Characterization of focal leptomeningeal extension in patients with brain metastases: A novel entity

Abstract

Background

Classical leptomeningeal disease (cLMD) impacts the entire craniospinal axis, but isolated parenchymal metastases sometimes display focal leptomeningeal extension (fLME) without diffuse LMD elsewhere. We assessed whether patients with fLME can be managed with stereotactic radiation (SRS/SRT) as opposed to whole brain/craniospinal radiation without excess development of marginal recurrences or subsequent cLMD.

Methods

We identified 796 patients with 2,354 newly-diagnosed brain metastases (BrM) without cLMD at diagnosis managed at a tertiary center between 2007 and 2022. Each metastasis was assessed for fLME, defined as isolated leptomeningeal extension of an intact BrM without cytologic or radiographic evidence of cLMD. Multivariable Fine and Gray’s models were constructed for the primary outcomes of local recurrence and cLMD development.

Results

Among 796 patients, 138 (17.3%) displayed evidence of fLME, corresponding to 185 of 2,354 (7.9%) BrM. Patients with versus without fLME did not display excess local recurrences (1-year rate: 4.5% vs. 8.2%, respectively, P = .14; multivariable HR 0.56 [95% CI 0.30-1.07], P = .08), including in lesions managed with SRS/SRT (1-year rate: 4.4% vs. 4.9%, respectively; P = .63; multivariable HR 1.21 [95% CI 0.58-2.50], P = .61). The presence of fLME was not a significant predictor of subsequent cLMD (1-year rate: 5.2% vs. 5.2%, P = .99; multivariable HR 0.87 [95% CI 0.44-1.71], P = .68), including following treatment with SRS/SRT (1-year rate: 6.2% vs. 4.1%, P = .83; multivariable HR 0.92 [95% CI 0.36-2.39], P = .87).

Conclusions

We describe a novel entity, fLME, which displays patterns of intracranial failure similar to parenchymal BrM. Stereotactic approaches may be viable in this population.

Key Points.

• Focal LME represents a distinct oncologic entity that more closely resembles parenchymal brain metastases than classical LMD

• Focal LME may be managed with stereotactic approaches in lieu of WBRT or CSI without excess risk of local recurrence or cLMD

Importance of the Study

Leptomeningeal disease (LMD) has historically been considered a diffuse phenomenon impacting the entire craniospinal axis and has therefore been managed with palliative, large-field radiation or systemic therapies and often precludes affected patients from clinical trial enrollment. In this study, we describe a novel entity of “focal” leptomeningeal disease (fLME), characterized by isolated parenchymal metastases with focal extension into the surrounding leptomeninges without diffuse LMD elsewhere. We demonstrate that patients with fLME do not display increased risk of local recurrence or development of classical leptomeningeal disease post-radiation, even when managed stereotactically, when compared to those with parenchymal brain metastases, supporting deferral of more neurotoxic treatments, such as whole brain radiation in such patients. In this study, we characterize focal LMD as a distinct oncologic entity that resembles parenchymal brain metastases and that may be managed with stereotactic approaches without excess risk of local recurrence or classical LMD development.

Leptomeningeal disease (LMD) occurs when tumor cells disseminate into the cerebrospinal fluid (CSF) and infiltrate the leptomeninges of the brain and spinal cord.¹ Approximately 2-12% of patients with brain metastases (BrM) present with LMD at initial diagnosis of intracranial disease, with another up to 37% developing LMD subsequently, with the wide range of incidence likely related to inconsistent definitions of LMD.^2-5 LMD is associated with neurologic symptomatology, functional decline, and a very poor prognosis.^6-8 Patients with LMD have historically been excluded from clinical trials, with management primarily entailing palliative radiation to the whole brain (WBRT), large spinal segments, or entire craniospinal axis, as well as consideration of systemic and/or intrathecal chemotherapy with CSF diversion if indicated.^1,8-11

Although LMD has classically been considered a diffuse phenomenon impacting the entire craniospinal axis, isolated parenchymal metastases sometimes display focal extension into the surrounding leptomeninges without diffuse LMD elsewhere. However, a dearth of scientific literature depicting such “focal” leptomeningeal disease (fLME) exists. As a result, whether large field radiation, such as WBRT, is needed for such patients remains unknown.

