- Email: [email protected]
Hemorrhagic shock and acid gastric injury in the rat
GASTROENTEROLOGY
ALIMENTARY
1986;90:1103-10
TRACT
Hemorrhagic Shock and Acid Gastric Injury in the Rat Comparison of Gross and Histologic Findings MAKOTO
ITOH, GARY PAULSEN, and PAUL H. GUTH
Medical and Research Services, Veterans Administration Wadsworth Medical Center, LJ.C.L.A and Center for Ulcer Research and Education, Los Angeles. California
The purpose of this study was to compare the gross and histologic changes in the corpus and antrum in the rat hemorrhagic shock plus acid model of gastric injury, and to determine the effect of transfusion of shed blood or albumin on gastric lesion formation. The data indicate that (a) despite the gross appearance ofmore severe damage in the corpus, histologic damage was more severe in the antrum; [b] covering mucus and cell debris partly explain the difficulty in recognizing antral lesions grossly; (c) the severity of histologic injury was similar in transfused and nontransfused rats; and (d) transfusion of shed blood rendered corpus lesions more recognizable grossly but did not affect the severity of histologic injury. The latter findings raise questions about the pathogenetic importance of reperfusion in gastric ischemit injury and Ihe validity of using gross lesions as an index of gastric injury. Hemorrhagic shock in the experimental animal is a widely used model to study gastric mucosal injury (1,Z). In this model, the gastric lesions tend to cluster in the mucosa of the corpus and spare the antrum. There are no reports, however, that carefully evaluate the mucosal changes in the antrum. In addition, although retransfusion of blood withdrawn to produce hemorrhagic shock is routinely performed, the effect of the retransfusion on the histologic extent of the gastric lesions has not been evaluated. Recent Received March 21, 1985. Accepted October 25, 1985. Address requests for reprints to: Paul H. Guth, M.D., 691/111C. Veterans Administration Wadsworth Medical Center, Los Angeles. California 90073. This work was supported by National Institute of Arthritis, Metabolism, and Digestive Diseases grant AM25891 and Veterans Administration Research funds. Makoto Itoh’s present address is: Nagoya City University School of Medicine, Nagoya, Japan. The authors thank Anita Starlight for secretarial assistance. ‘@Z 1986 by the American Gastroenterological Association OOlS-5085/86/$3.50
data, based solely on gross evaluation of lesions, suggest that this is an ischemia-reperfusion injury, with damage being due to oxygen-derived free radicals formed during the reperfusion period (3). The purpose of this study of the rat hemorrhagic shock model was to (a) evaluate histologically the severity of gastric mucosal injury in both the corpus and antrum, (b) compare the severity of the gross findings with the histologic findings, and [c) determine the effect of retransfusion on gastric lesion formation.
Materials Animal
and Methods
Preparation
Male Sprague-Dawley rats, weighing 200-250 g and fasted overnight, were anesthetized with 50 mgikg of pentobarbital sodium intraperitoneally. Tracheostomy was performed and PE-250 tubing was inserted into the trachea to maintain an open airway. The abdomen was then opened and the gastroesophageal junction was ligated. After ligation of the duodenum at a site 10-15 mm distal to the pylorus, a small duodenotomy was performed between the pylorus and the site of duodenal ligation. A cannula was then passed through the duodenotomy into the stomach and the gastric lumen was cleansed with warm saline. A carotid artery was cannulated with PE-90 tubing for monitoring blood pressure and withdrawal and reinfusion of blood. Body temperature was monitored by a rectal thermometer and maintained at ~-37°C by an external lamp. After the blood pressure stabilized, I ml/l00 g body wt of 0.1 N HCl was instilled into the stomach via the gastric cannula. The gastric cannula was then removed and the pylorus was ligated.
Experimental
Design
Studies in rats subjected to hemorrhagic shock and no retransfusion. Five minutes after the intragastric instillation of 0.1 N HCl, blood was withdrawn from the carotid artery over a 2-min period into a syringe containing 0.6 ml
1104 ITOH ET AL.
GASTROENTEROLOGY
of heparinized saline (100 U heparin per milliliter of saline). The mean systemic blood pressure was reduced to ~30 mmHg by bleeding and was maintained there for 40 min in one animal group [group 2) and for 20 min in another animal group (group 4). The rat was then killed by thoracotomy and the stomach was removed. One minute before killing, 1 ml of 1% Evans blue was injected via the carotid artery to enhance the contrast of gastric lesions. The stomach was opened along the greater curvature, pinned out on cardboard, and fixed in 10% formalin. Control animals underwent the same procedure except for the withdrawal of blood (group 1, 40 min; group 3, 20 min). Studies in rats subjected to hemorrhagic shock and reinfusion. The procedure was the same as described above except that 20 min after reducing systemic blood pressure to ~30 mmHg by bleeding, the heparinized shed blood was retransfused [group 5) and, 20 min after the retransfusion, the rat was killed. In two additional studies, which were designed to study the effect of restitution of circulating volume by colloids with and without heparin instead of blood, 7% albumin containing 0.6 ml of heparinized saline (100 U heparin per milliliter of saline) (group 6) or nonheparinized albumin (group 71, in amounts equal to the shed blood, were retransfused instead of the shed blood. Tissue
Preparation
After the area of the gastric lesions was measured (as described later), three strips of tissue were removed at right angles to the lesser curvature: from across the proximal and distal anterior wall of the corpus and across the middle of the antrum. The corpus specimens involved the entire width of the anterior wall from the greater curvature to the lesser curvature, and the specimen from the antrum involved the entire width of the antrum. The tissue specimens were dehydrated, embedded in paraffin, sectioned at 6 pm, and stained with hematoxylin and eosin. In animals from groups 4 and 5, additional sections were cut and stained with the periodic acid-Schiff (PAS) stain.
stained with hematoxylin and eosin using a modification of the criteria of Lacy and Ito (4). Damage to the corpus mucosa was defined as follows: 0, all gastric mucosal cells appeared intact and had normal shape, location, appearance, and density; I, surface mucous cell damage-these cells were vacuolated, had pyknotic nuclei, and brightly stained or lysed cytoplasm; II, extensive surface cell damage plus disruption and exfoliation of cells lining the gastric pits (in some areas, the first or second parietal cells lining the glands were also involved); III, damage extending well beyond the gastric pits but involving ~50% of the thickness of the gastric mucosa; and IV, extensive gastric mucosal damage involving >50% of the thickness of the gastric mucosa. Damage to the antral mucosa was defined as follows: 0, all cells were intact and normal in appearance; I, damage involved the surface epithelial cells and the uppermost two or three cells lining the glands; II, damage greater than I but involving ~50% of the thickness of the of the mucosa; and III, damage involving >50% thickness of the antral mucosa. Using an ocular micrometer, the length of each type of cellular damage and the section length were determined for each specimen. The percentage of the mucosal length with each grade of damage was then calculated. ESTlMATIONOFTHEAMOUNTOFMUCUSCOVERINGGASLESIONS. The thickness and length of mucus covering the surface of gastric lesions [types III and IV damage in the corpus and types II and III damage in the antrum) were measured on PAS-stained sections using an ocular micrometer in a microscope. Dehydration during the histologic preparation might cause mucus shrinkage artifacts; therefore, the measurements of mucus thickness represent relative values of mucus thickness among the different sections rather than true values. The thickness of the mucus was defined as follows: 0, no mucus; +,lOO pm. The length of each type of mucus covering gastric lesions and the total length of gastric lesions were determined for each section. The percentage of the mucosal damage covered with each type of mucus was then calculated.
