Managing hydrocephalus in patients with leptomeningeal disease: A multicenter retrospective analysis

Abstract

Leptomeningeal disease (LMD) represents a terminal condition of tumor cell seeding that can cause symptomatic hydrocephalus. With improved survival rates under systemic therapy, the role of cerebrospinal fluid (CSF) drainage through ventriculo-peritoneal shunt (VPS) or Rickham reservoir (RR) placement in LMD patients is gaining more relevance. This study aimed to compare outcomes of both modalities in a multicentric contemporary cohort. A retrospective analysis of medical charts in patients receiving VPS for LMD and malresorptive hydrocephalus in two neurosurgical centers between 2006 and 2021 yielded 64 patients. The most common underlying oncological conditions were breast (n = 32, 49%) and non-small cell lung cancer (NSCLC, n = 16, 25%). The median time between primary and LMD diagnosis was 23.3 months (11.2 to 43.4 months). Symptoms of intracranial hypertension were relieved in 79% of cases (n = 50) after shunting, with 42 (66%) and 32 patients (50%) receiving systemic and intrathecal therapy, respectively. A further multicenter analysis comparing patients receiving VPS with patients receiving RR (with regular tapping) included 155 patients (VPS: n = 80, 52%; RR: n = 75, 48%). Compared to VPS, RRs were associated with a lower surgical revision rate (8% vs. 24%, p = 0.009). There was no difference in median overall survival in VPS patients (118 days) compared to RR patients (80 days, p = 0.180). Given this data showing a short and comparable survival of patients under both modalities with a lower RR complication rate, a rationale for an initial Rickham implantation in LMD patients with hydrocephalus, with later VPS conversion for long-term surviving patients, could be contemplated.

Keywords: CSF diversion; Rickham reservoir; intrathecal therapy; leptomeningeal disease; ventriculoperitoneal shunt.

FIGURE 1

Study flow chart. The Bicentric analysis was carried out on data from Centers A and B that implant ventriculoperitoneal shunts (VPS) in patients with symptomatic intracranial hypertension and leptomeningeal disease. Center C employs both Rickham Reservoirs (RR) and VPS, and Center D implants only RR.

FIGURE 2

Patient characteristics of the bicentric cohort prior to ventriculoperitoneal shunt (VPS) implantation. (A) Onco‐plot of selected variables of the 64 patients with leptomeningeal disease (LMD) included in the bicentric analysis. Rows represent variables, and each column represents one patient. M + = Metastasis of primary tumor to another organ system at initial diagnosis. CUP, Cancer of Unknown Primary; GI‐Tumors, Gastrointestinal tumors; NSCLC = Non‐small Cell Lung Cancer. All therapies included pertain to the primary tumor and not LMD. (B) Elapsed time (days) from primary tumor to LMD diagnosis in n = 64 patients. (C) Stacked pie charts showing the number of patients with 1, 2, 3 or more organs with tumor metastases at LMD diagnosis (outer circle), the number of patients with solid brain metastases (=mets, middle circle), and the overall percentage of patients that had received a craniotomy for resection of brain metastases (inner circle). (D) Number of solid brain metastases in the bicentric cohort of 64 patients receiving VPS. (E) Clinical and radiological features of the patients in the bicentric cohort before shunting. CSF, cerebrospinal fluid; MRI, magnetic resonance imaging. (F) Violin plot of the cell count (cells/μl) in n = 64 patients of the bicentric VPS cohort. G: Protein concentration (g/L) in n = 64 patients of the bicentric VPS cohort.

FIGURE 3

Surgical and survival outcomes of patients receiving VPS for LMD (Leptomeningeal Disease). (A) Time (days) lapsing between LMD diagnosis and VP‐shunting in the n = 64 patients of the bicentric cohort. (B) Violin plots of pre‐ and post‐operative Karnofsky Performance Score = KPS of VPS patients. Note that patients with in‐hospital death (n = 9) were excluded (ns = non‐significant, Wilcoxon matched‐pairs signed rank test). (C) Overview of the postoperative course of VPS patients, including rate of symptom relief, post‐operative discharge (also to other departments = dept.) and further systemic and intrathecal therapy. (D) Swimmer's plot of the course of disease (across) in n = 64 patients (down, chronologically ordered from 2006 to 2021) in the bicentric cohort highlighting the time (in months) from primary tumor diagnosis to LMD (green), from LMD to shunt (red) and from shunting to death (or censorship, blue). (E) Survival of LMD patients after VP‐shunting. (F) Survival (days) in patients after VPS (n = 64) with stratification based on Karnofsky Performance Score (KPS < 70 vs. KPS > 60 (p < 0.001, Log‐rank [Mantel‐Cox] test). (G) Survival after shunting in patients receiving postoperative intrathecal therapy (i.t.) compared to patients without postoperative intrathecal therapy (p = 0.0518, Log‐rank [Mantel‐Cox] test). (H) Survival of patients with preoperative CSF protein concentrations higher than the median in this cohort (0.65 g/L) compared to patients with lower protein concentrations (p = 0.4998, Log‐rank [Mantel‐Cox] test).

FIGURE 4

Pooled multicenter retrospective analysis of Ventriculoperitoneal shunt (VPS) and Rickham reservoir (RR) patients. (A) Overview of the number of patients with the respective procedure performed included from each center. (B) Plot of patient age at LMD diagnosis showing no differences between the VPS (left, red) and RR cohorts (right, violet, two‐tailed Student's t test). (C) Bar chart comparing proportions of primary tumor entities in VPS and RR patients (chi‐square test). CUP, Cancer of Unknown Primary; GI‐Tumors, Gastrointestinal tumors; NSCLC, Non‐small Cell Lung Cancer. (D) Comparison of complications requiring surgical revision after VPS vs. after RR placement (Fisher's exact test). SDH = Subdural Hematoma. (E) Survival of LMD patients after VP‐shunting (n = 80) vs. RR placement (n = 75), p = 0.180 (Log‐rank (Mantel‐Cox) test). (F) Survival (days) in patients with breast cancer as a primary tumor after VPS (n = 39) vs. RR (n = 41), p = 0.431 (Log‐rank (Mantel‐Cox) test). (G) Survival (days) stratified according to primary neoplasm, p = 0.341 (Log‐rank [Mantel‐Cox] test).