Central Nervous System–Specific Outcomes of Phase 3 Randomized Clinical Trials in Patients With Advanced Breast Cancer, Lung Cancer, and Melanoma

Abstract

This cohort study determines the proportion of phase 3 clinical trials for patients with metastatic breast cancer, lung cancer, and melanoma that included patients with brain metastases and/or evaluated central nervous system–specific end points.

Brain metastases (BrM) are among the most common neurologic complications of cancer. In fact, up to 40% of patients with a metastatic malignant tumor will develop BrM during their lifetime.¹ Historically, patients with BrM have been excluded from clinical trials due to concerns about poor prognosis and increased risk of neurologic toxic effects.² This has resulted in a scarcity of evidence for the efficacy of systemic therapies in the central nervous system (CNS). Our objective was to determine the proportion of phase 3 clinical trials investigating a systemic therapy intervention for patients with metastatic breast cancer, lung cancer, and melanoma that included patients with BrM and/or evaluated CNS-specific end points.

Methods

A search was conducted on January 19, 2020, using the online, publicly accessible ClinicalTrials.gov database to identify eligible studies. Details regarding search terms, inclusion/exclusion criteria, and statistical methods are outlined in the Supplement. Patient consent was waived due to the deidentified nature of the data analyzed. The Sunnybrook Health Sciences Centre Research Ethics Board determined that ethical review was not required for the use of publicly available data.

Results

Of 223 included trials, 52 (23%) excluded all patients with preexisting BrM while 124 (56%) permitted enrollment of patients with BrM under certain conditions, the most common being stable/nonprogressing BrM (n = 83, 67%), prior treatment for BrM (n = 62, 50%) and lack of neurologic symptoms (n = 49, 40%). Exclusion of patients with BrM decreased over 5-year increments from 2000 to 2019 (P = .006); for example, 8 of 16 studies (50%) in 2000 through 2004 excluded patients with BrM, compared with 3 of 27 studies (11%) in 2015 through 2019. Factors associated with exclusion of patients with BrM are outlined in Table 1. There was no association between inclusion of patients with brain metastases in trials and reporting of CNS-specific end points.

Table 1. Trial Factors Associated With Exclusion of Patients With Brain Metastases.

Trial characteristic	Categorization	No.	No. (%) with complete exclusion of patients with brain metastases	P value	No. (%) with conditional exclusion of patients with brain metastases	P value
Industry funding	Yes	205	46 (22)	.29	112 (55)	.32
	No	18	6 (33)		12 (67)
Year	Before 2000	2	1 (50)	.006	1 (50)	<.001
	2000-2004	16	8 (50)		5 (31)
	2005-2009	92	27 (29)		37 (40)
	2010-2014	86	13 (15)		59 (69)
	2015-2019	27	3 (11)		22 (81)
Disease site	Breast	81	32 (40)	<.001	28 (35)	<.001
	Lung	117	15 (13)		78 (67)
	Melanoma	25	5 (20)		18 (72)
Cooperative group study	Yes	21	7 (33)	.25	14 (67)	.28
	No	202	45 (22)		110 (54)
Systemic therapy	Cytotoxic chemotherapy	37	12 (32)	.21	15 (41)	.10
	Targeted therapy	182	40 (22)		106 (58)
	Both	4	0		3 (75)
Locationa	Asia	21	4 (19)	.96	8 (38)	.04
	Europe	9	2 (22)		4 (44)
	North America	21	4 (19)		17 (81)
	Intercontinental	160	37 (23)		88 (55)
	No location provided	12	5 (42)		7 (58)
Stage	III and IV	154	40 (26)	.16	82 (53)	.29
	Stage IV only	69	12 (17)		42 (61)
Results published	Yes	161	39 (24)	.61	90 (56)	.89
	No	62	12 (21)		34 (55)
CNS-specific outcomes	Yes	13	4 (31)	.51	9 (69)	.31
	No	210	48 (23)		115 (55)

^aP value calculated based on studies with known location.

Of 223 trials, CNS-specific outcomes were evaluated in 13 (6%; 95% CI, 3.1-9.5) studies. The most commonly reported CNS-specific outcomes included the presence vs absence of CNS progression (7 of 13, 54%), time to CNS progression (7 of 13, 54%), CNS-specific response (4 of 13, 31%), and duration of CNS response (3 of 13, 23%). Among studies that included CNS-specific outcomes, there was considerable heterogeneity in the selection and definition of these end points. Only 2 studies (15%) addressed the effect of systemic therapy on neurologic quality of life and only 1 study assessed neuro-cognition.

