Analysis of cell death in the trochlear nucleus of the chick embryo: calibration of the optical disector counting method reveals systematic bias.

Detection of changes in numbers of neurons is essential for an understanding of neuronal development, function, and death. Optical disector counting is claimed to be the most efficient technique to estimate accurate numbers of neurons in microscopic sections. We calibrated the optical disector by comparison with three-dimensional reconstructions from serial sections and determined how accurate this technique is relative to conventional profile counting methods. The calibration was performed on the trochlear nucleus in developing chicks. Optical disector estimates, when obtained as generally recommended, were about 25% lower than the actual number of neurons. This underestimate was caused by a nonuniform (bimodal) distribution of neuronal nuclei in paraffin and plastic (glycolmethacrylate) sections, but not in cryosections. The density of neurons in the core of the paraffin and plastic sections was substantially lower than in the upper and lower margins of these sections. Accurate estimates of neuronal numbers were obtained with a modified optical disector method that sampled the entire extent of tissue sections. Previous estimates of numbers of trochlear neurons in the developing chick have been controversial. The modified (calibrated) optical disector method revealed that the number of trochlear neurons decreased from about 1,600 at day 8.5 of incubation (embryonic day, [E] 8.5) to about 900 at the time of hatching. Numbers of pyknotic nuclei peaked at E6 and at E9, revealing an additional early, but postproliferative, period of cell death. Taken together, these data emphasize the need for calibration of stereological counting techniques and the need to examine sampling strategies for potential bias.