Here, we describe the incidence and outcomes of patients with fLME. We hypothesized that such patients might be successfully managed with stereotactic radiation without excess development of adverse outcomes such as marginal misses/local recurrence or classical/global LMD (cLMD) relative to intact BrM devoid of fLME. Given the deleterious outcomes associated with WBRT,^12,13 avoidance of this modality when viable could benefit patients.

Methods

We identified 999 patients with newly diagnosed intracranial metastases secondary to a solid primary managed within Brigham and Women’s Hospital/Dana-Farber Cancer Institute (Boston, MA) between 2007 and 2022. All magnetic resonance imaging (MRI) of the brain were centrally reviewed by an attending radiation oncologist specializing in BrM who assessed for the presence versus absence of cLMD at or after initial diagnosis of intracranial disease; cLMD was defined as (1) diffuse MRI-based oncologic disease in the leptomeninges, which generally manifested as subarachnoid enhancement in the cranial nerves, folia of the cerebellum, supratentorial sulci, ependyma, spinal cord, or cauda and/or (2) positive CSF cytology. Unexplained ventricular enlargement over time or symptoms suggestive of hydrocephalus were also accounted for when assessing for cLMD. Patients with cLMD at diagnosis of intracranial disease (N = 203) were excluded from all analyses except for a simple analysis to characterize overall survival in this population. Among the remaining 796 patients, who harbored a total of 2,354 BrM, each metastasis was assessed for fLME, which was defined as focal leptomeningeal extension of an intact brain metastasis without evidence of cLMD or a single, isolated focal leptomeningeal deposit without features suggesting widespread cLMD (Supplementary Figure S1). All MRIs were reviewed by one of two radiation oncologists with expertise in the management of CNS tumors; for any equivocal cases, both MDs reviewed the case to ensure agreement on the designation of fLME vs. not.

Stereotactic radiosurgery and stereotactic radiotherapy (SRS/SRT) were generally prescribed to 18-20 Gy in 1 fraction and 30 Gy in 5 fractions, respectively; the most common WBRT regimen was 30 Gy in 10 fractions and was typically utilized for patients with >10-15 brain metastases, small cell histology, or high brain metastasis velocity.

All patients must have had high-quality MR imaging to be assessed for fLME, including T1-weighted post-contrast imaging with thin sections and ideally 1 mm isotropic voxels, such as inversion recovery gradient echo pulse sequences like MPRAGE, SPGR, or BRAVO, or post-gadolinium 3D T1-weighted fast spin echo pulse sequences like SPACE, CUBE, or VISTA. Local recurrence following initial brain-directed therapy was ascribed based on RANO-BM guidelines¹⁴ and defined as definitive radiographic enlargement of ≥20% in maximal cross-sectional diameter on MRI, in the absence of radiation necrosis, or use of salvage brain-directed treatment for suspected recurrent disease. Delineation of recurrence versus necrosis was made by a single investigator (AAA) with expertise in brain metastases using pathologic, clinical, and radiographic factors. Pathology derived from neurosurgical resection served as the gold standard to ascribe radiation necrosis versus tumor recurrence. For patients who did not undergo neurosurgical resection, a clinical diagnosis of radiation necrosis was ascribed using serial MRIs if a treated metastasis displayed enlargement that ceased to worsen over time or improved in the absence of changes in central nervous system-active systemic therapy or additional radiation to the site in question. Tumor recurrence was ascribed if shrinkage of a lesion in question occurred after a change to a central nervous system-active systemic therapy or after additional radiation was administered. MRI-based imaging-based factors suggesting necrosis entailed a lesion displaying an irregular, hazy enhancing border or one with a hypointense center with or without speckled internal foci of enhancement. Factors indicative of tumor progression included a smooth, solid, nodular lesion with distinct borders. When indicated, advanced imaging modalities such as dual- phase PET/CT, MRI perfusion, MRI spectroscopy, and Treatment Response Assessment Maps were pursued to assist with delineation of necrosis versus progression.

