Neurologic Symptom Phenotypes in Breast Cancer Brain Metastases by Molecular Subtype

Location

Philadelphia, PA

Start Date

17-4-2026 1:30 PM

End Date

17-4-2026 2:30 PM

Description

Purpose

To describe symptom phenotypes in metastatic breast cancer (MBC) brain metastases and examine their relationship to molecular subtype.

Background/Significance

Patients with MBC who develop brain metastases experience substantial symptom morbidity and poor outcomes. While neurologic symptoms are classically understood as reflecting lesion localization, emerging evidence suggests that tumor subtype may influence spatial patterns of central nervous system progression. A formal cluster analysis enumerating symptom phenotypes in MBC brain metastases has not previously been reported.

Methods

A retrospective cohort study leveraged electronic health record (EHR) data from patients treated at an academic cancer center. Molecular subtype, anatomic lesion location, and new-onset neurologic symptoms in the 90 days preceding diagnosis of CNS involvement were extracted and coded as binary indicators across symptom domains. Latent class analysis was used to identify symptom phenotypes. Associations between subtype, lesion localization, and symptom class were examined using likelihood ratio testing and post hoc comparisons.

Findings and Interpretations

A three-class solution best characterized patients’ symptom experiences (N = 425).

  • Class 1 (n = 129) contained proportionally more triple-negative breast cancer (TNBC) disease and was characterized by prominent cognitive, mood, fatigue, and nausea symptoms.

  • Class 2 (n = 185) contained proportionally more HR+/HER2− disease and was characterized by pain, sensory, visual, and balance symptoms.

  • Class 3 (n = 111) contained proportionally more HER2+ disease and was characterized by seizures, motor deficits, and speech disturbances.

Overall, molecular subtype did not independently predict symptom phenotype membership (p = 0.92). Consistent with published literature, TNBC lesions were significantly overrepresented in the frontal lobe (p = 0.031, Cramér’s V = 0.128) and parietal lobe (p = 0.049, Cramér’s V = 0.119), while HER2+ lesions were significantly overrepresented in the cerebellum (p < 0.001, Cramér’s V = 0.188). All associations demonstrated modest effect sizes. Limitations include the high proportion of cases with multifocal disease, edema-related mass effect, and limited granularity of EHR-derived symptom data.

Discussion

Molecular subtype did not independently predict symptom phenotype membership (p = 0.92). Symptom science emphasizes the patient’s symptom experience as a meaningful clinical and biological signal, suggesting that neurologic symptom patterns may reflect complex interactions between lesion location, multifocal disease burden, and secondary effects such as edema or mass effect rather than tumor biology alone. Notably, previously reported subtype-specific spatial patterns of brain metastases were reproduced, indicating that tumor biology may influence lesion distribution without fully determining patients’ neurologic symptom experiences.

For clinicians, systematic symptom assessment may provide clinically meaningful signals for CNS involvement, particularly in settings where advanced neuroimaging is limited. Recognition of symptom patterns may support earlier evaluation for brain metastases and facilitate timely referral for diagnostic imaging and multidisciplinary management. Future studies incorporating prospective symptom assessment and imaging-based lesion mapping may further clarify relationships between tumor biology, lesion distribution, and neurologic symptom expression.

Embargo Period

5-21-2026

Comments

Presented by Marina Petruzzi

This document is currently not available here.

COinS
 
Apr 17th, 1:30 PM Apr 17th, 2:30 PM

Neurologic Symptom Phenotypes in Breast Cancer Brain Metastases by Molecular Subtype

Philadelphia, PA

Purpose

To describe symptom phenotypes in metastatic breast cancer (MBC) brain metastases and examine their relationship to molecular subtype.

Background/Significance

Patients with MBC who develop brain metastases experience substantial symptom morbidity and poor outcomes. While neurologic symptoms are classically understood as reflecting lesion localization, emerging evidence suggests that tumor subtype may influence spatial patterns of central nervous system progression. A formal cluster analysis enumerating symptom phenotypes in MBC brain metastases has not previously been reported.

Methods

A retrospective cohort study leveraged electronic health record (EHR) data from patients treated at an academic cancer center. Molecular subtype, anatomic lesion location, and new-onset neurologic symptoms in the 90 days preceding diagnosis of CNS involvement were extracted and coded as binary indicators across symptom domains. Latent class analysis was used to identify symptom phenotypes. Associations between subtype, lesion localization, and symptom class were examined using likelihood ratio testing and post hoc comparisons.

Findings and Interpretations

A three-class solution best characterized patients’ symptom experiences (N = 425).

  • Class 1 (n = 129) contained proportionally more triple-negative breast cancer (TNBC) disease and was characterized by prominent cognitive, mood, fatigue, and nausea symptoms.

  • Class 2 (n = 185) contained proportionally more HR+/HER2− disease and was characterized by pain, sensory, visual, and balance symptoms.

  • Class 3 (n = 111) contained proportionally more HER2+ disease and was characterized by seizures, motor deficits, and speech disturbances.

Overall, molecular subtype did not independently predict symptom phenotype membership (p = 0.92). Consistent with published literature, TNBC lesions were significantly overrepresented in the frontal lobe (p = 0.031, Cramér’s V = 0.128) and parietal lobe (p = 0.049, Cramér’s V = 0.119), while HER2+ lesions were significantly overrepresented in the cerebellum (p < 0.001, Cramér’s V = 0.188). All associations demonstrated modest effect sizes. Limitations include the high proportion of cases with multifocal disease, edema-related mass effect, and limited granularity of EHR-derived symptom data.

Discussion

Molecular subtype did not independently predict symptom phenotype membership (p = 0.92). Symptom science emphasizes the patient’s symptom experience as a meaningful clinical and biological signal, suggesting that neurologic symptom patterns may reflect complex interactions between lesion location, multifocal disease burden, and secondary effects such as edema or mass effect rather than tumor biology alone. Notably, previously reported subtype-specific spatial patterns of brain metastases were reproduced, indicating that tumor biology may influence lesion distribution without fully determining patients’ neurologic symptom experiences.

For clinicians, systematic symptom assessment may provide clinically meaningful signals for CNS involvement, particularly in settings where advanced neuroimaging is limited. Recognition of symptom patterns may support earlier evaluation for brain metastases and facilitate timely referral for diagnostic imaging and multidisciplinary management. Future studies incorporating prospective symptom assessment and imaging-based lesion mapping may further clarify relationships between tumor biology, lesion distribution, and neurologic symptom expression.