Improved assessment of isolated islet tissue volume using digital image analysis

Abstract

%¹Correspondence should be addressed to: Claudy J.-P. Mullon, Circe Biomedical Inc., 99 Hayden Avenue, Lexington, MA, 02173.\. Accurate and consistent measurement of tissue volume is critical to performing many types of islet research; however, conventional visual determination of isolated islet yields through a microscope is heavily operator dependent. An improved method of islet volume determination using digital image analysis (DIA) was developed to remove operator bias and automate the islet counting process. A series of 140 porcine islet isolations were used to evaluate the DIA method in three separate stages. In Stage 1 (n = 29 isolations), the conventional and DIA methods were correlated with two other independent islet quantitation methods: insulin extraction, and DNA extraction. It was found that volumes determined by DIA correlated more closely with insulin content and DNA content than did conventionally determined volumes. In Stages 2 and 3 (n = 54 and 57 isolations, respectively), it was shown that an increase in the number of fields analyzed by DIA did not significantly improve the quality of the correlations. Inclusion of very small tissue (<50 μm in diameter), which is ignored in the conventional protocol affected yields by less than 10% and did not significantly improve the correlation with insulin or DNA content. Quantitation of isolated islet tissue volume using DIA has been shown to be rapid, consistent, and objective. In the laboratory, use of this method as the standard for islet volume measurement will allow more meaningful comparison of experimental results between centers. In the clinic, its use will allow more accurate dosing of transplanted tissue.

Introduction

A promising therapy for patients with Type I diabetes is the replacement of their lost β-cell function by transplantation of allo- or xenogeneic islets of Langerhans. This type of treatment offers significant advantages over conventional insulin therapy. Most notably, it offers the potential of largely eliminating problems with patient compliance, leading to better blood glucose regulation and fewer of the serious long-term side effects associated with the disease (35). Significant successes have been achieved with whole-pancreas transplants in humans (34), but this approach requires a major operation and life-long chemical suppression of the immune system. An alternate, less invasive approach is to transplant only the insulin-producing pancreatic islets as naked grafts (6); however, immunosuppression is still required. Other approaches, which attempt to circumvent the need for immunosuppression, include the use of immunoprotective devices such as microcapsules 8, 10, 31, diffusion chambers 20, 30, or perfusion chambers 22, 33. Treatment with allografts has been hampered by the scarcity of donor organs, fueling an increased interest in the use of xenogeneic tissue. Pigs are a promising source of islets for β-cell replacement therapy due to their abundance as a food source and the similarity of porcine insulin to human insulin. However, the use of porcine donors also poses special challenges due to the difficulty in isolating large numbers of porcine islets and because of their unique morphology 4, 28.

Regardless of the species from which islets are obtained, the ability to reliably quantitate the amount of tissue being used is critical to properly planning, performing and interpreting experiments, especially when results are compared between laboratories. This is of particular significance in the clinical situation, where dosing of transplanted islets is often adjusted based on patient weight (5). Knowledge of the number of transplanted islets is not sufficient to define a dose because an islet preparation will contain cell clusters of varying sizes. Instead, an accurate measure of the volume of transplanted tissue is required. The need for a common basis in which to express islet quantities has lead to the nearly universal adoption of the islet equivalent (IE) as the standard unit of islet volume. One IE is defined as the volume of a perfectly spherical islet with a diameter of 150 μm (29).

The conventional method for assessing isolated islet yields is to visually perform a quantitation through a microscope using a calibrated eyepiece reticule. The operator must identify, count, and size the islets in a sample of isolated tissue and then convert this information to an overall yield in terms of IE. Because isolated islets are clusters of hundreds or thousands of cells, accurate determination of both their size and their number is important in arriving at a realistic estimation of the volume of isolated tissue. For each islet, the operator is required to use the reticule to estimate an average radius, which is later converted to a volume based on the assumption of sphericity. Because volume varies with the cube of radius, small errors, or biases in radius determination can lead to significant errors in volume estimation.

Expression of islet volume in terms of a number of IE has the advantage of being based on direct observation of the isolated cells and provides an intuitively simple expression of the amount of tissue. However, several alternate methods for quantifying islet yields have been proposed, each of which has advantages and limitations. Assays that quantitate the total protein or DNA content (19) give a representation of the total amount of tissue present; however, neither technique is islet specific and, therefore, impurities in the preparation will skew the results. Extractable insulin content has also been used to express islet yields (16). This technique is β-cell specific; however, it is not clear that the amount of stored insulin can be expected to be constant in islets from different donors, because organ procurement and islet isolation conditions may cause varying degrees of insulin release. Measurement of islet zinc content (17) has been proposed, but again, it has not been shown that zinc content of islets is similar among different pancreases. Quantitation of islet volume using flow cytometry has also been investigated (14), but requires relatively expensive equipment, and the stresses of flow may lead to fragmentation of fragile pig islets.

Our laboratory has used a combination of volume estimation, DNA content, and insulin content to express yields of porcine islet isolations. Use of a combination of techniques makes it possible to avoid some of the limitations inherent in the individual methods when used by themselves. Estimation of islet size and conversion to tissue volume is especially problematic in the case of porcine islets because they are typically fragmented or irregular in shape after isolation. In the case of market-ready donors (∼6 months old), islets in the native gland tend to be loosely structured with a weak or nonexistent peri-insular capsule 7, 38. As a result, islets isolated from young donor pigs are typically small and irregularly shaped 12, 28. This morphology makes it difficult to accurately estimate diameters and exacerbates the problem of operator-introduced bias.

In an effort to achieve more objective and consistent measurements of islet tissue volume (IE), we have recently developed an automated counting and sizing system using digital image analysis (DIA). Isolated islets are identified according to preset parameters and are then objectively counted and sized based on the number of pixels that make up the area they present in the field of view. The DIA system has also been used to automate certain elements of the quantitation process, making it possible to analyze a large number of images in a practical time frame. Increasing the number of images analyzed increases the effective sample size and, therefore, reduces the error introduced by sample variability. DIA techniques are becoming increasingly popular in the field of cytometry 2, 21, 23, 37, 40 due to their consistency, objectivity, and accuracy, as well as their ability to automate tedious cell observation procedures 18, 24, 25, 36. Other investigators 1, 41 have also recently applied DIA methods to the evaluation of isolated islet tissue, an area that presents special challenges due to the complex morphology of the islet cluster.

The goal of any quantitation method is to develop a consistent predictive tool. One way to assess the predictive value of a method is to study how well it correlates with other quantitation techniques. We have examined the correlation between tissue volume estimation and two other independent quantitation methods: DNA extraction, and insulin extraction. The study was performed in three stages. The study presented evaluated the applicability of DIA methods to the quantitation of isolated porcine islets, investigated the effect of increasing the sample size on the quality of the correlation between quantitation methods, and examined the effects of increasing the sample size.

Quantitation of isolated porcine islets using digital image analysis was shown to be superior to the conventional visual method in terms of objectivity, ease of use, and correlation with other quantitation methods. Adoption of this technique by the wider islet transplant community would constitute a significant advance in the ability to perform and compare results from experiments and clinical studies.