Construction and evaluation of a 180-day readmission prediction model for chronic heart failure patients based on sCD40L

doi:10.21203/rs.3.rs-3142916/v1

Download PDF

Article

Construction and evaluation of a 180-day readmission prediction model for chronic heart failure patients based on sCD40L

https://doi.org/10.21203/rs.3.rs-3142916/v1

This work is licensed under a CC BY 4.0 License

Version 1

posted

You are reading this latest preprint version

Background

The high readmission rate of patients with chronic heart failure can cause waste of medical resources and economic losses. Establishing an effective heart failure readmission model can effectively alleviate medical pressure and improve the quality of treatment.

Methods

In this study, we conducted a comprehensive analysis of clinical and laboratory data from 248 patients with chronic heart failure who received treatment at our medical center between January 2021 to January 2023. We also measured sCD40L levels to determine their association with readmission due to heart failure during follow-up. To analyze the data, we employed various statistical methods including independent sample t-test, correlation analysis, univariate COX regression, and LASSO COX regression. Using these techniques, we organized the data and constructed a predictive model that was both trained and validated.

Results

We developed a nomogram to assess the likelihood of readmission within 180 days for patients with chronic heart failure. Our findings revealed that monocytes, creatinine, sCD40L, and hypertension history were all independent risk factors for 180-day heart failure readmissions. Additionally, our model's area under the curve was 0.731 in the training dataset and 0.704 in the validation dataset.

Conclusion

This study provides new insights for predicting readmission within 180 days for patients with chronic heart failure. And sCD40L is an important predictive indicator for readmission of heart failure patients within 180 days, and clinical doctors can develop appropriate treatment plans based on sCD40L.

Health sciences/Health care/Prognosis

Health sciences/Cardiology/Cardiovascular biology

Health sciences/Risk factors

sCD40L

chronic heart failure

readmission

prediction model

COX regression

Heart failure (HF) is a major cause of morbidity and mortality globally(1). Large-scale data indicate an increasing prevalence of HF, high hospitalization costs, and increased frequency of hospitalization, which places a huge burden on both individual and social healthcare and may become a more serious challenge in the future(2). Many people worldwide have high blood pressure, diabetes, obesity, kidney disease, and other risk factors for heart disease(3). Some researchers have estimated that the incidence of HF will continue to increase over the coming decades(2).

Currently, HF is difficult to treat, and most patients require readmissions. This requires considerable time, money, and energy. Therefore, a prediction model for HF readmission that fits the characteristics of clinical practice is needed. Studies predicting the risk of readmission in HF patients have been based on clinical parameters such as brain natriuretic peptide (BNP) or N-terminal brain natriuretic peptide (NT-proBNP) levels, renal insufficiency, comorbidities, and previous hospitalization history(4). However, in terms of "risk prediction,” classic cardiovascular risk markers such as soluble CD40 ligand (sCD40L), have rarely been explored.

As a key immune factor, sCD40L is primarily expressed in immune cells (5) and activated platelets (6), and its role in vascular inflammation cannot be ignored (7). Many studies have confirmed that the serum expression level of sCD40L is positively correlated with cardiovascular events, and higher sCD40L levels are associated with coronary microvascular dysfunction (8). Some studies have suggested that sCD40L can be used as a diagnostic marker for systemic atherosclerosis (9). Clinical studies have shown that acute or chronic HF patients have elevated sCD40L levels (10). In addition, few studies have evaluated the risk of readmission in chronic HF patients based on sCD40L levels. Therefore, an analysis of readmission risk in HF patients based on sCD40L expression levels may provide a better risk assessment in a clinical setting.

2.1 Study population

This study included chronic HF patients who visited the Cardiology Department of Shaoxing People's Hospital between January 2021 and January 2023. The inclusion criteria were as follows: (1) age > 18 years (2) patients diagnosed to have chronic HF based on the guidelines of the European Heart Association, and patients with extent of HF between class II and IV according to the New York Heart Association (NYHA); (3) those with a history of congestive HF for at least a month. Patients with the following conditions were excluded: (1) end-stage liver or renal failure; (2) hematological system diseases; (3) history of cancer or tumor resection; (4) rheumatic immune system diseases; (5) those with incomplete clinical data or demographic characteristics; (6) patients readmitted to other hospitals; and (7) those who have undergone cardiac stent surgery (Fig. 1). A total of 207 patients who met the inclusion criteria were included in the study. The study was conducted in accordance with the Declaration of Helsinki and approved by the Shaoxing People’s Hospital Ethics Review Committee. Informed consent was obtained from all subjects before their participation.