TRIC MUCOSAL
Statistical Gross
and Histologic
Evaluation
of Gastric
Lesions Measurement of the area of gross gastric lesions. The area of gastric lesions (in square millimeters) was measured using a dissecting microscope with x 10 magnification and a l-mm’ grid eyepiece. Lesion measurements were made without knowing which animal group the stomach came from. The area of the corpus and antrum was measured by planimetry. The mucosal lesions in the corpus and antrum were then expressed as percent of the total corpus and antral area, respectively. Histologic analysis. Histologic tissue sections from all the groups were coded, randomized, and examined for gastric mucosal injury and mucus on the surface of gastric lesions. The code was not broken until all tissue sections were read. ESTIMATION OF GASTRIC MUCOSAL INJURY. The severity of gastric mucosal injury was evaluated on the sections
Vol. 90, No. 5, Part 1
Analysis
The Wilcoxon rank sum test was used throughout except for comparison of histologic scores in the upper and lower corpus (Table 2). In the latter instance, because data from the two areas were obtained in each rat, Student’s paired t-test was used. A probability level of < 0.05 was considered significant.
Results Gross Gastric Study
Lesions
shock and no transto hemorrhagic shock for 40 20 min (group 4) but not shed blood, there were very could be clearly recognized the surface of the mucosa
with hemorrhagic
fusion. In rats subjected min [group 2) or for receiving transfusion of few gastric lesions that as breaks (erosions) in
HISTOLOGY
May 1SRfj
Figure
1. Photograph of the stomach of a rat subjected to 20 min of hypotension [plus intragastric acid) but not receiving transfusion of shed blood. The corpus mucosa appears abnormal but it is difficult to recognize breaks in the muLosa [erosions]. The antrum is covered by a white (mucus?) laycar.
(Figure 1). Therefore the areas without a normal mucosal appearance were regarded as gastric “lesions” (Table 1). According to this definition, there were extensive lesions in the corpus and antrum in both groups 2 and 4 (the mean percent area involved with lesions ranging from 40% to 80%) but not in their respective controls (Wilcoxon rank sum test, p < 0.05). Comparison of the gastric lesions in the corpus with those in the antrum showed that the percent of gastric lesions in the corpus and the percent in the antrum were almost the same in group 2, but corpus lesions in group 4 were significantly greater than antral lesions (Wilcoxon rank sum test, p < 0.05). Study with hemorrhagic shock and reinfu-
Table
1. Gross
OF ISCHEMIC
GASTKIC: IN]LJRY
sion. In the animals subjected to hemorrhagic shock for 20 min and receiving transfusion of heparinized shed blood (group 5), heparinized albumin (group 6), and nonheparinized albumin (group 7), the gastric mucosal lesions, unlike those in group 2 and 4 rats without transfusion, could be recognized clearly as breaks (erosions) in the surface of the mucosa (Figure 2). Only these areas were measured as gastric lesions. In this respect, the gastric lesions estimated macroscopically in the studies with infusion cannot be equated with those in groups 2 and 4. The mean percent of mucosal area involved with lesions in groups 5, 6, and 7 was 19%, 20%, and 19%, respectively, in the corpus, and 10% 9%, and 9% respectively, in the antrum. There were no significant differences in the formation of gastric lesions either in the corpus or in the antrum among these three groups. In all studies, however, corpus lesions were significantly greater than antral lesions (Wilcoxon rank sum test, p < 0.05).
Histologic
Evaluation
of Gastric
Lesions
Comparison of histologic scores in the upper and lower corpus. In all groups, there were no significant differences (Wilcoxon rank sum test) between histologic scores in the proximal and distal corpus for any type of damage, except for types I and II in group 7 (Table 2). This finding indicates that hemorrhagic shock-induced gastric mucosal injury affects the proximal and distal corpus mucosa equally. Therefore, the upper and lower corpus scores for each type of damage in each rat were averaged, and further data analysis is based on these combined scores. Study with hemorrhagic shock and no transfusion. In animals subjected to either 40 min [group 2) or 20 min (group 4) of hypotension, there was greater damage of all types, except for type I, in both corpus and antrum in comparison with their respective controls [Table 3). This was especially true of
Gastric Lesions Reinfusion Shed No reinfusion
Corpus Antrum
1105
blood
No reinfusion
Group 1 (II= 7)
Group 2 (n = 6)
Group 3 (II= 6)
Group 4 (n = 6)
Group 5 (n = 7)
0.1 + 0.1 0
80 ? 9b 76 L 8b
0.4 2 0.2 4k2
64 ? 5"' 40 k 9"
19 +- 5' 10 t 4
Albumin heparin
of +
Group 6 (11 7) 20 2 4' 952
_
Albumin alone Group 7 (n = 7) 19 f 3’ 9f2
group 2, hypotension for 40 min; groups 4-7, hypotension the mean Groups 1 and 3, no hypotension: for 20 min. Each value represents -t standard error of the percent of the area involved with gastric lesions-areas without a gross normal appearance in nonreinfused rats (groups l-4) vs.clear-cut erosions in reinfused rats (groups 5-7).Significantly different from the respective control: ” p < 0.05," p < 0.01. ’Significant difference between corpus and antrum, p < 0.05.Wilcoxon rank sum test was used for statistical analvsis.
1106 ITOH ET AL.
Figure
GASTROENTEROLOGY
Vol. 90, No. 5, Part 1
Study with hemorrhagic shock and reinfusion. In the three reinfusion groups, transfusion of heparinized shed blood (group 5), heparinized albumin (group 6), and nonheparinized albumin (group i’), the extent of damage of all types in both corpus and antrum was similar and there were no significant differences among these groups (Kruskal-Wallis analysis of variance with contrasts) (Table 3). The extent of damage of all types in these three groups also showed no significant differences from those in group 4 (20 min of hypotension and no retransfusion) either in the corpus or the antrum. Deep damage in the antrum again was greater than in the corpus, varying from 52% to 55% versus 8% to 20%. This was especially true of the deepest damage (type III in the antrum and type IV in the corpus), with damage being significantly greater in the antrum in groups 5 and 7. Evaluation of mucus covering gastric lesions. This evaluation was performed on PASstained sections from animals in groups 4 and 5 (20 min of hypotension without and with reinfusion of shed blood) (Table 4). The percent of lesion length covered with thick [lo0 pm (+ + +)] or moderately thick [lo-100 pm (+ +)] mucus was greater in group 4 than in group 5 rats in both the corpus and the antrum, with greater coverage in the antrum than in the corpus. The converse was found with regard to lesions covered with no mucus (0) or thin [lo pm (+)] mucus. None of these differences, however, were statistically significant. When the percent of lesion length covered with ++ mucus and that covered with + + + mucus were combined, however, mucus coverage was significantly greater (Wilcoxon
2. Photograph of the stomach of a rat subjected to XI min of hypotension (plus intragastric acid) and receiving transfusion of shed blood. The corpus mucosal lesions can be recognized clearly as breaks in the surface of the mucosa (erosions). A few small lesions can be seen in the antrum.
deep damage, i.e., types III and IV in the corpus and types II and III in the antrum (which corresponds to type IV in the corpus). Deep damage was significantly greater in the antrum (types II and III) than in the corpus (types III and IV) in both groups 2 and 4. The extent of damage of all types was similar in groups 2 and 4 and there were no statistically significant differences between these two groups.
