Human gastric myoelectric activity and gastric emptying following gastric surgery and with pacing

Human gastric myoelectric activity and gastric emptying following gastric surgery and with pacing

GASTROENTEROLOGY 1992;103:1811-1818 Human Gastric Myoelectric Activity and Gastric Emptying Following Gastric Surgery and With Pacing MICHAEL P. HOC...

819KB Sizes 16 Downloads 157 Views



Human Gastric Myoelectric Activity and Gastric Emptying Following Gastric Surgery and With Pacing MICHAEL P. HOCKING, STEPHEN B. VOGEL, and CHARLES A. SNINSKY Division of Gastroenterology, Hepatology, and Nutrition, Departments Florida College of Medicine: and Gainesville Veterans Administration

Postoperative gastric myoelectric activity, gastric emptying, and clinical course were correlated in 17 patients at high risk of developing gastroparesis after gastric surgery. In addition, an attempt was made to pace the stomach with an electrical stimulus and determine the effect of pacing on early postoperative gastric emptying. Gastric dysrhythmias (bradygastria, slow wave frequency ~2 cycles/min; tachygastria, slow wave frequency >4 cycles/min) persisted beyond the first postoperative day in 6 patients (35%). Delayed gastric emptying was identified by a radionuclide meal in 15 patients (88%), but symptoms of gastroparesis developed in only 6 of 15 (40%). Patients with postoperative gastroparesis had more frequent dysrhythmias than asymptomatic patients (67% vs. 18%), but these differences were not significant, although we cannot exclude a type II statistical error. Gastric rhythm was entrained in 10 of 16 patients (63%). Pacing increased the gastric slow wave frequency (3.1 vs. 4.1cycles/ min; P < 0.01)but did not improve gastric emptying (gastric retention at 60 minutes, 86% f 6% for control and 60% + 2% for paced). In conclusion, gastric dysrhythmias do not appear to play a major role in the development of postsurgical gastroparesis. Although gastric rhythm could be entrained in the majority of patients, pacing did not improve gastric emptying overall.


he stomach, like the heart, is governed by an electrical rhythm. Alterations in gastric rhythm may disturb gastric function, and spontaneous disturbances in gastric rhythm have been reported in association with impaired gastric emptying.‘-” Gastric dysrhythmias have been observed in the postoperative period,‘,’ but it is unknown whether they play a role in the development of postoperative gastroparesis. The incidence ofpostoperative gastroparesis varies with preoperative risk factors. Risk factors believed to predispose to gastroparesis include gastric outlet obstruction, diabetes mellitus, and longterm Hz-blocker therapya

of Surgery and Medicine, University of Medical Center, Gainesville, Florida

Gastric rhythms can be altered by an electrical stimulus, and electrical stimulation has been reported to speed gastric emptying, although this has not been studied in the early postoperative period.“-14 The purpose of this study was to evaluate gastric myoelectric activity and gastric emptyingprospectively in a group of patients at high risk of developing postoperative gastroparesis. In addition, we attempted to entrain the gastric rhythm with electrical stimuli and observed the effect of pacing on early postoperative gastric emptying. Materials and Methods This protocol was approved by the University of Florida Institutional Review Board and the Veterans Administration Subcommittee for Clinical Research, and informed consent was obtained from each patient. Seventeen patients undergoing primary or revisional gastric surgery at Shands Teaching Hospital or the Gainesville Veterans Administration Medical Center were included in the study. The clinical histories and procedures performed are listed in Table 1. Fifteen of the 17 patients (88%) had one or more risk factors for postoperative gastroparesis. These included preoperative gastric outlet obstruction or gastroparesis in 12 patients (71%), diabetes mellitus in 1 patient, and long-term Hz-blocker therapy in 10 patients (59%). Three pairs of stainless steel cardiac conductive sutures (0 Flexon; Davis & Geck, Danbury, CT) were placed at the time of surgery (Figure 1). The electrodes were placed at equidistant intervals along the greater curvature of the stomach or gastric remnant measured from the distal extent of the stomach. In patients undergoing vagotomy and drainage, the most distal electrodes were situated 5-7 cm proximal to the pylorus; in patients with partial gastrectomy, the distal electrodes were placed 2-5 cm proximal to the site of gastrectomy. The sutures were passed through the seromuscular layer of the stomach, and approximately 5-7 mm of each electrode was exposed and the suture secured to the stomach by “feathering” the polymer coating. The two sutures comprising each bipolar electrode pair were placed 5-10 mm apart and exteriorized via puncture 0 1992 by the American Gastroenterological OOlE-5085/92/$3.00