Statistical Methodology

Baseline characteristics among patients with vs. without fLME were compared using Fisher’s exact test, the unpaired t-test, and Wilcoxon rank sum test, as appropriate.¹⁵ Multivariable logistic regression was used to assess predictors of fLME versus not at presentation with intracranial metastatic disease using the following covariates: age, sex, Charlson comorbidity index, Karnofsky performance status, the presence versus absence of neurologic symptoms, the presence versus absence of progressive extracranial disease, the number of prior systemic therapy regimens for metastatic disease, the number of brain metastases identified, the size of the largest brain metastasis identified, and primary cancer type. Cumulative incidence curves, Gray’s test, and multivariable Fine and Gray’s competing risks regression were used to characterize the association between fLME at diagnosis and both local recurrence and cLMD development after initial management with death treated as a competing risk. The Fine and Gray models were adjusted for the same covariates as for the logistic regression except that the presence versus absence of fLME was added into the model. In addition, the model for local recurrence was performed on a per-metastasis level using a sandwich estimator to account for intrapatient correlations and the largest metastasis covariate was replaced width the size of individual metastasis evaluated. Overall survival was assessed with Kaplan-Meier curves and the log-rank test. The assumption of proportional hazards was verified. This study was approved by the Dana-Farber/Harvard Cancer Center Institutional Review Board; written informed consent was waived.

Results

Among 796 patients, 138 (17.3%) displayed fLME, corresponding to 185 of 2,354 BrM (7.9%). Patients with fLME versus parenchymal-only BrM differed with respect to age (mean 60 vs. 63 years, P = .02), number of BrM (median 4 vs. 2, P < .001), and primary tumor site (fLME was most common in breast cancer (31%) and least common in melanoma (9%), P < .001); no significant differences in sex, Karnofsky performance status (KPS), or Charlson co-morbidity index (CCI) were observed (Table 1). On a per-lesion level, metastases with fLME vs. not were larger (median 11 mm vs. 8 mm, P < .001) and per-metastasis management varied with regard to neurosurgical resection (9% vs. 13%), stereotactic radiation as monotherapy (SRS/SRT, 69% vs. 58%), and WBRT (23% vs. 29%), respectively, P = .02, although numeric differences were small in magnitude. A multivariable logistic regression analysis displaying predictors of fLME is presented in Table 2. A higher number of prior systemic therapy regimens for metastatic disease (OR 1.18 [95% CI 1.07-1.31], P = .001) and breast cancer, as opposed to non-small cell lung cancer, as an underlying primary (OR 2.02 [95% CI 1.13-3.59], P = .02) were associated with a greater adjusted odds of presenting with fLME.

Table 1.

Baseline characteristics in patients with and without focal leptomeningeal extension at presentation with intracranial disease

	No Focal LMD (N = 658)	Focal LMD (N = 138)	P
Age, years, number (SD)	63 (12)	60 (13)	.02
Sex, N (%)			.25
Male	259 (39)	47 (34)
Female	399 (61)	91 (66)
Charlson Comorbidity Index, N (%)			.68
0	570 (87)	122 (88)
≥1	88 (13)	16 (12)
Karnofsky Performance Status, N (%)			.09
<90	283 (43)	48 (35)
90-100	375 (57)	90 (65)
Neurologic Symptoms, N (%)			.92
No	240 (36)	51 (37)
Yes	418 (64)	87 (63)
Progressive Extracranial Disease, N (%)			1.00
No	323 (49)	68 (49)
Yes	335 (51)	70 (51)
Number of Prior Systemic Therapy Regimens, median (IQR)	0 (0-1)	1 (0-2)	<.001
Number of Brain Metastases, median (IQR)	2 (1-5)	4 (2-9)	<.001
Largest Metastasis, mm, median (IQR)	19 (9-31)	17 (12-25)	0.93
Primary Tumor Site, N (%)			<.001
NSCLC	269 (41)	43 (31)
Melanoma	62 (9)	13 (9)
Breast	97 (15)	43 (31)
Other	230 (35)	39 (28)

Abbreviations: IQR, Inter-Quartile Range; LMD, Leptomeningeal Disease; mm, millimeters; N, Number; NSCLC, Non-Small Cell Lung Cancer; SD, Standard Dweviation.

Table 2.

