Abstract

Aims—To arrive at a reproducible sampling technique for counting apoptotic cells in tissue sections of invasive breast cancer that can serve as a protocol for further clinical studies.

Methods—In 4 μm thick haematoxylin and eosin stained tissue sections of 12 breast cancers, apoptotic cells, recognised by strict morphological criteria, were counted in consecutive fields of vision at ×1000 magnification in a marked area in the most poorly differentiated region of tumour. These counts were regarded as the gold standard. Subsequently, in a systematic sampling experiment, the number of fields that had to be counted to derive an acceptable coefficient of variation (CV) was determined. To compare counts at different magnifications, all fields of vision were also counted at ×630 and ×400. The intra- and inter-observer reproducibility was tested by repeated measurements at these magnifications in 10 systematically selected fields of vision.

Results—Apoptosis seemed to be a rare event, affecting, on average, about 1% of tumour cells. Noticeable clustering of apoptotic cells was observed. The systematic sampling experiment showed that at ×1000 magnification, the CV was improved by counting up to 20 fields. When comparing ×400 and ×630 magnifications with the ×1000 magnification, the correlation coefficients were 0.88 and 0.87, respectively. However, the lower magnifications yielded lower counts. With regard to reproducibility, the intra-observer correlation coefficient was 0.91 at ×630 and 0.76 at ×400. The inter-observer correlation coefficient was 0.77 at ×630 and 0.68 at ×400.

Conclusions—Apoptotic cells can be counted readily in haematoxylin and eosin stained tissue sections. However, a systematic sampling protocol must be followed and cells should be counted at a relatively high magnification to obtain acceptable reproducibility. The suggested protocol will permit further correlative and prognostic studies and the monitoring of the effects of treatment.