Table 2. Comparison of Histologic Scores in the Upper and Lower Corpus Reinfusion Shed No reinfusion Group 1 (n = 7) 0
I II III IV
No reinfusion
blood
heparin
+
Albumin alone
Group 2
Group 3
Group 4
Group 5
Group 6
Group 7
(n = 6)
(n = 6)
(n = 6)
(n = 7)
(n = 7)
(n = 7)
17 ? 11
17 r
27 2 8
3k3
72 2 10
16 *
72-4
63 k 16
45 k 8
18 k 5
27 ? 10
53 2 8
26 2 7
24 k 11
10
Albumin
of
28 k 12
42 k 7
32 -t- 7
41 2 10
121 12 k 6
6 2 10 12 ” 6 9?3
25 * 9
7
723
14 + 7 11 2 9
22 + 6
23 2 7
19 k 7
18 k 4
29 + 7
40 2 14
38 + 5”
69 ? 7
45 + 8
44 2 8
64 -r- 8
41 t
8
35 + 10
29 + 11
39 -t 70
0
33 2 11
0
12 t
7
16 + 9
14 + 7
0
17 2 6
0
20 2 9
11 + 5
11 + 5
0
3+2
0
5k3
0
9+9
0
928
0 lkl
3+3 14 + 12
323 12 + 5 0 0
Groups I and 3, no hypotension; group 2, hypotension for 40 min; groups 4-7, hypotension for 20 min. Each value represents mean 2 standard error. The upper and lower values in each row represent the values for the upper and the lower corpus, respectively. a Significant difference between upper and lower corpus paired values (p < 0.05, paired t-test).
May 1986
Table
HISTOLOGY
3. Histologic
Evaluation
OF ISCHEMIC
GASTRIC
Shed No reinfusion
I II 111
Group 2 (n = 6)
Group 3 (n == 6)
23 2 9 46 + 8 29 2 9
523 23 * 40
68 -t 12 25 + 10 623
13 + 3
0
42 ? 9 24 +- W
IV 0
6?5 0
24 t 9 77 * 9
I II
0
III
0
3 * 3”
Albumin alone
8 f 5”
20 r
10
12 -+ 5
12 + 7
25 t
5
M 2 9
28 t
5
57 + 7b
41 -+ 8
27 i- 9
52 ?
7
15 + 6”
14 ? 6
13 t
t
0.5
5
7 + 6
Group 7 (n = 7)
8+3 0
49 1- 17
18 + 16
0
17 t
14
28 2 16
46 ?
18 2 11
45 t
12
22 t
13
20 2 11
11 + 3
12 *
7
17 t
8
11 2 6
52 + l6”.
43 t
151
35 i
15
42 2 18”
16
525
27 t- 8” 70 f 9%”
-
Group 5 (n = 7)
0.5
Group 6 (n = 7)
of
Group 4 (n = 6)
725
0
,4lbumin heparin -___
blood
No reinfusion
Group 1 (n = 7)
221 0
1107
of Gastric Lesions Reinfusion
0
INfURY
(I
Groups 1 and 3. no hypotension; group 2, hypotension for 40 min; groups 4-7, hypotension for 20 min. Each value represents the mean + standard error of the percent of the mucosal length involved with gastric lesions. Significantly different from the respective control: different from respective corpus type IV lesion: ’p CC0.05, d p < 0.01.Wilcoxon rank sum test was ‘l p < 0.05, h p < 0.01. Significantly used for statistical analysis.
rank sum test, p < 0.05) in the antrum in group 4 (no reinfusion] than group 5 (83.6% * 16.4% vs. 30.2% -+ 13.7%). In addition, there also was a statistically significant difference between the corpus and antrum in group 4 (83.6% 2 16.4% vs. 30.5% -+ 17.0%). An example of this is shown in Figures 3A and 3B. In group 5 this difference did not quite reach statistical significance (30.2% ? 13.7% vs. 1.0% + 1.0%).
Table
4. Mucus Covering
Gastric
Lesions
Group 4: 20 min of hypotension with no reinfusion
Group 5: 20 min of hypotension with reinfusion of shed blood
0 antrum corpus
29 t 16
525 64 r
18”
78 + 14
12 t
12
41 t
+
antrum corpus
15
21 ‘-’ 14
523
f-t
antrum corpus
55 t
16
30 f
21 t
15
l-+1
28 -+ 14
0
14
+ -t +
antrum corpus + + and
antrum corpus
927
0
+ + + 84 -c 16”,” 31 +- 17
30 + 14h 1+1
Group 4: antral lesions, n = 5; corpus lesions, n = 6. Group 5: antral lesions, n = 6; corpus lesions, n = 7. Upper and lower rows in each set of paired figures represent percent of antral and corpus lesions covered with mucus. Significant difference between the antrum and corpus: ” p < 0.05, h p < 0.01, respectively. Significantly different from group 5 value: c p < 0.05. Wilcoxon rank sum test was used for statistical analysis.
Discussion In experimental studies of ischemic mucosal injury in the stomach, the animal is routinely subjected to withdrawal and retransfusion of blood. It has been pointed out in this model that the gastric mucosal injury tends to cluster in the corpus mucosa and affect the antral mucosa less severely. In this respect it resembles stress ulcer in humans. To explain this finding, several hypotheses have been proposed. These include differences in energy metabolism (5-71, mucosal blood flow (2,8,9), and disposal capacity of influxing H+ between the corpus and antral mucosa (10). Those investigations have been based on the gross appearance of gastric injury. Histologic studies were not performed. It was surprising, therefore, that the present histologic studies revealed more marked deep injury in the antrum than in the corpus. In all groups of rats subjected to 20 min of hypotension, whether reinfused or not and regardless of the fluid infused, mean percent histologic damage extending beyond the superficial mucosa in the different groups varied from 52% to 63% in the antrum versus 8% to 20% in the corpus. Thus the rat antrum, contrary to general belief, is more susceptible to acid injury during ischemia than the corpus. The present study did not provide any indication of why this is so. However, it did shed some light on why it is so difficult to recognize antral lesions by macroscopic examination alone. A mucus layer covered 71% of the length of the antral lesions but only 22% of the length of corpus lesions in rats subjected to the usual procedure of 20 min of hypotension plus retransfusion (Table 4, group 5). Thirty percent of the length of
1108
Figure
ITOH ET AL.
GASTROENTEROLOGY
3. Photomicrograph of PAS-stained sections from the corpus (A) and the antrum (B) of a group 4 retransfusion). A. There is extensive mucosal damage including approximately one-third of the mucosa. This lesion is only partially covered by mucus and cell debris. B. There is extensive mucosal the full thickness of the antral mucosa. This is entirely covered by a thick layer of mucus and cell
Vol. 90, No. 5, Part 1
rat (hypotension and no thickness of the corpus damage involving nearly debris.
