Adaptive Molecular Subtyping in Treated Breast Cancer: Reporting Paired Tissue and Plasma Findings

In a single-center, retrospective cohort of 32 patients, investigators reported therapy-associated biomarker and genomic shifts seen through longitudinal profiling of tumor tissue and plasma, supporting a model they describe as “adaptive molecular subtyping” rather than proposing a new formal classification system.

The team assembled a 32-patient, single-institution dataset with longitudinal sampling to compare baseline diagnostic material with post-treatment assessments when available, with paired plasma testing when available. The reporting emphasis is within-patient molecular “drift” under treatment pressure across compartments, rather than relying on a single diagnostic snapshot. As presented, subtype-defining signals may change measurably over time and may not be fully captured by baseline testing alone.

The authors describe three specimen categories: diagnostic core biopsy tissue, surgical resection specimens after therapy when available, and plasma collected for cell-free DNA/circulating tumor DNA evaluation. Tissue profiling included immunohistochemistry for ER/PR, HER2 immunohistochemistry scored as 0/1+/2+/3+ (with SISH recorded for equivocal IHC 2+), and Ki67 recorded as the percentage of positively stained tumor cells. All 32 diagnostic core biopsy specimens underwent baseline tissue next-generation sequencing using the Oncomine Comprehensive Assay Plus (Ion Torrent). For longitudinal comparisons, receptor conversion was defined as a change in ER and/or PR status between the core biopsy and surgical specimen, HER2 score variability as any change in IHC score between time points, and Ki67 variation as an absolute within-patient change in percentage points (ΔKi67 = surgical minus baseline). They also describe “boundary crossing” for HER2 as shifts that move a case across a clinically relevant categorization threshold (e.g., 3+ versus non–3+), aligning definitions with their focus on within-individual change.

Within the subset with paired baseline biopsy and post-treatment surgical material (n=18), the authors report multiple forms of tissue biomarker drift. Progesterone receptor discordance was observed in 6/18 paired comparisons (33.3%) and was described as strictly directional loss. HER2 IHC score discordance was reported in 6/18 cases (33.3%), and boundary crossing to or from HER2-positive status was reported in 2/18 cases (11.1%). Ki67 remodeling was reported in 14/18 cases (77.8%), with a median within-patient ΔKi67 of −5 percentage points. In aggregate, these paired-tissue findings were presented as post-treatment phenotypic variability affecting subtype-defining markers within this cohort.

Plasma testing added a second compartment with different feasibility constraints and a distinct mix of detected alterations. The authors report that plasma ctDNA results were considered informative in 17/32 patients (53.1%), while 8/32 tests (25.0%) were reported as invalid and 7/32 (21.9%) were unavailable or missing, reflecting limitations of sample/assay yield in this dataset. Among the full cohort, they describe plasma-only, clinically relevant or biologically meaningful findings in 2/32 cases (6.3%), including an FGFR1 amplification detected in plasma and BRCA1/2 variants reported in plasma in a case where pre-treatment tissue molecular evaluation was unavailable. They also note interpretive constraints specific to ctDNA, including frequent non-definitive variant classification (e.g., UPV/VUS) and the need to read plasma results in the context of assay validity and available tissue phenotyping, positioning plasma as complementary rather than substitutive in their reporting framework.

Across results and discussion, the authors connect these longitudinal, multi-compartment observations to an adaptive model that treats subtype assignment as time- and treatment-dependent rather than fixed at diagnosis. They emphasize that discordance across measurements may reflect tumor heterogeneity, sampling differences, and biological evolution under therapy.

Stated limitations include the retrospective, single-center design, modest cohort size, incomplete availability of paired surgical specimens, and constraints on ctDNA informativity and variant classification, with calls for prospective validation in larger cohorts with standardized sampling. The authors’ overall conclusion is that breast cancer molecular profiles can be dynamic under therapy, and that longitudinal assessment across tissue and blood can document these shifts over time.

Key Takeaways:

• In the paired baseline–surgery subset, the authors observed post-treatment changes in subtype-defining tissue markers, including PR loss, HER2 IHC score shifts, and frequent Ki67 remodeling.
• Plasma ctDNA was informative in just over half of patients in this cohort, and a small number of alterations were reported only in plasma (including FGFR1 amplification and BRCA1/2 variants).
• The authors frame these multi-compartment, longitudinal findings as support for an adaptive molecular subtyping concept and describe the need for prospective studies to validate and operationalize longitudinal profiling.