Ommaya Reservoir Use in Pediatric ALL and NHL: A Review at St. Jude Children’s Research Hospital

Abstract

Purpose

The intraventricular route of chemotherapy administration, via an Ommaya Reservoir (OmR) improves drug distribution in the central nervous system (CNS) compared to the more commonly used intrathecal administration. We retrospectively reviewed our experience with intraventricular chemotherapy, focused on methotrexate, in patients with Acute Lymphoblastic Leukemia (ALL) and Non-Hodgkin Lymphoma (NHL).

Methods

Twenty-four patients (aged 7 days – 22.2 years) with 26 OmR placements were identified for a total of 25,009 OmR days between 1990 and 2019. Methotrexate cerebrospinal fluid (CSF) concentrations (n=124) were analyzed from 59 courses of OmR therapy in 15 patients. Twenty-one courses involved methotrexate dosing on day 0 only, whereas 38 courses involved booster dosing on days 1, 2, or both. We simulated the time CSF methotrexate concentrations remained > 1 μM for 3 days given various dosing regimens.

Results

CSF methotrexate exposure was higher in those who concurrently received systemic methotrexate than via OmR alone (p < 10⁻⁷). Our simulations showed that current intraventricular methotrexate boosting strategy for patients ≥ 3 years of age maintained CSF methotrexate concentrations ≥ 1 μM for 72 hours 40% of the time. Alternatively, other boosting strategies were predicted to achieve CSF methotrexate concentrations ≥ 1 μM for 72 hours between 46 to 72% of the time.

Conclusions

OmR were able to be safely placed and administer intraventricular methotrexate with and without boost doses in patients from 7 days to 22 years old. Boosting strategies are predicted to increase CSF methotrexate concentrations ≥ 1 μM for 72 hours.

INTRODUCTION

The central nervous system (CNS) is a sanctuary site of leukemia and lymphoma that can be challenging to treat because chemotherapy does not cross the blood brain barrier efficiently. Intrathecal chemotherapy administration via lumbar puncture (LP) is the most common method to increase drug concentrations in the cerebral spinal fluid (CSF). The intraventricular route, via an Ommaya reservoir (OmR), presents an advantage of improved drug distribution [1–3]; however, its placement remains uncommon due to concern for complications, such as infection, catheter misplacement, hemorrhage, and leukoencephalopathy [4–8].

One of the first chemotherapy medications infused via OmR was methotrexate for patients with resistant meningeal leukemia [9]. Despite the use of OmR for intraventricular chemotherapy administration since the 1960s, dosing strategies have not been well established. Pediatric dosing of intraventricular methotrexate is reported in the range of 6.25 mg/m² to 15 mg/m² [1, 10–17], as well as flat dosing strategies ranging from 1–6 mg, including some with pharmacokinetic derived strategies. Strother et al. used a modified concentration x time method to expose CNS blasts to a methotrexate concentration of 1 μM for 72 hours by redosing methotrexate based on CSF concentrations [18]. Contemporary practice for “triple therapy” (methotrexate, hydrocortisone, cytarabine) via intraventricular route is to reduce the intrathecal doses of methotrexate, hydrocortisone, and cytarabine by 50%. [19]

Data compiled by Bleyer has informed age-based intrathecal dosing to account for the fact that CSF volume does not correlate with body surface area and reaches adult volumes by age 3. [20] Reports of OmR administration of methotrexate in patients less than 3 years of age are limited and none reported in patients less than 12 months [10, 11, 13, 17, 18, 21]. We report our experience with administration of chemotherapy via OmR to patients with acute lymphoblastic leukemia (ALL) and non-Hodgkin lymphoma (NHL) at St. Jude Children’s Research Hospital (SJCRH) with an emphasis on methotrexate. The objectives of the study were to identify reasons for OmR placement, describe our methotrexate dosing approach and pharmacokinetic data, and describe OmR-related complications.

