Reproducibility of microvessel counts in breast cancer specimens

L P Marson, K M Kurian, W R Miller, J M Dixon

Research output: Contribution to journalArticle (Academic Journal)peer-review

14 Citations (Scopus)


Assessment of tumour vascularity in core biopsy specimens may be a useful predictor of response to primary therapy. This study addresses practical methodological issues regarding accuracy of tumour vascularity assessments in different breast cancer specimens. Issues addressed in the study are variation caused by (i) inherent observer variation in the method, (ii) tumour heterogeneity and (iii) previous surgical manipulation of tumours. Microvessel counts were performed by two observers on separate occasions and by two different observers. Counts were performed on core biopsies and tumour sections taken simultaneously (n = 16) and with an intervening time interval (n = 21). In addition core biopsies were obtained from the same tumour on two separate occasions (n = 10). A highly significant correlation was found in counts performed by the same observers at different times and between two different observers. No significant correlation was found in counts of core biopsies and tumour sections taken either simultaneously or subsequently. No correlation was found between counts of sequential core biopsies. Study findings suggest that, although microvessel counts may be assessed reproducibly by the same and different observers, counts performed in core biopsies do not accurately reflect those of overall tumour, limiting their potential as predictive or prognostic markers.
Original languageEnglish
Pages (from-to)1088-93
Number of pages6
JournalBritish Journal of Cancer
Issue number6
Publication statusPublished - Nov 1999


  • Reproducibility of Results
  • Humans
  • Breast
  • Adult
  • Prognosis
  • Breast Neoplasms
  • Biopsy
  • Observer Variation
  • Neovascularization, Pathologic
  • Female


Dive into the research topics of 'Reproducibility of microvessel counts in breast cancer specimens'. Together they form a unique fingerprint.

Cite this