📄 medicine

Baseline gene expression and dynamic endocrine therapy use in ER+ breast cancer: associations with recurrence in a real-world cohort

This study suggests that in a real-world cohort of ER+ breast cancer patients, specific baseline gene expression profiles are associated with recurrence risk only in those treated with selective estrogen receptor modulators, whereas switching endocrine therapy is primarily driven by intolerance rather than tumor biology.

Original authors: Veronica Jones, Yongzhe Wang, Christine Quinones, Dana Aljaber, Eva Nelson, Aritro Nath, Irene Kang, Hope Rugo, Lisa Yee, Victoria Seewaldt, Joanne Mortimer

Published 2026-06-30

📖 4 min read☕ Coffee break read

CC BY 4.0

Original authors: Veronica Jones, Yongzhe Wang, Christine Quinones, Dana Aljaber, Eva Nelson, Aritro Nath, Irene Kang, Hope Rugo, Lisa Yee, Victoria Seewaldt, Joanne Mortimer

Original paper licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine breast cancer treatment as a long journey where the goal is to stop a specific type of tumor (Estrogen Receptor-positive, or ER+) from growing back. For decades, doctors have had two main "vehicles" to drive this journey: SERMs (like Tamoxifen) and AIs (Aromatase Inhibitors).

Currently, choosing which vehicle to start with is like picking a car based on the driver's age and how well they handle the steering wheel (menopausal status and side effects), rather than looking at the specific engine of the tumor itself. This study asked: Does the tumor's internal "blueprint" (its gene expression) tell us which car will actually stop the tumor from coming back?

Here is what the researchers found, broken down into simple concepts:

1. The "Blueprint" Matters, But Only for Some Drivers

The researchers looked at 74 patients and analyzed the "blueprints" (RNA genes) of their tumors before treatment started. They wanted to see if certain genes made the tumor more likely to return, depending on which drug was used first.

The Discovery: They found a specific set of five genes (TP53, WNT7B, UBE2T, BAG1, and ACTR3B) that acted like a "danger signal."
The Catch: This danger signal only lit up for patients who started with SERMs. If a patient had high levels of these genes and took a SERM, their risk of the cancer coming back was much higher.
The Contrast: For patients who started with AIs, having these same "danger genes" didn't seem to increase the risk of recurrence. It's as if the AI vehicle had a special shield that neutralized the threat these genes posed, while the SERM vehicle did not.

2. The "Switching" Problem

In the real world, patients often have to change their medication because of side effects, much like a driver switching cars because the first one is too bumpy or uncomfortable.

The Reality: About 41% of the patients in this study had to switch drugs at least once.
The Reason: Most switches happened because of side effects, specifically joint pain and general discomfort.
The Finding: The researchers looked to see if the tumor's "blueprint" predicted who would need to switch. The answer was mostly no. The genes that predicted the cancer coming back did not predict who would get joint pain or have to stop the drug. This suggests that while biology drives the cancer's behavior, the body's reaction to the drug (tolerability) is a separate, messy story that isn't easily read from the tumor's genes yet.

3. The Big Picture

The study concludes that we are currently driving blind regarding the tumor's biology. We pick a drug based on the patient's age and how they feel, but the tumor's internal machinery might actually be better suited for one drug over the other.

The Analogy: Imagine you have a locked door (the cancer). You have two keys (SERM and AI). Currently, you pick a key based on the color of your shirt (menopausal status). This study suggests that for some doors, one key is clearly better than the other, but only if you look at the lock's internal pins (the specific genes) first.
The Limitation: This was a small study (74 people), so these findings are like "clues" or "hypotheses" rather than final rules. The researchers are saying, "We found a pattern here that needs to be tested on a much larger group of people to be sure."

In summary: The paper argues that we should eventually stop choosing breast cancer drugs just based on age and side effects. Instead, we should look at the tumor's genetic "blueprint" to see if it will respond better to one type of drug than the other, potentially preventing the cancer from returning. However, the study also admits that predicting who will suffer from side effects (and need to switch drugs) is still a mystery that the tumor's genes haven't solved yet.

Technical Summary: Baseline Gene Expression and Dynamic Endocrine Therapy Use in ER+ Breast Cancer

Problem Statement
Current clinical decision-making for selecting initial endocrine therapy (ET) in estrogen receptor-positive (ER+) breast cancer relies primarily on menopausal status and tolerability rather than tumor biology. Despite the heterogeneity of recurrence risks within the ER+ subtype, no established biomarker guides the choice between Selective Estrogen Receptor Modulators (SERMs) and Aromatase Inhibitors (AIs). Furthermore, while ET is the cornerstone of adjuvant treatment, up to 30% of tumors exhibit resistance, and treatment is frequently complicated by toxicities (e.g., joint pain, vasomotor symptoms) that drive discontinuation or switching. The interplay between baseline tumor molecular profiles, specific ET classes, recurrence risk, and treatment switching patterns in real-world settings remains poorly understood.

Methodology
The study utilized a retrospective real-world cohort of 74 patients with HER2-negative, ER-positive breast cancer treated at a national cancer center between 2012 and 2016.