May 1Wi
antral lesions was covered with a mucus layer >lO pm (+ + or + + + mucus) versus only 1% for corpus lesions. The mucus covering, by masking the breaks in the antral mucosa, explains, in part, why it is difficult to recognize erosions there. Perhaps if different criteria were used, antral injury could be identified grossly. However, nearly all investigators look for the clear-cut erosions seen in the corpus. In most studies of gastric mucosal injury only gross examination is used to evaluate the extent of mucosal injury. The findings of the present study indicate how misleading this can be. Similarly, in a histologic study of ethanol-induced gastric mucosal injury, Lacy and Ito (4) observed extensive surface mucus cell damage that could not be seen on gross examination. In addition, they were able to characterize histologically the gross “necrotic” injury (hyperemia and hemorrhage as well as necrosis). These two studies clearly demonstrate the necessity of histologic as well as gross evaluation of damage in gastric mucosal injury studies. Another purpose of these studies was to cletermine the effect of retransfusion of shed blood on gastric lesion formation. In animals subjected to hypotensive shock and no retransfusion of shed blood, it was difficult to recognize gross lesions. In contrast, in shocked animals that were reinfused with blood or albumin. lesions were readily identifiable as breaks in the mucosa (erosions). This was especially true in the corpus. In retransfused animals (group 5), the extent of deep histologic damage (types III and IV), 140/o, was in good agreement with gross corpus damage, 19%. This was not true of nonretransfused rats (group 4), in whom the gross corpus “lesion” scores markedly overestimated the true deep histo64% vs. 22%. Study of PAS-stained logic damage, sections revealed that + + or + + + mucus (>lO pm thick) covered 31% of the corpus lesions in the nonretransfused rats but only 1% in the retransfused rats. As suggested by Mersereau and Hinchey (ll), perhaps retransfusion, by restoring mucosal blood flow, results in a washing away of mucus that covers lesions, thus rendering the lesions more recognizable. In addition, gross and histologic damage is similar in both the corpus and antrum whether the retransfusion is with shed blood or an equal volume of heparinized or nonheparinized albumin. Using a stereomicroscope and a gastric chamber technique in a hemorrhagic shock-retransfusion study, Mersereau and Hinchey (11) reported that gastric mucosal lesions were formed earlier and were larger in the antrum than in the corpus. Although they did not study histologic features of the mucosal injury, their observations support the histologic findings in the present study. In all the studies reporting that the antrum is spared from gross gastric
HISTOLOGY OF ISCHEMICCXSTRIC INJURY
1109
injury, as in the present study, no in vivo microscopic observations were made during lesion formation as was done by Mersereau and Hinchey. (11)suggested that the onset Mersereau and Hinchey of ulcerations was surface cell necrosis of gastric mucosa brought about during hypotensive shock and, on retransfusion of shed blood, the necrotic cells are flushed off into the gastric lumen by the returned gastric mucosal blood flow. Thus, the retransfusion of shed blood renders the gastric lesions induced during hypotensive shock recognizable as breaks in the surface of the mucosa. The finding that histologic damage was as severe in the nontransfused (group 4) as in the transfused rats (group 5,Table 3) also was surprising. Recently there has been considerable evidence that, in many “ischemic” tissues, injury occurs not during the period of ischemia but during the reperfusion period. This has been clearly demonstrated in the cat intestine (12,13). Histologic studies revealed much greater mucosal damage if an intestinal segment that was rendered ischemic for 2 h (12) or 3 h (13) was use of allopurinol to reperfused for 1 h. Furthermore, inhibit xanthine oxidase, the enzyme that catalyzes superoxide formation, or superoxide dismutase to scavenge superoxide radicals. prevented the histologic damage on reperfusion (14). The concept is that during the ischemic period the necessary changes for free-radical formation develop--xanthine dehydrogenase changes to the xanthine oxidase form and adenosine triphosphate breaks down to hypoxanthine. With reperfusion of the previously ischemic tissue, adequate oxygen becomes available and there is a burst of free-radical formation (hypoxanthine + O2 in the presence of xanthine oxidase --+ xanthine + 0,) that is responsible for the tissue injury. We recently reported that administration of allopurinol or superoxide dismutase significantly decreased gross gastric mucosal injury in the hypotension-acid model (3). This appeared to provide strong evidence for a role of free radicals in ischemic gastric mucosal injury. That the gross and histologic findings in this study do not correlate well and that the damage occurs during the ischemic period (with the role of retransfusion being to render the gross lesions more recognizable), raise serious questions. First, do allopurinol and superoxide dismutase decrease the histologic mucosal injury? Second, if they do, is this protection seen in nonretransfused rats? Third, if the answer to the previous question is yes, are oxygen metabolites really involved in the mechanism of this tissue injury? There still could be a role for oxyradicals in gastric ischemic injury. The precise tissue oxygen tension needed to set the stage for oxyradical formation (xanthine oxidase and hypoxanthine formation) and the amount of oxygen needed for the
1110
ITOH ET AL.
final reaction are not known. It is conceivable that under the conditions of our study, the partial pressure of O2 in tissue fell sufficiently to cause formation of xanthine oxidase and hypoxanthine, and yet there was still a sufficiently high tissue oxygen tension for oxyradical formation. Additional studies will have to be performed to answer the specific questions raised and to test this hypothesis.
References 1. Starlinger M, Jakesz R, Matthews JB, Yoon Ch, Schiessel R. The relative importance of HCO,and blood flow in the protection of rat gastric mucosa during shock. Gastroenterology 1981;81:732-5. 2. Leung FW, Itoh M, Hirabayashi K, Guth PH. Role of blood flow in gastric and duodenal mucosal injury in the rat. Gastroenterology 1985;88:281-9. 3. Itoh M, Guth PH. Role of oxygen-derived free radicals in ischemic gastric injury in the rat. Gastroenterology 1985;88: 1162-7. 4. Lacy ER, Ito S. Microscopic analysis of ethanol damage to rat gastric mucosa after treatment with a prostaglandin. Gastro1982;83:619-25. enterology 5. Menguy R, Masters YF. Gastric mucosal energy metabolism and stress ulceration. Ann Surg 1974;180:538-48.
GASTROENTEROLOGY
Vol. 90, No. 5, Part
1
6. Menguy R, Masters YF. Mechanism of stress ulcer: influence of hypovolemic shock on energy metabolism in the gastric mucosa. Gastroenterology 1974;66:46-55. 7. Menguy R, Masters YF. Mechanism of stress ulcer. III. Effects of hemorrhagic shock on energy metabolism in the mucosa of the antrum, corpus, and fundus of the rabbit stomach. Gastroenterology 1974;66:1168-76. 8. Manabe T, Suzuki T, Honjo I. Changes of upper gastrointestinal blood flow after hemorrhage in rabbits. Surgery 1977; 81:446-52. 9. Shiraji S, Mueller TM, Hardy BM. Canine gastric acid secretion and blood flow measurement in hemorrhagic shock. Gastroenterology 1977;73:75-8. 10. Kivilaakso E, Fromm D, Silen W. Relationship between ulceration and intramural pH of gastric mucosa during hemorrhagic shock. Surgery 1978;84:70-7. 11. Mersereau WA, Hinchey EJ. Effect of gastric acidity on gastric ulceration induced by hemorrhage in the rat, utilizing a gastric chamber technique. Gastroenterology 1973;64:1130-5. 12. Schoenberg MH, Muhl E, Sellin D, Younes M, Schildberg FW, Haglund U. Posthypotensive generation of superoxide free radicals-possible role in the pathogenesis of the intestinal mucosal damage. Acta Chir Stand 1984;150:301-9. 13. Parks DA, Granger DN. Reperfusion injury in the ischemic small intestine (abstr). Fed Proc 1984;lOlO. 14. Parks DA, Bulkley GB, Granger DN, Hamilton SR, McCord Ischemic radicals.