MATERIALS AND METHODS

A retrospective review was performed in patients receiving treatment at SJCRH from January 1, 1990 to September 1, 2019. Patients were included if they had a diagnosis of ALL or NHL and an OmR placed for CNS-directed chemotherapy. All research was approved by the institutional review board at St. Jude Children’s Research Hospital.

Patients were identified using medical and pharmacy records for intraventricular chemotherapy administrations, CSF methotrexate concentrations, plasma methotrexate concentrations, and International Classification of Diseases (ICD) codes. Data collected from the health record included: dates of OmR placement and removal, reason for placement, type of chemotherapy, date, and time administered via OmR, CSF and plasma methotrexate concentrations, date, and time of systemic methotrexate administration (if applicable), complications relating to OmR, and reason for OmR removal. Complications were identified based on documentation in the health record. Patients were determined to have an infection related to OmR if they had a positive CSF culture and received systemic antimicrobials. Leukoencephalopathy was determined by reviewing magnetic resonance imaging (MRI) reports if patient had one performed in addition to symptoms documented in clinical notes. Imaging reports from MRI or computerized tomography (CT) were reviewed for other evidence of complications including but not limited to changes in placement of OmR, hemorrhage, or white matter changes. Total OmR days included the time an OmR was in place until removal, death, or end of study period. Procedures for obtaining CSF and administering medication via OmR are included in the supplement.

Sample Collection and Assay

Patients ≥ 3 years of age may have received triple intraventricular methotrexate, hydrocortisone, cytarabine (MHA) followed by boost doses of methotrexate using a SJCRH-derived algorithm based on CSF methotrexate concentrations obtained 24, 48, and 72 hours after intraventricular administration of methotrexate 6 mg (TABLE 1). Algorithm was designed to maintain CSF methotrexate concentration > 1 μM for at least 72 hours. The OmR drugs given over a 7-day period represented a course of OmR therapy. The decision to use this algorithm was driven by the treatment protocol or by the clinical team. Patients <3 years of age were not eligible to receive algorithm-based dosing and thus received individualized dosing following discussion with the clinical team. Plasma and CSF methotrexate levels were analyzed using FDA-approved commercially available immunoassays. From 1990 to 2017 methotrexate was measured by fluorescence polarization immunoassay (FPIA) utilizing the TDxFlx analyzer (Abbott Laboratories, Abbott Park, IL). Samples between 2017 to 2019 were assayed using a different method and were not included in this analysis. The limit of quantification for the FPIA was 0.03 μM.

TABLE 1:

Dosing algorithms used for simulations evaluating ability to maintain CSF methotrexate > 1 μM for at least 3 days

Time (hours)	MTX CSF concentration (μM)	St. Jude algorithm MTX dose (mg)	Modified version 1 MTX dose (mg)	Modified version 2 MTX dose (mg)	Strother dosing algorithm [18] MTX dose (mg)	Fixed dosing MTX dose (mg)
	Dependent upon CSF concentrations					independent of MTX concentration
0		6	6	6	6	6
24	≥ 15	0	0	0	0	6
	≥ 8 and < 15	3	0	0	0
	≥ 4 and < 8	4	4	4	0
	< 4	6	6	6	6
48	≥ 4	0	0	0	0	6
	> 1 and < 4	2	6	4	2
	≤ 1	2	6	4	4
Over 72 hours	Range of total MTX doses possible	6 to 14 mg	6 to 18 mg	6 to 16 mg	6 to 16 mg	18 mg

Abbreviations: CSF, cerebrospinal fluid; SJCRH, St. Jude Children’s Research Hospital; MTX, methotrexate