Discussion

The majority of phase 3 clinical trials for patients with advanced breast cancer, lung cancer, and melanoma excluded or restricted enrollment of patients with BrM. Furthermore, collection of CNS-specific outcomes is rarely specified in trial protocols, highlighting challenges (Table 2) in assessing CNS response to systemic therapies.

Table 2. Common Barriers to Including Patients With Brain Metastases (BrM) in Clinical Trials.

Barrier	Example(s)	Potential solution(s)
Natural history of BrM is unpredictable	Complications such as hemorrhage can interfere with the evaluation of drug efficacy	Enroll patients based on a predefined eligibility quotient3
Patients with BrM have variable prognoses	Patients with multiple BrM have a worse prognosis than those with solitary lesions Perception of relative futility of treating patients with BrM due to poor survival Concurrent BrM and leptomeningeal disease may confound patient outcomes	Stratification by the Graded Prognostic Assessment3 or by the number of BrM (eg, 1 vs >1 brain metastasis), but larger sample sizes may be required Include strict baseline MRI protocols to rule out leptomeningeal disease and to assess the burden of parenchymal BrM
Discordant phenotype between primary malignant tumor and BrM	Discordance in the molecular profile of BrM with that of the primary tumor may be problematic for molecularly targeted clinical trials	Investigate molecular underpinnings of BrM through preclinical models and tissue biobanks Develop noninvasive methods to sample and assess BrM (eg, liquid biomarker assays)
Evaluation of treatment response in the CNS is challenging after treatment with radiotherapy	Regression of a BrM may be due to a delayed treatment effect postradiotherapy Progression of a BrM postradiotherapy (particularly SRS) may be attributed to radiation necrosis	Use brain-specific response criteria Develop better techniques to distinguish radiation necrosis vs disease progression
The CNS activity of certain systemic therapies is unknown	High risk of CNS progression if systemic therapies are unlikely to have CNS activity	Incorporate CNS end points into early-phase clinical trials and/or perform CNS substudies to determine intracranial efficacy/penetration Incorporate close monitoring of the CNS radiographically; patients with CNS progression with stability of extracranial disease should be given an opportunity to stay on trial while receiving local therapy to the brain Exclude patients with leptomeningeal disease and those with untreated or unstable parenchymal BrM
Differential effect of systemic therapy on CNS and extracranial disease	It may be difficult to ascertain extracranial benefit of systemic therapy if patients withdraw from the study due to CNS progression	Patients with CNS progression with stability of extracranial disease should be given an opportunity to stay on trial while receiving local therapy to the brain Select appropriate overall and intracranial-specific outcomes Ensure adequate monitoring of CNS disease among patients with BrM at baseline
Logistical barriers associated with neurologic-specific end point assessments	Evaluating neurologic-specific quality of life and neurocognition may be time consuming and costly	Consider self-administered patient-reported outcome assessments

Abbreviations: CNS, central nervous system; MRI, magnetic resonance imaging; SRS, stereotactic radiosurgery.

Because this study involved a search of the ClinicalTrials.gov website, it does not capture reporting of CNS-specific outcomes as part of exploratory analyses in future publications. Whether trials appropriately excluded patients with brain metastases (eg, based on the clinical study purpose) also is not captured. Nevertheless, in accordance with recommendations from the American Society of Clinical Oncology and Friends of Cancer Research, patients with treated and/or stable BrM should be routinely included in clinical trials of all phases, while patients with active BrM should be evaluated for inclusion based on specified criteria.^3,4 Unfortunately, several potential barriers to including patients with BrM in clinical trials exist; examples derived from the experience of individual experts and the Response Assessment in Neuro-Oncology (RANO) group^5,6 are provided in Table 2.

It is notable that varying definitions of CNS-specific end points were used across included trials. For example, 3 different criteria were used to evaluate CNS-specific response (Response Evaluation Criteria in Solid Tumors 1.1, World Health Organization, and RANO-BM criteria), which differ with respect to the definition of target lesions, the maximum number of target lesions in the brain, and the definition of progression vs response. It is important to unify the definitions of CNS-specific end points across trials, particularly in the modern era of stereotactic radiosurgery and immunotherapy. Finally, since CNS-penetrating systemic therapies may impact neurologic-specific quality of life and neurocognition, efforts must be made to measure these patient-reported outcomes using validated tools.

Supplement.

eAppendix. Methods