Patient characteristics of HGGs with LMS

This study enrolled 114 patients, including 76 (66.7%) males and 38 (33.3%) females with a mean age of 41.5 ± 15.4 years. Patients were followed up for a median time of 16.5 (2.0–117.0) months postoperatively. LMS could occur during glioma diagnosis or recurrence or progression, which was defined as primary LMS and secondary LMS. Primary LMS consisted of 42 (36.8%) patients, including three with anaplastic astrocytomas, two with pleomorphic xanthoastrocytomas, 29 with GBMs, one with epithelioid GBM, five diffuse midline gliomas, and two gliosarcomas. Secondary LMS consisted of 72 (63.2%) patients, including 41 with GBMs, two with diffuse midline gliomas, two with epithelioid GBMs, one with gliosarcoma, five with anaplastic oligodendrogliomas, and 21 with anaplastic astrocytomas at the initial glioma diagnosis. Progression to glioblastoma at LMS diagnosis was confirmed in 14 patients originally diagnosed with anaplastic astrocytoma or oligodendroglioma. Five patients were unable to obtain the latest pathology diagnosis due to lacking surgery or biopsy after LMS diagnosis. Table 1 and 2 respectively shows the clinical, radiological, and pathological characteristics at initial glioma diagnosis and at LMS diagnosis. IDH mutation information was unavailable and pathological tissue was not obtained in six patients because they only received an Ommaya reservoir in our hospital.

Table 1 Clinical, radiological and pathological characteristics of patients at initial glioma diagnosisTable 2 Clinical, radiological and pathological characteristics of patients at LMS diagnosisRadiographic characteristics of patients with LMS

According to the radiographic features of dissemination, we categorize it into three types. Brain MRI was obtained in all cases and spinal MRI was obtained in 41 patients. This cohort included 50 (43.9%), 27 (23.7%), and 37 (32.5%) with types Ia, Ib, and II, respectively. Nodular LMS occurred in 18 (15.8%) patients. Secondary LMS was identified in 16 (14.0%) patients with stable disease at the initial tumor site. Hydrocephalus occurred in 27 (23.7%) patients at the time of LMS diagnosis.

Management of patients after glioma diagnosis and LMS diagnosis

Table 1 outlines the management strategies after glioma diagnosis. All patients underwent surgical intervention at the time of glioma diagnosis, including biopsies in 14 (12.3%), gross total resections in 52 (45.6%), and subtotal resections in 48 (42.1%) patients. Subsequent adjuvant therapy was performed in all patients, included radiotherapy in 90 (78.9%), chemotherapy in 98 (86.0%), intrathecal methotrexate in 12 (10.5%), antiangiogenic therapy in 24 (21.1%) patients, etc. The Stupp protocol was performed on 71 patients. At least two surgical treatments were performed on 73 patients. In the secondary LMS group, 55 (76.4%) cases had ventricular entry during initial resection, 15 (20.8%) had no ventricular entry, and 2 (2.8%) cases had no records.

Table 2 outlines the management strategies after LMS diagnosis. Tumor resection was performed on 70 (61.4%) patients after the LMS diagnosis, among them, 27 (11 with primary lesions, 12 with local recurrent lesions and four with disseminated lesions) underwent GTR. Fifteen (13.2%) patients received the Ommaya reservoir, while only 12 patients underwent subsequent intrathecal methotrexate (MTX) treatment. Operations were not performed on 21 (17.4%) patients who directly underwent subsequent adjuvant treatment. Subsequent adjuvant treatment after LMS diagnosis was administered in 81 (71.1%) patients, including radiotherapy (27, 23.7%), chemotherapy (69, 60.5%), intrathecal MTX (12, 10.5%), antiangiogenic therapy (23, 20.2%), and clinical trials (4, 3.5%). All 27 cases with hydrocephalus underwent VP shunt. Among the patients with secondary LMS, there were 56 patients with local recurrence and LMS.

