Causal interpretation requires appropriate study d

Authors: S J Richardson A Willcox A J Bone A K Foulis N G Morgan

Publish Date: 2009/04/29

Volume: 52, Issue: 7, Pages: 1452-1453

PDF Link

Abstract

To the Editor We fully agree with P C Priest 1 that the validity of controls is critical for the correct interpretation of the immunohistochemical data and can confirm that careful attention was paid to this aspect of our study 2 In particular we would emphasise the following pointsFirst the number of pancreas samples available from human patients who died shortly after receiving a diagnosis of type 1 diabetes is extremely small and our collection represents by far the largest sample worldwide As such the samples constitute a unique resource for the analysis of the molecular pathology of type 1 diabetes in man The pancreases used in the study were collected from autopsies carried out at a wide range of hospitals across the UK over a 25 year period dating from 1960 and the collection has previously been described in detail for example see 3Second because of their varied origin a nonuniform interval between death and post mortem and variations in the methodology used to fix the specimens it was necessary to select a wide range of nondiabetic groups to accommodate and control for these variablesThird the issue of neonatal controls is critically important and contrary to the suggestion of P C Priest this group is absolutely required for correct interpretation of the data For immunohistochemical analysis negative controls are at least as important as positive controls for determining the specificity of immunostaining and we have employed both Thus we have been able to demonstrate that the islet cells of a group of neonates who were infected with cultureproven enterovirus stained positively for vp1 thereby revealing that the virus has tropism for these cells in neonates By contrast the islets of a second group of neonates who should not have encountered the virus were uniformly negative Hence this provides important confirmation that the immunostaining pattern for vp1 seen in the islets of human patients correlates with the expected pattern of infection As a result this helps to exclude the possibility that the antiserum was staining a nonviral antigen in the endocrine pancreasFourth as noted in our analysis approximately 10 of control pancreases spanning a wide range of ages displayed some evidence of vp1 immunopositivity in the islet cells This occurred at a much lower frequency in terms of the proportion of infected beta cells than was seen in patients with recentonset type 1 diabetes but nevertheless these results imply that enteroviral infection of beta cells is not uncommon in humans As such it is clear that the majority of such infections do not result in autoimmune diabetes Rather it appears that in a certain subgroup of susceptible individuals enteroviral infection can initiate changes within the beta cells that ultimately provoke an autoimmune response It is now important to characterise these changes in molecular terms in order to understand the role of persistent enteroviral infection in mediating beta celldirected autoimmunity in such individuals

Keywords: