Ultrastructural differences between diabetic and idiopathic gastroparesis

Abstract The ultrastructural changes in diabetic and idiopathic gastroparesis are not well studied and it is not known whether there are different defects in the two disorders. As part of the Gastroparesis Clinical Research Consortium, full thickness gastric body biopsies from 20 diabetic and 20 idiopathic gastroparetics were studied by light microscopy. Abnormalities were found in many (83%) but not all patients. Among the common defects were loss of interstitial cells of Cajal (ICC) and neural abnormalities. No distinguishing features were seen between diabetic and idiopathic gastroparesis. Our aim was to provide a detailed description of the ultrastructural abnormalities, compare findings between diabetic and idiopathic gastroparesis and determine if patients with apparently normal immunohistological features have ultrastructural abnormalities. Tissues from 40 gastroparetic patients and 24 age- and sex-matched controls were examined by transmission electron microscopy (TEM). Interstitial cells of Cajal showing changes suggestive of injury, large and empty nerve endings, presence of lipofuscin and lamellar bodies in the smooth muscle cells were found in all patients. However, the ultrastructural changes in ICC and nerves differed between diabetic and idiopathic gastroparesis and were more severe in idiopathic gastroparesis. A thickened basal lamina around smooth muscle cells and nerves was characteristic of diabetic gastroparesis whereas idiopathic gastroparetics had fibrosis, especially around the nerves. In conclusion, in all the patients TEM showed abnormalities in ICC, nerves and smooth muscle consistent with the delay in gastric emptying. The significant differences found between diabetic and idiopathic gastroparesis offers insight into pathophysiology as well as into potential targeted therapies.


Introduction
Gastroparesis is a disorder characterized by chronic delay in gastric emptying in the absence of obstruction. Gastroparesis is most commonly associated with diabetes (both type 1 and type 2) or is of unknown cause (idiopathic) [1,2]. Idiopathic and diabetic gastroparesis are assumed to have very different pathophysiology but presently there are no known cellular abnormalities that can separate the two disorders.
Most of our understanding of the cellular basis for gastroparesis has come from animal studies focused on diabetic gastroparesis [3][4][5][6][7][8][9][10] with limited human studies [11][12][13][14][15][16][17][18][19][20][21]. The lack of understanding of the cellular aetiology of gastroparesis is a significant limitation to developing targeted therapy. In response to the limited information known about gastroparesis and the limited therapies currently available, the National Institutes of Health established a Gastroparesis Clinical Research Consortium (GpCRC). As part of this consortium 40 full thickness gastric body biopsies (20 diabetic, 20 idiopathic) were collected from patients with diabetic and idiopathic gastroparesis and 20 site-, age-and sex-matched controls and then studied by immunohistochemistry [22]. The main finding of that study was that 83% of these patients had a cellular abnormality. The most common findings were loss of interstitial cells of Cajal (ICC), changes in enteric nerves and an immune cell infiltrate. None of the cellular abnormalities differentiated diabetic and idiopathic gastroparesis. Also in 17% of gastroparetics, no abnormality was found by light microscopy. In the same study, TEM examination was conducted on a limited number of the patients chosen because of their ICC and nerve abnormalities which confirmed the immunohistochemical findings [22].
The aim of the present study was to use TEM to provide a detailed description of the ultrastructural abnormalities in patients with diabetic and idiopathic gastroparesis, determine if there are distinguishing features at the ultrastructural level between diabetic and idiopathic gastroparesis and determine if patients with apparently normal immunohistological features have ultrastructural abnormalities.

