiao Tong University School of Medicine Affiliated Ninth People’s Hospital (SH9H-2021-T400-2). The Ethics Committee of Shanghai Jiao Tong University School of Medicine Affiliated Ninth People’s Hospital waived the informed consent.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.