Cohort Characteristics: Patients had tumors $\ge$ 2 cm or < 2 cm with biopsy-proven nodal metastasis, $\ge$ 10% ER positivity, and at least 3 years of follow-up. All received adjuvant ET (SERMs, SERDs, or AIs).
Molecular Profiling: Baseline tumor RNA was extracted from FFPE blocks and profiled using a custom NanoString nCounter panel of 145 genes (including PAM50 genes) covering pathways such as ER signaling, proliferation, EMT, PI3K, and immune response.
Statistical Analysis:
- Patients were stratified by initial ET class: AI-initiated (n=49) vs. SERM-initiated (n=25).
- Primary Outcomes: Time-to-recurrence (locoregional or distant disease) and time-to-switch (change in ET class).
- Models: Multivariable Cox proportional hazards models were fitted for each gene, adjusting for baseline prognostic factors (age, BMI, menopausal status, stage, grade, histology). Crucially, interaction terms between gene expression (continuous, standardized) and initial ET type were included to test if gene-recurrence associations differed by therapy class.
- Significance: Interactions were highlighted with a False Discovery Rate (FDR) < 0.05. Sensitivity analyses included unadjusted models and a postmenopausal-only subset.
- Switching Analysis: Patient flow and self-reported intolerance symptoms (e.g., joint pain) at the time of switching were analyzed descriptively.

Key Results

Gene-ET Interactions on Recurrence:
- Among SERM-initiated patients, high baseline expression of five specific genes—TP53, WNT7B, UBE2T, BAG1, and ACTR3B—was significantly associated with a markedly increased risk of recurrence (Hazard Ratios [HR] range: 2.53–7.07).
- In contrast, AI-initiated patients showed no comparable association between these genes and recurrence.
- The interaction terms for these five genes were statistically significant (FDR < 0.05), indicating that the prognostic value of these genes is dependent on the initial ET class. Specifically, high expression conferred a stronger adverse association in the SERM group (HR range for interaction: 5.33–8.09).
- Other genes (e.g., PIK3R5, IGF1, ITPR1) showed nominal associations suggesting lower risk with SERMs, while a broader set of proliferation-related genes (e.g., MELK, CCND1, MKI67) showed stronger adverse associations in the SERM group, though these did not survive multiple-testing correction.
Clinical Characteristics:
- Baseline clinical variables (menstrual status, tumor stage, grade) did not show strong independent associations with time-to-recurrence in this cohort when adjusted for gene expression.
- Menstrual status differed significantly between groups (64% of AI-initiated were postmenopausal vs. 24% of SERM-initiated), reflecting standard prescribing patterns.
Endocrine Therapy Switching:
- 41% of patients (30/74) switched ET at least once.
- Switching was predominantly driven by intolerance rather than disease progression. The most common reported symptoms were joint pain (30% of switchers) and unspecified side effects (37%).
- No consistent clustering of symptom patterns was observed, and gene-level associations with switching intensity were modest and not statistically significant after correction, suggesting switching is driven by complex, individualized tolerability factors rather than a single molecular signature.

Key Contributions

Identification of Differential Biomarkers: The study identifies a specific set of genes (TP53, WNT7B, UBE2T, BAG1, ACTR3B) whose high expression predicts poor outcomes specifically in patients treated with SERMs, but not in those treated with AIs. This challenges the notion of a universal prognostic signature for ER+ breast cancer independent of treatment modality.
Real-World Dynamics: By integrating longitudinal switching data and patient-reported symptoms, the study highlights that real-world ET trajectories are complex and often driven by tolerability, a factor rarely incorporated into current molecular biomarker models.
Mechanistic Hypotheses: The findings suggest that tumors with high expression of these specific genes may possess biological mechanisms (e.g., WNT pathway activation, altered ubiquitination, cytoskeletal remodeling) that render them less susceptible to receptor modulation (SERMs) compared to estrogen deprivation (AIs).

Significance and Claims
The authors frame these findings as hypothesis-generating rather than definitive clinical guidance. The study claims to:

Provide a framework for understanding how tumor-intrinsic molecular features interact with specific ET strategies to shape recurrence outcomes in routine practice.
Motivate the validation of these gene-therapy interactions in larger, independent cohorts.
Suggest that future precision medicine efforts must move beyond static models of endocrine sensitivity. Instead, next-generation signatures should integrate biological determinants of efficacy with susceptibility to adverse effects and adherence patterns to better balance efficacy with quality of life.
Highlight a gap in current literature: existing biomarkers rarely account for the dynamic nature of ET delivery (switching) or the impact of side effects on treatment continuity.

The paper explicitly acknowledges limitations, including the modest cohort size, the observational design (precluding causal inference), and the collinearity between initial ET class and menopausal status, which constrains the full separation of drug-class effects from menopausal-associated biology. Consequently, the authors urge caution in interpreting effect estimates and emphasize the need for validation.

1. The "Blueprint" Matters, But Only for Some Drivers

2. The "Switching" Problem

3. The Big Picture

Technical Summary: Baseline Gene Expression and Dynamic Endocrine Therapy Use in ER+ Breast Cancer

More like this