Prediction Score for Clinical Outcome of Chinese Patients with Cerebral Venous Thrombosis

Introduction

Cerebral venous thrombosis (CVT) accounts for 0.5–1% of all stroke cases, exhibiting an incidence rate of 1.3 to 1.6 per 100,000 population.^1,2 With the recent advancements in neuroimaging, the diagnosis of CVT has achieved greater accuracy despite the diverse clinical manifestations observed in patients.^2–4 Nevertheless, reports indicate a significant proportion of CVT patients experience poor clinical outcomes and elevated mortality rates.^5,6 For instance, a multicenter study by Ferro et al⁵ found that 13.4% of CVT patients across 21 Western countries encountered poor clinical outcomes, with 8.3% succumbing to the condition. Similar findings were reported by Wasay et al⁶ for nine Asian countries and in Germany where 56.1% of severe CVT patients had unfavorable outcomes and 34.2% died.⁷ These observations underscore the urgent need for a predictive method for the clinical outcomes of CVT.

Prior research, including our previous studies, has shown that the neutrophil-to-lymphocyte ratio (NLR), neuron-specific enolase (NSE), and diastolic blood pressure (DBP) can be significant predictors of CVT outcomes.^8–10 Although other studies has identified a range of prognostic markers for CVT,^5,11,12 there is still uncertainty as to whether these markers might yield contradictory results. Consequently, the development of a CVT outcome score based on the identified prognostic markers may offer enhanced predictive capabilities for the clinical outcome of CVT.

Previously reported CVT outcome scores, such as those by Ferro et al⁶ in 2009 and Bushnaq et al¹³ in 2018, present significant discrepancies between their versions. While Ferro et al⁶ focused on CVT patients in Portugal, France, and Brazil, Bushnaq et al¹³ recruited patients from New Mexico in the United States. It is posited that different nations and ethnic populations may have distinct prognostic markers, leading to unique CVT outcome scores. In alignment with this notion, the current study aims to establish a CVT outcome score specifically tailored for Chinese patients with CVT.

Methods

Study Sample

The study sample was derived from the CCC cohort (NCT03919305), a multicenter study conducted across 26 tertiary hospitals in mainland China.¹⁴ Patients diagnosed with CVT were prospectively enrolled between January 2021 and May 2022. The diagnosis of CVT was established on the basis of clinical symptoms and neuroimaging findings, as delineated previously.¹⁰

Data Extraction

Potential prognostic indicators, including age, sex, level of consciousness (LOC),¹⁵ new-onset epilepsy,¹⁶ mental disorders,⁶ malignancy,¹⁷ central nervous system (CNS) infections,⁵ deep venous system thrombosis,¹⁸ utilization of oral contraceptive pills (OCPs),¹⁹ DBP,¹⁰ papilledema,¹⁶ number of sinuses involved,¹⁶ occurrence of parenchymal lesions,²⁰ C-reactive protein (CRP),¹² fasting blood glucose,¹¹ NLR,⁸ NSE,⁹ platelet count (PLT),¹⁶ and serum sodium level¹⁶ were extracted from the CCC database for further analysis.

Outcome Measurement

CVT outcome was assessed using the modified Rankin scale (mRS) at six months post-onset. In alignment with prior studies, an mRS of ≤ 1 was characterized as a good outcome, and an mRS of > 1 as a poor outcome.^8,12 The binarized mRS outcomes were employed for additional analyses.

Statistical Analysis

Statistical evaluations were conducted employing SPSS version 19.0 (SPSS Inc., Chicago, IL, USA). Two-sample t-tests were utilized to examine differences in continuous variables between the groups, while the chi-square test was applied for categorical variables. The maximum value of the Youden index (sum of sensitivity and specificity minus one) in the receiver operating characteristic (ROC) curve²¹ was used to identify cut-off values. Univariate logistic regression analysis, followed by multivariate logistic regression modeling, was performed to identify prognostic markers in Chinese patients with CVT. The identified prognostic markers were implemented to construct CVT outcome scores, with one point designated to the lowest OR and multiple points allocated based on the multiplicities of the lowest OR.²² The efficacy of the CVT risk score was gauged through classification accuracy, as ascertained by the area under the ROC curve (AUC),²³ the distribution of the mRS score according to the CVT risk score, and a two-sample t-test. Statistical significance was determined as a two-sided p value of < 0.05.

