Basal breast cancer metastases show higher genomic radioresistance than matched primary tumors

In 15 paired basal breast cancers, genomic radiosensitivity index increased significantly in metastases, challenging assumptions of uniform radiosensitivity across tumor sites.

KEY POINTS

  • This transcriptomic study from Institut Curie, France, evaluated whether genomic radiosensitivity differs between primary breast tumors and metastatic deposits. The discovery dataset comprised 125 RNA-sequenced samples from 53 patients in the AURORA US cohort, including 44 primary tumors and 81 metastatic samples. An independent validation dataset from the TEX trial included 119 breast cancer deposits from 112 patients. Radiosensitivity was estimated using the ten-gene Radiosensitivity Index (RSI) and the derived Genomic-Adjusted Radiation Dose (GARD) framework.
  • Overall, genomic radiosensitivity did not differ significantly between primary tumors and metastases in paired analyses. Among 39 paired patients, median RSI was 0.466 in primary tumors versus 0.530 in metastatic samples, without a significant paired difference (p=0.169). RSI values were not significantly correlated between paired primary and metastatic specimens (Spearman ρ=0.19; p=0.234). In 22/39 patients (56.4%), metastatic RSI was higher than primary-tumor RSI, suggesting substantial individual heterogeneity rather than a consistent cohort-wide shift.
  • Basal breast cancers showed a significant increase in RSI after metastatic dissemination. Among 15 paired basal-subtype cases, 13 (86.7%) had higher RSI in metastatic deposits than in matched primary tumors, with a median paired increase of 0.195 (p=0.018). Because higher RSI indicates lower predicted intrinsic radiosensitivity, this pattern suggests relative genomic radioresistance in metastatic basal tumors. No significant paired RSI differences were observed in the HER2-enriched or Luminal A subgroups, although these included only six and four pairs, respectively.
  • Anatomical metastatic location was associated with RSI in the independent validation cohort. In AURORA, RSI differences across metastatic sites did not reach statistical significance (p=0.349). However, the TEX cohort demonstrated significant site-dependent variation (p=0.030), which persisted after adjustment for PAM50 subtype (p=0.040). Soft tissue metastases generally exhibited relatively high RSI values, while Luminal A metastatic samples showed comparatively low RSI values. These findings support heterogeneity in transcriptomic predictions but do not demonstrate differences in actual radiation response.
  • Paired nodal analyses suggested potentially lower predicted biological effect in regional lymph node metastases. Only seven patients had matched primary tumors and lymph node metastases, of whom five (71.4%) showed higher RSI in nodal disease. The difference was not statistically significant (p=0.353). For 40.05 Gy in 15 fractions, median GARD was 22.1 in primary tumors versus 16.2 in nodal metastases. Among six basal-subtype patients, median GARD was 23.6 versus 15.4; four of six had primary-tumor GARD above a previously proposed threshold of 21 while matched nodal GARD remained below it.
  • Exploratory GARD estimates varied with fractionation and tumor sampling. For 26 Gy in five fractions, median GARD was 17.7 in primary tumors versus 13.8 in nodal metastases, without a significant paired difference (p=0.272). The previously proposed GARD threshold of 21 has not been validated for ultrahypofractionation, limiting interpretation. In an additional analysis, primary specimens obtained after systemic treatment had higher median RSI than pretreatment specimens (0.574 versus 0.306; p=0.009), although this unadjusted association cannot establish treatment-induced radioresistance.
  • The study does not demonstrate that nodal metastases require higher radiation doses. The paired primary–nodal cohort contained only seven patients, and RSI/GARD values were calculated from bulk transcriptomic measurements rather than directly observed clonogenic responses. No patient-level radiotherapy outcomes, recurrence patterns, or treatment-response endpoints were available to validate the modeled differences. The assumption of a fixed β parameter and the limited applicability of existing GARD thresholds to hypofractionated regimens further constrain clinical interpretation.

CLINICAL TAKEAWAY

Genomic radiosensitivity estimates may differ between primary breast tumors and metastatic deposits, particularly in basal-subtype disease, suggesting that a single tumor specimen may not represent every irradiated compartment. However, paired nodal data are sparse, and the study provides no direct evidence that dose escalation or individualized regional nodal prescriptions improve outcomes. These findings support further investigation of multi-compartment radiogenomic profiling rather than immediate changes to breast or nodal radiotherapy planning.

SOURCE

Clinical Oncology

Browse more research Suggest a correction