2.2 Clinical and laboratory data

All blood samples were taken within 1–2 days after admission. Blood samples were obtained via venous puncture and collected in pre-cooled tubes containing EDTA. The supernatant was centrifuged and stored at -80°C. According to the manufacturer's protocol, an enzyme-linked immunosorbent assay kit (ab196268, Abcam) was used to measure serum CD40L levels.

2.3 Clinical outcomes

The main clinical outcome was that patients were admitted due to worsening chronic HF. The endpoint was determined by contacting each patient individually.

2.4 Statistical analysis

Continuous variables are represented as means and standard deviations (SDs), while categorical variables are represented as frequencies and percentages. A normality test was performed on the data distribution to select the appropriate test. Continuous symmetric variables were tested using an independent t-test, and categorical variables were tested using the Pearson chi-square test (Pearson χ2 test). The Spearman’s correlation test was used to determine the correlations between various variables. Evaluate the diagnostic ability of each parameter or model by analyzing the area under the curve (AUC) of the receiver operator characteristic (ROC) curve analysis.

The initial dataset has been randomly divided into training and validation datasets in a 7:3 ratio to create and validate a prediction model. The Least Absolute Shrinkage and Selection Operator (LASSO) method has been used to select meaningful indicators from the training dataset, followed by COX regression to establish the prediction model and construct a nomogram. A calibration curve was constructed in the training group to predict the similarity between the predicted probability and the actual observation probability. In addition, decision curve analysis evaluates the clinical practicality of the column chart by quantifying the net benefits under different threshold probabilities.

The R version 4.1.0 (The R Foundation for Statistical Computing, Vienna, Austria) was used for all data analyses and models. All statistical tests were two-tailed. P < 0.05 is considered statistically significant.

3.1 Baseline characteristics

The baseline characteristics of the study participants are shown in Table 1. We divided the readmission time of chronic HF patients into four subgroups according to quartiles and analyzed the relationship between readmission time and various indicators. Our results showed that patients with shorter readmission times had higher levels of serum creatinine (P = 0.036) and sCD40L (P = 0.004). Interestingly, the trend of patients' diabetes history was opposite, with a lower incidence of diabetes being observed in patients with shorter readmission times. However, there was no statistically significant difference in other features.