Pharmacokinetic Methods

The population pharmacokinetic and post-hoc individual parameter estimates (conditional modes) were determined using non-linear mixed effects modeling with the Stochastic Approximation Expectation-Maximization (SAEM) method (Monolix; version 5.1.0). A one-compartment pharmacokinetic model was used to describe CSF methotrexate data and a two-compartment model was used to describe plasma methotrexate data (Supplemental Figure 1). The influx of plasma methotrexate into the CSF was described by a first-order process. Plasma pharmacokinetic parameters were fixed to their population estimates when estimating the CSF methotrexate pharmacokinetics [22]. Plasma methotrexate pharmacokinetic parameters included clearance (L/hr/m²), volume (L/m²), and the intercompartmental rate constants (kcp, kpc [1/hrs]). CSF methotrexate pharmacokinetic parameters estimated included the efflux from the CSF (ke [1/day]), volume (V [L]), and plasma to CSF influx (kcsf [1/day]). Each patient’s data were divided into multiple courses where each course was the 7-day interval around a single dose or series of up to 3 OmR methotrexate doses. The inter-individual (IIV) and inter-occasion variability (IOV; variability between courses) of the CSF parameters were assumed to be log-normally distributed and a proportional residual error model was used with assumed normal distribution of the residuals.

The relationships between the pharmacokinetic parameters and covariates were described using the following model: θ=θ_Base*exp(b*covariate). A covariate was considered significant in the univariate analysis if the addition of the covariate to the model reduced the objective function value (OFV) at least 3.84 units (p < 0.05, based on the χ²test for the difference in the −2 log-likelihood between two hierarchical models that differ by 1 degree of freedom).

Pharmacokinetic Simulations

The time CSF methotrexate > 1 μM was simulated in each patient by using multiple parameters drawn from their conditional distribution (n=10 per patient) and five different CSF methotrexate dosing schedules described in TABLE 1. In all cases systemic methotrexate dosing was not included in the simulations.

Data Sharing Statement

Individual participant data will not be shared.

RESULTS

Patient characteristics

Twenty-four patients with 26 OmR placements between December 1992 and January 2019 were identified. The median age at first OmR placement was 11.3 years (range, 7 days to 22.2 years). The patient demographics and characteristics are summarized in TABLE 2. Nineteen patients had an Ommaya reservoir placed during initial therapy, 3 (15.8%) of which experienced deaths related to malignancy. Five patients had Ommaya reservoir placed for treatment of CNS relapse (B-ALL, n=2; T-ALL, n=2; diffuse large B-cell lymphoma, n=1), 4 (80%) of which experienced death related to malignancy.

TABLE 2:

Patient characteristics

Patient characteristics	n= 24 (%)
Gender
Male	17 (70.8)
Race
Caucasian	16 (66.7)
African American	6 (25)
Other	2 (8.3)
Median Age at first placement, years (range)	11.3 (7 days-22.2 years)
Primary diagnosis
ALL	16 (66.7)
B-cell	11
T-cell	5
Burkitt lymphoma	2 (8.3)
Diffuse large B-cell lymphoma	2 (8.3)
Blastic plasmacytoid dendritic cell neoplasm	1 (4.2)
T-lymphoblastic lymphoma	1 (4.2)
CNS lymphoproliferative disorder	1 (4.2)
Received CNS radiation	6 (25)
Median time from placement to chemotherapy dose, days (range)	3 (0–32)
Median duration of placement, days (range)	467 (34–6557)
Complications (patients may have more than one)
Nausea/Vomiting associated with taps	7 (29.2%)
Headache	6 (25%)
Pain and/or tenderness at OmR site	5 (20.8%)
Edema around OmR on imaging	5 (20.8%)
Anxiety with OmR tap	3 (12.5%)
Infection	3 (12.5%)
Coagulase negative Staphylococcus	1
Pseudomonas aeruginosa colonization leading to sepsis	1
Corynebacterium group G	1
CSF leak	2 (8.3%)
Hematoma	2 (8.3%)
Methotrexate encephalopathy	2 (8.3%)
Eyes burning after methotrexate instilled	1 (4.2%)
Arachnoiditis	1 (4.2%)
Blurred vision	1 (4.2%)