Multivariable logistic regression depicting predictors of focal leptomeningeal extension at diagnosis of intracranial involvement

	OR (95% CI)	P
Age (per year increase)	1.00 (0.98-1.01)	.60
Sex
Male	REF	-
Female	0.84 (0.54-1.32)	.45
Charlson Comorbidity Index
0	REF	-
≥1	0.84 (0.52-1.35)	.47
Karnofsky Performance Status
<90	REF	-
90-100	1.30 (0.86-1.97)	.22
Neurologic Symptoms		.71
No	REF	-
Yes	1.09 (0.69-1.73)
Progressive Extracranial Disease		.47
No	REF	-
Yes	0.87 (0.59-1.28)
Number of Prior Systemic Therapy Regimens (Per Increase of 1)	1.18 (1.07-1.31)	.001
Number of Brain Metastases (Per Increase of 1)	1.00 (0.99-1.01)	.43
Largest Metastasis (Per Increase of 1 mm)	0.99 (0.98-1.01)	.41
Primary Tumor Site
NSCLC	REF	-
Melanoma	1.12 (0.55-2.27)	.76
Breast	2.02 (1.13-3.59)	.02
Other	0.89 (0.54-1.47)	.65

Abbreviations: CI, Confidence Interval; mm, millimeters; NSCLC, Non-small cell lung cancer; OR, Odds Ratio; REF, Reference category.

Patients with versus without fLME were not more likely to present with seizures (14.5% vs. 15.0%, P = 1.00) and, post-initial management, were also not more likely to seize (1-year: 13.5% vs. 9.6%, P = .75, respectively), or develop new BrM (1-year: 57.9% vs. 51.9%, P = .57). Metastases displaying fLME versus not did not display higher risk of local recurrence after initial management (1-year: 4.5% vs. 8.2%, respectively P = .14; Figure 1A), respectively, including in the subset managed with SRS/SRT (1-year: 4.4% vs. 4.9%, respectively; P = .63; Figure 1B). On multivariable analysis on a per-metastasis the presence versus absence of fLME was not associated with local recurrence (HR 0.56 [95% CI 0.30-1.07], P = .08, Table 3), including among metastases managed with SRS/SRT (HR 1.21 [95% CI 0.58-2.50], P = .61). On a per-patient level, the presence of fLME at diagnosis did not predict subsequent development of cLMD (1-year rate: 5.2% vs. 5.2%, P = .99 among all patients and 6.2% vs. 4.1%, P = .83 among patients managed with SRS/SRT; Figure 2A and B). On multivariable analysis initial presentation with fLME was not associated with subsequent cLMD development (HR 0.87 [95% CI 0.44-1.71], P = .68, Table 4) including in the subset treated with SRS/SRT (HR 0.92 [95% CI 0.36-2.39], P = .87), as well as among a more modern cohort of patients diagnosed with brain metastases in 2021 or 2022 (HR: 0.63 [0.24-1.67], P = .36). Median survival for patients with vs. without fLME was not significantly lower (15.2 vs. 19.0 months, respectively; log-rank P = .06). Among a separate cohort of 203 patients with cLMD at diagnosis of intracranial disease, median survival was only 3.3 months.

Figure 1.

Cumulative incidence of local recurrence on a per-metastasis basis after initial intracranial management stratified by the presence vs. absence of focal leptomeningeal extension at diagnosis in the entire cohort of patients (A) and in the subset managed with stereotactic radiation therapy (B). Abbreviations: fLME, focal leptomeningeal extension.

Table 3.

Fine and Gray competing risks regression for local recurrence on a per-metastasis level

Variable	HR (95% CI)	P
Age (per year increase)	0.99 (0.97-1.01)	.33
Sex
Male	REF	-
Female	1.05 (0.62-1.75)	.87
Charlson Comorbidity Index
0	REF	-
≥1	1.59 (1.00-2.52)	.05
Karnofsky Performance Status
<90	REF	-
90-100	1.00 (0.63-1.59)	1.00
Neurologic Symptoms
No	REF	-
Yes	1.17 (0.66-2.07)	.60
Progressive Extracranial Disease
No	REF	-
Yes	0.51 (0.34-0.78)	.002
Number of Prior Systemic Therapy Regimens (Per Increase of 1)	0.98 (0.86-1.12)	.79
Number of Brain Metastases (Per Increase of 1)	1.00 (0.99-1.01)	.84
Metastasis Size (Per Increase of 1mm)	1.01 (1.00-1.03)	.05
Primary Tumor Site
NSCLC	REF	-
Breast	2.53 (1.18-5.44)	.02
Melanoma	3.51 (1.62-7.61)	.001
Other	2.19 (1.19-4.04)	.01
Focal LMD
No	REF	-
Yes	0.56 (0.30-1.07)	.08

Abbreviations: CI, Confidence Interval; HR, Hazard Ratio; LMD, Leptomeningeal Disease; mm, millimeters; NSCLC, Non-small cell lung cancer; REF, Reference category.