Table 1. Clinical Data of Study Participants Previous gastric surgery

Age Patient






2 3 4 5

71 62 65 33












43 68 61 48


11 12 13 14 15 16 17

Medical history Prepyloric ulcer, ulcerative colitis Duodenal carcinoma Gastric ulcer, diabetes Periampullary carcinoma Pyloric stenosis, Crohn’s disease Efferent limb stricture



Pyloric stenosis, inflammatory bowel disease Gastroesophageal reflux, chronic gastroparesis Recurrent duodenal ulcer Duodenal ulcer Duodenal ulcer Duodenal ulcer Marginal ulcer Stenotic gastroenterostomy

65 57 64


Duodenal ulcer Duodenal ulcer Prepyloric ulcer


Indication for surgery



None None None None

Carcinoma Intractability Carcinoma GO0



TV, A-Bll;



Present operation TV, A-B1



Revagotomy, gastrectomy, revision RY A-B1



Nissen fundoplication

PCV None None None TV, A-B1 TV, STG-Bll

Intractability GO0 GO0 Intractability COO GO0

None None None


TV, A-B1 TV, P TV, P TV, A-B11 Partial gastrectomy-Bll Revision of gastroenterostomy TV, P TV, GE TV, P

A, Antrectomy; Bl, Billroth I gastroduodenostomy; Bll, Billroth II gastrojejunostomy; GE, gastroenterostomy; GOO, Gastric outlet obstruction; P, pyloroplasty; PCV, parietal cell vagotomy; PPW, pylorous-preserving Whipple procedure; RY, Roux-en-Y gastrojejunostomy; TV, truncal vagotomy; W, Whipple procedure (pancreaticoduodenectomy).

sites in the left upper quadrant of the abdominal wall. The electrodes were removed by gentle traction prior to discharge from the hospital. Myoelectric recordings were obtained for approximately 1 hour daily, beginning on the first to third postoperative day (POD) and continuing until the patient was ready for discharge or was otherwise stable (POD 5-61). Recording time varied from 214 to 1830 minutes per patient (mean f SEM, 567 + 112 minutes). The recordings

were obtained using a physiographic recorder (Sensormedies R612, Anaheim, CA) with a time constant of 1 second and a paper speed of 0.25 mm/s. No attempt was made to interfere with clinical care, and thus the decision to remove the nasogastric tube, begin oral feeding, etc., was made by the patient’s clinician. In addition, analgesics or other medications were not restricted during the study period. The myoelectric recordings were visually analyzed and the slow wave frequency was determined by counting the slow waves for at least five l-minute intervals during each recording. A dysrhythmia was identified and defined as bradygastria if the slow wave frequency was ~2 cycles/ min and tachygastria if the slow wave frequency was >4 cycles/min and was sustained for a minimum of 1 minute. The timing of the gastric emptying studies was determined by the patient’s clinical course. When the patient was tolerating an oral diet, two solid-phase gastric emptying studies were obtained in an identical fashion in random order on consecutive days with and without electrical stimulation. The first solid emptying study was

obtained on POD 10 + 1.2 (range, POD 5-23). The meal consisted of an egg sandwich (275-300 kcal; 30 g carbohydrate, 12 g protein, 15 g fat) and was labeled with 0.5-1.0 mCu of ““‘Tc-sulfur colloid, as previously described.15 A single-camera technique was used, and delayed gastric emptying was defined as a half-time of gastric emptying >90 minutes. Electrical stimulation was delivered through the proximal electrode pair with a square wave stimulus provided by an experimental gastrointestinal pacing unit (GIX004; Medtronics, Minneapolis, MN) at an amplitude of 2 mA and a pulse width of 300 milliseconds. The frequency of the pacing stimulus varied from 0.3 to 1.6 cycles/min faster than the native gastric slow wave frequency just before pacing. Successful entrainment was determined when there was an increase in the slow wave frequency to that of the pacing stimulus and a phase locking of the slow wave with the pacemaker artifact (Figure 2). The pulse width, stimulus rate, and polarity of the electrodes were varied to improve successful entrainment. No attempt was made to control the polarity of the electrical stimulus. The day of the first pacing attempt was not rigorously controlled but was usually 2-3 days before the gastric emptying study. The electrical stimulus was usually applied approximately 15 minutes before the initiation of the gastric emptying study, with the meal administered once it was determined that the gastric rhythm had been entrained. Statistical analysis was performed with a statistical software package (True Epistat; Epistat Services, Richardson, TX) using Student’s t test for paired variables and x2 analy-

