There are many times when a pathologist needs to randomly sample tissue in a directed manner. This is to assure that similar tissue substructures are sampled in a manner that prevents biasing of results. If the problem was as straightforward as sampling ANY area, then random sampling algorithms would have already been developed for this. Unfortunately in real life, tissues and organs do not have such homogeneous histology that “directed random sampling” can be dispensed with. Consequently, the pathologist may restrict his/her sampling to centrilobular regions of liver, for example. However, having stated this, there are circumstances where biomarkers may be known to be uniformly distributed across wide regions of a tissue (eg. panlobular in liver) regardless of substructure, and in those cases sampling may be more random and expansive in nature.
It is extremely important to be consistent in sampling tissues across dose groups with regard to anatomy. For example, sampling similar regions of the median lobe of the liver in all animals in a study would provide a greater chance of yielding meaningful results. If multiple samples from a local region reveal similar analytical results for a specific biomarker, then that gives confirmation that sampling procedures may have been correct.
Sample size may also be important, and in some circumstances may depend upon the number or relative rarity of events or target structures within a given area of tissue. For example larger areas (or even entire tissue sections) may need to be evaluated in the case of counting apoptotic TUNL-stained cells, proliferating Ki67 cells or other specific cell types.
Selection of sample region could also be important if there are regional differences in expression of toxicologic changes. For example, since the distribution and/or severity of hepatocellular necrosis may vary in different hepatic lobes, it may be necessary to select a specific lobe of the desired representative intensity for quantitation.
How to do directed random sampling:
We will provide some suggestions using the Aperio ImageScope viewer. The trick is to start with a magnification while viewing a digital slide that is high enough to be able to see that you are looking at the right tissue type (e.g the correct lobe in the live example above, or the cancerous non-necrotic area in the xenograft example below). But not so high that you can see the cells or objects you want to measure, and thus bias your results.
Let’s walk through an example in a xenograft. You want to subsample microvascular architecture in a xenograft in only the cancer areas, avoiding the necrotic and normal regions. You are looking to get an idea of variability through the xenograft. So you first view the image at 2x, and then in ImageScope you go to the Tools –> Options menu, and select the Annotations tab, and set the fixed size regions to 300 x 300 microns. Then holding the Control key down you can drop these “directed random annotations” anywhere, being able to avoid the necrotic areas, but not being able to see the vascular architecture. If you really want to be fancy, Hold the control shift keys down together and you can move them all around a bit together and really get “random”, or just the control key to move one at a time. It is really nicely programmed, and easy for even an old pathologist like me to use. :)
When you run Microvessel on these regions, you will get statistics on each region, and also on the layer (all the regions together) as a whole. Then you can copy the data over to SAS or JMP and run stats on them.
Of course, you could just use Genie and then run it on all the neoplastic non-necrotic regions! Or draw the regions yourself for 2 days.
Share this Post:
Flagship Biosciences Inc.
Flagship Biosciences, the industry leader in tissue image analysis, advances personalized medicine by quantifying and simplifying complex pathology. The company’s comprehensive “fit for purpose” image analysis platform transforms conventional, subjective methods of histopathology with clear actionable data to speed global drug development and approval.