Figure 2.

Cumulative incidence of development of classical leptomeningeal disease after initial intracranial management among patients with vs. without focal leptomeningeal extension at diagnosis in the entire cohort of patients (A) and in the subset managed with stereotactic radiation therapy as initial therapy (B). Abbreviations: fLME, focal leptomeningeal extension.

Table 4.

Fine and Gray competing risks regression for development of classical LMD on a per-metastasis level

Variable	HR (95% CI)	P
Age (per year increase)	0.99 (0.97-1.01)	.50
Sex
Male	REF	-
Female	1.01 (0.57-1.79)	.97
Charlson Comorbidity Index
0	REF	-
≥1	0.73 (0.38-1.39)	.33
Karnofsky Performance Status
<90	REF	-
90-100	0.63 (0.35-1.13)	.12
Neurologic Symptoms
No	REF	-
Yes	0.94 (0.50-1.77)	.86
Progressive Extracranial Disease
No	REF	-
Yes	0.38 (0.21-0.67)	<.001
Number of Prior Systemic Therapy Regimens (Per Increase of 1)	1.08 (0.96-1.22)	.20
Number of Brain Metastases (Per Increase of 1)	1.00 (0.99-1.01)	.49
Largest Metastasis (Per Increase of 1mm)	1.01 (0.99-1.02)	.60
Primary Tumor Site
NSCLC	REF	-
Breast	1.28 (0.58-2.82)	.53
Melanoma	2.28 (1.11-4.69)	.03
Other	1.00 (0.51-1.98)	1.00
Focal LMD
No	REF	-
Yes	0.87 (0.44-1.71)	.68

Abbreviations: CI, Confidence Interval; HR, Hazard Ratio; LMD, Leptomeningeal Disease; mm, millimeters; NSCLC, Non-small cell lung cancer; REF, Reference category.

Discussion

We describe a novel pattern of leptomeningeal disease that, to our knowledge, has not been well-characterized in prior literature; fLME appears to be a distinct oncologic entity that displays patterns of intracranial failure and prognosis resembling that of parenchymal BrM as opposed to cLMD. On a national-level, it is likely that some patients with fLME are managed similarly to patients with cLMD, including routine use of WBRT, given the lack of distinction of these entities in the literature and/or concern for more global dissemination; our study suggests that stereotactic approaches may be viable in fLME given the lack of excess risk of local recurrence or development of cLMD.

Prior reports of fLME have been sparse and tangible conclusions from these studies have been hindered by variable definitions of fLME, limited outcome measures, lack of comparison groups, and small sample sizes.^16,17 As a result, fLME is not considered as an independent oncologic entity.15 Our characterization of fLME expands upon prior pilot work and represents the largest experience to date while also quantifying the most critical outcomes of local recurrence and cLMD.

Avoidance of WBRT may carry significant benefit for patients. Prior randomized trials evaluating local brain-directed treatment with or without WBRT in patients with a limited number of brain metastases have consistently demonstrated that WBRT does not improve overall survival.^3,4,12,18,19 In addition, WBRT is associated with deleterious effects such as decrease in quality of life and neurocognitive function. Lastly the local control achieved with SRS/SRT is superior to that achieved with WBRT.²⁰ Accordingly, when SRS/SRT can be utilized in place of WBRT, patients can benefit on multiple levels. The low rates of local recurrence and cLMD that we observed with SRS/SRT in patients with fLME add further support to the notion that, when viable, targeted approaches may be beneficial for such patients. It should be noted that although stereotactic approaches may be viable for patients with fLME, such focal treatment techniques should generally not be considered for patients with cLMD.