Abbreviations: ALL acute lymphoblastic leukemia, CNS central nervous system, OmR Ommaya reservoir, CSF cerebrospinal fluid

There were 398 OmR medication administrations (including methotrexate boost doses), with most administrations being methotrexate either alone or in combination with cytarabine, hydrocortisone, or both (n=344). This represented 335 courses of OmR therapy. Initial intraventricular chemotherapy was administered a median of 3 days after OmR placement (range, 0–32 days). The median number of OmR drug administrations per patient was 14 (range, 3–43). Other medications administered via OmR were rituximab (n=7), topotecan (n=13), and liposomal cytarabine (n=14). Doses of all medications administered via OmR are included in the supplement (Supplemental Table 1). OmR doses of cytarabine ranged from 3 to 50 mg either given alone or in combination with methotrexate, hydrocortisone, or both. Hydrocortisone doses up to 24 mg were administered.

All 24 patients received methotrexate via OmR with day 0 doses ranging from 1 to 6 mg. All patients ≥3 years of age (n=19) received an initial (day 0) methotrexate dose of 6 mg; two patients at two years of age received an initial dose of 5 mg. Patients ≤ 1 year of age (n=3) received various methotrexate doses: 1 mg (n=1, age range 67–138 days old), 1.5 mg (n=1, age range 55–203 days), and 3 mg (n=1, age range 12–140 days).

Eleven patients received 49 courses where CSF methotrexate concentrations were assessed to determine if day 1 and day 2 methotrexate doses were indicated based on the SJCRH algorithm (TABLE 1). Four patients < 3 years of age received 10 courses with CSF methotrexate concentrations. Eighteen patients (75%) received intraventricular methotrexate in combination with 56 courses of high-dose systemic methotrexate. Intraventricular methotrexate was either given the same day as high-dose methotrexate (50 courses), within 24 hours after high-dose methotrexate started (5 courses) or within 24 hours before high-dose methotrexate started (1 course).

Placement, removal, complications

Patients had an OmR in place for a median of 467 days (range: 34–6,557 days). The most common reason cited for OmR placement was spinal stenosis or lipomatosis (n=10), including one patient with achondroplasia. Patients also had an OmR placed for other spinal abnormalities that resulted in obstruction or concern for obstruction of CSF flow (n=3); or they had multiple failed LP attempts (n=3). Three patients had an OmR placed upon evidence of CNS relapse. Additional reasons for placement are listed in Supplemental Table 2.

A total of 25,009 OmR days were evaluated for complications. The most common complications associated with OmR taps and chemotherapy administration were nausea/vomiting (29.2%) and headache (25%). Infections with positive CSF culture occurred in 3 patients with the time from OmR placement to a positive CSF culture being 39, 43, and 392 days. Time from last OmR access to infection was 3 days in two patients and 6 days in one patient. Other complications are described in TABLE 2. Leukoencephalopathy/white matter changes were seen on imaging in 12 patients; however, we were unable to attribute these changes directly to OmR. One patient had a seizure while OmR was in place; the cause of seizure was related to temporal injury and not OmR or chemotherapy. Temporal lobe hemorrhage from a venous stroke present prior to OmR placement.

Two patients (8.2%) with ALL had likely methotrexate encephalopathy. One patient presented with slurred speech and hand weakness; the onset of symptoms was 10 days from high dose methotrexate 5.2 g/m² (methotrexate via OmR last given 17 days prior to symptoms). The other patient had altered mental status that occurred one day after getting triple therapy with MHA via OmR and during 24-hour high dose methotrexate infusion (5 g/m²). Methotrexate infusion was stopped early 15 hours into the infusion. This patient also received three consecutive days of methotrexate via OmR 5–7 days prior to symptom onset. Both patients continued to receive subsequent methotrexate doses via OmR without further episodes.

Fifteen patients had documentation of OmR removal. Among the reasons of OmR removal, completion of therapy was most common (n=10). Two patients had the OmR removed due to infection; Pseudomonas aeruginosa and coagulase negative staphylococcus, respectively. Other reasons for removal include hematoma (n=1), change around site seen on MRI (n=1) and CSF leak (n=1).

