Suc B, Msika S, Piccinini M, Fourtanier G, Hay J, Flamant Y, Fingerhut A, Fagniez P, Chipponi J, for the French Associations for Surgical Research. Octreotide in the Prevention of Intra-abdominal Complications Following Elective Pancreatic ResectionA Prospective, Multicenter Randomized Controlled Trial. Arch Surg. 2004;139(3):288-294. doi:10.1001/archsurg.139.3.288
Prophylactic administration of octreotide acetate decreases the rate of postoperative intra-abdominal complications (IACs) after elective pancreatic resection.
Single-blind, controlled, randomized trial.
Multicenter (N = 20) trial in France.
Of 230 randomized patients undergoing pancreatoduodenectomy and pancreatic enteric anastomosis or distal pancreatectomy for either malignant or benign tumor or chronic pancreatitis, 122 were allotted intraoperatively to receive octreotide; 108 served as controls.
All 230 patients were analyzed. Both groups were comparable except that significantly more patients in the octreotide group had biological glue injected into the main pancreatic duct alone (P<.001) or reinforcing the pancreatic enteric anastomosis (68% [83/122] vs 39% [42/108]; P = .002). Fewer patients (P = .08) in the octreotide group sustained 1 or more IACs (22% vs 32%). In subgroup analysis, octreotide significantly reduced the rate of patients sustaining 1 or more IACs when the main pancreatic duct diameter was less than 3 mm (P<.02), when pancreatojejunostomy was performed (P<.02), or both (P<.02). No significant differences were found regarding IAC severity. Twenty-three patients (10%) died postoperatively, 16 (70% of deaths) of whom had 1 or more IACs. The only independent risk factor for IACs found on multivariate analysis was pancreatoduodenectomy compared with distal pancreatectomy (P<.01) (odds ratio, 3.54 [95% confidence interval, 1.44-8.65]).
Our results suggest that octreotide is not necessary for all patients undergoing pancreatic resection; it could be useful when the main pancreatic duct is less than 3 mm in diameter and when pancreatoduodenectomy is completed by pancreatojejunostomy.
Mortality and morbidity rates after pancreatic resection range from 0%1 to 10.2%2 and from 20%1 to 40%,3 respectively. The most common major complication is pancreatic fistula,1,3 usually related to persistent pancreatic secretion, which hinders the healing of the pancreatic stump. Octreotide and vapreotide, synthetic somatostatin analogues, inhibit the exocrine secretion of the pancreas4 and therefore might lower the rate of postoperative pancreatic fistula by decreasing the secretion of the pancreatic remnant. Of the 7 large (>100 patients) reported controlled trials5- 11 that tested octreotide5- 10 or vapreotide11 in this setting, 4 European multicenter studies5- 8 have shown a significant decrease in the rate of postoperative intra-abdominal complications (IACs), pancreatic fistula, and extra-abdominal complications (EACs) in patients undergoing pancreatic resection treated with octreotide. On the other hand, the 3 American, high-volume (2 single-center9,10 and 1 multicenter11) studies found no statistically significant difference in the postoperative course between the 2 groups whether or not they were treated with octreotide9,10 or vapreotide.11 Whether and how somatostatin analogues really work, however, remain a matter of debate, because the drugs act not only on pancreatic secretion but also on gastric, biliary, and jejunal secretions and intestinal motility.4 We therefore undertook a large, prospective, multicenter randomized trial to determine whether octreotide acetate decreased the rate of patients with IACs and/or their severity7 and to highlight the patients at risk who might benefit most from this inhibitor of pancreatic secretion.
Between August 1, 1993, and January 31, 1998 (54 months), 230 consecutive patients (131 men and 99 women; sex ratio, 1.32; mean ± SD age, 56.3 ± 13 years; age range, 16-81 years) undergoing either pancreatoduodenectomy followed by anastomosis or distal pancreatectomy were enrolled in this single-blind (without placebo), multicenter trial. Twenty surgical units (13 university and 7 general hospitals) participated, but all centers did not begin or finish on the same date. The median number of patients included per center was 8 (range, 2-31), and the median duration of inclusion was 41 months (range, 1-55 months), attesting to a preponderance of low-volume centers, defined as 5 patients per year2 or fewer than 10 patients per year.12
Patients 18 years or older were included whether or not they had preoperative biliary drainage or intraoperative septic factors (infected bile or pancreatic juice). Pancreatic resection was performed for either malignant or benign disease (pancreatic, biliary, ampullar, or duodenal tumor or chronic pancreatitis) (Table 1). Patients were not included if they underwent either resection for acute pancreatitis or trauma or other types of resection, such as simple tumor excision, total or central pancreatectomy, pancreatoduodenectomy without immediate pancreatodigestive anastomosis, and duodenum-preserving pancreatectomy.