Mimickers of cLMD are important for practitioners to be aware of. Infection, infarction, and immunotherapy-related effects can all yield enhancement in the leptomeningeal space.^21,22 Radiation therapy has no role in the management of such radiographic changes. In addition, post-surgical pachymeningal seeding, where tumor cells dispersed during neurosurgical resection recur along the pachymeninges, is often confused with classical leptomeningeal disease.²³ Like fLME, as discussed above, stereotactic radiation can often effectively salvage patients with pachymeningeal seeding; traditional WBRT is sometimes not required. Improved precision and delineation of these entities, including fLME, in the radiology domain might help minimize the risk of patients receiving WBRT unnecessarily.

We acknowledge that accurate identification of fLME may in some scenarios be subjective, especially in settings where a brain tumor-focused radiation oncologist is not present. This underscores the importance of engaging in multidisciplinary care whenever possible with neuro-oncology and neuro-radiology, and also the utility of incorporating additional clinical data whenever possible, as well, such as cytology and full cranio-spinal axis imaging. A lumbar puncture should strongly be considered for patients with a concern for fLME as the presence of tumor cells within the CSF would imply classical leptomeningeal disease, which is not amenable to focal treatment approaches. As fLME becomes better recognized as a distinct oncologic entity, future work should aim to develop quantitative criteria that can aid in its accurate and standardized identification. Furthermore, accurate assessment of fLME’s response to radiation is also challenging, as the delineation between necrosis and tumor recurrence after stereotactic-based radiation remains imperfect. When feasible, multi-disciplinary review, as well as consideration of advanced imaging modalities, such as MRI perfusion and dual-phase PET/CT, should be utilized to aide in this delineation. Of note, we found an increased risk of developing fLME among patients with breast cancer primaries; such patients also tend to display higher rates of pachymeningeal disease. Further study is needed to understand the relationship between breast cancer, pachymeningeal disease, and fLME. Another important limitation of the current work is that our cohort largely predated targeted therapies; in the modern era with better systemic therapies leading to improvements in extracranial disease control and hence longer survival, whether fLME correlates with an increased risk of developing global leptomeningeal disease in the long term is not known and should be the focus of future studies.

In conclusion, fLME seems to be a distinct oncology entity that behaves more like intact brain metastases than cLMD. Our work suggests that fLME should not exclude patients from BrM trials that prohibit cLMD, and taken together, our findings highlight a novel pattern of intracranial disease that appears biologically distinct from cLMD and that potentially can be viably managed with less toxic forms of intracranial treatment.

Supplementary Material

Supplementary material is available online at Neuro-Oncology (https://academic.oup.com/neuro-oncology).

Funding

None declared.

Conflicts of interest statement

A.A.A. reports research funding from Varian Medical Systems and NH TherAguix as well as consulting fees from Novartis and Seagen. P.Y.W. reports consulting for Alexion/Astra Zeneca, Black Diamond, Chimerix, Day One Bio, Fore Biotherapeutics, Genenta, Glaxo Smith Kline, Merck, Nerviano Medical Sciences, Novartis, Novocure, Rigel, Sapience, Servier, Telix, and research support from Astra Zeneca, Black Diamond, Bristol Meyers Squibb, Chimerix, Eli Lily, Erasca, Global Coalition For Adaptive Research, Kazia, MediciNova, Merck, Nerviano, Novartis, Philogen, Quadriga, Servier.

Author Contributions

Substantial contributions to the conception or design of the work (N.L., A.A.A.); acquisition, analysis, or interpretation of data for the work (N.L., L.K., R.R., S.T., D.C., P.J.C., H.E., D.A.H., P.Y.W., A.A.A.); drafting the work or reviewing it critically for important intellectual content (N.L., L.K., R.R., S.T., D.C., P.J.C., H.E., D.A.H., P.Y.W., A.A.A.); final approval of the version to be published (N.L., L.K., R.R., S.T., D.C., P.J.C., H.E., D.A.H., P.Y.W., A.A.A.); agreement to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved work (N.L., L.K., R.R., S.T., D.C., P.J.C., H.E., D.A.H., P.Y.W., A.A.A.).

Data Availability

Pending IRB review of requests, de-identified patient clinical and imaging data will be made available upon request.