Table 1

Patients characteristics.
Variables	First quartile (n = 62)	Second quartile (n = 62)	Third quartile(n = 61)	Fourth quartile(n = 63)	p value
sex:					0.442
Female	25 (40.3%)	20 (32.3%)	22 (36.1%)	17 (27.0%)
Male	37 (59.7%)	42 (67.7%)	39 (63.9%)	46 (73.0%)
age, years	58.0 (9.70)	57.1 (12.1)	60.4 (13.6)	55.7 (14.0)	0.19
WBC, *10¹²/L	5.22 (1.69)	5.40 (2.24)	5.27 (1.62)	5.96 (2.33)	0.141
Neutrophils,*10¹²/L	5.51 (3.24)	7.05 (9.90)	5.20 (2.73)	6.00 (8.10)	0.448
lymphocyte,*10¹²/L	4.35 (8.72)	4.76 (9.50)	4.59 (13.4)	4.77 (10.5)	0.996
monocyte,*10¹²/L	2.66 (3.43)	2.80 (3.09)	2.44 (2.77)	1.89 (2.65)	0.346
RBC, *10¹²/L	3.17 (1.67)	3.31 (2.74)	3.14 (1.72)	3.81 (2.69)	0.328
hsCRP, mg/L	2.94 (6.65)	4.35 (9.21)	3.71 (7.19)	3.83 (8.43)	0.804
AST, U/L	26.1 (11.0)	338 (2395)	27.5 (12.2)	29.1 (19.9)	0.374
LDH, U/L	230 (59.6)	495 (2009)	231 (79.1)	226 (107)	0.358
Albumin, g/L	37.5 (3.95)	37.1 (5.38)	37.1 (4.87)	37.6 (5.30)	0.917
Globulin, g/L	28.9 (4.27)	29.8 (4.50)	28.8 (4.10)	29.5 (5.31)	0.597
Creatinine, umol/L	86.8 (29.4)	109 (98.6)	106 (93.5)	77.0 (23.3)	0.036
Urea, umol/L	420 (115)	417 (137)	435 (142)	384 (130)	0.18
TG, mmol/L	1.33 (0.71)	1.20 (0.68)	1.24 (0.66)	1.36 (0.73)	0.534
HDL-C, mmol/L	1.08 (0.28)	1.08 (0.33)	1.05 (0.29)	1.04 (0.30)	0.881
LDL-C, mmol/L	2.54 (0.70)	2.47 (0.85)	2.39 (0.73)	2.65 (0.81)	0.268
Apo A1, g/L	1.13 (0.24)	1.09 (0.28)	1.09 (0.27)	1.09 (0.28)	0.759
Apo B, g/L	0.84 (0.24)	0.85 (0.29)	0.81 (0.22)	0.91 (0.28)	0.171
FBG, mmol/L	6.64 (3.54)	6.90 (3.64)	6.07 (2.36)	7.26 (3.41)	0.228
ALT, U/L	26.5 (17.5)	120 (705)	27.9 (32.6)	33.0 (29.8)	0.384
HBA1c, (%)	6.77 (1.76)	6.88 (1.63)	6.53 (1.36)	7.07 (2.06)	0.366
sCD40L, ng/ml	9.65 (2.02)	9.64 (2.00)	10.1 (1.93)	8.76 (2.25)	0.004
LVEF, %	49.1 (13.1)	49.7 (13.1)	47.3 (13.4)	53.5 (15.0)	0.081
LVFS, %	25.6 (8.16)	26.5 (7.87)	25.0 (8.95)	28.5 (9.25)	0.117
AF					0.067
No	33 (53.2%)	31 (50.0%)	28 (45.9%)	43 (68.3%)
Yes	29 (46.8%)	31 (50.0%)	33 (54.1%)	20 (31.7%)
Hypertension:					0.087
No	28 (45.2%)	30 (48.4%)	26 (42.6%)	40 (63.5%)
Yes	34 (54.8%)	32 (51.6%)	35 (57.4%)	23 (36.5%)
hyperlipidemia:					0.656
No	60 (96.8%)	58 (93.5%)	60 (98.4%)	60 (95.2%)
Yes	2 (3.23%)	4 (6.45%)	1 (1.64%)	3 (4.76%)
diabetes:					0.032
No	43 (69.4%)	36 (58.1%)	45 (73.8%)	32 (50.8%)
Yes	19 (30.6%)	26 (41.9%)	16 (26.2%)	31 (49.2%)
smoking:					0.473
No	48 (77.4%)	40 (64.5%)	43 (70.5%)	44 (69.8%)
Yes	14 (22.6%)	22 (35.5%)	18 (29.5%)	19 (30.2%)
drinking:					0.171
No	54 (87.1%)	47 (75.8%)	51 (83.6%)	46 (73.0%)
Yes	8 (12.9%)	15 (24.2%)	10 (16.4%)	17 (27.0%)
SBP,mmhg	136 (19.9)	131 (20.4)	132 (23.6)	127 (20.8)	0.153
DBP,mmhg	83.7 (15.4)	80.0 (13.4)	80.6 (13.4)	78.7 (13.6)	0.238
BMI (kg/m2)	24.1 (3.80)	23.4 (2.82)	23.8 (3.70)	24.3 (3.82)	0.491
Readmission day	36.2 (14.3)	80.4 (14.2)	131 (16.3)	275 (130)	<0.001
WBC, white blood cell; RBC, red blood cell; hsCRP, highly sensitive C-Reactive Protein; AST, aspartate amino transferase; LDH, lactate dehydrogenase; TG, triglyceride; HDL-C, high-density lipoprotein cholesterol; LDL-C, low-density lipoprotein cholesterol; Apo A1; Apolipoprotein A1; Apo B, Apolipoprotein B; FBG, fasting blood glucose; ALT, Alaninetransaminase; HBA1c, hemoglobin A1c; sCD40L, soluble CD40 ligand; LVEF, left ventricular ejection fraction; LVFS, left ventricular fractional shortening; AF, atrial fibrillation; SBP, systolic blood pressure; DBP, diastolic blood pressure; BMI, Body mass index. Discrepant p-values are represented in bold.

3.2 Correlation between sCD40L and various features

As a novel biomarker, sCD40L may be associated with the patient's readmission time. Therefore, we aimed to investigate the correlation between sCD40L and various clinical features. We divided chronic HF patients into four subgroups based on quartile values of sCD40L concentration (Table 2). Our results revealed significant differences between sCD40L and several serological indicators including white blood cells (P < 0.001), monocytes (P < 0.001), and red blood cells (P < 0.001). Additionally, sCD40L showed association with several blood biochemical indicators such as albumin (P < 0.001), blood uric acid (P = 0.002), LDL-C (P = 0.038), apo A1 (P = 0.018), APO B (P = 0.004), and ejection fraction (P < 0.001). In terms of demographic data, sCD40L was found to be related to the history of atrial fibrillation (P < 0.001), hypertension (P = 0.001), diabetes (P < 0.001), and alcohol consumption (P = 037). Interestingly, we observed a negative correlation between SCD40L and fasting blood glucose (P < 0.001) as well as HBA1C (P = 0.001).