The technical aspects of pancreatoduodenectomy were left to surgeons' preferences, but immediate anastomosis with the jejunum or the stomach was mandatory for inclusion eligibility. The pylorus could be preserved or not, and in the latter case, vagotomy was recommended.
Distal (tail or body tail) pancreatectomy was performed with or without preservation of the spleen. The pancreas remnant was either suture closed or anastomosed to a Roux-en-Y loop. In both pancreatoduodenectomy and distal pancreatectomy, fibrin glue (Tissucol; Baxter, Vienna, Austria) mixed with aprotinin, injected through a double-barreled syringe (Duplojet; Baxter) (3-5 mL), could be used to seal either the pancreatic duct or the pancreaticoenteric anastomosis.
Resection extended to other organs (colon, liver, superior mesenteric-portal venous confluence, kidney, adrenal gland, and diaphragm), and resection of nearby organs extended to the pancreas was allowed (Table 1 and Table 2). Abdominal drainage was recommended. All patients had single-shot intravenous antibiotic prophylaxis administered at anesthetic induction.
The main end point was the rate of patients with 1 or more postoperative IACs7 as diagnosed during the postoperative period (entire hospital stay and 30 days after discharge), during subsequent operation, and/or at autopsy. The IACs included pancreatic, biliary, or digestive tract fistula, intra-abdominal collections (either infected [abscess] or not), acute pancreatitis, and intra-abdominal or digestive tract hemorrhage. Pancreatic fistula was defined either chemically9 as fluid obtained through drains or percutaneous aspiration containing at least 4 times normal serum values of amylase for 3 days, irrespective of the amount of output and the date of appearance, or clinically and radiologically as anastomotic leaks demonstrated by fistulography or upper gastrointestinal tract hydrosoluble contrast studies in the case of pancreatogastrostomy. Biliary fistula was diagnosed by the distinctive color of discharge containing bilirubin or by fistulography. Gastrojejunal fistula was diagnosed by fistulography, contrast follow-through studies, or contrast-enhanced computed tomography (CT). Intra-abdominal collections were diagnosed by CT-guided needle aspiration, cultures, and amylase dosage of contents. Acute pancreatitis was defined chemically as increased values of serum amylase (4-fold normal), lipase (3-fold normal), and urine amylase (4-fold normal) dosed every other day for 8 days and confirmed by CT. Digestive tract hemorrhage was defined as blood exiting through the nasogastric tube and confirmed on fibroendoscopy or arteriography and intra-abdominal hemorrhage by exteriorization of blood through drains on arteriography or both. The secondary end point was EACs isolated or associated with IACs. Severity of IACs was evaluated by the mortality rate, the rate of subsequent surgical operations and/or percutaneous drainage for collections, and the duration of hospital stay.
Patients were randomized within 8 strata according to 3 variably combined risk factors: type of resection (pancreatoduodenectomy or distal pancreatectomy),7 type of pathologic finding (tumor or chronic pancreatitis),5,6 and texture of the pancreatic remnant parenchyma (normal or fibrotic).5,13 Normal parenchyma was defined as healthy, soft, or friable pancreatic tissue. Other risk factors included the diameter of the main pancreatic duct as measured with a ruler on the severed surface after pancreatic resection (≥3 mm or <3 mm13) and the type of pancreatic anastomosis after pancreatoduodenectomy (pancreatojejunostomy or pancreatogastrostomy).14,15
Random assignment, balanced every 4 patients within each stratum and in each center, was made in the operating room after resection but before restoration of digestive tract continuity or closure of the remnant stump in distal pancreatectomy once the surgeon had checked the inclusion and noninclusion criteria. Allotment to octreotide or not, as generated by computerized random number tables, was decided by a telephone call to the coordinating center,16 which collected and processed all data.
Octreotide acetate, 100 µg, was injected subcutaneously during the operation once randomization took place and administered every 8 hours at the same dose and by the same route for 10 days (300 µg/24 h).
Patients were not aware of the treatment arm to which they were allotted (octreotide injection was considered as part of postoperative treatment), but the surgeon performing the operation, obviously, was (single-blind study without placebo). Complications, however, were assessed by a physician who was unaware of the allotted treatment.
