Video summary

Puede una gota de sangre revelar un cáncer? El reto de los biomarcadores. Dra. Marcela Esquvel

Main summary

Key takeaways

Science and Nature

Scientific concepts, discoveries, and nature/biological phenomena

Breast cancer background and clinical challenge

  • Cancer definition & progression

    • Breast cancer is a neoplasm of breast tissue (e.g., ducts, lobules producing milk).
    • It arises when abnormal cells proliferate uncontrollably, can mutate, and may metastasize (spread to brain, bones, lungs, liver).

  • Epidemiology and survival dependence on stage

    • Breast cancer is a leading cancer in women globally.
    • Timely diagnosis is emphasized by WHO as a key determinant of survival.
    • Survival drops sharply when cancer is diagnosed at advanced/metastatic stages.

  • Current diagnostic methods

    • Mammography (screening; often for women ≥40)
    • Ultrasound
    • Thermography
    • Magnetic resonance imaging
    • Biopsy (gold standard; also enables molecular marker-based treatment guidance)

  • Limitation motivating liquid biopsy

    • When screening/diagnosis is delayed or access is limited, more cases are found at advanced stages.
    • The proposed approach is to search for tumor “traces” in blood as an alternative to direct tumor imaging.

Liquid biopsy: “tumor window” in blood

Blood can reflect tumor activity via circulating biomarkers such as:

  • Circulating tumor DNA (ctDNA): tumor-released DNA fragments in blood.
  • MicroRNAs: small RNAs involved in regulation of gene expression.
  • Extracellular vesicles: membrane “envelopes” carrying biochemical messages between cells.
  • Metabolites and metabolism-related proteins.
  • Autoantibodies: antibodies generated by the immune system against tumor-associated antigens (a natural immune response).

Types of biomarkers (functional categories)

  • Prognostic biomarkers: estimate likelihood of developing cancer.
  • Diagnostic biomarkers: indicate presence of cancer and sometimes cancer type.
  • Predictive biomarkers: forecast whether a therapy will work.
  • Pharmacodynamic/response biomarkers: indicate whether treatment is working.
  • Recurrence biomarkers: indicate whether cancer returns after treatment.

Why single-biomarker blood tests have failed (key scientific bottlenecks)

  • Low sensitivity/specificity in early stages

    • Example: CA 15-3 (K153) is clinically used but performs poorly for early detection:
      • ~7% sensitivity in stage 1
      • improves later (higher sensitivity at advanced/metastatic stages)
    • Even with high specificity thresholds, many true cases are still missed.

  • “Area under the curve” (AUC) ≠ real-world clinical performance

    • AUC reflects separation in selected study conditions, but does not guarantee diagnostic utility in real populations.
    • Real patients include:
      • other diseases/benign conditions
      • multiple cancers
      • different ages, medications, comorbidities, and genetic backgrounds

  • Translational gap

    • Promising lab biomarkers often fail to become clinical standards due to:
      • insufficient standardization
      • small sample sizes
      • inadequate controls
      • overly simplistic comparisons (e.g., cancer vs healthy only)
      • biological noise and assay variability

Case study presented: autoantibody biomarkers in breast cancer

A multi-step experimental/validation pipeline was used:

  • Goal: evaluate candidate autoantibodies as biomarkers for breast cancer using rigorous validation.

Phase 1: discovery + proof of concept

  • 2D electrophoresis to generate protein “dot maps”
  • Use patient sera to determine which proteins antibodies recognize
  • Select proteins uniquely recognized by breast cancer sera
  • Protein identification by mass spectrometry
  • Confirm identity using commercial antibodies
  • Initial candidates (3):
    • anti-PCA6
    • anti-RP94
    • anti-RP75

Phase 2: validation (Western blot-style reactivity / purified proteins)

  • Test purified proteins (PCA6/GRP75/GRP94 depending on assay identifications)
  • Result:
    • Two candidates were effectively ruled out due to lack of discriminatory antibody signals across groups
    • Remaining candidate:
      • anti-PCA6 with initially reported performance (e.g., ~90% specificity, AUC ~0.96 in breast cancer vs the relevant comparison group at that stage)

Development of a quantitative assay (ELISA)

  • Serum titration to quantify antibody levels
  • Build an ELISA to measure antibody intensity against PCA6

Key failure point: real-world-like validation

  • When additional relevant groups were included—especially other cancers—diagnostic separation degraded:
    • PCA6 antibody levels overlapped strongly between groups
    • The marker lost specificity and AUC effectiveness (“total overlap” and lost diagnostic capacity)

Why the signal disappeared (hypothesis)

  • Assay-dependent epitope masking
    • In Western blot formats, protein epitopes may be more accessible.
    • In ELISA, capture antibodies and immobilization can mask epitopes, reducing detectability.

Conclusion from the case

  • A biomarker can appear promising early but collapse under broader, more realistic comparisons.
  • Therefore, single-marker diagnosis from a blood drop is not yet reliable.

Toward solutions: multi-omics + machine learning

  • Single molecules are insufficient due to:

    • tumor and patient heterogeneity
    • immune response variation (including immune escape/editing discussed later)

  • Proposed future strategy:

    • Multi-omics panels combining multiple biomarker classes, e.g.:
      • DNA (including methylation signals)
      • vesicle content (e.g., vesicle-associated microRNAs)
      • autoantibodies
      • metabolites and other molecular fingerprints
    • Integrate clinical variables with biomarker panels
    • Use machine learning / AI to identify diagnostic signatures from combined data

Liquid biopsy already in use (mainly for monitoring)

While diagnostic blood biomarkers for general cancer screening are not yet clinically ready, ctDNA-based liquid biopsy is already used for:

  • Minimal residual disease detection
  • Assessing response to treatment
  • Other oncology monitoring tasks

Tumor immunoediting (raised in Q&A)

  • Concept: Tumors undergo immune editing, changing immunogenicity over time.
  • Implication: Antibody profiles may shift as tumor clones change.
  • Talk’s implied focus: rather than deeply modeling immunoediting in the biomarker study, the work emphasized detectability under controlled conditions and validation across stages/groups.

Methodology / workflow explicitly described (autoantibody case study)

  • Biomarker development strategy
    • Discovery of candidate autoantibodies
    • Proof-of-concept validation
    • Clinical validation in progressively more diverse cohorts
    • Assay development:
      • discovery via 2D electrophoresis protein dot maps
      • protein identification by mass spectrometry
      • confirmation using commercial antibodies
      • quantitative testing via ELISA after serum titration
    • Statistical evaluation using:
      • AUC/ROC-like curves
      • sensitivity/specificity under different thresholds
      • validation with more clinically realistic comparator groups (including other cancers)
    • Exploratory modeling:
      • machine learning decision tree using biomarkers + clinical variables
      • assessment of whether one marker provides true diagnostic value vs only partial informational value

List of researchers / sources featured (named in subtitles)

Featured speakers / researchers

  • Dr. Marcela Esquivel Velázquez (biotechnological engineer; biomedical sciences doctorate; breast cancer immunoproteomics)
  • Dr. Eduardo Liseaga (medical science researcher; proteomics, metabolomics; ceric biomarker validation; non-caloric sweeteners research)

Additional named author referenced

  • Mukergi (referenced via The Emperor of All Maladies)

Original video