Results

A total of 170 patients with CVT, comprising 75 males and 95 females, were studied with a mean age of 35.71 (standard error of mean [SEM] = 1.03). The most frequent occurrences were papilledema (55.88%), new-onset epilepsy (32.35%), and decreased LOC (14.12%). Among the patients, 95 (55.88%) had involvement in three or more sinuses, and 60 (35.29%) had parenchymal lesions. The specific data on the reported prognostic markers are delineated in Table 1.

Table 1 The Detailed Data of Reported Prognostic Markers in Chinese CVT Patients

The patients were subsequently classified into groups with good outcomes (n = 129) and poor outcomes (n = 26) based on the mRS scores at a six-month follow-up. Factors including age (P = 0.0331), DBP (P = 0.0436), NLR (P = 0.0006), NSE (P = 0.0001), the number of patients with decreased LOC (P = 0.0415), new-onset epilepsy (P = 0.0339), and parenchymal lesions (P = 0.0005) were found to be significantly different between the two groups (Table 1).

Univariate logistic regression analysis revealed age (P = 0.036), decreased LOC (P = 0.048), new-onset epilepsy (P = 0.038), DBP (P = 0.047), parenchymal lesions (P = 0.001), CRP level (P = 0.002), NLR (P = 0.012), and NSE level (P = 0.008) as significant prognostic markers in Chinese patients with CVT (Table 2).

Table 2 Univariate Logistic Regression Between Reported Prognostic Markers at Baseline and the Clinical Outcome at the 6-Month Follow-Up in Chinese CVT Patients

Cut-off values for age (27.5 years), DBP (79.5 mmHg), CRP (30 mg/L), NLR (6.6), and NSE (16.5 ng/mL) were ascertained using the Youden index’s maximum value (Figure 1). These continuous variables were transformed into binary variables for further multivariate modeling using the cut-off values. AUC analysis confirmed that age (AUC = 0.6263, P = 0.0426), DBP (AUC = 0.6434, P = 0.0291), NLR (AUC = 0.7667, P = 0.0027), and NSE (AUC = 0.6545, P = 0.0346) independently predicted CVT’s clinical outcomes with high specificity and sensitivity. Nevertheless, CRP’s predictive value (AUC = 0.5875, P = 0.2441) exhibited low sensitivity and specificity.

Figure 1 Analysis of the ROC curve to determine cut-off values. The ROC analysis was utilized to identify the cut-off values for age (A), DBP (B), CRP (C), NLR (D), and NSE (E).

A multivariate model was constructed including age, decreased LOC, new-onset epilepsy, DBP, parenchymal lesions, CRP, NLR, and NSE levels (Table 3). The significant variables for predicting the clinical outcome of Chinese CVT patients were identified as age > 27.5 years (P = 0.036), DBP > 79.5 mmHg (P = 0.003), NLR > 6.6 (P < 0.001), and NSE > 16.5 ng/mL (P = 0.015).

Table 3 Multivariate Logistic Regression Between Identified Prognostic Markers at Baseline and the Clinical Outcome at the 6-Month Follow-Up in Chinese CVT Patients

A CVT outcome score was established using four prognostic markers (age, DBP, NLR, and NSE) identified in the multivariate logistic analysis. Points were assigned as follows: one point for NSE > 16.5 ng/mL and age > 27.5 years; two points for DBP > 79.5 mmHg; and three points for NLR > 6.6 (Table 4). There was a positive correlation between the CVT outcome score at baseline and the mRS at six months follow-up (P < 0.001, Figure 2A). ROC analysis indicated that the CVT outcome score robustly predicted CVT’s clinical outcome (AUC = 0.8785, P = 0.0001, Figure 2B). Additionally, the Youden index identified 3.5 as the cut-off value for the CVT outcome score, so patients were categorized into CVT outcome scores of > 3 and ≤ 3. A two-sample t-test revealed that the mRS score in patients with a CVT outcome score of > 3 was substantially higher than that in patients with a CVT outcome score of ≤ 3 (Figure 2C).