Prognostic factors of OS in all patients

Death was recorded in 106 patients upon study completion. The median OS was 17.0 months. Univariate analysis demonstrated statistically significant associations between OS and pathology grade

III (HR: 0.343, 95% CI: 0.214-0.548, p = 0.000), KPS of ≥80 (HR: 0.480, 95% CI: 0.287-0.804, p = 0.005), GTR at glioma diagnosis (HR: 0.313, 95% CI: 0.204-0.480, p = 0.000), radiotherapy (HR: 0.354, 95% CI: 0.217-0.577, p = 0.000), chemotherapy (HR: 0.373, 95% CI: 0.214-0.651, p = 0.001), primary LMS (HR: 3.147, 95% CI: 2.046-4.841, p = 0.000), GTR after LMS diagnosis (HR: 0.511, 95% CI: 0.317-0.823, p = 0.006), non-adjuvant therapy after LMS diagnosis (HR: 2.044, 95% CI: 1.346-3.105, p = 0.001), IDH mutation (HR: 0.285, 95% CI: 0.162-0.501, p = 0.000), MGMT methylation (HR: 0.220, 95% CI: 0.062-0.788, p = 0.020) (Table 3). Multivariable analysis revealed that pathology grade III (HR: 0.043, 95% CI: 0.003-0.589, p = 0.018) and GTR after LMS diagnosis (HR: 0.058, 95% CI: 0.009-0.384, p = 0.003) were statistically significantly associated with longer OS, while non-adjuvant therapy after LMS diagnosis (HR: 30.58, 95% CI: 4.68-199.89, p = 0.000) was predictor of shorter OS (Table 3). The median OS in patients with pathology grade III and GTR after LMS diagnosis were longer than those with pathology grade IV, non-GTR after LMS diagnosis (31.5 vs. 15.0 months, p = 0.000; 26.0 vs. 15.0 months, p = 0.004, respectively; log-rank test; Fig. 3). The median OS in patients with non-adjuvant therapy after LMS diagnosis was shorter than patients with adjuvant therapy after LMS diagnosis (12.0 vs. 20.0 months, p = 0.001, log-rank test; Fig. 3).

Table 3 Overall survival by univariable and multivariable Cox analysesFig. 3

Comparison of OS and PLS by Kaplan–Meier curves in all patients. OS of (a) pathology grade, (b) GTR after LMS diagnosis and (c) non-adjuvant therapy after LMS diagnosis; PLS of (d) nodular LMS, (e) GTR after LMS diagnosis and (f) non-adjuvant therapy after LMS diagnosis

Prognostic factors of PLS in all patients

The median PLS was 6.0 months. Univariate analysis revealed KPS of ≥80 at LMS diagnosis (HR: 0.472, 95% CI: 0.229-0.744, p = 0.001), nodular LMS (HR: 0.468, 95% CI: 0.269-0.815, p = 0.007), MRI type Ia (HR: 0.600, 95% CI: 0.402-0.896, p = 0.012), GTR after LMS diagnosis (HR: 0.582, 95% CI: 0.365-0.927, p = 0.023), radiotherapy after LMS (HR: 0.603, 95% CI: 0.398-0.997, p = 0.048), chemotherapy after LMS (HR: 0.362, 95% CI: 0.237-0.551, p = 0.000), Intrathecal MTX (HR: 0.473, 95% CI: 0.245-0.913, p = 0.026) were associated with better PLS, while non-adjuvant therapy after LMS diagnosis (HR: 4.662, 95% CI: 2.887-7.528, p = 0.000) and MRI type II (HR: 2.217, 95% CI: 1.443-3.405, p = 0.000) were associated with shorter PLS (Table 4). Multivariable analysis revealed nodular LMS (HR: 0.530, 95% CI: 0.300-0.938, p = 0.029), GTR after LMS diagnosis (HR: 0.554, 95% CI: 0.346-0.885, p = 0.013), and non-adjuvant therapy after LMS diagnosis (HR: 4.273, 95% CI: 2.635-6.931, p = 0.000) were identified as independent prognostic factors on PLS (Table 4). The median PLS in patients with nodular LMS and GTR after LMS diagnosis were longer than those without GTR (17.0 vs. 6.0 months, p = 0.005; 9.0 vs. 6.0 months, p = 0.017, respectively; log-rank test; Fig. 3). The median PLS in patients with non-adjuvant therapy after LMS diagnosis was shorter than patients with adjuvant therapy after LMS diagnosis (3.0 vs. 8.5 months, p = 0.000, log-rank test; Fig. 3).