CSF methotrexate pharmacokinetics

Fifteen patients, all treated prior to 2017, had a total of 124 CSF methotrexate concentrations measured, for a total of 134 doses of methotrexate administration via OmR, representing 59 courses. Intervals between courses ranged from 1 to 8 weeks; one patient with pharmacokinetic studies that were performed almost 3 years apart was treated as two independent “patients” in the pharmacokinetic analysis. The measured CSF methotrexate concentrations along with the population pharmacokinetic parameter estimates and post-hoc individual pharmacokinetic parameters are summarized in Supplemental Tables 3-5. Additionally, goodness-of-fit plots along with plots showing the post-hoc individual estimated clearance subdivided by BSA and age are shown in Supplemental Figures 1-4.

Nine patients received at least one systemic dose of methotrexate around the time CSF methotrexate concentration was evaluated. A total of 19 courses included systemic and OmR doses; 40 courses included OmR dosing only. Of the 59 courses, 21 received doses on day 0 only, 25 received doses on day 0 and at 24 and 48 hours, 9 received doses on day 0 and at 24 hours, and 4 received doses on day 0 and at 48 hours. For those with systemic plus OmR methotrexate dosing, the median (range) OmR day 0 dose was 6 mg (1–6 mg) and systemic dose 5035 mg/m² (2800–8327 mg/m²); none of these patients received booster OmR doses. There was no difference in CSF methotrexate clearance for those who received concurrent systemic methotrexate and methotrexate via OmR vs those who received OmR doses only (p=0.47). However, CSF methotrexate exposure was higher in those who received concurrent systemic and OmR methotrexate vs OmR alone (FIGURE 1; p < 10⁻⁷, Supplemental Table 3).

FIGURE 1:

Cerebral spinal fluid (CSF) AUC (day*μM) subdivided by those with and without systemic dosing (normalized to the CSF dose by course---where a course was defined as a 7-day window after a CSF methotrexate dose)

Abbreviations: MTX, methotrexate; AUC, area under the curve; OCC, occasion

Pharmacokinetic Simulation Studies

Based on observations in the 54 courses given to 12 patients ≥ 3 years old, we estimated the ability of 4 different booster-based dosing algorithms, in addition to a fixed dose schedule of 6 mg on day 0, to maintain CSF methotrexate > 1 μM for at least 3 days (TABLE 1). The main difference between regimens in CSF methotrexate exposure occurred on day 3 when doses ranged from 2 to 6 mg depending on the algorithm and CSF methotrexate concentration (TABLE 3). The percentage of courses anticipated to require 1, 2, or 3 doses are summarized in FIGURE 2A, and the total CSF methotrexate dose/course (between 6 and 18 mg) estimated to be required to maintain CSF methotrexate concentrations in the desired range (TABLE 1) is summarized in FIGURE 2B. In all 4 dosing algorithms, between 59% to 65% of the simulated studies required 3 doses of CSF methotrexate with a total dose between 14 and 18 mg. Of these 4 algorithms, the modified SJCRH algorithm version 1 gave the highest percentage of studies with CSF methotrexate > 1 μM on day 3 (72%) and the longest time for CSF methotrexate to be > 1 μM (median 3.2, [2.3, 4.6] days [5^th-95^th percentile]) (FIGURE 3). PK was available in two individuals < 1 year old. One of these individual’s clearance was similar to the older individuals while the second individual’s clearance was substantially lower (~15x).

TABLE 3:

Summary results of simulations in patients ≥ 3 years of age.