Our objective was to decrease the rate of patients with 1 or more postoperative IACs from 40%3 to 20%1 in a 1-tailed explanatory analysis with an α risk set at .05 and 90% power (β risk = .10).17 Within this framework, 88 patients were required in each group (a total of 176 patients).17 The study was discontinued when the minimum number of patients was attained in both arms, even if 1 or several strata had not been completed within each center, possibly leading to unequal numbers of patients in each arm.
The validity of data was checked randomly (1 in 5 patients) by a quality control officer (surgical resident student in applied sciences research).
Percentages were compared with either the χ2 test or Fisher exact test as appropriate. Continuous variables were compared with the unpaired, 2-tailed t test, analysis of variance, or the Mann-Whitney U test as appropriate. The Mantel-Haenszel adjustment test18 was used to check whether statistically significant differences in preoperative or intraoperative patient characteristics influenced outcome or not. The center effect was studied, as should be done in all multicenter trials.17 Risk factors were analyzed according to univariate and multivariate analysis using the stepwise logistic regression model. Risk factors with P≤.25 in univariate analysis were retained for multivariate analysis.
The ethics committee of the coordinating center approved this study, and informed consent was obtained preoperatively for all patients. No financial support was provided for this trial.
Of 230 randomized patients, 122 were allotted to receive octreotide and 108 were controls. No patient was withdrawn after randomization.
No statistically significant differences were found between the 2 groups regarding the preoperative and intraoperative characteristics except for injection of biological glue into the main pancreatic duct alone or in association with the application of biological glue around the pancreaticoenteric anastomosis, both performed more often in the octreotide group (83 [68%] of 122 vs 42 [39%] of 108, P<.001 and P = .002, respectively) (Table 1 and Table 2).
Fewer patients in the octreotide group had 1 or more IACs (27 [22%] of 122; 95% confidence interval [CI], 15%-29%; vs 35 [32%] of 108; 95% CI, 23%-41% in the control group), but this difference, although close, failed to reach statistical significance (P = .08). Even after statistical adjustment18 for statistically significantly different patient characteristics (fibrin glue injected into the main pancreatic duct and/or around the pancreatic anastomosis) (Table 2), this difference remained nonsignificant. Differences in the rates of the other complications, analyzed one by one in both groups, were not found to be statistically significant (P = .79), notably the rate of pancreatic fistula (17% [21/122] in the octreotide group vs 19% [20/108] in the controls) (Table 3).
The differences found between the 2 groups in the postoperative mortality rate (12% [15/122] vs 7% [8/108]), the rate of subsequent operations (20% [24/122] vs 19% [21/108]), or the duration of hospital stay (19 vs 19.5 days) were not statistically significant (Table 3). Twenty-three patients (10%) died in the postoperative period. Sixteen patients (70% of the deaths) died with the following IACs: intra-abdominal hemorrhage (n = 4), pancreatic fistula with EACs (n = 5), and intra-abdominal hemorrhage and pancreatic (n = 4), biliary (n = 1), gastrojejunal (n = 1), or colonic fistulae (n = 1). Three of 16 of the deceased patients died before postoperative day 10. Seven patients (30% of all deaths) died with EACs alone, including 4 with cardiovascular damage (2 myocardial infarctions, 1 pulmonary embolism, and 1 stroke) and 3 with pneumonia, whereas 6 other patients had nonfatal EACs (3 with pulmonary and cardiac failure, 1 with catheter-related septicemia, and 1 with transient stroke). Two of these 6 EACs were associated with IACs. Mortality was higher in extended resections (4 [29%] of 14) than in nonextended resections (19 [9%] of 216). Forty-five patients (20% of all patients and 73% of patients with complications) underwent a subsequent operation and/or percutaneous drainage for 1 or more of the following IACs: intra-abdominal (n = 16) and/or digestive tract hemorrhage (n = 7), pancreatic (n = 2) and/or biliary (n = 2) fistula, and intra-abdominal abscess (n = 8) and/or collections (n = 12). There were more IACs in pancreatoduodenectomy than in distal pancreatectomy (53 [29%] of 177 vs 10 [19%] of 53) (Table 3 and Table 4), with lower rates in the octreotide group, but none of these differences were statistically significant (P = .20) (Table 4).
