Implications of Screening for Brain Metastases in Patients With Breast Cancer and Non–Small Cell Lung Cancer

Abstract

This study analyzes the value of magnetic resonance imaging of the brain for patients with cancers that frequently metastasize to the brain.

Brain metastases affect many patients with cancer.¹ Given the limited intracranial penetration of most systemic therapies, the size and number of brain metastases at diagnosis determines management, with more invasive or toxic therapies such as neurosurgical resection and whole brain radiation therapy (WBRT) used for bulky and multifocal disease, respectively. Consequently, consensus guidelines from the National Comprehensive Cancer Network recommend screening magnetic resonance imaging (MRI) of the brain for patients with stage II to IV non–small cell lung cancer (NSCLC), small cell lung cancer of any stage, and stage IIIC to IV melanoma—all cancers that frequently metastasize to the brain.²

Brain metastases are common in select patients with breast cancer as well: 8% to 11% of patients with de novo metastatic human epidermal growth factor receptor 2 (ERBB2/HER2)–positive or triple negative breast cancer harbor brain metastases,³ and 46% to 53% of such patients develop brain metastases during their clinical course.^4,5 Yet, the National Comprehensive Cancer Network does not recommend brain-directed screening for any patients with breast cancer,² a recommendation that is based only on expert consensus given the lack of definitive or prospective studies on this issue. To identify the potential value of brain-directed MRI screening to select patients with breast cancer, we compared the presentation, management, and outcome of patients with breast cancer and brain metastases (which are largely unscreened with brain MRI) with those of patients with NSCLC and brain metastases (which are largely screened with brain MRI at diagnosis of systemic malignant neoplasm).

Methods

We identified patients treated for newly diagnosed brain metastases between January 1, 2000, and December 31, 2015, at Dana-Farber/Brigham and Women's Cancer Center. We then compared intracranial disease burden at presentation, intracranial management, and intracranial outcomes between patients with breast cancer and patients with NSCLC. Using SAS, version 9.4 (SAS Institute Inc), we compared continuous outcome measures using the 2-tailed unpaired t test or Wilcoxon rank sum test with categorical outcome measures using Fisher exact test; we performed a multivariable Cox regression analysis for time-to-event outcomes, except for neurologic death, which was assessed using the Fine and Gray competing risks regression. A 2-sided P < .05 defined statistical significance. Proportional hazards were verified. This study was approved by the Dana-Farber/Harvard Cancer Center institutional review board, which waived the need for written or oral patient informed consent.

Results

Table 1 compares the baseline characteristics between 349 patients with breast cancer and 659 patients with NSCLC. Patients with breast cancer, compared with patients with NSCLC, presented with larger (median [interquartile range (IQR)] diameter, 17 [10-29] mm vs 14 [8-23] mm; P < .001) and more numerous (median [IQR], 3 [1-8] vs 2 [1-4]; P < .001) brain metastases. Patients with breast cancer were also more likely than patients with NSCLC to be symptomatic at presentation (265 [75.9%] vs 399 [60.5%]; P < .001), present with seizures (59 [16.9%] vs 75 [11.4%]; P = .01), harbor brainstem involvement (28 [8.0%] vs 28 [4.2%]; P = .02), have leptomeningeal disease at diagnosis (40 [11.5%] vs 14 [2.1%]; P < .001), and receive WBRT as initial management (209 [59.9%] vs 283 [42.9%]; P < .001;Table 1). After initial brain-directed therapy, no significant differences in recurrence or treatment-based intracranial outcomes were found between the 2 groups. However, neurological death was more common in patients with breast cancer than patients with NSCLC, both as a percentage of total deaths (103 [37.3%] vs 98 [19.9%]; P < .001; Table 1) and as a time-to-event based outcome (hazard ratio, 1.54; 95% CI, 1.10-2.17; P = .01; Table 2).

Table 1. Baseline Characteristics, Intracranial Management, and Outcomes for Patients With Newly Diagnosed Brain Metastases Secondary to Breast Cancer vs Non–Small Cell Lung Cancer.

Parameter	Patients With Breast Cancer (N = 349)	Patients With NSCLC (N = 659)	P Value
General Clinical Characteristics
Age, mean (SD), y	52 (11)	62 (11)	<.001j
Sex, No. (%)
Male	10 (2.9)	283 (42.9)	<.001k
Female	339 (97.1)	376 (57.1)
Stage at initial diagnosis of cancer, No. (%)a
I	53 (15.2)	60 (9.1)	<.001k
II	112 (32.1)	40 (6.1)
III	118 (33.8)	120 (18.2)
IV	66 (18.9)	439 (66.6)
KPSb
Mean (SD)	81 (10)	83 (9)	.02l
Median (IQR)	80 (80-90)	80 (80-90)
Brain-Related Characteristics at Diagnosis of BM
Largest BM diameter, mm
Mean (SD)	20 (13)	17 (11)	<.001l
Median (IQR)	17 (10-29)	14 (8-23)
With BM, >3 cm, No. (%)c	62 (18.5)	81 (12)	.01k
No. of BM
Mean (SD)	11 (28)	5 (12)	<.001l
Median (IQR)	3 (1-8)	2 (1-4)
With >4 BM, No. (%)d	131 (38.5)	137 (20.9)	<.001k
Neurological symptoms, No. (%)e	265 (75.9)	399 (60.5)	<.001k
Seizures, No. (%)f	59 (16.9)	75 (11.4)	.01k
Leptomeningeal disease, No. (%)g	40 (11.5)	14 (2.1)	<.001k
Brainstem involvement, No. (%)	28 (8.0)	28 (4.2)	.02k
Initial Treatment for BM
Systemic therapy onlyh	56 (16.0)	79 (12.0)	<.001m
Craniotomy plus stereotactic radiation therapy	29 (8.3)	83 (12.6)
Craniotomy plus WBRT	20 (5.7)	58 (8.8)
Stereotactic radiation only	55 (15.8)	213 (32.3)
WBRT only	163 (46.7)	201 (30.5)
WBRT plus stereotactic radiation therapy	22 (6.3)	18 (2.7)
Craniotomy plus WBRT plus stereotactic radiation therapy	4 (1.1)	7 (1.1)
Outcomes After Initial Treatment for BM
Survival, median (95% CI), y	1.45 (1.29-1.65)	1.09 (0.98-1.20)	.06n
Neurological death, No. (%)i	103 (37.3)	98 (19.9)	<.001k