December 1992


Gastric Emptying

Figure 1. Schematic drawing of electrode(G) placement in a patient with a truncal vagotomy and pylorplasty.

sis. Statistical significance was defined as 95% confidence (P < 0.05). All values are expressed as the mean k SEM.

Delayed gastric emptying of the radionuclide meal was noted in 15 of 17 patients (88%). Only 3 of the 15 patients (20%) with delayed gastric emptying of the meal had dysrhythmias present during the study. Symptoms of gastroparesis (nausea, vomiting, inability to tolerate oral intake) occurred in only 6 of 15 patients (40%) with delayed emptying of the radionuclide meal. Symptoms were severe in 4 patients (24%), as defined by the inability to tolerate oral intake adequate to maintain hydration by POD 24. Two additional patients had moderate symptoms and were able to tolerate oral intake by POD 6 but required prokinetic drugs for symptomatic relief. Mechanical factors were excluded, and delayed gastric emptying eventually resolved in all patients without further operative intervention. Four of the 6 patients with postoperative gastroparesis (67%) had dysrhythmias identified on postoperative recordings compared with 2 of 11(18%) without symptoms of gastroparesis. Of the 4 patients with dysrhythmias present for more than 20% of total recording time, 2 (50%) had symptoms of gastroparesis. However, the patient with the greatest percentage of postoperative dysrhythmias (63%) had no symptoms of gastroparesis and was discharged tolerating a regular diet on POD 8. Pacing

Results The electrodes were well tolerated, and there were no complications related to the electrodes. Myoelectric

Pacing was first attempted a mean of 7.1 f 0.7 postoperative (range, POD 3-13). Gastric



Gastric slow wave frequency ranged from 0.8 to 9 cycles/min. The slow wave frequency varied from patient to patient and from day to day in each patient. Excluding dysrhythmias, mean slow wave frequency was 3.0 f 0.1 cycles/min (range, 2.5-3.6 cycles/min per patient). Dysrhythmias were identified in 8 of the 17 patients (47%). The dysrhythmias consisted of bradygastria in 1 patient, tachygastria in 4, and bradygastria and tachygastria in 3 (Figure 3A and B). Dysrhythmias occurred in 0 of 3 patients after partial gastrectomy without vagotomy, 2 of 6 patients 1 of 2 paafter vagotomy and partial gastrectomy, tients who underwent neither; and 5 of 6 patients after vagotomy and drainage. Of the 8 patients with an identified dysrhythmia, 2 had dysrhythmias present during only one recording session (the first postoperative day), 4 had dysrhythmias identified on two to five separate occasions, and 2 had dysrhythmias during nine or more sessions. The percentage of total recording time occupied by dysrhythmias ranged from 6% to 63%.

GI t 7cm +

G2 0.1 mV 1


2 min

Figure 2. Myoelectric recording showing entrainment of the gastric slow wave frequency via an electrical stimulus applied to Gl, beginning at the first arrow. Gastric slow wave frequency was 3.5 cycles/min before stimulation, and the stimulus was applied at a rate of 4.5 cycles/min. The pacemaker is turned off at the second arrow. Note that the slow wave frequency remains at the paced rate for approximately one minute before returning to the prepaced frequency (third arrow).




c G2 4

0.1 mV




0.1 mV


G3 0


B 2 min





2 min

Figure 3. (A) Myoelectric recording in a patient 3 days after truncal vagotomy and pyloroplasty showing tachygastria at a rate of approximately 4.6 cycles/mitt. Note that the tachygastria is first observed in the most distal lead and propagates proximally. Arrows

indicate the onset of tachygastria in each lead. (B)Myoelectric recording in a patient 4 days after revision of a gastroenterostomy with a previous truncal vagotomy and antrectomy, showing a bradyarrhythmia at 1.7cycles/min (black arrows) alternating with a second rhythm at 1.6 cycles/min (white arrows). Note the different configuration and polarity of the slow waves in the first electrode. The dashed lines show propagation of the slow waves distally.