Table 4 Post-LMS survival by univariable and multivariable Cox analysesPrognostic factors of OS in GBM subgroup

In univariable analysis, KPS of ≥80 (HR: 0.467, 95% CI: 0.230-0.946, p = 0.035), GTR at glioma diagnosis (HR: 0.449, 95% CI: 0.268-0.750, p = 0.002), radiotherapy (HR: 0.471, 95% CI: 0.258-0.859, p = 0.014), chemotherapy (HR: 0.157, 95% CI: 0.075-0.331, p = 0.000), Intrathecal MTX (HR: 0.431, 95% CI: 0.202-0.916, p = 0.029), GTR after LMS diagnosis (HR: 0.545, 95% CI: 0.295-1.004, p = 0.052), IDH mutation (HR: 0.327, 95% CI: 0.129-0.830, p = 0.019) had better survival, while primary LMS (HR: 1.837, 95% CI: 1.099-3.070, p = 0.020) and non-adjuvant therapy after LMS diagnosis (HR: 3.830, 95% CI: 2.117-6.929, p = 0.000) had shorter survival (Table 5). Multivariable analysis revealed that GTR after LMS diagnosis (HR: 0.431, 95% CI: 0.227-0.821, p = 0.010), primary LMS (HR: 4.209, 95% CI: 2.270-7.804, p = 0.000) and non-adjuvant therapy after LMS diagnosis (HR: 7.879, 95% CI: 3.821-16.245, p = 0.000) were independent prognostic factors on OS (Table 5). The median OS in patients with GTR after LMS diagnosis was longer than the patients without GTR (25.0 vs. 14.0 months, p = 0.044, log-rank test; Fig. 4). The median OS in patients with primary LMS and non-adjuvant therapy after LMS diagnosis were shorter than the others (12.0 vs. 18.0 months, p = 0.016; 6.5 vs. 19.0 months, p = 0.000, respectively, log-rank test; Fig. 4).

Table 5 Overall survival by univariable and multivariable Cox analyses of GBM subgroupFig. 4

Comparison of OS and PLS by Kaplan–Meier curves in GBM subgroup. OS of (a) primary LMS, (b) GTR after LMS diagnosis and (c) non-adjuvant therapy after LMS diagnosis; PLS of (d) KPS (≥80) at LMS diagnosis, (e) Chemotherapy after LMS and (f) Intrathecal MTX

Prognostic factors of PLS in GBM subgroup

In GBM subgroup, univariate analysis demonstrated that KPS of ≥80 at LMS diagnosis (HR: 0.459, 95% CI: 0.259-0.811, p = 0.007), nodular LMS (HR: 0.481, 95% CI: 0.259-0.895, p = 0.021), MRI type II (HR: 2.061, 95% CI: 1.158-3.668, p = 0.014), GTR after LMS diagnosis (HR: 0.571, 95% CI: 0.314-1.039, p = 0.067), chemotherapy after LMS (HR: 0.106, 95% CI: 0.050-0.222, p = 0.000), Intrathecal MTX (HR: 0.441, 95% CI: 0.200-0.972, p = 0.042) and primary LMS (HR: 0.576, 95% CI: 0.350-0.948, p = 0.030) were associated with better PLS, while non-adjuvant therapy after LMS diagnosis (HR: 4.662, 95% CI: 2.887-7.528, p = 0.000) was associated with shorter PLS (Table 6). Multivariable analysis revealed that KPS of ≥80 at LMS diagnosis (HR: 0.472, 95% CI: 0.256-0.870, p = 0.016), chemotherapy after LMS (HR: 0.105, 95% CI: 0.048-0.229, p = 0.000) and Intrathecal MTX (HR: 0.382, 95% CI: 0.150-0.974, p = 0.044) were independent prognostic factors of PLS (Table 6). The median PLS in patients with KPS of ≥80 at LMS diagnosis, chemotherapy after LMS and Intrathecal MTX was longer than those opposites (7.0 vs. 5.0 months, p = 0.004; 12.0 vs. 3.0 months, p = 0.000; 18.0 vs. 6.0 months, p = 0.032, respectively, log-rank test; Fig. 4).

Table 6 Post-LMS survival by univariable and multivariable Cox analyses of GBM subgroup