Table 4 Determination of CVT Outcome Score for Chinese Patients

Figure 2 Validation of the CVT outcome score. (A) The CVT outcome score at baseline exhibited a positive correlation with mRS at a 6-month follow-up period. (B) A CVT outcome score of 3.5 was identified as a cut-off value to predict the clinical outcome of CVT. (C) Patients with a CVT risk score greater than 3 displayed a significantly higher mRS score compared to those with scores less than or equal to 3. *P<0.05.

Discussion

In the present study, both univariate logistic analysis and multivariate logistic analysis were utilized to identify age, DBP, NLR, and NSE as prognostic indicators for CVT. A CVT outcome score, encompassing the aforementioned four variables, was constructed to assess the clinical outcomes in Chinese patients diagnosed with CVT.

Previous findings from two multicenter CVT cohorts, ISCVT and VENOST, have shown that older age correlates with adverse CVT outcomes.^5,24 In our investigation, the cut-off age was established at 27.5 years, contrasting with 37 years in the ISCVT study, and 50 years in the VENOST study.^5,24 This discrepancy is likely attributed to the unique study populations, as neither ISCVT nor VENOST included patients with CVT from clinical centers within mainland China.

NSE, recognized as a biomarker for brain injury released upon neuronal cell membrane damage,²⁵ was identified in our previous work as an effective predictor for CVT severity and outcome.⁹ The prognostic role of NSE was further validated in this study through multivariate logistic analysis. Furthermore, a DBP reading greater than 79.5 mmHg emerged as a significant prognostic factor for CVT following multivariate logistic analysis. An elevated DBP is associated with increased intracranial pressure, which may consequently induce heightened inflammatory reactions, oxidative stress, disruptions to the blood-brain barrier, and even cerebral herniation.^26,27

Inflammation is recognized as playing a pivotal role in CVT. The NLR serves as a straightforward parameter to gauge inflammatory status and is often deemed more robust than isolated changes in neutrophils or lymphocytes.^28,29 Earlier studies revealed a correlation between NLR and adverse clinical outcomes in CVT.^8,12 The findings of the current study reinforced the significance of NLR in forecasting the clinical outcomes of CVT. Additionally, SII, defined as PLT × NLR, emerges as a novel inflammatory indicator for evaluating host immune and inflammatory conditions. The SII has previously been reported to act as a prognostic marker for CVT.³⁰ In this particular investigation, we established that NLR, rather than PLT, anticipated the clinical outcomes of patients suffering from CVT. This observation has been proposed to be integral to the prognostic efficacy of SII in CVT patients.

The baseline CVT outcome score was found to be positively associated with the mRS score at six months post-onset. Furthermore, the CVT outcome score efficiently predicted the clinical outcome of CVT, pinpointing a cut-off value of 3.5. Since the CVT outcome score is an integer, a score exceeding 3 signifies a poor clinical outcome, whereas a score of 3 or less is indicative of a favorable clinical result. Endovascular therapy is currently considered when patients exhibit deterioration despite anticoagulation measures.² Research has shown that CVT patients at a higher risk of undesirable clinical outcomes may benefit from early endovascular therapy, thereby preventing thrombus formation.³¹ The CVT outcome score could serve as a vital tool in identifying CVT patients at risk for poor clinical outcomes, thereby enabling timely implementation of endovascular therapy.

The scope of this study was limited by a relatively modest sample size. However, we aim to corroborate the CVT outcome score using a more extensive cohort in future research.

Conclusions

This study successfully developed a CVT outcome score that encompasses age, DBP, NLR, and NSE for the assessment of Chinese patients. The correlation analysis revealed that the baseline CVT outcome score was positively associated with the mRS score at a follow-up period of six months. This finding demonstrates that a CVT outcome score greater than 3 may serve as an effective marker to identify patients with CVT who are at an elevated risk for poor clinical outcomes. Consequently, patients exhibiting CVT outcome scores above 3 should receive prioritized attention, including additional care and early intervention measures, to mitigate the risk of clinical deterioration. However, the relatively small number of samples obtained is a limitation of this study. Therefore, the results should be interpreted more cautiously.