Prevalence, treatment patterns, and survival of patients with brain metastases from small cell lung cancer: A retrospective study using the TriNetX Oncology Database

Abstract

Background

Brain metastases (BM) portend increased morbidity and mortality in patients with small cell lung cancer (SCLC). We aimed to characterize the prevalence, timing, treatment patterns, and survival outcomes of BM associated with SCLC over the past decade.

Methods

Data from 4014 patients with histologically confirmed SCLC were extracted from the TriNetX Oncology database. Clinical and demographic variables were compared between patients with and without BM using Chi-squared and t-tests. Kaplan–Meier and Cox regression analyses were used to evaluate overall survival (OS), after propensity score matching cohorts for age at diagnosis, sex, cancer stage at diagnosis, extracranial metastases, and cancer-directed therapy.

Results

Among 4014 patients with SCLC, 35.0% had BM (9.9% synchronous, 21.2% metachronous, 3.9% precocious). Patients who developed BM were younger (P < .001) at SCLC diagnosis, more likely Black/African American (P = .0068), and presented with more advanced cancer stages (P < .001) than patients who did not develop BM. The median BM-free survival from the time of SCLC diagnosis was 27.9 months. Patients with BM received higher rates of cancer-directed therapies than those without BM. Synchronous BM was associated with lower OS than metachronous BM after the diagnosis of SCLC (HR[95% CI] = 1.56[1.32–1.83]), but there was no difference in OS after the BM diagnosis. OS did not differ between patients with BM and patients with extracranial metastases only, following the diagnosis of metastatic disease.

Conclusions

Our findings support that independently of the chronicity of BM diagnosis, patients with SCLC have poor survival once the diagnosis of BM is conferred.

Key Points.

• One-third of patients with SCLC were diagnosed with BM.

• The probability of developing BM increases over time in patients with SCLC.

• Patients with SCLC have poor survival once the diagnosis of BM is conferred, regardless of the chronicity of BM diagnosis in the course of their disease.

• SCLC patients with BM have more complex oncologic management than patients without BM.

Importance of the Study.

Small cell lung cancer (SCLC) frequently metastasizes to the brain, significantly impacting patient survival. Although synchronous brain metastases (BM) are documented in national cancer registries, BM that develop after the diagnosis of primary cancer (metachronous) are often not captured in these databases. This study offers a comprehensive examination of BM development throughout the disease course of SCLC, based on a large multi-institutional billing-claims-based database. Furthermore, our data support the evolving risk of metachronous BM over the disease trajectory, which is consistent with prior research. Our data highlight the importance of continued compliance with intracranial screening protocols, adherence to monitoring, and new treatment paradigms aimed at delaying the onset of BM in SCLC patients. Finally, our findings support proactive strategies to mitigate the impact of BM, emphasizing the need for vigilant surveillance and timely intervention to improve patient outcomes.

Small cell lung cancer (SCLC) accounts for approximately 15% of lung and bronchus cancers.¹ SCLC is highly aggressive, marked by rapid growth, early metastatic spread, and a dismal 5-year survival rate of 7%.¹ Brain metastases (BM) are a common occurrence in patients with SCLC.² BM may be identified at various times within the disease course of SCLC, including precociously (prior to the diagnosis of SCLC), synchronously (within 2 months of the diagnosis of SCLC), and metachronous (after 2 months of the diagnosis of SCLC).³ Nationwide cancer registries, such as the National Cancer Institute’s Surveillance, Epidemiology, and End Results (SEER) Program and the American College of Surgeons’ National Cancer Database (NCDB), only report synchronous BM and thus, exclude the majority of BM in this patient population.^4,5 Other attempts to evaluate the frequency of BM in patients with cancer include autopsy, institutional, or city-wide population studies.^3,4,6 However, many of these studies are decades old, reflect the practice of either a single institution or region, or lack longitudinal oncologic data.^3,4,6

With improved systemic therapies that prolong overall survival in patients with lung cancer, the incidence of BM, patient mortality secondary to intracranial metastasis, and healthcare burden are likely to rise.⁷ Consequently, updated epidemiological studies of BM are necessary to inform surveillance and diagnostic guidelines, treatment strategies, and the initiation of preventative measures. Until recently, with the release of the TriNetX Oncology database, there was no major national database with longitudinal data on both synchronous and metachronous BM associated with histologically confirmed SCLC in the United States. In this retrospective study, we used the TriNetX Oncology database, a large, multi-institutional claims-based oncology database to report the prevalence, BM-free survival, treatment patterns, and survival outcomes of patients with SCLC over the past decade.

Materials and Methods

Patient Selection and Recorded Variables

Study data were retrieved from the TriNetX Oncology database (TriNetX Inc., Cambridge, MA), a curated claims-based database released in August 2023 that contains de-identified clinical tumor registry data from 17 centers of cancer care excellence in the United States. This data also includes cancer-specific data, including histology for all patients and TNM staging for most patients.

At the time of data analysis, the TriNetX Oncology Dataset included 663 281 patients with histologically confirmed primary tumors. We identified 64 560 patients diagnosed with lung and bronchus neoplasms using International Statistical Classification of Diseases and Related Health Problems, Tenth Revision (ICD-10) code C34 (Figure 1). To confirm that BM were associated with lung and bronchus neoplasms exclusively, 4653 patients diagnosed with primary extra-pulmonary tumors were excluded from this study. We excluded an additional 25 807 patients diagnosed greater than 10 years ago, to more accurately reflect modern diagnostic and treatment approaches for lung and bronchus neoplasms.³ Patients with histologically confirmed SCLC were identified using the TriNetX Curated small cell carcinoma variable, which includes patients with the following ICD-O morphology codes—small cell carcinoma, NOS (8041/3); combined small cell carcinoma (ICD-O 8045/3); oat cell carcinoma (8042/3), small cell carcinoma, fusiform cell (8043/3); and small cell carcinoma, intermediate cell (8044/3).

