Revisiting prognostic factors in glioma with leptomeningeal metastases: a comprehensive analysis of clinical and molecular factors and treatment modalities

Patient characteristics

This study included 226 glioma patients with LM (mean age: 56.1 ± 14.2 years, 82 females and 144 males), with a median follow-up period of 18.8 (interquartile range [IQR] 10.1–28.7) months. There were 3 patients with oligodendroglioma (1.3%; one with grade 2 and two with grade 3); 11 patients with IDH-mutant astrocytoma (4.9%; seven with grade 2, one with grade 3, and three with grade 4); 200 patients with IDH-wildtype astrocytoma (88.5%; two with grade 2, 19 grade 3, and 179 grade 4), and 12 patients with H3 K27M-altered diffuse midline glioma (6.3%). Seventy-one patients (31.4%) had MGMT promoter methylation. Based on the date of LM diagnosis, 108 (47.8%) and 118 (52.8%) patients had initial and recurrent LMs, respectively. The median interval between initial glioma and LM diagnoses was 11.4 (IQR 6.6–20.8) months in recurrent LM patients.

Among patients with LM, 222 of 226 patients (98.2%) who underwent brain MRI was positive on brain MRI, and 69 of 101 patients (68.3%) who underwent spine MRI was positive on spine MRI, while 12 of 61 patients (19.7%) who underwent CSF cytology was positive. According to the MRI criteria, 181 (80.1%) and 45 (19.9%) patients had disseminated and subependymal LMs, respectively. The median OS was 17.0 (IQR 9.7–67.1) months, and 151 patients (66.8%) died. The median OS significantly differed according to the molecular subgroups of glioma (log-rank P < 0.001). The Kaplan–Meier curve of OS according to the molecular subgroups is in Supplementary Fig. 1.

Comparison of patients with LM at initial diagnosis and recurrence

Characteristics of patients with initial and recurrent LM are summarized in Table 1. The age at initial diagnosis was higher in initial LM than recurrent LM patients (58.8 vs. 53.6, P = 0.005). Patients with initial LM had a significantly higher proportion of grade 4 gliomas than those with recurrent LM (94.4% vs. 77.1%, P = 0.001).

Table 1 Characteristics of glioma patients with LM

Among patients with initial LM, 104 (96.3%) and 105 (97.2%) patients received chemotherapy and radiation therapy, respectively, at initial diagnosis. Among patients with initial LM who received radiation therapy, 15 (13.9%), 11 (10.2%), and 79 (73.1%) patients received craniospinal irradiation, whole ventricular radiation therapy, and localized radiation therapy, respectively. Antiangiogenic therapy was not performed at initial diagnosis. When recurrence was diagnosed, repeated radiation therapy was administered to five (4.6%) patients, with one (0.9%) receiving whole ventricular radiation therapy. Antiangiogenic therapy was administered to 21 (19.4%) patients with initial LM when recurrence was diagnosed, while one (0.9%) patient received experimental therapy with lenvatinib plus pembrolizumab on recurrence.

Among patients with recurrent LM, 14 (11.9%) and 17 (14.4%) patients received chemotherapy and radiation therapy, respectively, at LM diagnosis at recurrence. Among patients with recurrent LM who received radiation therapy on LM diagnosis, six (35.3%) and 11 (64.7%) patients received whole ventricular radiation therapy and localized radiation therapy, respectively. Antiangiogenic therapy was administered to 57 (48.3%) patients when recurrent LM was diagnosed. Two (1.7%) patients received experimental therapy with lenvatinib plus pembrolizumab, while one (0.8%) patient received belvarafenib treatment.

The median OS was significantly shorter in patients with initial LM than in those with recurrent LM (12.2 [IQR 6.9–23.2] vs 20.6 [IQR 12.4–40.3] months; log-rank test P < 0.001).