Table 2

Baseline characteristics of patients according to quartile levels of sCD40L
Variables	First quartile (n = 62)	Second quartile (n = 63)	Third quartile(n = 61)	Fourth quartile(n = 62)	p value
sex:
Female	21 (33.9%)	30 (47.6%)	18 (29.5%)	15 (24.2%)	0.038
Male	41 (66.1%)	33 (52.4%)	43 (70.5%)	47 (75.8%)
age, years	52.4 (11.8)	60.2 (13.8)	61.2 (11.6)	57.3 (11.0)	<0.001
WBC, *10¹²/L	6.31 (2.40)	5.60 (1.68)	4.77 (2.08)	5.16 (1.45)	<0.001
Neutrophils,*10¹²/L	5.43 (9.98)	5.88 (8.27)	6.60 (2.84)	5.86 (2.59)	0.816
lymphocyte,*10¹²/L	3.51 (11.9)	2.86 (6.69)	7.29 (14.2)	4.89 (7.61)	0.096
monocyte,*10¹²/L	0.75 (1.69)	1.88 (3.31)	4.00 (2.86)	3.19 (2.86)	<0.001
RBC, *10¹²/L	4.54 (1.14)	3.84 (2.52)	2.33 (2.71)	2.70 (1.66)	<0.001
hsCRP, mg/L	1.81 (1.89)	5.97 (10.0)	3.95 (10.2)	3.07 (5.96)	0.026
AST, U/L	27.7 (21.3)	31.0 (27.1)	338 (2415)	29.4 (13.4)	0.386
LDH, U/L	200 (88.6)	242 (83.5)	501 (2025)	244 (80.4)	0.33
Albumin, g/L	40.6 (4.48)	37.4 (4.91)	35.0 (4.57)	36.3 (3.74)	<0.001
Globulin, g/L	29.6 (4.52)	29.4 (4.75)	29.3 (4.70)	28.6 (4.32)	0.674
Creatinine, umol/L	75.1 (32.0)	95.9 (69.9)	103 (88.2)	104 (79.2)	0.087
Urea, umol/L	376 (106)	417 (144)	399 (137)	462 (125)	0.002
TG, mmol/L	1.63 (0.92)	1.11 (0.55)	1.21 (0.59)	1.18 (0.54)	<0.001
HDL-C, mmol/L	1.07 (0.28)	1.11 (0.34)	1.05 (0.30)	1.03 (0.26)	0.56
LDL-C, mmol/L	2.72 (0.79)	2.37 (0.67)	2.38 (0.80)	2.58 (0.81)	0.038
Apo A1, g/L	1.17 (0.25)	1.14 (0.28)	1.06 (0.28)	1.04 (0.24)	0.018
Apo B, g/L	0.94 (0.29)	0.78 (0.22)	0.82 (0.23)	0.87 (0.27)	0.004
FBG, mmol/L	8.21 (4.15)	5.86 (2.76)	6.65 (3.28)	6.17 (2.25)	<0.001
ALT, U/L	30.3 (27.9)	30.2 (32.6)	117 (711)	30.7 (29.6)	0.429
HBA1c, (%)	7.54 (2.43)	6.33 (1.26)	6.72 (1.54)	6.65 (1.18)	0.001
sCD40L, ng/ml	6.95 (0.67)	8.75 (0.37)	9.94 (0.49)	12.5 (0.59)	<0.001
LVEF, %	64.4 (4.36)	57.8 (9.77)	44.0 (4.92)	33.2 (5.35)	<0.001
LVFS, %	35.1 (3.34)	31.1 (6.40)	22.3 (2.71)	17.0 (6.06)	<0.001
AF					<0.001
No	48 (77.4%)	22 (34.9%)	30 (49.2%)	35 (56.5%)
Yes	14 (22.6%)	41 (65.1%)	31 (50.8%)	27 (43.5%)
Hypertension:					0.001
No	42 (67.7%)	25 (39.7%)	22 (36.1%)	35 (56.5%)
Yes	20 (32.3%)	38 (60.3%)	39 (63.9%)	27 (43.5%)
hyperlipidemia:					0.159
No	60 (96.8%)	63 (100%)	57 (93.4%)	58 (93.5%)
Yes	2 (3.23%)	0 (0.00%)	4 (6.56%)	4 (6.45%)
diabetes:					<0.001
No	20 (32.3%)	49 (77.8%)	45 (73.8%)	42 (67.7%)
Yes	42 (67.7%)	14 (22.2%)	16 (26.2%)	20 (32.3%)
smoking:					0.143
No	44 (71.0%)	49 (77.8%)	45 (73.8%)	37 (59.7%)
Yes	18 (29.0%)	14 (22.2%)	16 (26.2%)	25 (40.3%)
drinking:					0.037
No	49 (79.0%)	58 (92.1%)	45 (73.8%)	46 (74.2%)
Yes	13 (21.0%)	5 (7.94%)	16 (26.2%)	16 (25.8%)
SBP,mmhg	130 (18.6)	134 (23.5)	135 (21.2)	126 (20.7)	0.059
DBP,mmhg	81.3 (11.8)	81.6 (14.9)	79.2 (12.6)	80.9 (16.5)	0.801
BMI (kg/m2)	24.0 (3.71)	24.2 (3.43)	23.0 (3.05)	24.4 (3.90)	0.136
Readmission day	173 (151)	107 (65.8)	128 (120)	118 (83.1)	0.005
WBC, white blood cell; RBC, red blood cell; hsCRP, highly sensitive C-Reactive Protein; AST, aspartate amino transferase; LDH, lactate dehydrogenase; TG, triglyceride; HDL-C, high-density lipoprotein cholesterol; LDL-C, low-density lipoprotein cholesterol; Apo A1; Apolipoprotein A1; Apo B, Apolipoprotein B; FBG, fasting blood glucose; ALT, Alaninetransaminase; HBA1c, hemoglobin A1c; LVEF, left ventricular ejection fraction; LVFS, left ventricular fractional shortening; AF, atrial fibrillation; SBP, systolic blood pressure; DBP, diastolic blood pressure; BMI, Body mass index. Discrepant p-values are represented in bold.

