Preservation and storage
One issue is preservation after sectioning. While the production of large batches of microarray sections is most efficient, it raises a separate problem of antigenic loss due, presumably, to tissue oxidation. While this may not be the case if you use ZN-Formalin, regular buffered formalin is more commonly used and does not prevent oxidation after sectioning. Others and we have found loss of antigenicity if sections are stored for as little as a week prior to immunostaining. Work is underway to quantify this loss. It appears that loss is an oxidative process since the loss appears to be insensitive to storage temperature or retrieval conditions. For tissue microarrays, we have found that this loss can be prevented by sectioning without water (using the tape transfer system), removal of the degreasing agent after tape release by a short incubation in xylene, and finally re-coating the slides in paraffin prior to storage and storage in a nitrogen safe. We have also begun experimenting with preservatives (BHA or BHT) in the paraffin.
Are the small histo-spots representative of a whole section?
The major potential limitation of this technique is tissue volume. Skeptics claim that the amount of material analyzed is too small and potentially not representative of the entire tumor. In a reproducibility study, the Sauter group found identical Kaplan Meier curves were generated by analysis of 4 unique sets of spots from the same patients (Torhorst et al, AJP 2001). Thus although any given histo-spot may be negative on a given array, the statistical power of analysis of hundreds or thousands of cases eliminates the affect of variability of a single data point in the ultimate conclusions. Our own study to assess the number of histo-spots required to obtain an equivalent result to a tissue section using the standard breast cancer prognostic markers (Estrogen and progesterone receptors and HER2 oncogene) shows that analysis of only 2 histo-spots results in >95% accuracy (see figure in Camp et al, Lab Invest, 2000). More recently, numerous other studies have found similar results.