In univariate analysis (Table 4), compared with patients having fibrotic parenchyma, significantly more patients with normal pancreatic parenchyma had 1 or more IACs (46 [32%] of 146 vs 16 [19%] of 84, P = .02). In subgroup analysis, octreotide significantly reduced the number of patients sustaining 1 or more IACs when the main pancreatic duct diameter was less than 3 mm (P<.02), when pancreatojejunostomy was performed (P<.02) (Table 4), or both (P<.02). These P values remained statistically significant even after statistical adjustment17 for use of biological glue. However, neither of these factors was an independent risk factor in multivariate analysis (Table 4). Multivariate analysis showed that the only independent risk factor of IACs was pancreatoduodenectomy compared with distal pancreatectomy (P<.01), with an odds ratio of 3.54 (95% CI, 1.44-8.65).
No adverse effects could be directly attributed to octreotide injections. All patients, except for 3 who died before day 10, received their full 10-day course of octreotide. In case of pancreatic or digestive tract fistula, octreotide was prolonged until closure, subsequent operation, or death. No center effect was found, even though more than two thirds of the extended resections were performed in one center.
Our results show that the usefulness of octreotide is somewhere between the conclusions of the European studies, which advocate its routine use, and those of the North American trials, which concluded that it was useless. Octreotide decreased, but not significantly, the rate of patients with 1 or more IACs (P = .08). In subgroup analysis, however, octreotide significantly decreased the rate of patients with 1 or more IACs when the main pancreatic duct was less than 3 mm in diameter, when pancreatojejunostomy was performed, or both (P<.02). On the other hand, severity and, in particular, mortality were not influenced by octreotide.
As in the series by Lowy et al,9 single blinding without placebo was used in our study for 2 reasons. First, because the main end point was objective complications and not subjective criteria linked to patients and/or surgeons such as pain, double blinding with placebo was not deemed necessary17; second, double-blind studies with placebo are more complex and more expensive10 to conduct in the multicenter setting, and we had no financial support for this study.
The difference in the number of patients in each arm (122 octreotide vs 108 controls) occurred because randomization was performed according to 8 strata and balanced every 4 patients per stratum and per center: some centers entered fewer than 4 patients in each of the 8 strata and the study was closed before randomization was complete in each stratum and center.
Regarding the main end point, we chose to use the number of patients with 1 or more IACs and not simply the rate of pancreatic fistula or the crude overall number of complications,10 because pancreaticoenteric fistula can be obscured by, associated with, promoted by, or revealed by other IACs.5 Tabulation of all IACs ensures that no atypical manifestation of pancreatic fistula is ignored. In addition, there is no consensus regarding the exact definition of pancreatic fistula,19 rendering comparisons confusing or impossible; octreotide acts on all secretions4 and therefore can reduce any of the IACs in relation to the secretion; and the IAC rate is better than the overall complication rate (IACs and/or EACs) because most EACs are secondary to or associated with IACs.6 Conversely, some EACs can occur without IACs and therefore are not related to technical considerations.
Our rate (27% [62/230]) of patients with 1 or more IACs (Table 3) was the same as in the study by Lowy et al9 and well within the range reported in 4 other controlled trials (23%,6,8 28%,11 and 29%7) but lower than that reported in 2 other studies.5,10 In one study, the rate was 37%10 because of inclusion of delayed gastric emptying, excluded from other studies, and tabulation of both IACs and EACs. In another study, the rate was 44%5 because of inclusion of duodenum-preserving resections, well known for their higher rate of IACs.20
In our series, the level of significance did not attain the fatidic P<.05, apparently in accordance with the 3 American studies.9- 11 However, a possible explanation might be that the expected improvement (reducing the IAC rate from 40% to 20%) was not the same as the observed improvement (32% to 22%) (Table 3). To have been able to conclude with such an improvement rate, 318 patients per group (ie, 636 patients in all) would have been necessary17; none of the controlled studies in the literature, including ours, came near to this number of patients.
Moreover, contrary to the 7 previous studies,5- 11 ours was the only one that included patients with fibrin glue injection in the main pancreatic duct and around the pancreaticoenteric anastomosis. However, 3 controlled trials21- 23 have shown that fibrin glue, used in either of these 2 settings, was ineffective in preventing IACs. Likewise, the adjustment technique used17 here, as well as multivariate analysis and the critical appraisal of the literature, confirmed that fibrin glue did not influence outcome. Octreotide as a prophylactic agent, even when used in conjunction with other potentially active methods (eg, fibrin glue), did not significantly lower the complication rate. Our overall pancreatic fistula rate was 18% (41/230) (Table 3), very close to that found by Pederzoli et al6 (17.9%) and within the range reported in other trials (10%,10 13.9%,7 16.3%,8 23%,11 24.5%,9 and 27.6%5).