3.3 Feature construction and regression Analysis

In univariate COX regression analysis (Table 3), the following parameters were associated with readmission: age (Odd radio (OR): 1.01; 95% Confidence interval (CI): 1.00–1.02; P = 0.017), WBC (OR: 0.85; 95% CI: 0.78–0.91; P = 0.011), monocyte (OR: 1.07; 95% CI: 1.03–1.10; P < 0.001), RBC (OR: 0.88; 95% CI: 0.81–0.94; P < 0.001), albumin (OR: 0.95; 95% CI: 0.93–0.98; P < 0.001), creatinine (OR: 1.01; 95% CI: 1.00–1.01; P = 0.007), urea (OR: 1.01; 95% CI: 1.00–1.01; P = 0.002), triglyceride (OR: 0.75; 95% CI: 0.62–0.91; P = 0.003), Apolipoprotein B (OR: 0.49; 95% CI: 0.30–0.83; P = 0.007), HBA1c (OR: 1.18; 95% CI: 1.11–1.25; P < 0.001), LVEF (OR: 0.96; 95% CI: 0.95–0.97; P < 0.001), LVFS (OR: 0.94; 95% CI: 0.92–0.96; P < 0.001), history of atrial fibrillation (OR: 1.71; 95% CI: 1.31–2.23; P < 0.001), and history of hypertension (OR: 1.41; 95% CI: 1.10–1.83; P = 0.007).

Table 3

Univariate COX regression analysis
Variable	Unadjusted HR (95%CI)	p value
SEX
Male	Reference
Female	0.87 (0.66–1.13)	0.288
age	1.01 (1.00-1.02)	0.017
WBC	0.85 (0.78–0.91)	0.011
Neutrophils	1.01 (0.99–1.02)	0.233
lymphocyte	1.00 (0.99–1.02)	0.268
monocyte	1.07 (1.03–1.10)	< 0.001
RBC	0.88 (0.81–0.94)	< 0.001
hsCRP	1.01 (0.99–1.02)	0.202
AST	1.00 (0.99-1.00)	0.432
LDH	1.00 (0.99-1.00)	0.317
Albumin	0.95 (0.93–0.98)	< 0.001
Globulin	1.00 (0.97–1.03)	0.77
Creatinine	1.01 (1.00-1.01)	0.007
Urea	1.01 (1.00-1.01)	0.002
TG	0.75 (0.62–0.91)	0.003
HDL-C	1.04 (0.67–1.60)	0.875
LDL-C	0.85 (0.72–1.01)	0.07
Apo A1	0.79 (0.49–1.27)	0.326
Apo B	0.49 (0.30–0.83)	0.007
FBG	0.97 (0.89–1.06)	0.097
ALT	1.00 (0.99-1.00)	0.491
HBA1c	1.18 (1.11–1.25)	< 0.001
LVEF	0.96 (0.95–0.97)	< 0.001
LVFS	0.94 (0.92–0.96)	< 0.001
AF
No	Reference	< 0.001
Yes	1.71 (1.31–2.23)
Hypertension
No	Reference	0.007
Yes	1.41 (1.10–1.83)
hyperlipidemia
No	Reference
Yes	0.96 (0.50–1.80)	0.887
diabetes
No	Reference
Yes	0.85 (0.37–1.33)	0.394
smoking
No	Reference
Yes	0.85 (0.64–1.14)	0.233
drinking	1.07 (0.52–2.20)	0.86
No	Reference
Yes	0.91 (0.66–1.24)	0.537
SBP	1.01 (1.00-1.01)	0.285
DBP	1.01 (1.00-1.01)	0.164
BMI	1.01 (0.96–1.04)	0.912
WBC, white blood cell; RBC, red blood cell; hsCRP, highly sensitive C-Reactive Protein; AST, aspartate amino transferase; LDH, lactate dehydrogenase; TG, triglyceride; HDL-C, high-density lipoprotein cholesterol; LDL-C, low-density lipoprotein cholesterol; Apo A1; Apolipoprotein A1; Apo B, Apolipoprotein B; FBG, fasting blood glucose; ALT, Alaninetransaminase; HBA1c, hemoglobin A1c; LVEF, left ventricular ejection fraction; LVFS, left ventricular fractional shortening; AF, atrial fibrillation; SBP, systolic blood pressure; DBP, diastolic blood pressure; BMI, Body mass index. Discrepant p-values are represented in bold.

