Development of an evidence-based approach to external quality assurance for breast cancer hormone receptor immunohistochemistry: comparison of reference values

Arch Pathol Lab Med. 2011 Jul;135(7):874-81. doi: 10.5858/2010-0380-OAR1.1.

Abstract

Context: External quality assurance and proficiency testing programs for breast cancer predictive biomarkers are based largely on traditional ad hoc design; at present there is no universal consensus on definition of a standard reference value for samples used in external quality assurance programs.

Objective: To explore reference values for estrogen receptor and progesterone receptor immunohistochemistry in order to develop an evidence-based analytic platform for external quality assurance.

Design: There were 31 participating laboratories, 4 of which were previously designated as "expert" laboratories. Each participant tested a tissue microarray slide with 44 breast carcinomas for estrogen receptor and progesterone receptor and submitted it to the Canadian Immunohistochemistry Quality Control Program for analysis. Nuclear staining in 1% or more of the tumor cells was a positive score. Five methods for determining reference values were compared.

Results: All reference values showed 100% agreement for estrogen receptor and progesterone receptor scores, when indeterminate results were excluded. Individual laboratory performance (agreement rates, test sensitivity, test specificity, positive predictive value, negative predictive value, and κ value) was very similar for all reference values. Identification of suboptimal performance by all methods was identical for 30 of 31 laboratories. Estrogen receptor assessment of 1 laboratory was discordant: agreement was less than 90% for 3 of 5 reference values and greater than 90% with the use of 2 other reference values.

Conclusions: Various reference values provide equivalent laboratory rating. In addition to descriptive feedback, our approach allows calculation of technical test sensitivity and specificity, positive and negative predictive values, agreement rates, and κ values to guide corrective actions.

MeSH terms

  • Breast Neoplasms / diagnosis*
  • Breast Neoplasms / metabolism
  • Carcinoma / diagnosis*
  • Carcinoma / metabolism
  • Evidence-Based Practice
  • Female
  • Humans
  • Immunohistochemistry / standards*
  • Quality Control
  • Receptors, Estrogen / analysis*
  • Receptors, Estrogen / metabolism
  • Receptors, Progesterone / analysis*
  • Receptors, Progesterone / metabolism
  • Reference Values
  • Sensitivity and Specificity

Substances

  • Receptors, Estrogen
  • Receptors, Progesterone