Figure 1.

Flow diagram detailing patient cohort selection from the TriNetX Oncology database.

We identified diagnoses, procedures, and medications with the 10^th revision of the International Statistical Classification of Diseases (ICD-10) codes, Current Procedural Terminology (CPT) codes, and RxNorm codes, respectively. The list of billing codes utilized for each variable may be found in Supplementary Table S1. Deaths were determined by the vital status code “deceased,” according to the Social Security Death Index. Patients were staged according to The Eighth Edition of the American Joint Committee on Cancer Cancer Staging Manual.⁸

We identified patients with BM using ICD-10 code C79.3. The BM cohort included patients who had the C79.3 billing code at any time. Precocious BMs were defined as BM diagnosed at any time prior to the diagnosis of SCLC. Synchronous BMs were defined as BM diagnosed within 2 months of the diagnosis of SCLC.³ Metachronous BMs were defined as BM diagnosed after 2 months of the diagnosis of SCLC.³ The non-BM cohort was defined as patients with SCLC diagnosed between 2014 and 2024 who did not have the C79.3 billing code at any time. The extracranial metastases-only cohort included patients with SCLC diagnosed between 2014 and 2024 who had an ICD-10 billing code for a secondary malignancy at any site, excluding BM. The no metastases cohort included patients with SCLC diagnosed between 2014 and 2024 who did not have an ICD-10 billing code for a secondary malignancy at any site or BM.

This study was approved by the Institutional Review Board at Johns Hopkins University and informed patient consent was waived (IRB00324993).

Statistical Analysis

All statistics were performed using the TriNetX online platform. Clinical and demographic characteristics were evaluated using the Chi-square test for categorical variables and the independent sample t-test for continuous variables. To evaluate associations between demographic variables and the odds of receiving a BM diagnosis, groups were propensity score matched for age and stage at SCLC diagnosis and compared with logistic regression analysis.

The Kaplan–Meier estimate was utilized to determine the BM-free survival probability following the diagnosis of SCLC. BM-free survival was defined as the time from the diagnosis of SCLC to the diagnosis of BM. Patients who were lost to follow-up or died were censored. The probability of BM was determined with the following equation: (1—BM-free survival).

For the survival analyses, the TriNetX platform uses the R Survival package v3.2-3 and validates results by comparison with SAS version 9.4. Statistical significance was defined as P < .05. Cohorts were propensity score matched for age at diagnosis of SCLC, sex, cancer stage at diagnosis, extracranial metastases, thoracic surgery after SCLC diagnosis, radiation therapy after SCLC diagnosis, and receipt of antineoplastic medication after SCLC diagnosis. Propensity score matching was performed using logistic regression and greedy nearest-neighbor matching to create matched cohorts with a caliper width of 0.1 pooled standard deviations. Survival times were calculated from the date of SCLC diagnosis to the date of death/last follow-up visit, from the date of BM diagnosis to the date of death/last follow-up visit, or from the date of first extracranial metastasis diagnosis to date of death/last follow-up visit. Survival probabilities were estimated using the Kaplan–Meier method. The log-rank test was used to evaluate differences in survival. Hazard ratios and 95% confidence intervals (CI) were calculated using Cox regression analyses after propensity score matching.

Results

Prevalence of BM Diagnoses and Overall Cohort Characteristics

We identified a total of 4014 patients diagnosed with SCLC during the study period, of whom 1404 (35.0%) had a BM diagnosis and 2610 (65.0%) did not have a BM diagnosis. Most BM diagnoses were metachronous (n = 850, 21.2%), followed by synchronous (n = 397, 9.9%), and precocious (n = 157, 3.9%). The overall cohort characteristics are presented in Table 1. The average ± SD age at diagnosis of SCLC of the overall cohort was 66.4 ± 9.4 years. The SCLC cohort included 1812 (45.1%) female, 95 (2.4%) Hispanic, 2883 (71.8%) White, 429 (10.7%) Black/African American, 43 (1.1%) Asian, 20 (0.5%) American Indian or Alaska Native, and 10 (0.3%) Native Hawaiian or Other Pacific Islander patients. Risk factors for SCLC were prominent in the overall cohort—2284 (56.9%) patients had an ICD-10 code for either tobacco use or nicotine dependence and 838 (20.9%) had a family history of any primary malignancy. Seventy-four percent of the total SCLC cohort had at least 1 year of follow-up and 44.6% had at least 5 years of follow-up. The primary tumor was located in the lung parenchyma for 89.7% of patients and in the bronchi for 10.3% of patients. Staging data at diagnosis was available for 79.3% (n = 3184) of the SCLC cohort. At diagnosis, 30 (0.8%) patients had stage 0 SCLC (ie, carcinoma in situ) at diagnosis, 232 (5.8%) patients had stage 1 SCLC, 148 (3.7%) patients had stage 2 SCLC, 817 (20.4%) patients had stage 3 SCLC, and 1957 (48.8%) patients had stage 4 SCLC. TriNetX does not categorize patients as limited versus extensive stages. Four-hundred eighty-nine (12.2%), 1274 (31.7%), and 1191 (29.7%) patients were diagnosed with adrenal, bone, and liver metastases, respectively, during the disease course.