Compared with the control groups, octreotide significantly (P<.05) decreased the rate of pancreatic fistula in 4 European studies.5- 8 In our series, as in the 3 other American studies,9- 11 the difference found in the rate of pancreatic fistula was not statistically significant (Table 3). As in most of the studies in the literature, including the single-center studies,9,10 we studied both left and right resections of the pancreas and benign and malignant disease (only Friess et al8 limited their patient population to chronic pancreatitis).
As already suggested,24 discrepancy in outcome (IACs and fistula) might be related to the timing of the first injection of octreotide. In the European studies,5- 8 the first injection took place at least 1 hour before the operation started. In the study reported by Yeo et al,10 the definition was less strict, meaning that some of the injections could have been given less than 1 hour before operation. In the series by Lowy et al,9 the injection took place at the end of the resection, after the completion of pancreatojejunostomy, or after the end of operation. In our series, octreotide was injected during the operation after resection took place so as not to exclude patients because they no longer met the inclusion criteria. However, in the last negative American trial on vapreotide, the injection took place 2 hours before the operation.11
The results of several of these trials must be interpreted with caution because of serious methodological weaknesses.24,25 One major methodological weakness found in the 4 European5- 8 and 2 American10,11 trials was that randomization took place before the operation, and therefore 12%,8 17%,6 22%,7 and 24%5 to a high of 42%10 and 46%11 of patients were withdrawn after randomization without (except for 1 study11) intention-to-treat analysis. Several of the other methodological weaknesses in the study by Yeo et al10 have been highlighted elsewhere.24,25 Therefore, the true power of the test in these studies10,11 was low, always less than 80%, except for the study by Lowy et al (85%)9 and our study (90%). In the study by Lowy et al,9 allocation was not based on a true randomization but in fact on the number of patient medical records, an incorrect method because the allocated treatment is known in advance.17 Moreover, in this same study, patients received preoperative and/or intraoperative radiation,9 known to act negatively on pancreatic secretion, which could have obscured the action of octreotide, as confirmed in their multivariate analysis. Taking all these weaknesses into consideration, the conclusions of some of these controlled studies could be misleading.
In the controlled trials, overall postoperative mortality ranged from 0.5%10 to 6.9%.7 Our overall mortality rate, however, was higher (10%) but close to that found in recent reports, calculated for those not classified as high-volume or very high-volume centers (ie, 9.8%12 and 14.8%).2 One possible reason for this higher rate might be that extended resections, reputed to be associated with higher morbidity and mortality,14 were included in our series, whereas they were not in the others.5- 10 If these extended procedures were not counted in our series, the mortality rate would have been 8.7%. Another reason for this higher mortality might be the multicenter character of our study. Because more low-volume centers rather than high-volume centers participated in our trial, as opposed to the high-volume American centers, there might have been less selection bias due to highly selected referral patterns of low-risk patients.26 However, we were unable to correctly compare low- and high-volume centers within our study, because the true volume per center was masked by differences in duration of participation for each center and the variable mix of high- and low-risk patients. Moreover, not all surgeons in each center participated in the trial. The best way to interpret and differentiate between outcomes in high- and low-volume centers would be to use a predictive risk factor score, such as the Charlson index cited in 2 recent studies2,12 or the p-POSSUM.27,28 However, the multicenter setting (low- and high-volume centers) makes the results of our study more realistic and plausible for the overall surgical population. Finally, when octreotide was compared with controls, whether in our study or in the 7 other trials, none showed any significant decrease in mortality in the somatostatin analogue group.
In our series, the rate of subsequent operations and/or percutaneous drainage was nearly identical in the 2 groups (20% [24/122] for octreotide vs 19% [21/108] for controls, respectively; Table 3) but was recorded more often (difference not statistically significant) in the octreotide arm in 2 of the American studies.9,10 These rates were not available in the other studies.5- 8,11 Five trials5- 8,11 studied the hospital stay in addition to ours. The differences (average or median) found between the 2 groups were not statistically significant.