rhythm could be entrained in 10 of 16 patients (63%). Pacing could not be performed in 1 patient because electrical stimulation caused painful rectus abdominus muscle contractions. The mean POD of the first pacing attempt was not different in patients who could be paced (6.9 f 1.0) and those who could not (7.8 + 1.0). One of the 10 patients who could be paced initially could not be paced during his subsequent solid gastric emptying study, and another could not tolerate solid food. Overall, pacing increased the gastric slow wave frequency from 3.1 + 0.1 cycles/min to 4.2 f 0.2 cycles/min (P < O.Ol), but solid gastric emptying was unchanged (gastric retention at 60 minutes, 86% + 6% for control and 90% +- 2% for paced). Only 1 of the 8 patients with baseline and paced solid emptying studies had dysrhythmias during the control emptying study. Discussion Gastric dysrhythmias occurred in 35% of patients after POD 1. This is a conservative estimate because recordings were obtained for only 1 hour daily. Others have reported that all patients have gastric dysrhythmias within the first 24 hours if they are monitored continuously.7 However, these abnormalities resolved spontaneously within 7-25 hours in patients not undergoing gastrectomy or vagotomy. The dysrhythmias that we observed occurred most

frequently in patients undergoing truncal vagotomy. The vagus is thought to suppress ectopic pacemakers, evidenced by the development of multifocal ectopic pacemakers in the early postoperative period after vagotomy.‘6~17 These pacemakers generally arise in the gastric antrum, probably related to the increased excitability of antral smooth muscle cells. In fact, the antrum can be driven to a higher slow wave frequency with an electrical stimulus than the gastric corpus.‘*-21 In addition, the absolute refractory period, i.e., the time during which the smooth muscle cannot be stimulated by a premature electrical stimulus, is shorter in the antrum than in the corpus.22 Gastric emptying was delayed by radionuclide scan in 88% of our patients. It is not surprising that most patients had evidence of gastric retention of the test meal because the emptying studies were performed in the immediate postoperative period, with the patients often still receiving analgesic medication. However, symptoms of gastroparesis developed in only 35% of patients. This is similar to the findings in diabetes mellitus, in which delayed gastric emptying of solids is often asymptomaticz3 Symptoms were considered severe in 24% of patients, as defined by the inability to tolerate oral intake adequate to maintain hydration by POD 24. The latter represents 2SD beyond the mean time for patients undergoing a gastrojejunostomy to tolerate a liquid diet sufficient



to sustain nonoliguric renal output, as reported by Fromm et alz4 Although the 24% incidence of severe clinical gastroparesis seems somewhat high, we specifically studied patients at high risk to develop postoperative gastroparesis, and this number is within the range reported by others.Q,25 Tachygastria has been implicated in the etiology of idiopathic gastroparesis. Because tachygastria usually arises from a distal antral focus, slow waves arising from the ectopic focus are propagated in an orad direction and suppress the normal pacemaker.‘8*‘Q Any contractions that occur will also travel in an orad direction, in conjunction with the gastric slow waves, and thus potentially interfere by shortening with gastric emptying.” Furthermore, the gastric slow wave cycle, tachygastria allows less time for action potentials to develop and thus weakens any contractions that do occur.4 Although gastric dysrhythmias occurred more frequently in patients with postoperative symptoms, the difference in incidence did not reach statistical significance. Because of the small numbers involved, the chance for a type II statistical error exists. Nevertheless, we believe that dysrhythmias did not play a major contributory role in the development of gastroparesis. Severe gastroparesis developed in 1 patient with no identified dysrhythmias, and the patient with the highest frequency of dysrhythmias, which occupied nearly two thirds of his total recording time, had absolutely no symptoms of gastroparesis. The concept of gastric pacing is not new. Based on the success of cardiac pacing, Bilgutay et al. proposed gastrointestinal “pacing” via a transluminal stimulator nearly 30 years ago.26 A 7-lo-mA current was delivered at a frequency of 50 Hz, or 1000 times the normal human gastric slow wave frequency. The authors reported more rapid resolution of postoperative ileus with electrical stimulation, and the stimulator was marketed commercially. However, other investigators were unable to show any beneficial effect of gastrointestinal stimulation on the duration of postoperative ileus or on upper gastrointestinal motor activity, and clinical interest in the concept of gastrointestinal pacing waned.27-2Q In 1972, Kelly and LaForce showed that delivery of an electrical stimulus at a frequency and rhythm similar to those of the native slow wave frequency could consistently entrain the gastric rhythm in a canine mode13’ In 1976, Sarna et al. showed that orad pacing of the canine stomach slowed gastric emptying of solids and liquids and postulated that pacing might be clinically useful for the treatment of abnormal motility states.31 In 1985, Waterfall et al. reported the orad and