Abbreviations: BM, brain metastases; IQR, interquartile range; KPS, Karnofsky performance status score; NSCLC, non–small cell lung cancer; WBRT, whole brain radiation therapy.

^aBased on the American Joint Committee on Cancer Cancer Staging Manual, 8th ed.⁶

^bKPS score range: 0-100, with the lowest score indicating “dead” and the highest score indicating “normal/no complaints.”

^cPercentage accounts for missing data: 13 patients with breast cancer and 2 patients with NSCLC.

^dPercentage accounts for missing data: 9 patients with breast cancer and 2 patients with NSCLC.

^ePercentage accounts for missing data: 1 patient with NSCLC.

^fPercentage accounts for missing data: 1 patient with breast cancer and 1 patient with NSCLC.

^gPercentage accounts for missing data: 1 patient with breast cancer.

^hDefined as any patient who received systemic therapy within 30 d of a diagnosis of BM without employment of brain-directed radiation therapy for at least 30 additional days.

ⁱReflects a percentage of total deaths.

^jUnpaired, 2-tailed t test used to compare breast cancer with NSCLC.

^kFisher exact test used to compare breast cancer with NSCLC.

^lWilcoxon rank sum test used to compare breast cancer with NSCLC.

^mχ² Test used to compare breast cancer with NSCLC.

ⁿUnadjusted Cox regression analysis used to compare breast cancer with NSCLC.

Table 2. Intracranial Outcomes for Patients With Brain Metastases Secondary to Breast Cancer vs Non–Small Cell Lung Cancer Following Brain-Directed Radiation.

Parameter	HR (95% CI)a	P Value
Neurological death	1.54 (1.10-2.17)	.01b
Time to new BM in patients receiving upfront brain-directed stereotactic radiation therapy	1.02 (0.71-1.46)	.93c
Time to salvage WBRT in patients receiving upfront brain-directed stereotactic radiation therapy	0.74 (0.46-1.20)	.22c
Time to salvage brain-directed stereotactic radiation therapy in patients receiving upfront brain-directed stereotactic radiation therapy	1.51 (0.97-2.34)	.07c
Time to salvage craniotomy	1.18 (0.69-2.00)	.55c
Time to leptomeningeal disease in patients without leptomeningeal disease at diagnosis	1.05 (0.57-1.92)	.88c
Time to seizure in patients without seizures at diagnosis	1.13 (0.73-1.76)	.58c
All-cause mortality	0.95 (0.80-1.13)	.58c

Abbreviations: BM, brain metastases; HR, hazard ratio; WBRT, whole brain radiation therapy.

^aAll HRs reflect breast cancer vs NSCLC; larger HRs reflect a shorter time to the end point in question; all multivariable models adjusted for age, sex, Karnofsky performance status score, and year of treatment.

^bThe Fine and Gray competing risk regression was used to compare breast cancer with NSCLC.

^cMultivariable Cox regression analysis was performed to compare breast cancer and NSCLC.

Discussion

Patients with breast cancer presented with more advanced intracranial disease than did patients with NSCLC and more frequently required WBRT as initial management. However, after initial brain-directed therapy, no differences in recurrence or salvage therapy–based outcomes between the 2 cohorts were noted. This finding suggests that intracranial disease in patients with breast cancer was not more aggressive or resistant to treatment but rather was diagnosed at a later stage.

Brain-directed MRI screening for patients who harbor malignant neoplasms with potential for intracranial involvement is important given the impact of neurological compromise on quality of life. In addition, early identification of intracranial disease facilitates less invasive or less toxic approaches, such as stereotactic radiosurgery or careful use of promising systemic agents rather than WBRT or neurosurgical resection.

The limitations of our study include its retrospective nature, inherent selection bias, and potential clinical confounders. In addition, 84 patients with breast cancer (24.0%) were screened for brain metastases (eg, patient/provider preference, clinical trial)—representing a higher percentage than in previous studies, which screened less frequently⁵—whereas some patients with NSCLC were not, including patients diagnosed with NSCLC due to symptomatic brain metastases. Our results are, therefore, biased toward the null. Despite these limitations, this study strongly supports further investigation into MRI screening of the brain among select patients with metastatic breast cancer.