JM.
injury in the cat small intestine: role of superoxide 1982;82:9-15. Gastroenterology
ALIMENTARY
1986;90:1103-10
TRACT
Hemorrhagic Shock and Acid Gastric Injury in the Rat Comparison of Gross and Histologic Findings MAKOTO
ITOH, GARY PAULSEN, and PAUL H. GUTH
Medical and Research Services, Veterans Administration Wadsworth Medical Center, LJ.C.L.A and Center for Ulcer Research and Education, Los Angeles. California
The purpose of this study was to compare the gross and histologic changes in the corpus and antrum in the rat hemorrhagic shock plus acid model of gastric injury, and to determine the effect of transfusion of shed blood or albumin on gastric lesion formation. The data indicate that (a) despite the gross appearance ofmore severe damage in the corpus, histologic damage was more severe in the antrum; [b] covering mucus and cell debris partly explain the difficulty in recognizing antral lesions grossly; (c) the severity of histologic injury was similar in transfused and nontransfused rats; and (d) transfusion of shed blood rendered corpus lesions more recognizable grossly but did not affect the severity of histologic injury. The latter findings raise questions about the pathogenetic importance of reperfusion in gastric ischemit injury and Ihe validity of using gross lesions as an index of gastric injury. Hemorrhagic shock in the experimental animal is a widely used model to study gastric mucosal injury (1,Z). In this model, the gastric lesions tend to cluster in the mucosa of the corpus and spare the antrum. There are no reports, however, that carefully evaluate the mucosal changes in the antrum. In addition, although retransfusion of blood withdrawn to produce hemorrhagic shock is routinely performed, the effect of the retransfusion on the histologic extent of the gastric lesions has not been evaluated. Recent Received March 21, 1985. Accepted October 25, 1985. Address requests for reprints to: Paul H. Guth, M.D., 691/111C. Veterans Administration Wadsworth Medical Center, Los Angeles. California 90073. This work was supported by National Institute of Arthritis, Metabolism, and Digestive Diseases grant AM25891 and Veterans Administration Research funds. Makoto Itoh’s present address is: Nagoya City University School of Medicine, Nagoya, Japan. The authors thank Anita Starlight for secretarial assistance. ‘@Z 1986 by the American Gastroenterological Association OOlS-5085/86/$3.50
data, based solely on gross evaluation of lesions, suggest that this is an ischemia-reperfusion injury, with damage being due to oxygen-derived free radicals formed during the reperfusion period (3). The purpose of this study of the rat hemorrhagic shock model was to (a) evaluate histologically the severity of gastric mucosal injury in both the corpus and antrum, (b) compare the severity of the gross findings with the histologic findings, and [c) determine the effect of retransfusion on gastric lesion formation.
Materials Animal
and Methods
Preparation
Male Sprague-Dawley rats, weighing 200-250 g and fasted overnight, were anesthetized with 50 mgikg of pentobarbital sodium intraperitoneally. Tracheostomy was performed and PE-250 tubing was inserted into the trachea to maintain an open airway. The abdomen was then opened and the gastroesophageal junction was ligated. After ligation of the duodenum at a site 10-15 mm distal to the pylorus, a small duodenotomy was performed between the pylorus and the site of duodenal ligation. A cannula was then passed through the duodenotomy into the stomach and the gastric lumen was cleansed with warm saline. A carotid artery was cannulated with PE-90 tubing for monitoring blood pressure and withdrawal and reinfusion of blood. Body temperature was monitored by a rectal thermometer and maintained at ~-37°C by an external lamp. After the blood pressure stabilized, I ml/l00 g body wt of 0.1 N HCl was instilled into the stomach via the gastric cannula. The gastric cannula was then removed and the pylorus was ligated.
Experimental
Design
Studies in rats subjected to hemorrhagic shock and no retransfusion. Five minutes after the intragastric instillation of 0.1 N HCl, blood was withdrawn from the carotid artery over a 2-min period into a syringe containing 0.6 ml
1104 ITOH ET AL.
GASTROENTEROLOGY
of heparinized saline (100 U heparin per milliliter of saline). The mean systemic blood pressure was reduced to ~30 mmHg by bleeding and was maintained there for 40 min in one animal group [group 2) and for 20 min in another animal group (group 4). The rat was then killed by thoracotomy and the stomach was removed. One minute before killing, 1 ml of 1% Evans blue was injected via the carotid artery to enhance the contrast of gastric lesions. The stomach was opened along the greater curvature, pinned out on cardboard, and fixed in 10% formalin. Control animals underwent the same procedure except for the withdrawal of blood (group 1, 40 min; group 3, 20 min). Studies in rats subjected to hemorrhagic shock and reinfusion. The procedure was the same as described above except that 20 min after reducing systemic blood pressure to ~30 mmHg by bleeding, the heparinized shed blood was retransfused [group 5) and, 20 min after the retransfusion, the rat was killed. In two additional studies, which were designed to study the effect of restitution of circulating volume by colloids with and without heparin instead of blood, 7% albumin containing 0.6 ml of heparinized saline (100 U heparin per milliliter of saline) (group 6) or nonheparinized albumin (group 71, in amounts equal to the shed blood, were retransfused instead of the shed blood. Tissue
Preparation
After the area of the gastric lesions was measured (as described later), three strips of tissue were removed at right angles to the lesser curvature: from across the proximal and distal anterior wall of the corpus and across the middle of the antrum. The corpus specimens involved the entire width of the anterior wall from the greater curvature to the lesser curvature, and the specimen from the antrum involved the entire width of the antrum. The tissue specimens were dehydrated, embedded in paraffin, sectioned at 6 pm, and stained with hematoxylin and eosin. In animals from groups 4 and 5, additional sections were cut and stained with the periodic acid-Schiff (PAS) stain.
stained with hematoxylin and eosin using a modification of the criteria of Lacy and Ito (4). Damage to the corpus mucosa was defined as follows: 0, all gastric mucosal cells appeared intact and had normal shape, location, appearance, and density; I, surface mucous cell damage-these cells were vacuolated, had pyknotic nuclei, and brightly stained or lysed cytoplasm; II, extensive surface cell damage plus disruption and exfoliation of cells lining the gastric pits (in some areas, the first or second parietal cells lining the glands were also involved); III, damage extending well beyond the gastric pits but involving ~50% of the thickness of the gastric mucosa; and IV, extensive gastric mucosal damage involving >50% of the thickness of the gastric mucosa. Damage to the antral mucosa was defined as follows: 0, all cells were intact and normal in appearance; I, damage involved the surface epithelial cells and the uppermost two or three cells lining the glands; II, damage greater than I but involving ~50% of the thickness of the of the mucosa; and III, damage involving >50% thickness of the antral mucosa. Using an ocular micrometer, the length of each type of cellular damage and the section length were determined for each specimen. The percentage of the mucosal length with each grade of damage was then calculated. ESTlMATIONOFTHEAMOUNTOFMUCUSCOVERINGGASLESIONS. The thickness and length of mucus covering the surface of gastric lesions [types III and IV damage in the corpus and types II and III damage in the antrum) were measured on PAS-stained sections using an ocular micrometer in a microscope. Dehydration during the histologic preparation might cause mucus shrinkage artifacts; therefore, the measurements of mucus thickness represent relative values of mucus thickness among the different sections rather than true values. The thickness of the mucus was defined as follows: 0, no mucus; +,
TRIC MUCOSAL
Statistical Gross
and Histologic
Evaluation
of Gastric
Lesions Measurement of the area of gross gastric lesions. The area of gastric lesions (in square millimeters) was measured using a dissecting microscope with x 10 magnification and a l-mm’ grid eyepiece. Lesion measurements were made without knowing which animal group the stomach came from. The area of the corpus and antrum was measured by planimetry. The mucosal lesions in the corpus and antrum were then expressed as percent of the total corpus and antral area, respectively. Histologic analysis. Histologic tissue sections from all the groups were coded, randomized, and examined for gastric mucosal injury and mucus on the surface of gastric lesions. The code was not broken until all tissue sections were read. ESTIMATION OF GASTRIC MUCOSAL INJURY. The severity of gastric mucosal injury was evaluated on the sections
Vol. 90, No. 5, Part 1
Analysis
The Wilcoxon rank sum test was used throughout except for comparison of histologic scores in the upper and lower corpus (Table 2). In the latter instance, because data from the two areas were obtained in each rat, Student’s paired t-test was used. A probability level of < 0.05 was considered significant.