Materials and methods
We used tissue from 60 full-thickness gastric biopsies, 58 of which had been previously studied by immunohistochemistry [22]. In the previous study, we did not have tissue processed for electron microscopy on two patients and therefore two new patients for whom we had both immunohistochemistry and electron microscopy processed tissue were added. Tissue was collected from the anterior aspect of the stomach, midway between the greater and lesser curvatures where the gastroepiploic vessels meet. The anatomy of individual stomachs varies but, in general, the region where the gastroepiploic arteries meet is about 9 cm proximal to the pylorus. Tissues were obtained from 20 diabetic and 20 idiopathic gastroparetic patients undergoing surgery for placement of a gastric electrical stimulator and from 20 age-and sex-matched patients undergoing gastric bypass surgery for obesity (Table 1) following IRB approved protocols. A technician in the operating room ensured that ischaemic time was minimized and the tissue properly cut, stored and transported. The control tissue was selected from patients that did not have diabetes, did not report gastrointestinal symptoms and had a screening H and E slide read as normal. To determine if the 20 controls used were different from non-obese controls, thus causing any significant histological changes, we prospectively collected full thickness gastric biopsy specimens from 4 non-obese, non-gastroparetic subjects [gastro-esophageal junction cancer with no previous weight loss (Ͻ5 lbs), and no previous chemotherapy or radiation therapy] and compared histological changes with the above control group.
For electron microscopy, four strips for each patient, 1 mm ϫ 10 mm long and containing the muscularis propria plus a small portion of the tunica submucosa, were immediately cut after the full thickness biopsy was obtained and fixed for 6 hrs in a solution of 2% glutaraldehyde 0.1M in cacodylate buffer, pH 7.4. After four rinses in the cacodylate-buffered solution containing 0.22M sucrose, the strips were post-fixed for 1 hr in 1% OsO4 in 0.1M phosphate buffer. After a rinse for 30 min. in ddH2O, the strips were en bloc stained in 2% uranyl acetate for 30 min. at 55ЊC and rinsed again in ddH2O for 10 min. Dehydration was carried out in graded ethanol and the strips then embedded in Spurr using flat moulds to obtain full-thickness sections with the circular muscle cut in cross-section. Semi-thin sections, obtained with an LKB NOVA ultramicrotome (Stockholm, Sweden), were stained with a solution of toluidine blue in 0.1M borate buffer and then observed under a light microscope. Circular muscle and myenteric plexus rich areas away from the strips' edges with no apparent signs of mal-fixation or processing artefacts were selected. Ultra-thin sections of these selected areas were obtained with the LKB NOVA ultramicrotome using a diamond knife and stained with a saturated solution of uranyl acetate in methanol (50:50) per 12 min. at 45ЊC, followed by an aqueous solution of concentrated bismuth subnitrate per 10 min. at room temperature. At least 10-20 ultrathin sections from all four strips of each patient were examined under a JEOL 1010 electron microscope (JEOL Ltd., Tokyo, Japan) and photographed.

Results
The ultra-structural features of age-and sex-matched patients undergoing gastric bypass surgery used as controls were similar to those seen in the non-obese patients undergoing gastric surgery for carcinoma and those reported in literature as normal features [23,24] (Fig. 1A-E). Conversely, all of the gastroparetic patients had ultrastructural abnormalities, both those with at least one immunohistochemical abnormality and those with no apparent abnormalities on immunohistochemistry (Table 1).

Diabetic gastroparesis
As compared to controls, 17 of the 20 (85%) diabetic gastroparetics had ultrastructural changes in ICC morphology. Morphological abnormalities consisted of the presence of several intracytoplasmic vacuoles ( Fig. 2A), mitochondria with a clear matrix (Fig. 2B) and an extended rough endoplasmic reticulum (Fig. 2C). All of these abnormalities were not always all present in the same cell and normal ICC were also present in some of these patients, together with the altered ones. Apoptotic features (compact chromatin filling the entire nucleus or as dark clumps close to remnants of the nuclear envelope, cytoplasm either completely empty or containing few swollen mitochondria, increased lysosomes) were also observed in all of the patients with ultrastructural changes (Fig. 3A). In contrast to controls, for all diabetic gastroparetics with ultrastructural changes, none of the ICC visualized were in contact with nerve  endings and rarely had cell-to-cell contact with smooth muscle cells ( Fig. 2A) and other ICC. Amounts of flocculent and filamentous material, similar to a basal lamina, were occasionally found around ICC ( Fig. 2B and C). Both, intramuscular ICC and ICC in the myenteric plexus region had similar ultrastructural changes. In all of the patients, the majority of the smooth muscle cells appeared normal and only a few (no more than 2-3 cells/section) showed morphological alterations, such as vacuoles, abundant lipofuscin bodies (Fig. 3B), swollen mitochondria (Fig. 4A), lamellar bodies, disordered arrangement of myofilaments, with the myofilaments no more parallel to each other or to the major axis of the cell (Fig. 4B). Gap junctions (Fig. 4C) were present in both   normal and abnormal cells, although the connective tissue was filled with thin collagen fibrils that distanced the smooth muscle cells form each other (Figs 4B and 5A). Except for one patient (# 10), the basal lamina was patchily (Fig. 5B) or continuously ( Fig. 5A and C) markedly thickened (up to 0.6 m versus the 0.06 m of controls) in all diabetic gastroparetics.
In most of the subjects (except #6, that had only a few altered nerve endings), in each section, the nerve bundles contained several altered nerve endings. Abnormal nerve endings were large and either empty (Fig. 6A) or, especially in the patient 10, filled with filaments and microtubules (Fig. 6B). In the latter patient, the basal lamina had a normal thickness and elastic fibres attached the nerve endings to the smooth muscle cells (Fig. 6B) which was a finding in patients with idiopathic gastroparesis (see below) but not in controls. All other patients had a thick basal lamina and/or a thick sheath made by collagen fibrils (Figs 5B and 6C) around nerve fibres that appeared to be characteristic for diabetic gastroparesis. Free nerve endings (Fig. 6D), although few, were present and some of them, as in controls, were in close contact with smooth muscle cells.
The connective stroma always had an abnormal appearance. In particular, in many of the patients, even in the youngest ones, there was a large quantity of collagen fibrils not assembled in bundles, and the basal lamina was always either patchily or continuously thickened. Mast cells were frequently seen.

