Krag et al., NEJM 1998 — “The Sentinel Node in Breast Cancer: A Multicenter Validation Study.”

  • What did they asked?
    • Can a radioisotope-guided sentinel lymph node (SLN) biopsy accurately predict the status of the axilla in breast cancer across multiple surgeons and practice settings? 
  • Methods:
    • Population and setting:
      • 443 women with breast cancer:
        • Treated by 11 surgeons at varied centers
      • Everyone underwent complete axillary lymph node dissection (ALND) so SLNB performance could be verified
    • Technique:
      • Peritumoral / pericavity injection of:
        • 4 mL technetium-99m sulfur colloid (≈1 mCi / 37 MBq):
          • Intraoperative hand-held gamma probe to find “hot spots,” then SLN excision followed by ALND
  • Key performance metrics:
    • Identification (“hot-spot”) rate:
      • 93% (413 / 443)
    • Accuracy vs full axillary histology:
      • 97% (392 / 405)
    • Sensitivity:
      • 89% (101/114):
        • False-negative rate ≈ 11% among node-positive cases
      • Specificity / PPV:
        • 100% (no false positives)
      • NPV:
        • 96% (291/304)
    • Anatomic insight:
      • SLNs were outside the axilla in 8% and outside level I in 11%
      • 3% of positive SLNs were in non-axillary locations
  • Why it mattered?
    • Provided the first large, multicenter, surgeon-diverse validation:
      • Showing SLN biopsy is a highly accurate predictor of axillary status:
        • Establishing the foundation for replacing routine ALND in cN0 patients and enabling the de-escalation pathway:
          • Later confirmed by trials like NSABP B-32 and Z0011
  • Nuances and caveats the paper raised:
    • Learning curve / variability:
      • Success varied by surgeon and patient factors, underscoring the need for technique standardization and training
    • Technique scope:
      • Study used radioisotope only (no blue dye in the protocol):
        • Which many centers later combined with blue dye to further enhance identification and lower FNR:
          • Background reviews consistently show higher detection and lower FNR with combined mapping

Leave a comment