Results Gross Gastric Study
Lesions
shock and no transto hemorrhagic shock for 40 20 min (group 4) but not shed blood, there were very could be clearly recognized the surface of the mucosa
with hemorrhagic
fusion. In rats subjected min [group 2) or for receiving transfusion of few gastric lesions that as breaks (erosions) in
HISTOLOGY
May 1SRfj
Figure
1. Photograph of the stomach of a rat subjected to 20 min of hypotension [plus intragastric acid) but not receiving transfusion of shed blood. The corpus mucosa appears abnormal but it is difficult to recognize breaks in the muLosa [erosions]. The antrum is covered by a white (mucus?) laycar.
(Figure 1). Therefore the areas without a normal mucosal appearance were regarded as gastric “lesions” (Table 1). According to this definition, there were extensive lesions in the corpus and antrum in both groups 2 and 4 (the mean percent area involved with lesions ranging from 40% to 80%) but not in their respective controls (Wilcoxon rank sum test, p < 0.05). Comparison of the gastric lesions in the corpus with those in the antrum showed that the percent of gastric lesions in the corpus and the percent in the antrum were almost the same in group 2, but corpus lesions in group 4 were significantly greater than antral lesions (Wilcoxon rank sum test, p < 0.05). Study with hemorrhagic shock and reinfu-
Table
1. Gross
OF ISCHEMIC
GASTKIC: IN]LJRY
sion. In the animals subjected to hemorrhagic shock for 20 min and receiving transfusion of heparinized shed blood (group 5), heparinized albumin (group 6), and nonheparinized albumin (group 7), the gastric mucosal lesions, unlike those in group 2 and 4 rats without transfusion, could be recognized clearly as breaks (erosions) in the surface of the mucosa (Figure 2). Only these areas were measured as gastric lesions. In this respect, the gastric lesions estimated macroscopically in the studies with infusion cannot be equated with those in groups 2 and 4. The mean percent of mucosal area involved with lesions in groups 5, 6, and 7 was 19%, 20%, and 19%, respectively, in the corpus, and 10% 9%, and 9% respectively, in the antrum. There were no significant differences in the formation of gastric lesions either in the corpus or in the antrum among these three groups. In all studies, however, corpus lesions were significantly greater than antral lesions (Wilcoxon rank sum test, p < 0.05).
Histologic
Evaluation
of Gastric
Lesions
Comparison of histologic scores in the upper and lower corpus. In all groups, there were no significant differences (Wilcoxon rank sum test) between histologic scores in the proximal and distal corpus for any type of damage, except for types I and II in group 7 (Table 2). This finding indicates that hemorrhagic shock-induced gastric mucosal injury affects the proximal and distal corpus mucosa equally. Therefore, the upper and lower corpus scores for each type of damage in each rat were averaged, and further data analysis is based on these combined scores. Study with hemorrhagic shock and no transfusion. In animals subjected to either 40 min [group 2) or 20 min (group 4) of hypotension, there was greater damage of all types, except for type I, in both corpus and antrum in comparison with their respective controls [Table 3). This was especially true of
Gastric Lesions Reinfusion Shed No reinfusion
Corpus Antrum
1105
blood
No reinfusion
Group 1 (II= 7)
Group 2 (n = 6)
Group 3 (II= 6)
Group 4 (n = 6)
Group 5 (n = 7)
0.1 + 0.1 0
80 ? 9b 76 L 8b
0.4 2 0.2 4k2
64 ? 5"' 40 k 9"
19 +- 5' 10 t 4
Albumin heparin
of +
Group 6 (11 7) 20 2 4' 952
_
Albumin alone Group 7 (n = 7) 19 f 3’ 9f2
group 2, hypotension for 40 min; groups 4-7, hypotension the mean Groups 1 and 3, no hypotension: for 20 min. Each value represents -t standard error of the percent of the area involved with gastric lesions-areas without a gross normal appearance in nonreinfused rats (groups l-4) vs.clear-cut erosions in reinfused rats (groups 5-7).Significantly different from the respective control: ” p < 0.05," p < 0.01. ’Significant difference between corpus and antrum, p < 0.05.Wilcoxon rank sum test was used for statistical analvsis.
1106 ITOH ET AL.
Figure
GASTROENTEROLOGY
Vol. 90, No. 5, Part 1
Study with hemorrhagic shock and reinfusion. In the three reinfusion groups, transfusion of heparinized shed blood (group 5), heparinized albumin (group 6), and nonheparinized albumin (group i’), the extent of damage of all types in both corpus and antrum was similar and there were no significant differences among these groups (Kruskal-Wallis analysis of variance with contrasts) (Table 3). The extent of damage of all types in these three groups also showed no significant differences from those in group 4 (20 min of hypotension and no retransfusion) either in the corpus or the antrum. Deep damage in the antrum again was greater than in the corpus, varying from 52% to 55% versus 8% to 20%. This was especially true of the deepest damage (type III in the antrum and type IV in the corpus), with damage being significantly greater in the antrum in groups 5 and 7. Evaluation of mucus covering gastric lesions. This evaluation was performed on PASstained sections from animals in groups 4 and 5 (20 min of hypotension without and with reinfusion of shed blood) (Table 4). The percent of lesion length covered with thick [lo0 pm (+ + +)] or moderately thick [lo-100 pm (+ +)] mucus was greater in group 4 than in group 5 rats in both the corpus and the antrum, with greater coverage in the antrum than in the corpus. The converse was found with regard to lesions covered with no mucus (0) or thin [lo pm (+)] mucus. None of these differences, however, were statistically significant. When the percent of lesion length covered with ++ mucus and that covered with + + + mucus were combined, however, mucus coverage was significantly greater (Wilcoxon
2. Photograph of the stomach of a rat subjected to XI min of hypotension (plus intragastric acid) and receiving transfusion of shed blood. The corpus mucosal lesions can be recognized clearly as breaks in the surface of the mucosa (erosions). A few small lesions can be seen in the antrum.
deep damage, i.e., types III and IV in the corpus and types II and III in the antrum (which corresponds to type IV in the corpus). Deep damage was significantly greater in the antrum (types II and III) than in the corpus (types III and IV) in both groups 2 and 4. The extent of damage of all types was similar in groups 2 and 4 and there were no statistically significant differences between these two groups.