Subsequently, we devided the dataset into training dataset and validation dataset. As shown in Table S1, most features had no statistical differences. We then used LASSO COX regression to identify the best feature variables, selecting 4 out of 34 variables (Fig. 2) in training dataset. We considered monocytes (OR: 1.04; 95% CI: 1.00-1.09; P = 0.051), serum creatinine (OR: 1.02; 95% CI: 1.00-1.02; P = 0.042), sCD40L (OR: 1.07; 95% CI: 1.00-1.14; P = 0.033), and a history of hypertension (OR: 1.30; 95% CI: 1.01–1.58; P = 0.048) as independent risk factors for readmission in patients with chronic heart failure (Table 4). We then constructed a nomogram based on these features to predict the probability of readmission in patients with chronic HF within 30 days, 60 days, and 180 days (Fig. 3). The nomogram showed the best predictive ability of the model for readmission at 12 months, with AUC of 0.524, 0.629, and 0.731 in the training dataset respectively and the area in the validation dataset was 0.526, 0.627, and 0.7.4 at 30, 60, and 180 days, respectively (Fig. 4). To evaluate the consistency between actual readmission and predicted probability, we plotted a calibration curve, as shown in Fig. 5. In both the training and validation datasets, the predicted and actual readmission of the model remained relatively consistent. To evaluate the clinical effectiveness of the model, we conducted a decision curve analysis, which showed good net benefit under the probability of column line events in both the training and validation dataset, with higher net benefit in the validation dataset (Fig. 6).

Table 4

LASSO COX regression analysis
Variable	Hazard Radio (95% CI)	P-value
monocyte	1.04 (1.00-1.09)	0.051
Creatinine	1.02 (1.00-1.02)	0.042
sCD40L	1.07 (1.00-1.14)	0.023
Hypertension
No	Reference
Yes	1.30 (1.01–1.59)	0.048
Discrepant p-values are represented in bold.

3.4 Model performance

In Fig. 7, we present the area under the curve for different models or parameters. Our results show that when monocytes, creatinine, and a history of hypertension were used to model readmission in patients with chronic heart failure, the area under the curve was 0.686. However, when sCD40L was added to this model, the area under the curve significantly improved to 0.731, indicating a higher accuracy in predicting readmission.

In this study, we examined the relationship between sCD40L and readmission in patients with chronic heart failure, as well as its association with biochemical, demographic, and imaging data. Notably, our study is the first to establish a link between sCD40L levels and readmission in patients with chronic heart failure. Using LASSO analysis, we identified monocytes, creatinine, sCD40L, and hypertension history as independent risk factors for readmission in these patients. Additionally, we developed a nomogram based on sCD40L to predict the probability of readmission for chronic heart failure.

The clinical features of HF patients requiring readmission have been explored, and studies have shown that hemoglobin, blood sodium, SBP, heart rate, blood urea nitrogen, age, troponin, and BNP are risk factors for readmission within 30 days (11). Earlier studies have constructed predictive models for the readmission of HF patients, with age, NYHA, BMI, blood creatinine, and inflammatory markers as key factors, and the area under the curve (AUC) of the model reached 0.739 (95% CI: 0.673–0.800) (12). Studies have also used least absolute shrinkage and selection operator (LASSO) regression to screen for risk factors of readmission within 180 days in HF patients. Age, acute HF, emergency, beta-blocker usage, and blood urea nitrogen were ultimately included in the criteria, and the model's AUC reached 0.75 (95% CI: 0.69–0.81) (13). In our model, the AUC was 0.646 without the use of sCD40L. However, when sCD40L was included as a predictive factor, the AUC reached 0.764, indicating a significant improvement in the prediction.