Predictors of OS in all patients with LM

In all patients, univariable analysis showed that higher KPS (hazard ratio [HR] = 0.95, P < 0.001), chemotherapy (HR = 0.07, P < 0.001), radiation therapy (HR = 0.12, P < 0.001), and antiangiogenic therapy (HR = 0.34, P = 0.002) were predictors of longer OS, while older age (HR = 1.02, P < 0.001), male sex (HR = 1.97, P < 0.001), histological grade 4 (P < 0.001), IDH wildtype (HR = 5.94, P < 0.001), MGMT promoter unmethylation (HR = 2.05, P < 0.001), and LM at initial diagnosis (HR = 1.97, P < 0.001) were predictors of shorter OS. Multivariable analysis revealed that chemotherapy (HR = 0.08, P < 0.001) and antiangiogenic therapy (HR = 0.43, P < 0.001) were predictors of longer OS, while male sex (HR = 1.48, P = 0.038) and LM at initial diagnosis (HR = 1.75, P = 0.002) were predictors of shorter OS among patients with LM (Table 2). Simon-Makuch curves according to chemotherapy, antiangiogenic therapy, male sex, and LM at initial diagnosis are shown in Fig. 2.

Table 2 Univariable and multivariable time-dependent Cox analyses of glioma patients with LMFig. 2figure 2

Simon–Makuch curves according to A sex, B LM at initial diagnosis, C chemotherapy, and D antiangiogenic therapy in all patients with LM. LM leptomeningeal metastases

Predictors of OS in patients with initial LM

Among patients with initial LM, univariable analysis revealed that higher KPS (HR = 0.95, P < 0.001), chemotherapy (HR = 0.17, P = 0.001), radiation therapy (HR = 0.09, P < 0.001), and antiangiogenic therapy (HR = 0.34, P = 0.002) were predictors of longer OS, while older age (HR = 1.03, P = 0.008), male sex (HR = 2.07, P = 0.015), IDH wildtype (HR = 24.94, P = 0.040), and MGMT promoter unmethylation (HR = 1.90, P = 0.034) were predictors of shorter OS. Multivariable analysis showed that higher KPS (HR = 0.96, P < 0.001), chemotherapy (HR = 0.18, P = 0.002), and antiangiogenic therapy (HR = 0.34, P = 0.004) were predictors of longer OS, while male sex (HR = 2.17, P = 0.012) was a predictor of shorter OS (Table 3). Adjusted and unadjusted Kaplan–Meier curves according to KPS, chemotherapy, antiangiogenic therapy, and male sex are shown in Supplementary Fig. 2.

Table 3 Univariable and multivariable Cox analyses of patients with initial LMPredictors of OS in patients with recurrent LM

Among patients with recurrent LM, univariable analysis revealed that longer interval between initial glioma and LM diagnoses (HR = 0.93, P < 0.001), gross total resection (HR = 0.60, P = 0.025) and chemotherapy (HR = 0.15, P = 0.001) were predictors of longer OS, while male sex (HR = 1.83, P = 0.009), histological grade 4 (P = 0.001), IDH wildtype (HR = 4.32, P = 0.001), and MGMT promoter unmethylation (HR = 2.00, P = 0.005) were predictors of shorter OS. Multivariable analysis showed that longer interval between initial glioma and LM diagnoses (HR = 0.88, P < 0.001) and chemotherapy (HR = 0.04, P < 0.001) were predictors of longer OS, while male sex (HR = 1.86, P = 0.014) was a predictor of shorter OS (Supplementary Table 1). Adjusted and unadjusted Kaplan–Meier curves according to interval between initial glioma and LM diagnoses, chemotherapy, and male sex are shown in Supplementary Fig. 3.

Predictors of OS in IDH-wildtype glioblastoma patients with LM

When identical analyses were performed in IDH-wildtype glioblastoma patients (entire patients with LM [n = 179], initial LM [n = 82], and recurrent LM [n = 97]), a similar trend of results to that of entire glioma patients was observed. Supplementary Tables 2–4 show the univariable and multivariable results.

留言 (0)

沒有登入
gif