Overall, 310 acute/subacute CVT patients were assessed for eligibility, of which, 62 patients were excluded: 50 were not severe CVT, 10 were younger than 16, 2 patients had lung cancer. In the end, 248 patients were recruited, harboring 93/248 (37.50%) severe CVT patients treated with steroid and 155/248 (62.50%) severe CVT patients treated without steroid (Table 1). In 248 patients, the mean age was 36.32 ± 15.99 years, and 157 were women (63.30%). Baseline demographic and clinical variables related to the definition of severe CVT were balanced between the two groups through PSM analysis. Finally, 170 severe CVT patients were enrolled: 85 patients in the steroid group and 85 in the nonsteroid group. Flowchart of inclusion and exclusion criteria in Fig. 1.
Baseline characteristics and outcome in 248 severe CVT patients before PSM
Univariate analysis revealed that acquired thrombophilia factors significantly differed between the steroid and nonsteroid groups (60/93 [64.50%] vs. 77/155 [49.70%], P = 0.02). The proportion of cases of coma (GCS score < 9) was significantly higher in the steroid group than in the nonsteroid group (42/93 [45.16%] versus 48/155 [30.96%], P = 0.04), but the proportions of other symptoms and signs, including seizure, motor deficits, and aphasia, had no significant different.
Compared with those in the nonsteroid group, the imaging characteristics showed more significant edema (92/93 [98.92%] vs. 139/155 [89.67%], P = 0.01), infarction (88/93 [94.60%] vs. 121/155 [78.06%], P = 0.00), and hemorrhage (58/93 [62.40%] vs. 62/155 [40.00%], P = 0.00) in the steroid group. Detailed baseline characteristics are presented in Table 1.
At the 6 months follow-up, 197/248 (79.43%) survivors achieved favorable functional outcomes (mRS, 0–2), but there were no significant differences between the steroid and nonsteroid groups (76/93[81.72%] versus 121/155[78.06%], P = 0.49). A total of 220 survivors had assessable follow-up neuroimaging data (15 patients died [3 in the steroid group and 12 in the nonsteroid group], 7 had no follow-up neuroimaging data [4 in the steroid group and 3 in the nonsteroid group], and 6 had isolated cortex venous thrombosis [4 in the steroid group and 2 in the nonsteroid group]); 189 (85.90%) survivors exhibited venous sinus vessel recanalization (partial or complete), but there was no significant difference between the steroid and nonsteroid groups (72/82 [87.80%] vs. 117/138 [84.78%], P = 0.26).
Among the 248 severe CVT patients (93 in the steroid group and 155 in the nonsteroid group), 15 died (3 in the steroid group and 12 in the nonsteroid group). Among the 233 survivors, 76 (32.61%) had chronic headaches, but the steroid group had a lower likelihood of chronic headaches (16/90 [17.78%] vs. 60/143 [41.95%], P = 0.00), 27 survivors (10.88%) had seizures, and 63 (27.04%) survivors had motor deficits, but there was no significant difference between the steroid and nonsteroid groups.
Concerning steroid-related complications, 6 (2.57%) survivors had concurrent lower extremity deep venous thrombosis, 2 (0.86%) had pulmonary embolism, and 2 (0.86%) had a concurrent infection, but there were no significant differences in the likelihood of these symptoms between the steroid and nonsteroid groups. The follow-up information for all CVT survivors is presented in Table 2.
Baseline characteristics and Outcome of severe CVT patients after PSM analysis
After PSM analysis, we controlled for age, sex, NIHSS score, mental disorders, coma (GCS ≤ 9), edema, infarction, and cerebral hemorrhage. Baseline characteristics such as sex (female), age, hereditary or acquired thrombophilia, clinical symptomatology, and neuroimaging features were not significantly different between the steroid and nonsteroid groups (Table 3).
After PSM analysis, 170 severe CVT patients were eligible for matched cohort (85 patients in the steroid and nonsteroid groups each). At the 6 months follow-up, 7 died (2 in the steroid group and 5 in the nonsteroid group), and patients treated with steroid had a significantly higher likelihood of mRS (0–2) (73/85 [85.88%] versus 64/85[75.29%], P = 0.03) and a lower likelihood of residual headache (16/83 [19.28%] versus 38/80 [ 47.50%], P = 0.00, Table 4). Among the survivors, steroid-related safety measures, including CVT recurrence (5/83 [6.02%] vs 5/80 [6.25%], P = 0.95), lower extremity deep venous thrombosis (3/83 [3.61%] vs 1/80 [1.25%], P = 0.33), pulmonary embolism (1/83 [1.20%] vs 1/80 [1.25%], P = 0.98), and infection (1/83 [1.20%] vs 1/80 [1.25%], P = 0.98), were not significantly different between the steroid and nonsteroid groups. None of the patients experienced spontaneous fracture or osteonecrosis and gastroduodenal ulcer during follow-up, the detailed results are presented in Table 4.
Independent Predictors of the favorite outcome of severe CVT patients in the matched cohort
On the other hand, we further analyzed the associated risk factors with favorite outcome (mRS 0–2) of the matched cohort using multivariable logistic regression analysis. Together with age and sex(females), risk factors of CVT (sex-specific factors, hereditary thrombophilia factors, acquired thrombophilia factors), the location of the thrombus (superior sagittal sinus, lateral sinus, deep venous system, cortical vein), NIHSS score, mental status disorders, coma, edema, venous infarction, hemorrhage, received steroid treatment, sinus recanalization status as covariates, and mRS(0–2) as the dependent variable were entered into the regression model. Finally, the logistic model produced significant findings (χ2 = 65.47, P < 0.00). Among the independent variables included in the model, age, mental status disorders, coma (GCS < 9), edema, hemorrhage and steroid treatment were statistically significant. We calculated odds ratios (ORs) and 95% confidence intervals (CIs) for the retained variables. In the multivariable logistic regression model, age, mental disorders at diagnosis, coma (GCS < 9) at diagnosis, edema at diagnosis, and hemorrhage at diagnosis were independent predictors of poor outcome(mRS > 2), and only steroid treatment (OR 4.69, 95% CI 1.63–13.48, P = 0.00) was independent predictor of favorite outcome(mRS,0–2) in the matched cohort (Table 5).
Clinical outcomes distribution after PSM analysis
The mRS distribution at 6 months follow-up were also assessed. The median duration of follow-up was 6 (IQR 5.5-7.0) months in both groups. The mRS distribution is shown in Fig. 2. Overall, compared with the nonsteroid group, patients in the steroid group had a better functional outcome on the mRS. The frequencies of mRS (0–2) were higher in the steroid group (85.88% vs 75.29%, P = 0.03), but mortality was not significantly different between groups (2.35% vs 5.88%, P = 0.25, in Table 4).