sCD40L, an inflammatory mediator, is mainly released by activated platelets (6) and is a strong predictor of cardiovascular events. In animal models, previous studies have shown that CD40L promotes platelet aggregation, monocyte recruitment, and exosmosis by promoting the formation of ultra-large von Willebrand factor multipliers, leading to atherosclerosis (14). In a myocardial infarction model, recombinant sCD40L can specifically bind alphaIIbbeta3 and activated platelets, inducing platelet diffusion (15). Some studies have explored the relationship between cardiovascular disease and sCD40L levels. In a study by Christopher, an increase in sCD40L levels in acute coronary syndrome leads to an increased risk of recurrent myocardial infarction and death (16). Hamdi also obtained similar results, indicating that higher levels of sCD40L were associated with the 1-year all-cause mortality rate of ST-segment elevation myocardial infection in patients receiving primary percutaneous coronary intervention (17). Recent studies have shown that changes in the c allele of CD40's single nucleotide polymorphism can cause a pro-inflammatory phenotype in endothelial cells (18). In summary, several studies have confirmed that sCD40L is closely associated with the risk of cardiovascular diseases. Therefore, further research is required to confirm the relationship between sCD40L levels and cardiovascular disease prognosis.

We also conducted a correlation analysis between sCD40L levels and various parameters. Our findings revealed a positive correlation between sCD40L and males, which is consistent with existing literature. This may be due to the higher smoking rate among men, as smoking is a key factor leading to inflammatory reactions(19, 20). Additionally, we observed a positive correlation between sCD40L and some inflammatory markers and harmful metabolites such as hsCRP, uric acid, and LDL-C, which is in line with previous reports. However, we found a negative correlation between sCD40L levels and FBG, HBA1C, diabetes history. This could be because we did not collect information on the medication of patients for controlling blood sugar levels. Moreover, we noted a positive correlation between sCD40L levels and EF/FS. These are indicators used to diagnose heart failure and reflect its severity.Taken together, our results suggest that high sCD40L levels are associated with readmission risk in patients with chronic heart failure.

Finally, we constructed a nomogram based on sCD40L to predict the risk of readmission within 180 days for patients with chronic heart failure. Our model has demonstrated good accuracy and can provide clinicians with valuable insights to develop a more comprehensive plan for managing heart failure patients.

Our study had some limitations. First, sCD40L levels were measured from the blood samples collected from patients 2–3 days after admission. During sample processing, there may be degradation of sCD40L, which may cause errors in the experimental results. Second, although this was a prospective cohort study, there was no causal relationship between sCD40L levels in chronic HF patients and the time of readmission. Finally, owing to the limitations of the study, our sample size was relatively small, and we need to expand the sample size further to verify the accuracy of the model.

The level of sCD40L was related to readmission results in HF patients. Patients with high sCD40L levels had an increased risk of readmission within 180 days and exhibited significant associations with monocytes, creatinine, and a history of hypertension. Consequently, the model has good predictive performance in stratifying chronic patients HF and in timely clinical decision making.

Conflict of Interest

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Author Contributions

FP and LM: conceptualization, methodology, design of the research, writing, and original draft preparation. YX: data collection and analysis. HL: experimental data collection. HL: experimental data analysis. PZ: software validation and result interpretation. YX: figures preparation. YX and LM: reviewing and revising and editing. All authors approved the final version of the manuscript.

Funding

This study is supported by the National Natural Science Foundation of China (No. 82000252); Zhejiang Province Medical and Health Science and Technology Program (2021RC032); Medical and Health Science and Technology Plan Project of Shaoxing City (2020A13018); Postdoctoral program of Shaoxing People's Hospital(2021BSQDJ01); The 551-talent project of Zhejiang Province.

Acknowledgments

None

Data Availability Statement

All procedures performed in our study involving human participants were in accordance with the ethical standards of the Ethical Committee of the Shaoxing People’s Hospital and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards.

The datasets used and/or analysed during the current study available from the corresponding author on reasonable request.