Idiopathic gastroparesis
There was a marked dropout of ICC in all idiopathic gastroparetics. Ultrastructural damage to ICC was more marked than in diabetic gastroparesis. Except for patients 4, 14 and 15 that had few (no more than 2-3 cells in 10 sections examined) patchily distributed normal residual ICC (Fig. 7A) all others had abnormal ICC features. The remaining cells were often hard to recognize as ICC as they had few mitochondria which was in contrast to normal ICC and all had a clear matrix and few cristae, many lamellar bodies and extremely  large vacuoles that filled the cytoplasm ( Fig. 7B and C). As in diabetic gastroparesis, none of the ICC was in contact with nerve endings and rarely with each other and the smooth muscle cells (Fig. 7A). Also, ICC with apoptotic features similar to those seen in diabetic gastroparetics were seen in all patients with idiopathic gastroparesis. Both, intramuscular ICC and ICC in the myenteric plexus region had similar ultrastructural changes. The majority of the smooth muscle cells were normal, gap junctions were always present and some showed caveolae arranged in long rows (Fig. 8A). Only a few smooth muscle cells (again 2-3cells/section) showed morphological alterations (Fig. 8B) and these findings were similar to those observed in diabetic gastroparesis.
Nerve structures in all patients showed markedly altered morphology (Fig. 9A-D). In neuronal cell bodies, mitochondria had cristae with abnormal shape and orientation (Fig. 9A). Neurofilaments were chaotically arranged (no more oriented parallel to each other and to the major axis of the axon and often forming whorls) in both interganglial and intramuscular axons ( Fig. 9A and D). Some of the intramuscular nerve endings were large and empty (Fig. 9C) and others contained lamellar bodies and synaptic vesicles sequestered in membranous envelopes ( Fig. 9B and C). Nerve bundles containing normal nerve endings as well as free nerve endings were rarely seen. Glial cells ensheathing axons were also altered, with cytoplasm filled with lysosomes, and vacuoles ( Fig. 9B-D).
The connective stroma showed fibrosis with a marked increase in collagen fibrils, which were particularly abundant around the nerve structures (Fig. 9C). Conversely, the basal lamina had a normal thickness everywhere which distinguished patients with idiopathic gastroparesis from diabetic gastroparesis with the exception of 2 patients (4 and 18) with a diagnosis of idiopathic gastroparesis who had a thick basal lamina patchily distributed around smooth muscle cells, ICC and nerve endings. Mast cells were seen but less commonly than in diabetic gastroparesis.