Dosing algorithm*	% studies > 1 μM at:			Time CSF MTX > 1 μM (days). Median (5th, 95th percentile)	72 hour MTX concentration (μM) Median (5th, 95th percentile)
	24 hours	48 hours	72 hours
SJCRH	81%	82%	40%	2.9 (2.1, 3.8)	0.75 (0.11, 6.1)
Modified version 1	81%	79%	72%	3.2 (2.3, 4.6)	1.7 (0.25, 17.5)
Modified version 2	81%	79%	60%	3.1 (2.3, 4.3)	1.2 (0.21, 11.8)
Strother	81%	70%	46%	3.0 (2.3, 4.0)	0.98 (0.22, 6.6)
Fixed dosing: 6 mg X 3	81%	82%	81%	3.4 (2.3, 5.0)	2.9 (0.33, 28.3)

Abbreviations: MTX=methotrexate; CSF=cerebrospinal fluid; SJCRH=St. Jude Children’s Research Hospital

^*See Table 1 for definition of dosing schedules.

FIGURE 2:

A) Percentages of simulated courses that would require 3 doses (on days 0, 1, and 2); 2 doses (on days 0 and 1 or on days 0 and 2); and 1 dose only on day 0. B) Percentage of simulated courses that would require a total course CSF methotrexate dose of 6 mg, 8–10 mg, 11–13 mg, 14–16 mg, or 18 mg.

FIGURE 3:

Simulation of cerebrospinal fluid (CSF) methotrexate versus time for the modified St. Jude Children’s Research Hospital dosing algorithm version 1 (Day 0: 6 mg; Day 1: 6 mg if CSF methotrexate < 4 μM, 4 mg if CSF methotrexate is between 4 and 8 μM, 0 mg otherwise; Day 2: 6 mg if CSF methotrexate < 4 μM, 0 mg otherwise) in patients ≥ 3 years of age (54 studies from 12 patients). The simulations are based on pharmacokinetic parameters drawn from each patient’s conditional distribution (n=10 per study). The black curve is the median and the blue and grey shaded regions are the quartile range and the 5^th-95^th percentile range of the simulated studies in each of the four dosing groups.

Abbreviations: MTX, methotrexate

DISCUSSION

Our retrospective data in 24 patients from 7 days to 22 years old shows that we were able to both safely place an OmR and administer intraventricular methotrexate. By systematically reviewing our data, our current intraventricular methotrexate boosting strategy for patients ≥ 3 years of age maintained CSF concentrations of ≥ 1 μM for 72 hours in only 40% of studies. By simulating results, we predict that more aggressive dosing strategies could achieve CSF methotrexate concentrations ≥ 1 μM for 72 hours in >70% of patients. While fixed dosing had a higher percentage of individuals with concentrations ≥ 1 μM for 72 hours many of these levels were significantly >1 μM (TABLE 3) and there is some evidence that higher concentrations of methotrexate could have a higher risk of neurotoxicity [23]. However prospective studies are needed to confirm the safety of this approach.

Although our sampling was limited, we were able to simulate the time CSF methotrexate concentrations remained > 1 μM for 3 days using various dosing regimens in patients ≥ 3 years old. The goal to keep CSF methotrexate concentration > 1 μM is based on findings from an in vitro study of Hryniuk and Bertino; higher rates of deoxyuridine incorporated in DNA occurred with methotrexate concentrations less than 1 μM [24]. Keefe et al. also demonstrated higher cytotoxicity of murine leukemia cells when exposed to methotrexate at a concentration of 1 μM for 36 or 42 hours compared to a methotrexate concentration of 100 μM over 3 or 6 hours [25]. The need for extended exposure of methotrexate is based on findings that the leukemia cell cycle was 2–3 times longer than normal cells and that cell generation times up to 70 hours were observed [26]. Since these were in vitro studies, more data is needed to determine if maintaining CSF concentration >1 μM for 3 days correlates with reduced rates of CNS relapse in humans.