Table 1.

Overall Cohort Characteristics (N = 4014)

Characteristic	N	%
Age at diagnosis of SCLC—mean ± SD	66.4 ± 9.4
Sex
Male	1765	43.97%
Female	1812	45.14%
Unknown	437	10.89%
Ethnicity
Hispanic	95	2.37%
Not Hispanic	3291	81.99%
Unknown	628	15.65%
Race
White	2883	71.82%
Black or African American	429	10.69%
Asian	43	1.07%
American Indian or Alaska Native	20	0.50%
Native Hawaiian or Other Pacific Islander	10	0.25%
Other race	23	0.57%
Unknown	606	15.10%
Risk factors	0	0.00%
Smoking (tobacco use or nicotine dependence)	2284	56.90%
Family history of primary malignancy	838	20.88%
Stage1
Stage 0	30	0.75%
Stage 1	232	5.78%
Stage 2	148	3.69%
Stage 3	817	20.35%
Stage 4	1957	48.75%
Missing data	830	20.68%
T1
TX	985	24.54%
T0	33	0.82%
T1	679	16.92%
T2	667	16.62%
T3	519	12.93%
T4	993	24.74%
Missing data	138	3.44%
N1
NX	711	17.71%
N0	595	14.82%
N1	355	8.84%
N2	1480	36.87%
N3	738	18.39%
Missing data	136	3.39%
M1
M0	1060	26.41%
M1	2014	50.17%
Missing data	940	23.42%
Metastatic sites2
Brain	1404	34.98%
Precocious	157	3.91%
Synchronous	397	9.89%
Metachronous	850	21.18%
Adrenals	489	12.18%
Bone	1274	31.74%
Liver	1191	29.67%

¹TNM staging was determined at the time of diagnosis of primary SCLC.

²Metastases developed throughout the disease course.

Imaging Rates Within the First 2 Months of SCLC Diagnosis

Of the overall SCLC cohort, 1688 patients (42.1%) received intracranial screening within the first 2 months of the diagnosis of SCLC, including 1351 (33.7%) who received brain MRI and 589 (14.7%) who received computed tomography (CT) of the head. Six percent of patients who received head CT had a pacemaker, likely preventing them from receiving brain MRI. Rates of intracranial screening within the first 2 months of SCLC diagnosis varied by the stage at diagnosis. Rates of intracranial imaging (both head CT and/or brain MRI) within the first 2 months of SCLC diagnosis were 19.1%, 25.0%, 29.6%, and 47.4% in patients diagnosed with stages 1, 2, 3, and 4 SCLC, respectively. Fifty-nine percent of patients with missing staging data received intracranial imaging within 2 months of SCLC diagnosis. For each stage, brain MRI was more commonly utilized than head CT. Interestingly, rates of intracranial imaging within the first 2 months of SCLC diagnosis were higher among White patients than nonwhite patients (48.5% vs. 38.4%, P < .0001). Unsurprisingly, rates of intracranial imaging within the first 2 months of the diagnosis of SCLC were higher in patients who had a BM diagnosis. Patients with a BM diagnosis received brain MRI (43.9% vs. 28.2%, P < .0001), and CT of the head (17.2% vs. 13.3%, P < .0001) at higher frequencies than patients who did not receive a BM diagnosis.

We evaluated the impact of receiving a brain MRI within the first 2 months of SCLC on OS. Patients who received intracranial imaging within the first 2 months of the diagnosis of SCLC and patients who did not receive this imaging were propensity score matched for age at SCLC diagnosis, stage of SCLC at diagnosis, antineoplastic treatment, and extracranial metastases. OS from the diagnosis of BM was compared with Kaplan–Meier analysis. We noted no difference in OS after the diagnosis of BM between patients who received intracranial imaging within the first 2 months of the diagnosis of SCLC and patients who did not receive intracranial imaging within the first 2 months of the diagnosis of SCLC (HR [95%CI]: 1.08 [0.95-1.22]).

For extracranial screening, 1320 (32.9%) of the overall cohort received CT of the abdomen and pelvis and 1004 (25.0%) received diagnostic position emission tomography (PET) with concurrent CT. Rates of extracranial screening imaging varied, with patients without a BM diagnosis receiving abdominal and pelvic CT at higher rates than patients with a BM diagnosis (35.8% vs. 27.5%, P < .001), while patients with a BM diagnosis received higher rates of PET/CT than patients without a BM diagnosis (30.6% vs. 22.0%, P < .001).