Discussion
There are two main findings that emerge from this study. The first is that TEM can identify cellular changes in all patients studied with diabetic and idiopathic gastroparesis. Even in the patients with no apparent immunohistological changes on light microscopy, ultrastructural evidence for cellular damage was present. TEM therefore offers an additional tool to study these patients and can provide additional information over light microscopy. The finding that 100% of patients with delayed gastric emptying have cellular changes in the muscle wall of the stomach suggests that these cellular changes may be responsible for  the electrical and contractile abnormalities associated with gastroparesis. The second finding to emerge from this study is that there are significant differences between the ultrastructural defects in diabetic and idiopathic gastroparesis as highlighted in Table 2.
All tissues examined in this study had ICC, nerve or smooth muscle cell changes suggestive of injury consistent with the diagnosis of gastroparesis with documented delayed gastric emptying. These abnormalities, however, had a different appearance between the two disorders. Nineteen of 20 diabetic gastroparesis patients had a thickened basal lamina around smooth muscle cells and nerves. In contrast, tissues from 18 of 20 patients with idiopathic gastroparesis did not have the thickened basal lamina around smooth muscle cells and nerves but had more intense fibrosis than those from diabetic gastroparesis. Nerve damage was much more prominent in idiopathic gastroparesis with both nerve cell bodies and nerve fibres affected to a greater degree. Unlike in diabetic gastroparesis, glial cells were also abnormal in idiopathic gastroparesis.
Interstitial cells of Cajal were affected in both diabetic and idiopathic gastroparesis. Loss of ICC, already described in delayed gastric emptying associated with diabetes in mice, rats and human light microscopy studies [11,[13][14][15][16][17][18][19]21], was confirmed and the findings extended to idiopathic gastroparesis. Moreover, TEM allowed the determination that residual ICC were rarely in contact with each other and smooth muscle cells and never to nerve endings. This is in sharp contrast to what is seen in control tissues. Given the role ICC play in the control of gastric motility [25] we can reasonably hypothesize that these findings will have a significant impact on the coordinated motor activity in both types of disorders.
Transmission electron microscopy also confirmed previous findings of abnormalities of nerve tissue in both disorders; the loss and/or alterations of synaptic vesicles we presently observed are in agreement with the loss of expression of several neurotransmitters and the neuronal nitric oxide synthase, previously demonstrated [14][15][16]19] in diabetic and/ or idiopathic gastroparesis [11,13,14,17,21]. As outlined above, neuronal cell body, nerve fibres and glial cells were markedly altered in idiopathic gastroparesis, while only nerve ending abnormalities were seen in the diabetic patients.
Smooth muscle cell abnormalities were not commonly seen in diabetic and idiopathic gastroparesis and gap junctions were maintained in all patients studied. However, contractile activity of the muscle coat may still be compromised because the smooth  muscle cells were encased in a less elastic stroma. The TEM findings of markedly increased thin, scattered collagen fibrils that were not organized in bundles may explain why under light microscopy fibrosis (presence of thick bundles of fibrils) was rarely found [26]. The separation of nerves from ICC, and ICC and nerves from smooth muscle seen on TEM in both disorders suggests that there may be also impaired neurotransmission and transmission of the ICC electrical signal to smooth muscle affecting contractility. Furthermore, the marked changes in the basal lamina seen in diabetic gastroparesis, with the basal lamina either continuously or patchily thickened was similar to what has been described in other organs affected by diabetes such as blood vessels, the retina and the kidney. Its functional implication is a likely decreased ability to exchange metabolites leading to cell damage. Similarly, the fibrillar sheath observed in idiopathic gastroparesis around smooth muscle cells, ICC and nerves may also compromise smooth muscle cell metabolism and cause cell sufferance (presence of vacuoles, lipofuscinic and lamellar bodies, disordered arrangement of myofilaments).
In conclusion, in this large ultrastructural study of prospectively collected tissue from both diabetic and idiopathic gastroparesis multiple cellular abnormalities were seen. TEM not only confirmed the light microscopy findings but also allowed two new observations to be made. All 40 patients with gastroparesis had cellular abnormalities when the tissue is examined under TEM. Also, while light microscopy does not differentiate well between diabetic and idiopathic gastroparesis, TEM does. In particular, a thick basal lamina was strongly associated with diabetic gastroparesis, while more severe damage to ICC and neurons was associated with idiopathic gastroparesis. The known fairly rapid turnover of ICC [25,27,28] and the sparing of synaptic vesicles offer hope that diabetic gastroparesis may be reversible with therapy. This is borne out by animal studies that have shown the reversibility of the global cellular and neuromuscular or motility changes [3,4,29,30]. Conversely, in idiopathic gastroparesis, the severe nerve tissue (neuronal cell bodies, nerve endings and glial cells) impairment may play a larger role in development of gastroparesis. These insights should help differentiate the two disorders and also provide direction for the development of therapy targeted to each disorder.