aborad entrainment of the human stomach.‘o They noted that the stomach exhibited “memory,” with the gastric rhythm remaining at the paced frequency for a variable period after the electrical stimulus was withdrawn. We noted similar findings (see Figure 2). Although we were able to document entrainment of the gastric slow wave in the majority of patients, we observed no beneficial effects on gastric emptying. Nor did we observe any beneficial effect of electrical stimulation at or near the native slow wave frequency in the patients whose gastric rhythms were not entrained. There are three preliminary reports regarding the effect of gastric pacing on gastric emptying.“-l3 In those studies, improved gastric emptying was reported with pacing in patients with diabetic gastroparesis and in a dog model consisting of a vagotomy and infusion of glucagon. Our interpretation of this discrepancy is that gastric pacing may be of benefit only in a subpopulation of patients with gastroparesis. This is not surprising because pacing is reported to generate slow waves but not action potentials.20*30 The reason for our inability to pace all patients is unknown but might be related to when the pacing attempts were initiated. Miedema et al. reported that 9 of 10 patients with gastric electrodes placed at the time of cholecystectomy could be paced on POD 1.14 However, by POD 3 only 4 of 10 patients could be paced. The authors postulated that this decreased ability to entrain the gastric rhythm was related to a decreased sensitivity to pacing caused by alterations in the neurohormonal milieu, specifically decreases in serum catecholamine levels. Because of the variable time of initiation of the first pacing attempt in the current study, we cannot comment on this phenomenon further. However, the mean POD of the first pacing attempt was not different in patients who could be paced and those who could not. In summary, we found that gastric dysrhythmias occurred after POD 1 in more than one third of patients. Although radionuclide gastric emptying was slowed in the early postoperative period in the majority of patients, only a minority of these patients developed symptoms of gastroparesis. Dysrhythmias did not appear to play a major role in the development of postsurgical gastroparesis. Finally, although the majority of patients could be paced with an electrical stimulus, entrainment of the gastric slow wave frequency did not improve gastric emptying overall. Because of the relatively small number of patients involved and the limitations imposed by a clinical study, we regard our results as preliminary. Nevertheless, we feel that electrical pacing of the stomach using the parameters studied is unlikely to be of benefit in treating postsurgical gastroparesis.




References 1. Telander

RL, Morgan KG, Kreulen DL, Schmalz PF, Kelly KA, Szurszewski JH. Human gastric atony with tachygastria and gastric retention. Gastroenterology 1978;75:497-501. 2. You CH, Lee KY, Chey WY, Menguy R. Electrogastrographic study of patients with unexplained nausea, bloating, and vomiting. Gastroenterology 1980;79:311-314. 3. You CH, Chey WY, Lee KY, Menguy R, Bortoff A. Gastric and small intestinal myoelectric dysrhythmia associated with chronic intractable nausea and vomiting. Ann Intern Med

16. 17.

18. 19.

1981;95:449-451. 4. You CH, Chey WY. Study of electromechanical




8. 9.

10. 11.

activity of the stomach in humans and in dogs with particular attention to tachygastria. Gastroenterology 1984;86:1460-1468. Reynolds RPE, Bardakjian BJ, Diamant NE. A case of antral tachygastria: symptomatic and myoelectric improvement with gastroenterostomy and domperidone therapy. Can Med Assoc J 1983;128:826-829. Geldof H, Van der Schee EJ, Van Blankenstein MV, Grashuis JL. Electrographic study of gastric myoelectrical activity in patients with unexplained nausea and vomiting. Gut 1986;27:799-808, Sarna SK, Bowes KL, Daniel EE. Post-operative gastric electric control activity (ECA) in man. In: Daniel EE, ed. Proceedings of the 4th International Symposium on Gastrointestinal Motility. Vancouver: Mitchell, 1987. Stoddard CJ, Smallwood RH, Duthie HL. Electrical arrhythmias in the human stomach. Gut 1981;22:705-712. Smale BF, Copeland JG, Reber HA. Delayed gastric emptying after operation for obstructing peptic ulcer disease: the influence of cimetidine. Surgery 1984;96:592-598. Waterfall WE, Miller D, Ghista DN. Electrical stimulation of the human stomach (abstr). Dig Dis Sci 1985;30:799. Bellahsene B-E, Schirmer B, Updike 0, McCallum R. Effect of electrical stimulation on gastric emptying (abstr). Dig Dis Sci