BM-Free Survival

We used the Kaplan–Meier estimate to investigate BM-free survival by stage from the time of initial diagnosis of SCLC (Figure 2). Over the 10-year follow-up period, BM were diagnosed in 35 (18.5%), 19 (18.3%), 112 (22.1%), and 826 (39.8%) of patients with initial stages 1, 2, 3, and 4 SCLC, respectively. Among patients without staging data, 412 (49.6%) developed BM. As expected, patients with initial stage 4 SCLC had higher odds of developing BM than patients with initial stages 1–3 (OR: 1.69, 95% CI: 1.61–1.75), after propensity score matching for age at diagnosis of SCLC, thoracic surgery status, radiation therapy, and systemic therapies. Overall, the median BM-free survival from the time of SCLC diagnosis was 27.9 months. By stage, the median BM-free survival time was 110 months after the diagnosis of SCLC in patients with initial stage 3 SCLC and 9.3 months in patients with initial stage 4 SCLC. The median BM-free survival was not reached in patients with initial stages 1 and 2 SCLC (the KM survival curve did not cross 50%). The probability [95% CI] of patients being diagnosed with a BM 1-year post initial diagnoses of stage 1, 2, 3, and 4 SCLC were 8.2% [4.7%–13.9%], 6.8% [3.1%–14.6%], 14.9% [11.7%–19.0%], and 55.3% [52.6%–58.1%], respectively. Notably, the probability of receiving a BM diagnosis further increased over the study period. The probability [95%CI] of patients being diagnosed with a BM 5 years post initial diagnoses of stage 1, 2, 3, and 4 SCLC were 31.4% [22.7%–42.4%], 32.4% [21.2%–47.7%], 40.1% [33.8%–47.1%], and 68.1% [64.5%–71.6%], respectively.

Figure 2.

Kaplan–Meier survival curve comparing brain metastases-free survival of patients with initial stages 1-3 SCLC to patients with initial stage 4 SCLC. Index event = diagnosis of lung/bronchus cancer. TriNetX does not provide survival tables; therefore, we are unable to report the number of patients censored at each time point.

Comparison of Demographic and Clinical Characteristics Between Patients With a BM Diagnosis and Patients Without a BM Diagnosis

We evaluated differences in demographic and clinical characteristics between patients with a BM diagnosis (at any time) and patients without a BM diagnosis (Table 2). Patients with BM were diagnosed with SCLC younger than patients without a BM diagnosis (64.3 ± 8.8 vs. 67.7 ± 9.5, P < .0001). We also observed racial and ethnic differences between these groups. The BM diagnosis population had a higher frequency of Black/African American patients (12.7% vs. 9.6%, P = .0068) and Native Hawaiian or Other Pacific Islander patients (0.7% vs. 0.0%, P < .0001) than the cohort without a BM diagnosis. Risk factors for SCLC, such as tobacco use and/or nicotine dependence (69.4% vs. 50.2%, P < .0001) and family history of cancer (25.7% vs. 18.3%, P < .0001) were more prevalent in the cohort with a BM diagnosis. After propensity score matching for age at SCLC diagnosis and stage at diagnosis, tobacco use and/or nicotine dependence remained independently associated with higher odds of receiving BM diagnosis (OR [95% CI]: 1.2 [1.1–1.4]).

Table 2.

Clinical and Demographic Characteristics by Brain Metastases Status

Characteristic	BM diagnosis (N = 1404)	No BM diagnosis (N = 2610)	P
Age at diagnosis of SCLC—mean ± SD	64.3 ± 8.8	67.5 ± 9.5	<.0001
Sex
Male	636 (45.30%)	1129 (43.26%)	.1853
Female	634 (45.16%)	1178 (45.13%)	.7636
Unknown	134 (9.54%)	303 (11.61%)	.0523
Ethnicity
Hispanic	33 (2.35%)	62 (2.38%)	.9317
Not Hispanic	1182 (84.19%)	2109 (80.80%)	.0237
Unknown	189 (13.46%)	439 (16.82%)	.0152
Race
White	1002 (71.37%)	1881 (72.07%)	.5091
Black or African American	178 (12.68%)	251 (9.62%)	.0068
Asian	21 (1.5%)	22 (0.84%)	.612
American Indian or Alaska Native	10 (0.71%)	10 (0.38%)	.2507
Native Hawaiian or Other Pacific Islander	10 (0.71%)	0 (0%)	<.0001
Other race	188 (13.39%)	418 (16.02%)	.1393
Unknown	0 (0%)	0 (0%)
Risk factors
Smoking (tobacco use or nicotine dependence)	975 (69.44%)	1309 (50.15%)	<.001
Family History of primary malignancy	361 (25.71%)	477 (18.28%)	<.001
Stage1
Stage 0	10 (0.71%)	20 (0.77%)	.2507
Stage 1	47 (3.35%)	185 (7.09%)	<.001
Stage 2	34 (2.42%)	114 (4.37%)	.0011
Stage 3	199 (14.17%)	618 (23.68%)	<.001
Stage 4	826 (58.83%)	1131 (43.33%)	<.001
Missing data	288 (20.51%)	542 (20.77%)	.3718
T1
TX	315 (22.44%)	670 (25.67%)	.0148
T0	20 (1.42%)	13 (0.5%)	.6724
T1	201 (14.32%)	478 (18.31%)	.0003
T2	229 (16.31%)	438 (16.78%)	.9232
T3	178 (12.68%)	341 (13.07%)	.8312
T4	378 (26.92%)	615 (23.56%)	.0542
Missing data	83 (5.91%)	55 (2.11%)	<.001
N1
NX	223 (15.88%)	488 (18.7%)	.0373
N0	167 (11.89%)	428 (16.4%)	.0016
N1	110 (7.83%)	245 (9.39%)	.0718
N2	501 (35.68%)	979 (37.51%)	.2366
N3	306 (21.79%)	432 (16.55%)	.0002
Missing data	97 (6.91%)	39 (1.49%)	<.001
M1
M0	260 (18.52%)	800 (30.65%)	<.001
M1	845 (60.19%)	1169 (44.79%)	<.001
Missing data	299 (21.30%)	641 (24.56%)	.1853
Metastatic sites2
Adrenals	331 (25.85%)	158 (6.05%)	<.001
Bone	682 (61.25%)	592 (22.68%)	<.001
Liver	559 (53.42%)	632 (24.21%)	<.001

Results are presented as N (%) unless otherwise specified.