Bleyer et al. were the first to evaluate a concentration x time regimen, administering 1 mg intraventricular methotrexate every 12 hours for 6 doses [10]. This approach increased the number of times the OmR was accessed, but they found lower rates of methotrexate neurotoxicity compared to patients that received a single dose of 12 mg/m² (maximum dose 15 mg) [10]. Strother et al. took a different dosing approach but used the same goal of keeping CSF methotrexate > 1 μM for 72 hours; initial methotrexate doses varied from 4 to 6 mg and patients were given additional doses if the 24 or 48-hour methotrexate concentration was < 1 μM [18].

Based on simulation results for patients ≥ 3 years old in TABLE 3, the SJCRH methotrexate algorithm and Strother algorithm had similar results, with 40% versus 46% of studies > 1 μM at 72 hours. Our modified version 1 algorithm had better results compared to the above two algorithms, with 72% of studies achieving concentrations > 1 μM at 72 hours. With this algorithm patients could receive methotrexate doses totaling 16–18 mg over 3 days. We acknowledge that the safety of the simulated dosing algorithms other than the SJCRH algorithm in TABLE 1 were not evaluated as part of this study.

Nausea/vomiting was the most frequent complication seen in our study. Due to this common occurrence, it is recommended to administer oral or intravenous anti-emetics prior to intra-Ommaya chemotherapy administration. Three patients had anxiety when the OmR was accessed, showing the need for early psychological interventions prior to taps to ensure the patient is comfortable.

Using only systemic and intrathecal methotrexate, the incidence of methotrexate-induced neurotoxicity in ALL patients was found to be 3.8% on the SJCRH Total XV study [27]. In this review we found a higher incidence of methotrexate encephalopathy of 8.2%, but instances could not be traced specifically to OmR administration. Eighteen patients in our study received concurrent intraventricular and systemic high dose methotrexate, two of whom developed likely methotrexate-induced encephalopathy. Like the experience in the Total XV study, these patients were able to receive additional methotrexate doses without re-emergence of symptoms [27]. Given that neurotoxicity has been associated with higher CSF methotrexate concentrations, lower intraventricular methotrexate doses (< 6 mg) or increased use of leucovorin rescue may be needed to decrease methotrexate neurotoxicity when patients are expected to achieve high CSF concentrations, especially when systemic and OmR methotrexate are concurrently administered [23]. NHL patients in our study received higher systemic methotrexate doses than those with ALL, we only saw ALL patients presenting with concern for encephalopathy despite the lower systemic dose.

There are important limitations that should be considered when interpreting our findings. First, our study was retrospective, which resulted in an inability to collect leucovorin rescue data for all patients, an inability to confirm technical consistency when administering intra-Ommaya chemotherapy and obtaining methotrexate CSF concentrations from Ommaya reservoir, and the possibility of toxicity underestimation. In addition, because only two individuals were < 3 years of age, we were unable to make any conclusions regarding dosing in these young patients. Since all methotrexate concentrations were obtained prior to 2017, they were measured by FPIA utilizing the TDxFlx analyzer (Abbott Laboratories, Abbott Park, IL), and may differ from methotrexate concentrations measured by other immunoassays. Lastly, delayed CSF methotrexate clearance has been reported in patients with CNS disease [21]. While we were unable to evaluate the impact of CNS disease on methotrexate CSF clearance in our patients, three had an OmR placed for relapse and were included in the pharmacokinetic evaluation of methotrexate CSF concentrations. Among these 3, none had unusually high CSF concentrations, providing reassurance that clearance was not affected. We were not able to correlate CSF methotrexate concentrations to neurocognitive or survival outcomes.

Ommaya reservoirs continue to be utilized for patients with CNS relapse or for who are unable to receive IT chemotherapy due to anatomical abnormalities and medical reasons. Methotrexate and other antineoplastics can safely be administered via Ommaya reservoir. Our current intraventricular methotrexate boosting strategy for patients ≥ 3 years of age-maintained CSF methotrexate concentrations ≥ 1 μM for 72 hours in only 40% of studies and predicted that other dosing strategies could achieve CSF methotrexate concentrations ≥ 1 μM for 72 hours in >70% of patients.