Table 2. Comparison of Histologic Scores in the Upper and Lower Corpus Reinfusion Shed No reinfusion Group 1 (n = 7) 0
I II III IV
No reinfusion
blood
heparin
+
Albumin alone
Group 2
Group 3
Group 4
Group 5
Group 6
Group 7
(n = 6)
(n = 6)
(n = 6)
(n = 7)
(n = 7)
(n = 7)
17 ? 11
17 r
27 2 8
3k3
72 2 10
16 *
72-4
63 k 16
45 k 8
18 k 5
27 ? 10
53 2 8
26 2 7
24 k 11
10
Albumin
of
28 k 12
42 k 7
32 -t- 7
41 2 10
121 12 k 6
6 2 10 12 ” 6 9?3
25 * 9
7
723
14 + 7 11 2 9
22 + 6
23 2 7
19 k 7
18 k 4
29 + 7
40 2 14
38 + 5”
69 ? 7
45 + 8
44 2 8
64 -r- 8
41 t
8
35 + 10
29 + 11
39 -t 70
0
33 2 11
0
12 t
7
16 + 9
14 + 7
0
17 2 6
0
20 2 9
11 + 5
11 + 5
0
3+2
0
5k3
0
9+9
0
928
0 lkl
3+3 14 + 12
323 12 + 5 0 0
Groups I and 3, no hypotension; group 2, hypotension for 40 min; groups 4-7, hypotension for 20 min. Each value represents mean 2 standard error. The upper and lower values in each row represent the values for the upper and the lower corpus, respectively. a Significant difference between upper and lower corpus paired values (p < 0.05, paired t-test).
May 1986
Table
HISTOLOGY
3. Histologic
Evaluation
OF ISCHEMIC
GASTRIC
Shed No reinfusion
I II 111
Group 2 (n = 6)
Group 3 (n == 6)
23 2 9 46 + 8 29 2 9
523 23 * 40
68 -t 12 25 + 10 623
13 + 3
0
42 ? 9 24 +- W
IV 0
6?5 0
24 t 9 77 * 9
I II
0
III
0
3 * 3”
Albumin alone
8 f 5”
20 r
10
12 -+ 5
12 + 7
25 t
5
M 2 9
28 t
5
57 + 7b
41 -+ 8
27 i- 9
52 ?
7
15 + 6”
14 ? 6
13 t
t
0.5
5
7 + 6
Group 7 (n = 7)
8+3 0
49 1- 17
18 + 16
0
17 t
14
28 2 16
46 ?
18 2 11
45 t
12
22 t
13
20 2 11
11 + 3
12 *
7
17 t
8
11 2 6
52 + l6”.
43 t
151
35 i
15
42 2 18”
16
525
27 t- 8” 70 f 9%”
-
Group 5 (n = 7)
0.5
Group 6 (n = 7)
of
Group 4 (n = 6)
725
0
,4lbumin heparin -___
blood
No reinfusion
Group 1 (n = 7)
221 0
1107
of Gastric Lesions Reinfusion
0
INfURY
(I
Groups 1 and 3. no hypotension; group 2, hypotension for 40 min; groups 4-7, hypotension for 20 min. Each value represents the mean + standard error of the percent of the mucosal length involved with gastric lesions. Significantly different from the respective control: different from respective corpus type IV lesion: ’p CC0.05, d p < 0.01.Wilcoxon rank sum test was ‘l p < 0.05, h p < 0.01. Significantly used for statistical analysis.
rank sum test, p < 0.05) in the antrum in group 4 (no reinfusion] than group 5 (83.6% * 16.4% vs. 30.2% -+ 13.7%). In addition, there also was a statistically significant difference between the corpus and antrum in group 4 (83.6% 2 16.4% vs. 30.5% -+ 17.0%). An example of this is shown in Figures 3A and 3B. In group 5 this difference did not quite reach statistical significance (30.2% ? 13.7% vs. 1.0% + 1.0%).
Table
4. Mucus Covering
Gastric
Lesions
Group 4: 20 min of hypotension with no reinfusion
Group 5: 20 min of hypotension with reinfusion of shed blood
0 antrum corpus
29 t 16
525 64 r
18”
78 + 14
12 t
12
41 t
+
antrum corpus
15
21 ‘-’ 14
523
f-t
antrum corpus
55 t
16
30 f
21 t
15
l-+1
28 -+ 14
0
14
+ -t +
antrum corpus + + and
antrum corpus
927
0
+ + + 84 -c 16”,” 31 +- 17
30 + 14h 1+1
Group 4: antral lesions, n = 5; corpus lesions, n = 6. Group 5: antral lesions, n = 6; corpus lesions, n = 7. Upper and lower rows in each set of paired figures represent percent of antral and corpus lesions covered with mucus. Significant difference between the antrum and corpus: ” p < 0.05, h p < 0.01, respectively. Significantly different from group 5 value: c p < 0.05. Wilcoxon rank sum test was used for statistical analysis.
Discussion In experimental studies of ischemic mucosal injury in the stomach, the animal is routinely subjected to withdrawal and retransfusion of blood. It has been pointed out in this model that the gastric mucosal injury tends to cluster in the corpus mucosa and affect the antral mucosa less severely. In this respect it resembles stress ulcer in humans. To explain this finding, several hypotheses have been proposed. These include differences in energy metabolism (5-71, mucosal blood flow (2,8,9), and disposal capacity of influxing H+ between the corpus and antral mucosa (10). Those investigations have been based on the gross appearance of gastric injury. Histologic studies were not performed. It was surprising, therefore, that the present histologic studies revealed more marked deep injury in the antrum than in the corpus. In all groups of rats subjected to 20 min of hypotension, whether reinfused or not and regardless of the fluid infused, mean percent histologic damage extending beyond the superficial mucosa in the different groups varied from 52% to 63% in the antrum versus 8% to 20% in the corpus. Thus the rat antrum, contrary to general belief, is more susceptible to acid injury during ischemia than the corpus. The present study did not provide any indication of why this is so. However, it did shed some light on why it is so difficult to recognize antral lesions by macroscopic examination alone. A mucus layer covered 71% of the length of the antral lesions but only 22% of the length of corpus lesions in rats subjected to the usual procedure of 20 min of hypotension plus retransfusion (Table 4, group 5). Thirty percent of the length of
1108
Figure
ITOH ET AL.
GASTROENTEROLOGY
3. Photomicrograph of PAS-stained sections from the corpus (A) and the antrum (B) of a group 4 retransfusion). A. There is extensive mucosal damage including approximately one-third of the mucosa. This lesion is only partially covered by mucus and cell debris. B. There is extensive mucosal the full thickness of the antral mucosa. This is entirely covered by a thick layer of mucus and cell
Vol. 90, No. 5, Part 1
rat (hypotension and no thickness of the corpus damage involving nearly debris.