Bold-type face under P indicates statistical significance.

¹TNM staging was determined at the time of diagnosis of primary SCLC.

²Metastases developed throughout the disease course.

Prior to or at the same time as receiving a BM diagnosis, 10.5% of patients in the BM cohort were diagnosed with adrenal metastases, 26.0% were diagnosed with bone metastases, and 24.0% were diagnosed with liver metastases. An additional 15.3%, 35.3%, and 29.4% of patients in the BM cohort were diagnosed with adrenal, bone, and liver metastases after the diagnosis of BM, respectively. Overall, patients with a BM diagnosis were diagnosed with adrenal (25.9% vs. 10.1%, P < .001), bone (61.3% vs. 37.7%, P < .001), and liver (53.4% vs. 40.3%, P < .001) metastases at significantly higher rates than patients with extracranial metastases only. After propensity score matching the BM cohort and the extracranial metastases-only cohort for age at SCLC diagnosis, cancer stage at diagnosis, thoracic surgery status, radiation therapy, and antineoplastic therapy, the odds of developing extracranial metastases remained higher in the BM cohort (OR: 1.34, 95% CI: 1.15–1.57, P = .0002).

Impact of BM on Oncologic Management

Finally, we evaluated the impact of BM on the oncologic management and survival of patients with SCLC. Rates of surgical procedures, antineoplastic medication utilization, and radiation utilization in patients with BM, extracranial metastases only, and no metastases are presented in Table 3. There were no significant differences in the rates of thoracic surgery between patients with a BM diagnosis versus no BM diagnosis or between patients with a BM diagnosis versus extracranial metastases only. Patients with a BM diagnosis received higher rates of intracranial stereotactic radiosurgery (9.5%) than craniotomy or craniectomy (4.4%). After propensity score matching for age at SCLC diagnosis and initial stage, patients with BM had higher odds of receiving radiation (OR: 4.11, 95% CI: 2.56–6.62, P < .001) and antineoplastic medication (OR: 1.51, 95% CI: 1.30–1.76, P < .001) than patients with extracranial metastases only. In patients diagnosed with BM, the most utilized chemotherapeutic agents were etoposide (49.9%) and carboplatin (38.5%). Approximately 16% of patients with a BM diagnosis received atezolizumab, an immune checkpoint inhibitor. Overall, patients with BM diagnoses received chemotherapy, tyrosine kinase inhibitors, and immune checkpoint inhibitors at higher rates than patients without BM or patients with extracranial metastases only, prior to developing metastatic disease and after the development of metastatic disease.

Table 3.

Therapy Received From Diagnosis of SCLC to End of Follow-up (N = 4014)

Therapy	BM (N = 1404)	Extracranial metastases only (N = 1569)	No metastases (N = 1041)
Excision/resection procedures on the lung and pleura	69 (4.91%)	76 (4.84%)	104 (9.99%)
Brain-directed surgery
Craniectomy or craniotomy	62 (4.42%)	0 (0%)	0 (0%)
Stereotactic radiosurgery (cranial)	134 (9.54%)	0 (0%)	0 (0%)
Radiation (any)	713 (50.78%)	375 (23.9%)	197 (18.92%)
Stereotactic body radiation therapy	92 (6.55%)	30 (1.91%)	24 (2.31%)
Antineoplastic medication	807 (57.48%)	710 (45.25%)	246 (23.63%)
Platinum based
carboplatin (Paraplatin)	617 (43.95%)	551 (35.12%)	151 (14.51%)
cisplatin (Platinol AQ)	141 (10.04%)	130 (8.29%)	76 (7.3%)
Taxanes
paclitaxel (Taxol)	59 (4.2%)	31 (1.98%)	10 (0.96%)
docetaxel (Taxotere)	30 (2.14%)	10 (0.64%)	0 (0%)
Antimetabolites
gemcitabine (Gemzar)	23 (1.64%)	10 (0.64%)	10 (0.96%)
Topoisomerase inhibitors
etoposide (Etopophos, Toposar)	700 (49.86%)	622 (39.64%)	200 (19.21%)
topotecan (Hycamtin)	104 (7.41%)	45 (2.87%)	11 (1.06%)
irinotecan (Camptosar)	42 (2.99%)	19 (1.21%)	10 (0.96%)
doxorubicin HCl (Adriamycin)	20 (1.42%)	0 (0%)	10 (0.96%)
Alkylating agents
temozolomide (Temodar)	32 (2.28%)	10 (0.64%)	10 (0.96%)
lurbinectedin	71 (5.06%)	39 (2.49%)	10 (0.96%)
vinCRIStine (Vincasar PFS)	0 (0%)	(0%)	0 (0%)
Immunotherapy
pembrolizumab (Keytruda)	30 (2.14%)	10 (0.64%)	10 (0.96%)
nivolumab (Opdivo)	96 (6.84%)	50 (3.19%)	18 (1.73%)
atezolizumab (Tecentriq)	225 (16.03%)	155 (9.88%)	29 (2.79%)
Targeted therapy
erlotinib (Tarceva)	20 (1.42%)	0 (0%)	0 (0%)
bevacuzimab	20 (1.42%)	10 (0.64%)	10 (0.96%)

Treatment modalities by the timing of BM diagnosis are presented in Supplementary Table S2. Craniectomy/craniotomy was performed in 7.0%, 5.0%, and 3.6% of patients with precocious, synchronous, and metachronous BM, respectively. Stereotactic radiosurgery was performed in 8.3%, 7.8%, and 10.6% of patients with precocious, synchronous, and metachronous BM, respectively. Antineoplastic medication was received by 51.6%, 53.7%, and 60.4% of patients with precocious, synchronous, and metachronous BM, respectively.