Heidenreich PA, Bozkurt B, Aguilar D, Allen LA, Byun JJ, Colvin MM, et al. 2022 AHA/ACC/HFSA Guideline for the Management of Heart Failure: Executive Summary: A Report of the American College of Cardiology/American Heart Association Joint Committee on Clinical Practice Guidelines. Circulation. 2022;145(18):e876-e94.
Wang H, Chai K, Du M, Wang S, Cai JP, Li Y, et al. Prevalence and Incidence of Heart Failure Among Urban Patients in China: A National Population-Based Analysis. Circ Heart Fail. 2021;14(10):e008406.
Tsao CW, Aday AW, Almarzooq ZI, Alonso A, Beaton AZ, Bittencourt MS, et al. Heart Disease and Stroke Statistics-2022 Update: A Report From the American Heart Association. Circulation. 2022;145(8):e153-e639.
Liu J, Liu P, Lei MR, Zhang HW, You AL, Luan XR. Readmission Risk Prediction Model for Patients with Chronic Heart Failure: A Systematic Review and Meta-Analysis. Iran J Public Health. 2022;51(7):1481-93.
Michel NA, Zirlik A, Wolf D. CD40L and Its Receptors in Atherothrombosis-An Update. Front Cardiovasc Med. 2017;4:40.
Erez O, Romero R, Hoppensteadt D, Fareed J, Chaiworapongsa T, Kusanovic JP, et al. Premature labor: a state of platelet activation? J Perinat Med. 2008;36(5):377-87.
Daub S, Lutgens E, Münzel T, Daiber A. CD40/CD40L and Related Signaling Pathways in Cardiovascular Health and Disease-The Pros and Cons for Cardioprotection. Int J Mol Sci. 2020;21(22).
Aslan G, Polat V, Bozcali E, Opan S, Çetin N, Ural D. Evaluation of serum sST2 and sCD40L values in patients with microvascular angina. Microvasc Res. 2019;122:85-93.
Pereira-da-Silva T, Ferreira V, Castelo A, Caldeira D, Napoleão P, Pinheiro T, et al. Soluble CD40 ligand expression in stable atherosclerosis: A systematic review and meta-analysis. Atherosclerosis. 2021;319:86-100.
Ueland T, Aukrust P, Yndestad A, Otterdal K, Frøland SS, Dickstein K, et al. Soluble CD40 ligand in acute and chronic heart failure. Eur Heart J. 2005;26(11):1101-7.
Eapen ZJ, Liang L, Fonarow GC, Heidenreich PA, Curtis LH, Peterson ED, et al. Validated, electronic health record deployable prediction models for assessing patient risk of 30-day rehospitalization and mortality in older heart failure patients. JACC Heart Fail. 2013;1(3):245-51.
Han Q, Ren J, Tian J, Yang H, Zhang Q, Wang R, et al. A nomogram based on a patient-reported outcomes measure: predicting the risk of readmission for patients with chronic heart failure. Health Qual Life Outcomes. 2020;18(1):290.
Gao S, Yin G, Xia Q, Wu G, Zhu J, Lu N, et al. Development and Validation of a Nomogram to Predict the 180-Day Readmission Risk for Chronic Heart Failure: A Multicenter Prospective Study. Front Cardiovasc Med. 2021;8:731730.
Popa M, Tahir S, Elrod J, Kim SH, Leuschner F, Kessler T, et al. Role of CD40 and ADAMTS13 in von Willebrand factor-mediated endothelial cell-platelet-monocyte interaction. Proc Natl Acad Sci U S A. 2018;115(24):E5556-e65.
André P, Prasad KS, Denis CV, He M, Papalia JM, Hynes RO, et al. CD40L stabilizes arterial thrombi by a beta3 integrin--dependent mechanism. Nat Med. 2002;8(3):247-52.
Heeschen C, Dimmeler S, Hamm CW, van den Brand MJ, Boersma E, Zeiher AM, et al. Soluble CD40 ligand in acute coronary syndromes. N Engl J Med. 2003;348(12):1104-11.
Pusuroglu H, Akgul O, Erturk M, Uyarel H, Bulut U, Akkaya E, et al. Predictive value of elevated soluble CD40 ligand in patients undergoing primary angioplasty for ST-segment elevation myocardial infarction. Coron Artery Dis. 2014;25(7):558-64.
Sultan CS, Weitnauer M, Turinsky M, Kessler T, Brune M, Gleissner CA, et al. Functional association of a CD40 gene single-nucleotide polymorphism with the pathogenesis of coronary heart disease. Cardiovasc Res. 2020;116(6):1214-25.
Wang H, Chen H, Fu Y, Liu M, Zhang J, Han S, et al. Effects of Smoking on Inflammatory-Related Cytokine Levels in Human Serum. Molecules. 2022;27(12).
Jefferis BJ, Whincup PH, Welsh P, Wannamethee SG, Rumley A, Lawlor DA, et al. Prospective study of circulating soluble CD40 ligand concentrations and the incidence of cardiovascular disease in a nested prospective case-control study of older men and women. J Thromb Haemost. 2011;9(8):1452-9.

No competing interests reported.

TableS1.docx

Download PDF

Version 1

posted

You are reading this latest preprint version

Construction and evaluation of a 180-day readmission prediction model for chronic heart failure patients based on sCD40L

Status:

Version 1

Abstract

Background

Methods

Results

Conclusion

Figures

1 Introduction

2 Materials and methods

3 Result

4 Discussion

5 Conclusion

Declarations

References

Additional Declarations

Supplementary Files

Status:

Version 1