May 1Wi
antral lesions was covered with a mucus layer >lO pm (+ + or + + + mucus) versus only 1% for corpus lesions. The mucus covering, by masking the breaks in the antral mucosa, explains, in part, why it is difficult to recognize erosions there. Perhaps if different criteria were used, antral injury could be identified grossly. However, nearly all investigators look for the clear-cut erosions seen in the corpus. In most studies of gastric mucosal injury only gross examination is used to evaluate the extent of mucosal injury. The findings of the present study indicate how misleading this can be. Similarly, in a histologic study of ethanol-induced gastric mucosal injury, Lacy and Ito (4) observed extensive surface mucus cell damage that could not be seen on gross examination. In addition, they were able to characterize histologically the gross “necrotic” injury (hyperemia and hemorrhage as well as necrosis). These two studies clearly demonstrate the necessity of histologic as well as gross evaluation of damage in gastric mucosal injury studies. Another purpose of these studies was to cletermine the effect of retransfusion of shed blood on gastric lesion formation. In animals subjected to hypotensive shock and no retransfusion of shed blood, it was difficult to recognize gross lesions. In contrast, in shocked animals that were reinfused with blood or albumin. lesions were readily identifiable as breaks in the mucosa (erosions). This was especially true in the corpus. In retransfused animals (group 5), the extent of deep histologic damage (types III and IV), 140/o, was in good agreement with gross corpus damage, 19%. This was not true of nonretransfused rats (group 4), in whom the gross corpus “lesion” scores markedly overestimated the true deep histo64% vs. 22%. Study of PAS-stained logic damage, sections revealed that + + or + + + mucus (>lO pm thick) covered 31% of the corpus lesions in the nonretransfused rats but only 1% in the retransfused rats. As suggested by Mersereau and Hinchey (ll), perhaps retransfusion, by restoring mucosal blood flow, results in a washing away of mucus that covers lesions, thus rendering the lesions more recognizable. In addition, gross and histologic damage is similar in both the corpus and antrum whether the retransfusion is with shed blood or an equal volume of heparinized or nonheparinized albumin. Using a stereomicroscope and a gastric chamber technique in a hemorrhagic shock-retransfusion study, Mersereau and Hinchey (11) reported that gastric mucosal lesions were formed earlier and were larger in the antrum than in the corpus. Although they did not study histologic features of the mucosal injury, their observations support the histologic findings in the present study. In all the studies reporting that the antrum is spared from gross gastric
HISTOLOGY OF ISCHEMICCXSTRIC INJURY
1109
injury, as in the present study, no in vivo microscopic observations were made during lesion formation as was done by Mersereau and Hinchey. (11)suggested that the onset Mersereau and Hinchey of ulcerations was surface cell necrosis of gastric mucosa brought about during hypotensive shock and, on retransfusion of shed blood, the necrotic cells are flushed off into the gastric lumen by the returned gastric mucosal blood flow. Thus, the retransfusion of shed blood renders the gastric lesions induced during hypotensive shock recognizable as breaks in the surface of the mucosa. The finding that histologic damage was as severe in the nontransfused (group 4) as in the transfused rats (group 5,Table 3) also was surprising. Recently there has been considerable evidence that, in many “ischemic” tissues, injury occurs not during the period of ischemia but during the reperfusion period. This has been clearly demonstrated in the cat intestine (12,13). Histologic studies revealed much greater mucosal damage if an intestinal segment that was rendered ischemic for 2 h (12) or 3 h (13) was use of allopurinol to reperfused for 1 h. Furthermore, inhibit xanthine oxidase, the enzyme that catalyzes superoxide formation, or superoxide dismutase to scavenge superoxide radicals. prevented the histologic damage on reperfusion (14). The concept is that during the ischemic period the necessary changes for free-radical formation develop--xanthine dehydrogenase changes to the xanthine oxidase form and adenosine triphosphate breaks down to hypoxanthine. With reperfusion of the previously ischemic tissue, adequate oxygen becomes available and there is a burst of free-radical formation (hypoxanthine + O2 in the presence of xanthine oxidase --+ xanthine + 0,) that is responsible for the tissue injury. We recently reported that administration of allopurinol or superoxide dismutase significantly decreased gross gastric mucosal injury in the hypotension-acid model (3). This appeared to provide strong evidence for a role of free radicals in ischemic gastric mucosal injury. That the gross and histologic findings in this study do not correlate well and that the damage occurs during the ischemic period (with the role of retransfusion being to render the gross lesions more recognizable), raise serious questions. First, do allopurinol and superoxide dismutase decrease the histologic mucosal injury? Second, if they do, is this protection seen in nonretransfused rats? Third, if the answer to the previous question is yes, are oxygen metabolites really involved in the mechanism of this tissue injury? There still could be a role for oxyradicals in gastric ischemic injury. The precise tissue oxygen tension needed to set the stage for oxyradical formation (xanthine oxidase and hypoxanthine formation) and the amount of oxygen needed for the
1110
ITOH ET AL.
final reaction are not known. It is conceivable that under the conditions of our study, the partial pressure of O2 in tissue fell sufficiently to cause formation of xanthine oxidase and hypoxanthine, and yet there was still a sufficiently high tissue oxygen tension for oxyradical formation. Additional studies will have to be performed to answer the specific questions raised and to test this hypothesis.
References 1. Starlinger M, Jakesz R, Matthews JB, Yoon Ch, Schiessel R. The relative importance of HCO,and blood flow in the protection of rat gastric mucosa during shock. Gastroenterology 1981;81:732-5. 2. Leung FW, Itoh M, Hirabayashi K, Guth PH. Role of blood flow in gastric and duodenal mucosal injury in the rat. Gastroenterology 1985;88:281-9. 3. Itoh M, Guth PH. Role of oxygen-derived free radicals in ischemic gastric injury in the rat. Gastroenterology 1985;88: 1162-7. 4. Lacy ER, Ito S. Microscopic analysis of ethanol damage to rat gastric mucosa after treatment with a prostaglandin. Gastro1982;83:619-25. enterology 5. Menguy R, Masters YF. Gastric mucosal energy metabolism and stress ulceration. Ann Surg 1974;180:538-48.
GASTROENTEROLOGY
Vol. 90, No. 5, Part
1
6. Menguy R, Masters YF. Mechanism of stress ulcer: influence of hypovolemic shock on energy metabolism in the gastric mucosa. Gastroenterology 1974;66:46-55. 7. Menguy R, Masters YF. Mechanism of stress ulcer. III. Effects of hemorrhagic shock on energy metabolism in the mucosa of the antrum, corpus, and fundus of the rabbit stomach. Gastroenterology 1974;66:1168-76. 8. Manabe T, Suzuki T, Honjo I. Changes of upper gastrointestinal blood flow after hemorrhage in rabbits. Surgery 1977; 81:446-52. 9. Shiraji S, Mueller TM, Hardy BM. Canine gastric acid secretion and blood flow measurement in hemorrhagic shock. Gastroenterology 1977;73:75-8. 10. Kivilaakso E, Fromm D, Silen W. Relationship between ulceration and intramural pH of gastric mucosa during hemorrhagic shock. Surgery 1978;84:70-7. 11. Mersereau WA, Hinchey EJ. Effect of gastric acidity on gastric ulceration induced by hemorrhage in the rat, utilizing a gastric chamber technique. Gastroenterology 1973;64:1130-5. 12. Schoenberg MH, Muhl E, Sellin D, Younes M, Schildberg FW, Haglund U. Posthypotensive generation of superoxide free radicals-possible role in the pathogenesis of the intestinal mucosal damage. Acta Chir Stand 1984;150:301-9. 13. Parks DA, Granger DN. Reperfusion injury in the ischemic small intestine (abstr). Fed Proc 1984;lOlO. 14. Parks DA, Bulkley GB, Granger DN, Hamilton SR, McCord Ischemic radicals.
JM.
injury in the cat small intestine: role of superoxide 1982;82:9-15. Gastroenterology