Impact of BM on Survival

The median survival from the diagnosis of SCLC in our entire cohort was 11.9 months. Overall, patients with a BM diagnosis had a median survival of 13.5 months from the diagnosis of SCLC and 7.2 months from the diagnosis of BM. Patients with precocious BM had a median survival time of 3.7 months from the diagnosis of SCLC and BM. Patients with synchronous BM had a median survival time of 9.5 months from diagnosis of SCLC and BM. Patients with metachronous BM had a median survival time of 18.9 months from the diagnosis of SCLC and 6.6 months from the diagnosis of BM. Patients with extracranial metastases only had a median survival of 9.2 months from the diagnosis of SCLC and 6.7 months from the diagnosis of metastatic disease. The results of propensity score-matched survival analyses are presented in Figure 3. From the diagnosis of SCLC, patients with synchronous BM had poorer OS compared to patients with metachronous BM (HR [95% CI] = 1.60 [1.36–1.89]), but there was no significant difference in OS following the diagnosis of BM between these groups (HR [95% CI] = 0.96 [0.82–1.12]). There was no significant difference in OS between patients with a BM diagnosis (with or without extracranial metastases) compared to patients with extracranial metastases only, starting from the diagnosis of BM or extracranial metastases, respectively (HR [95% CI] = 0.93 [0.84–1.04]).

Figure 3.

Overall survival analysis for SCLC cohort. Kaplan–Meier curves illustrating survival probability over time in (A) patients with brain metastases (BM) compared to patients with extracranial metastases only with diagnosis of BM as the index event for patients with a BM diagnosis and the diagnosis of the first extracranial metastasis for patients with extracranial metastases only, (B) patients with synchronous BM compared to patients with metachronous BM with diagnosis of lung/bronchus cancer as the index event, and (C) patients with synchronous BM compared to patients with metachronous BM with diagnosis of BM as the index event. For each comparison, cohorts were propensity-score matched for age at diagnosis of lung/bronchus cancer, sex, cancer stage at diagnosis, extracranial metastases, excision/resection procedures on the lungs/pleura, radiation therapy, and receipt of antineoplastic medication. Log-rank P-values were determined by Kaplan–Meier comparison of the propensity-score matched cohorts. Hazard ratios and 95% confidence intervals were calculated with Cox Regression analyses of the propensity-score matched cohorts.

Discussion

BM occurs in over 30% of patients with SCLC.² Currently, the National Comprehensive Cancer Network (NCCN) recommends that patients with SCLC receive brain MRI or CT with contrast at initial evaluation, followed by intracranial surveillance imaging every 3–4 months during the first year of disease, every 6 months during year 2, and as clinically indicated thereafter, regardless of prophylactic cranial irradiation (PCI) status.⁹ Regarding prevention of BM, the NCCN recommends that SCLC patients without poor performance status or cognitive impairment receive PCI after resolution of toxicities associated with initial systemic therapy.⁹ Radiation therapy, either whole-brain radiotherapy (WBRT) or stereotactic radiosurgery (SRS) is recommended to treat BM.⁹ As improvements in cancer detection and therapy have increased survival times in cancer patients, the frequency of BM diagnoses has correspondingly increased, necessitating updated estimates of BM rates to better inform screening and treatment guidelines in patients with SCLC. In this study, we report the prevalence, BM-free survival, treatment patterns, and survival of patients with BM from SCLC using TriNetX, a large, claims-based oncology database.

In the TriNetX Oncology database at the time of analysis, approximately 13% of patients with SCLC either presented with synchronous BM or were diagnosed with BM prior to identification of their primary cancer. These findings align with existing studies, which report that approximately 10% of SCLC patients present with BM at the time of diagnosis.^4,10 BM were detected later in the disease course for approximately 21% of the cohort. Our estimation of the prevalence of metachronous BM in SCLC is lower than other reported rates in the literature, which range from 17% to 50%. However, prior studies reporting on metachronous BM prevalence are limited to small institutional and autopsy studies, rather than large population-based studies.^11,12 The lower estimates from this study may also be attributable to decreases in intracranial MRI screening 2 years after diagnosis of SCLC.¹³ Prominent United States cancer registries, such as the National Cancer Institute’s SEER and the NCDB, are cross-sectional and only capture synchronous cases, thereby missing nearly 20% of patients with BM. Future updates of these registries could aim to incorporate data on metachronous BM.

Clinical and demographic characteristics in patients with and without a BM diagnosis were also assessed, with the aim of informing screening practices and identifying potential disparities.⁴ Overall, the majority of patients in our cohort presented with stage 4 SCLC, consistent with population-based studies reporting that the majority of SCLC patients are diagnosed with extensive stage disease.^14,15 The presence of extracranial metastases was common in the overall cohort, with nearly one-third of patients receiving a diagnosis of bone and/or liver metastases, and one-eighth receiving a diagnosis of adrenal metastases either at the time of diagnosis of SCLC or later during the disease course. These findings align with prior epidemiological studies identifying contralateral lung, brain, liver, and adrenals as the most common sites of metastases in SCLC, with a notable co-occurrence of BM with adrenal metastases.^1,16 In our study, 25.9% of patients with a BM diagnosis developed adrenal metastases, compared to only 10.1% of patients without a BM diagnosis.

