Flagship’s experience in pancreas and diabetes-related image analysis has an interesting history. Frank Voelker first initiated islet cell image analysis while at Novartis in Cambridge (NIBR) that was presented at a couple of ADA meetings back in 2005 (see the highly referenced Duttaroy A,Voelker F, Zhang X, et al, “The DPP-4 inhibitor vildagliptin increases beta-cell mass in rodents”, Diabetologia (2005);48(suppl 1): p.A178). This is what initially started Dr. Voelker’s interest in using whole section analysis.
Since that time a number of improvements have happened, both in the industry and at Flagship. Dual staining for insulin/glucagon is now more accessible, allowing the quantification of alpha and beta cells.
- Dual staining approaches allow quantification of alpha and beta cell mass (slides courtesy Cellestan)
Histology pattern recognition approaches are quite amenable to pancreas work, both on H&Es and IHCs. A poster at a diabetes meeting in 2008 showed that islets can be reliably detected on H&Es on whole slide sections using histology pattern recognition.
Our approaches the last couple of years have been classic whole section analysis — use pattern recognition to identify all islets on a section, then quantify alpha and beta cell mass or cell counts per islet or per pancreas as a single section count. Sometimes other stains are used, like insulin/ BRDU or Ki67 for beta cell proliferation, or dual stain combinations for ductal proliferation. or other approaches to measure apoptotic cells in the islets. The challenge has been how to extrapolate that to a 3D representation, and we have primarily simply reported endpoints assuming the 2D cross-section is representative of the entire pancreas.
More recently, our sampling approaches been vastly improved by working with Visiopharm, the leader in whole-slide stereology and particularly Dr. Jens Nyengaard and others working with the NewCAST system. We can now offer stereology approaches where sampling is conducted in a way where 2D samples become more representative of the 3D volume. The problem is not trivial, as islets range in sizes throughout the pancreas, and so careful study design must be followed.
We are also applying modified stereology approaches from Visiopharm’s Whole Slide Stereology and newCAST suite to be able to utilize the formation of new islets as an endpoint in research — it requires the ability to differentiate a new islet from sampling the endcap of a larger one.
Finally, we can also multiplex measurements using FACTS in islets, this is extremely interesting and an active area of research.