20. 21.



24. 25.

1976;56:1413-1419. 26. Bilgutay AM, Wingrove R, Griffen WO, Bonnabeau

27. 28. 29.

1987;32:902. 12. Bellahsene

B-E, Lind CD, Schirmer BD, Updike OL, McCallum RW. Acceleration of gastric emptying in “gastroparetic” dogs using electrical stimulation of the stomach (abstr). Gastroenterology 1988;94:A31. TL, Schirmer BD, Bellahsene BE, Updike OL, 13. Courtney McCallum RW. Gastric electrical stimulation as a possible new therapy for patients with severe gastric stasis (abstr). Gastroenterology 1991;100:A822. 14. Miedema BW, Sarr MG, Kelly KA. Pacing the human stomach. Surgery 1992;111:143-150. 15. Martin JL, Beck WJ, McDonald AP, Carlson GM, Mathias JR. 99m Tc-Labeled solid phase meal: a quantitative clinical mea-

surement of human gastric emptying. J Clin Gastroenterol 1983;5:315-319. Kelly KA, Code CF. Effect of transthoracic vagotomy on canine gastric electric activity. Gastroenterology 1969;57:51-59. Nelsen TS, Eigenbrodt EH, Keoshian LA, Bunker C, Johnson L. Alterations in muscular and electrical activity of the stomach following vagotomy. Arch Surg 1967;94:821-835. Code CF, Marlett JA. Canine tachygastria. Mayo Clin Proc 1974;49:325-332. Kim CH, Azpiroz F, Malagelada J-R. Characteristics of spontaneous and drug-induced gastric dysrhythmias in a chronic canine model. Gastroenterology 1986;90:421-427. Sarna SK, Daniel EE. Electrical stimulation of gastric electrical control activity. Am J Physiol 1973;225:125-131. Kelly KA. Differential responses of the canine gastric corpus and antrum to electric stimulation. Am J Physiol 1974; 226:230-234. Sarna SK, Daniel EE. Threshold curves and refractoriness properties of gastric relaxation oscillator. Am J Physiol 1974;226:749-755. Keshavarzian A, Iber FL, Vaaeth J. Gastric emptying in patients with insulin-requiring diabetes melitus. Am J Gastroenterol 1987;82:29-35. Fromm D, Resitarits D, Kozol R. An analysis of when patients eat after gastrojejunostomy. Ann Surg 1988;207:14-20. Donovan I, Alexander-Williams J. Postoperative gastric retention and delayed gastric emptying. Surg Clin North Am

30. 31.

RC Jr, Lillehei CW. Gastro-intestinal pacing: a new concept in the treatment of ileus. Ann Surg 1963;158:338-348. Quast DC, Beall AC Jr, DeBakey ME. Clinical evaluation of the gastrointestinal pacer. Surg Gynecol Obstet 1965;120:35-37. Moran JM, Nabseth DC. Electric stimulation of the bowel: a controlled clinical trial. Arch Surg 1965;91:449-451. Berger T, Kewenter J, Kock NG. Response to gastrointestinal pacing: antral, duodenal, and jejunal motility in control and postoperative patients. Ann Surg 1966;164:139-144. Kelly KA, LaForce RC. Pacing the canine stomach with electric stimulation. Am J Physiol 1972;222:588-594, SarnaSK, Bowes KL, DanielEE. Gastric pacemakers. Gastroenterology 1976;70:226-232.

Received March 18,1992. Accepted June 30,1992. Address requests for reprints to: Michael P. Hocking, M.D., Surgical Service (112G), Veterans Administration Medical Center, Gainesville, Florida 32608. Supported in part by the Medical Research Service of the Department of Veterans Affairs.