We also identified demographic associations. Patients diagnosed with BM were younger than those without a BM diagnosis, aligning with existing evidence that BM from SCLC tends to be more common among younger patients.¹⁷ One explanation may be that older patients succumb to their primary cancer or comorbid conditions before the detection of BM. Patients diagnosed with BM in our study were also more likely to identify as Black/African American and Native Hawaiian or Other Pacific Islander compared to those without a BM diagnosis. Our findings align with existing literature indicating a higher incidence of synchronous BM in Black/African American populations and support continued initiatives to promote earlier disease detection in this population.⁴ More frequent screening for BM may also be warranted in patients with a personal history of tobacco use/nicotine dependence or a family history of any cancer, which were also more prevalent in the cohort with a BM diagnosis in our study.

Based on the current NCCN recommendations for intracranial screening imaging, our findings suggest an underutilization of intracranial screening imaging at diagnosis in this population.¹³ Notably, we observed no difference in OS between patients who received brain MRI within the first 2 months of SCLC diagnosis and those who did not. While this lack of survival benefit has been demonstrated in prior studies, the detection of asymptomatic BM via surveillance brain MRI is associated with a reduced likelihood of patients experiencing neurologic death, lower odds of neurologic deficits after treatment of BM, and fewer neurosurgical interventions.^18,19 Due to the limitations of TriNetX, we were unable to evaluate the impact of staging brain MRI on these outcomes. However, improving treatment approaches to reduce the spread of SCLC to the brain may be more effective than increasing surveillance. Interestingly, we identified a racial disparity in intracranial screening, with a higher proportion of White patients receiving intracranial screening than nonwhite patients. While these results may reflect limited data availability in the TriNetX database, additional work is warranted to address barriers to accessing intracranial imaging.

The role of long-term intracranial surveillance in patients with SCLC is also supported by our data on the probability of receiving a BM diagnosis over the disease course. We found that across stages, the risk of BM development increases with prolonged survival. This pattern is particularly pronounced in advanced stages of SCLC; however, nearly 20% of patients with stages 1 and 2 SCLC were diagnosed with BM over the study period, which is consistent with prior retrospective studies.^20,21 Likewise, prior literature demonstrates that the brain is the most common site of distant recurrence in patients with advanced SCLC, with 40%–50% developing BM post-palliative chemotherapy.¹ Together, these findings emphasize the continued need for intracranial surveillance for long-term survivors with SCLC. Our findings also support a continued role for PCI in patients diagnosed with early-stage SCLC, though the risk of neurocognitive side effects must be considered.²² In addition to imaging-based screening, promising alternatives, such as cerebrospinal fluid liquid biopsy, are emerging as potential tools to guide follow-up and therapy selection for patients with metastatic disease.^23,24

A secondary aim of our study was to assess management strategies in patients with SCLC and BM. Thoracic surgery is only recommended in stage I SCLC. In our analysis, the overall resection rate across all stages of SCLC was 6.0%, and there were no differences in rates of thoracic surgery between patients with or without BM or other sites. This likely reflects the aggressive nature of SCLC, which often precludes surgical intervention.¹

Given the limited utility of surgery in SCLC, the standard treatment approach for limited-stage SCLC is concurrent chemoradiotherapy.¹ We found that patients with BM received systemic chemotherapy and radiation at higher rates compared to those without BM, including those who had extracranial metastases only. This is likely attributable to the concordance of BM and metastases to extracranial sites, which leads to more aggressive treatment approaches.¹² Patients diagnosed with BM were treated predominantly with platinum-etoposide, which has a well-established role in the management of SCLC.²⁵

Beyond the mainstays of chemotherapy and radiation, targeted therapy and immunotherapy have the potential to improve penetration of the blood–brain barrier, increase specificity for tumor cells, and ultimately improve survival for patients with BM.²⁶ However, patients with BM have been historically underrepresented in landmark clinical trials of emerging SCLC therapies.^27–29 We found that patients with BM diagnoses received tyrosine kinase inhibitors at higher rates than those without BM or with extracranial metastases only. This may be driven by EGFR-mutant non-SCLC that transforms to SCLC.³⁰ The most common immune checkpoint inhibitor was the anti-PD-L1 monoclonal antibody, atezolizumab, which was approved for treatment of stage 4 SCLC in 2019 and may improve survival in patients with non-SCLC and BM.³¹ Interestingly, temozolomide, which is an approved oral therapy for SCLC and has activity in SCLC BM, was received by less than 3% of our BM cohort.³² Future clinical trials should strive to include a larger representation of patients with BM, encompassing symptomatic and untreated cases, to provide more comprehensive insights into tailored treatment approaches for this patient population.