In our study, subgroup analysis showed that octreotide significantly reduced IACs when the diameter of the main pancreatic duct was less than 3 mm, in the case of pancreatojejunostomy, or both (P<.02) (Table 4). Small pancreatic duct size alone13 or associated with pancreatojejunostomy,14 found elsewhere to be risk factors of fistula,14 was not studied in the other controlled trials,5- 11 but subgroup analysis is known for its weaknesses and cannot converge to formal conclusions. This information should logically lead to a new controlled study in this selected population.17
Normal consistency of pancreatic remnant was a significant risk factor in univariate analysis (P = .02) (Table 1 4). In accordance with Sato et al,29 octreotide significantly decreased the rate of complications in 4 controlled studies6- 8,11 when the pancreatic remnant consistency was normal rather than fibrotic, although in our study, this difference was not statistically significant (24% [8/34] vs 36% [27/74], P = .20) (Table 4). Because the main pancreatic duct diameter might be a more preponderant factor than consistency, and ours was the only controlled trial in which the size of the main pancreatic duct was specifically studied, this might explain why this risk factor emerged.
As in other studies,5- 7 more patients had 1 or more IACs when they had pancreatic tumor (vs chronic pancreatitis), but the difference in univariate analysis was not statistically significant (Table 4). Finally, in our multivariate analysis, only pancreatoduodenectomy (vs distal pancreatectomy) was found to be an independent risk factor for increasing IACs, as shown previously.7
In conclusion, according to our results, and in comparison with those of the other controlled trials, the effectiveness of somatostatin analogues remains controversial. Octreotide cannot be recommended routinely to decrease the rate of IACs, as indicated in the European studies.5- 8 On the other hand, in contradiction with the results of the North American trials,9- 11 octreotide may be useful in certain high-risk patients. In accordance with our results, we advocate that patients scheduled to undergo pancreatic resection without prior radiation therapy9 and undergoing the operation in a low-volume center10,30 should receive octreotide, 100 µg, injected subcutaneously at least 1 hour before incision. In patients undergoing resection and anastomosis, a minimum of 300 µg/24 h of octreotide should be continued for at least 8 days when the main pancreatic duct diameter is less than 3 mm or, if the duct diameter is not measured, when the pancreatic parenchyma is normal. However, because these conclusions arose from subgroup analysis, caution is warranted in their application.
In the future, definitive results of the efficacy of octreotide could be reached by a controlled trial that includes high-risk patients, notably those with a main pancreatic duct of less than 3 mm. Another crucial point might be the evaluation of the best timing for the injection (before or during the anastomosis). Evaluation of other technical artifices proposed to decrease the rate of IACs and pancreatic fistula still remains to be tested by controlled trials, including preoperative radiation therapy,31 especially in the case of cancer, temporary transanastomotic external drainage,11,32,33 and other anastomotic techniques, either already in use or to come. Finally, because 8 well-conducted randomized trials have been performed, the time might be ripe for a true meta-analysis.
Corresponding author: Abe Fingerhut, MD, FRCS, Service de Chirurgie Digestive, Centre Hospitalier Intercommunal, Rue du Champs Gaillard, Poissy, CEDEX, France (e-mail: AbeFinger@aol.com). Reprints: French Associations for Surgical Research, 8 Avenue des Peupliers, 92270 Bois-Colombes, France.
Accepted for publication September 29, 2003.
The following surgeons participated in this study: Jean-Pierre Arnaud, MD, Gérard Lorimier, MD (Angers); Loïk de Calan, MD, Pierre Bourlier, MD (Tours); Jacques Chipponi, MD, Denis Pezet, MD (Clermont-Ferrand); Jacques Domergue, MD (Montpellier); Jean-Marie Hay, MD, Guy Zeitoun, MD, Yves Flamant, MD, Simon Msika, MD (Colombes); Jean Escat, MD, Gilles Fourtanier, MD, Bertrand Suc, MD, Jean-Luc Gouzi, MD, Bernard Pradère, MD (Toulouse); Daniel Jaeck, MD (Strasbourg); Paul Ténière, MD, Francis Michot, MD (Rouen); Pierre-Louis Fagniez, MD, Daniel Cherqui, MD, Nelly Rotman, MD (Créteil); André Elhadad, MD, Elias Habib, MD, Didier Brassier, MD (Aulnay-sous-Bois); Philippe Vicq, MD, Jean-Louis André, MD (Paris); Yves Laborde, MD (Pau); Abe Fingerhut, MD, Jean-Charles Etienne, MD (Poissy); Jean-Claude Le Néel, MD (Nantes); Philippe Oberlin, MD (Villeneuve-Saint-Georges); Gérard Kohlmann, MD (Corbeil); Pierre Quandalle, MD, Luc Gambiez, MD (Lille); Jean-Pierre Lenriot, MD, Jean-Christophe Paquet, MD (Longjumeau).