The Invisible Scars

How Genomic Bruises Reveal Cancer's Weakness

HRD testing isn't just about genes—it's about reading the chaotic aftermath of a cellular battle.

Article Navigation

Introduction

Every cell faces thousands of DNA breaks daily. Most are repaired flawlessly, but cancers harbor a critical flaw: homologous recombination deficiency (HRD), where cells lose the ability to fix double-strand breaks. This leaves behind unique "genomic scars"—permanent marks of DNA repair gone awry. Detecting these scars predicts whether tumors will respond to life-saving PARP inhibitors or platinum therapy. But how do scientists measure such damage? Abstract 240 (AACR 2023) unveiled a breakthrough: a unified method to quantify HRD scars across three technologies—SNP arrays, next-gen sequencing (NGS), and optical genome mapping (OGM) 1 . This is the story of seeing the invisible.

The Genomic Scars of HRD

When homologous repair fails, cells resort to error-prone backup systems. This genomic chaos leaves three signature scars:

Loss of Heterozygosity (LOH)

Chromosomal regions where one parental copy is deleted or duplicated.

Why it matters: LOH >15 Mb indicates defective HR 5 .

Telomeric Allelic Imbalance (TAI)

Unequal parental alleles stretching to chromosome ends.

The clue: TAI hotspots reveal replication stress in HRD tumors 5 8 .

Large-Scale Transitions (LST)

Chromosome breaks between adjacent segments >10 Mb.

The red flag: High LST counts correlate with BRCA loss 7 .

These scars combine into an HRD-sum score (LOH + TAI + LST). A score ≥42 defines HRD-positivity—a threshold validated in ovarian and breast cancers 3 7 .

The Experiment: Validating a Universal HRD Tool

Abstract 240's core innovation was the scarHRD R package, designed to calculate HRD scores from NGS data and match SNP array accuracy. Here's how it succeeded 2 :

Methodology

  1. Cohort: 139 triple-negative breast cancer patients (TCGA dataset).
  2. Tech Comparison:
    • SNP arrays (Affymetrix 6.0): Gold standard for LOH, TAI, LST.
    • NGS (Whole exome/WXS): Analyzed via scarHRD using Sequenza for copy-number segmentation.
  3. Cutoffs: LOH (>15 Mb), LST (breaks >10 Mb spaced <3 Mb apart), TAI (subtelomeric imbalances).
  4. Correlation Test: Pearson correlation between SNP- and NGS-derived HRD scores.

Results & Analysis

Table 1: Correlation of Scar Signatures Between SNP Arrays and NGS

Signature Pearson (r) R² p-value
LOH 0.73 0.53 <2.2e-16
TAI 0.84 0.70 <2.2e-16
LST 0.79 0.62 <2.2e-16
HRD-sum 0.87 0.75 <2.2e-16

The near-perfect correlation (r=0.87) proved NGS could replace SNP arrays for HRD scoring. Critically, HRD-sum distinguished BRCA-mutated from wild-type tumors with 80.8% accuracy (AUC=0.81) 2 .

[Chart: Correlation between SNP and NGS HRD scores]

Pan-Cancer Insights: HRD's Hidden Landscape

Abstract 240's framework enabled a massive pan-cancer HRD analysis. A 2025 study of 9,262 Asian patients revealed striking patterns 3 7 :

Table 2: HRD Prevalence Across Cancers

Cancer Type HRD-Positive (%) Top Altered HR Genes
Ovarian (OV) 69% BRCA1, RAD51D, PPP2R2A
Lung Squamous (LUSC) 51% BRCA1, RAD54L, BARD1
Stomach (STAD) 32% BRCA2, MRE11
Breast (BRCA) 26% BRCA1, PALB2
Prostate (PRAD) 8% BRCA2, ATM
Key Findings
  • Biallelic hits drive HRD: Tumors with two-hit BRCA1/2 alterations showed 99% HRD-positivity in ovarian cancer.
  • Beyond BRCA: RAD51D, BARD1, and PPP2R2A losses significantly elevated HRD scores 3 7 .
  • Age & Stage Matter: HRD scores rose in metastatic (median=25) vs. primary tumors (median=19) 3 .

[Chart: HRD prevalence across cancer types]

Optical Genome Mapping: The Game Changer

SNP arrays and NGS infer scars indirectly. Optical Genome Mapping (OGM) directly visualizes structural variants in ultra-long DNA molecules 4 9 :

How OGM Works

  1. Extract: Isolate intact DNA molecules >150 kb.
  2. Label: Attach fluorescent dyes to specific motifs (e.g., CTTAAG).
  3. Linearize: Feed DNA into nanochannels, stretching them for imaging.
  4. Analyze: Software detects label shifts, revealing SVs at 500-bp resolution.
Genome mapping visualization

Table 3: HRD Detection Technologies Compared

Method Resolution Key Strength Limitation
SNP Array >5 Mb Low cost, established Misses small SVs
NGS Panels 1-10 kb Integrates mutation + scar data Struggles with repeats
OGM 500 bp Captures balanced SVs & repeats Requires fresh/frozen tissue

OGM excels in detecting "dark genome" regions (e.g., telomeres, centromeres) invisible to NGS. In myeloid cancers, it found 34% more clinically relevant SVs than traditional cytogenetics 6 9 .

The Scientist's Toolkit

Key reagents and software driving HRD scar analysis 2 8 9 :

Tool Function Example/Use Case
scarHRD R Package Calculates LOH, TAI, LST from NGS data Validated on TCGA breast tumors
NxClinical Software Automates scar scoring from SNP/NGS data Used in FOCR HRD Harmonization Project
Bionano Saphyr OGM platform imaging structural variants Detects chromothripsis in AML
Ultra-HMW DNA Kits Isolate long DNA fragments (>150 kb) for OGM Critical for preserving molecule integrity
scarHRD R Package

Open-source tool for comprehensive HRD analysis from NGS data.

R View Documentation
OGM Workflow

Visual representation of optical genome mapping process.

Extraction
Labeling
Imaging
Analysis

Conclusion: The Future of Precision Oncology

Abstract 240's framework marks a turning point. By unifying SNP, NGS, and OGM, it offers a multi-platform solution to quantify HRD—vital for expanding PARP inhibitor use beyond BRCA-mutant cancers. Yet challenges persist: genomic scars reflect past HRD, and repair pathways can reawaken, causing resistance 5 .

Future Directions
  • Combining scar scores with functional assays (e.g., RAD51 foci tests)
  • Real-time HRD snapshots for dynamic monitoring
  • Improved algorithms for scar pattern recognition
  • Expansion of OGM applications in clinical settings

"In the scars of our genome lies the history of a battle—and the blueprint for victory."

The next frontier? Combining scar scores with functional assays (e.g., RAD51 foci tests) for real-time HRD snapshots. As OGM scales and algorithms improve, we may finally decode the full language of genomic scars, turning cancer's hidden weaknesses into cures.

References