In addition to systemic therapies, patterns in BM-specific therapies were also examined in this study. The mainstay of treatment for BM includes cranial irradiation, with surgical intervention considered for patients with large, symptomatic lesions or when the primary tumor is unknown.²⁵ Patients with BM diagnoses received higher rates of intracranial stereotactic radiosurgery (SRS) than craniotomy or craniectomy. This may also be due to the need for extended recovery following surgical resection relative to SRS, which may delay systemic treatment for SCLC. SCLC is highly radiosensitive compared to other primary cancers, so SRS is often effective without surgical resection and SRS is favored when treating small brain lesions and oligometastatic disease.³³ Furthermore, for patients who receive operative resection of BM, adjuvant SRS targeted to the resection cavity is utilized to treat residual microscopic tumors.³⁴ While WBRT was historically the standard postoperative treatment for surgically resected BM, many recent studies have shown that SRS is often preferred for limited BM due to reduced toxic effects on cognitive function and quality of life.³⁵ Because a CPT code does not exist for WBRT specifically, we were not able to compare the utilization of WBRT versus SRS in our cohort. Another major limitation of our study is that we were unable to specifically assess the prevalence of PCI, which is also part of the standard management in most patients with limited-stage SCLC who respond to initial treatment, due to the lack of a specific billing code for this procedure.¹²

Finally, our study continues to highlight BM as a poor prognostic factor in patients with SCLC. Median survival from the diagnosis of BM was 7 months, which is consistent with prior literature reporting median survival times of 5–8 months after BM diagnosis.^1,36,37 Patients with metachronous BM had a significantly longer survival time than patients with synchronous BM, reflecting earlier detection of primary SCLC in patients with metachronous BM and lead-time bias. Despite this, the chronicity of BM in SCLC did not affect median survival times after a diagnosis of BM was conferred. Similar findings were reported in our prior study comparing the prognosis of synchronous BM to metachronous BM across primary cancers that commonly metastasize to the brain.³ Notably, patients with metachronous BM also had longer survival times compared to patients with no BM diagnosis in this study. This may reflect the eventual development of BM in patients living with limited-stage SCLC who have relatively long survival times. Recent studies show that increased survival due to better management of primary cancers has also increased the incidence of BM.^38,39 Additionally, the American Society of Clinical Oncology (ASCO) and NCCN® guidelines recommend PCI for patients with SCLC who have good initial response to therapy.^9,40 Thus, some patients with metachronous BM may represent a group with good initial response to systemic treatment and prolonged survival who also underwent PCI to delay BM onset. Patients with metachronous BM may also have longer survival because of more thorough monitoring and a greater number of interventions that improve survival. Taken together, these findings support the continued development of early detection methods and treatment strategies to delay BM onset.

While the negative impact of BM on survival is well established, studies comparing survival rates between patients with BM versus extracranial metastases are inconclusive. Studies looking across cancer types generally report decreased survival in patients with BM compared to those with extracranial metastases^3,41; however, some studies in specific primary cancer types report that extracranial metastases confer a worse prognosis than BM.⁴² In our study, there was no significant difference in OS between patients with BM (with or without extracranial metastases) and patients diagnosed with extracranial metastases without BM from the onset of metastatic disease. Numerous factors unique to metastases from SCLC may account for this finding. BM associated with SCLC have been shown to have high initial responsiveness to radiotherapy, though they often relapse and become less responsive to therapy.⁴³ On the other hand, radiation therapy cannot fully address widespread bone and liver metastases due to normal tissue toxicity.⁴⁴ Extracranial metastases may also alter the response to systemic therapies. For example, liver metastases are associated with significantly shorter one-year survival compared to isolated BM in SCLC, which the authors suggest may be due to alterations in the metabolism of cytotoxic drugs commonly used in SCLC management.⁴⁵ Likewise, the development of liver metastasis has been shown to induce immune alterations that may compromise the effectiveness of immune checkpoint inhibitors.⁴⁶ Ultimately, extracranial metastases, like BM, represent a major contributor to mortality in patients with SCLC, and therapies preventing the development of all metastatic diseases from SCLC will be critical to improving survival in these patients.

We recognize limitations intrinsic to the TriNetX Oncology database and other healthcare billing claims-based databases. First, because claims-based databases are used primarily for billing and reimbursement purposes, rather than clinical management, some diagnoses and treatments may be missing or misclassified. Thus, the “no BM diagnosis” group may include patients who had BM but did not receive intracranial imaging; patients who had BM but may have not received the billing code for BM because they were never treated for BM; or patients who were diagnosed and/or treated at outside institutions whose billing was not captured by this database, or lastly those who had BM that were misclassified. The TriNetX Oncology database includes data from 17 centers of cancer care excellence, in which intracranial screening and treatment for BM may be more robust than other treatment centers. However, previous research has demonstrated that Black and Hispanic patients have disproportionately less access to high-quality neuro-oncology centers, which is mirrored by the underrepresentation of Black and Hispanic patients in our cohort.⁴⁷ Nonetheless, our study provides updated estimates of BM rates in SCLC patients, which addresses the limitations of prominent cancer registries in capturing metachronous cases.

Supplementary material

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

Funding

This study received no direct funding and author contributions were done voluntarily.

Conflict of interest statement

C.B.: is a consultant for Depuy-Synthes, Bionaut Labs, Galectin Therapeutics, Haystack Oncology, and Privo Technologies. C.B. is a co-founder of OrisDx and Belay Diagnostics. K.J.R.: Research funding from Accuray, Canon, Elekta AB, icotec; travel expenses from Elekta AB, Accuray, icotec, Brainlab; patent under development with Canon for radiogenomics; data safety monitoring board for BioMimetix.

Authorship statement

Conceptualization: J.R.T.. Methodology: J.R.T, M.P., and K.J.. Analysis: M.P.. Writing—original draft: M.P. and A.K. Writing—review & editing: All authors. Final manuscript approval: All authors.

Data availability

The data used for this study is available from the TriNetX Analytics Network. https://trinetx.com.