Preoperative fecal elastase-1 (FE-1) adds value in predicting post-operative pancreatic fistula: not all soft pancreas share the same risk - A prospective analysis on 105 patients

BACKGROUND

Scores predicting postoperative pancreatic fistula (POPF) after pancreaticoduodenectomy (PD) mainly use intraoperative predictors. The aim of this study is to investigate the role of pancreatic exocrine function expressed by fecal elastase (FE-1) as preoperative predictor of POPF.

METHODS

Patients scheduled for PD at the Department of General and Pancreatic Surgery, University of Verona Hospital, from April 2017 to July 2018 were prospectively enrolled. FE-1 was measured in a preoperative stool sample through an ELISA test.

RESULTS

The study population consisted of 105 patients. The POPF rate was 17.1%. Patients developing POPF showed high values of FE-1 (454 vs 155 mcg/g; p < 0.01), and FE-1 was an independent predictor of POPF (OR 1.008, CI 95% 1.003-1.014; p < 0.01), even considering only patients with a "soft" texture. A cut-off value of 260 mcg/g presented 100% sensitivity and 64.3% specificity (AUC 0.83) in predicting POPF. Approximately 30% of patients with a "soft" pancreatic texture presented with FE-1 < 260 mcg/g and did not develop POPF.

CONCLUSION

FE-1 is a promising tool to preoperatively assess the risk of POPF after PD. Further studies with larger populations are needed to potentially incorporate FE-1 into risk scores for PD with better stratification.

Nutritional support and therapy in pancreatic surgery: A position paper of the International Study Group on Pancreatic Surgery (ISGPS)

BACKGROUND

The optimal nutritional therapy in the field of pancreatic surgery is still debated.

METHODS

An international panel of recognized pancreatic surgeons and pancreatologists decided that the topic of nutritional support was of importance in pancreatic surgery. Thus, they reviewed the best contemporary literature and worked to develop a position paper to provide evidence supporting the integration of appropriate nutritional support into the overall management of patients undergoing pancreatic resection. Strength of recommendation and quality of evidence were based on the approach of the grading of recommendations assessment, development and evaluation Working Group.

RESULTS

The measurement of nutritional status should be part of routine preoperative assessment because malnutrition is a recognized risk factor for surgery-related complications. In addition to patient's weight loss and body mass index, measurement of sarcopenia and sarcopenic obesity should be considered in the preoperative evaluation because they are strong predictors of poor short-term and long-term outcomes. The available data do not show any definitive nutritional advantages for one specific type of gastrointestinal reconstruction technique after pancreatoduodenectomy over the others. Postoperative early resumption of oral intake is safe and should be encouraged within enhanced recovery protocols, but in the case of severe postoperative complications or poor tolerance of oral food after the operation, supplementary artificial nutrition should be started at once. At present, there is not enough evidence to show the benefit of avoiding oral intake in clinically stable patients who are complicated by a clinically irrelevant postoperative pancreatic fistula (a so-called biochemical leak), while special caution should be given to feeding patients with clinically relevant postoperative pancreatic fistula orally. When an artificial nutritional support is needed, enteral nutrition is preferred whenever possible over parenteral nutrition. After the operation, regardless of the type of pancreatic resection or technique of reconstruction, patients should be monitored carefully to assess for the presence of endocrine and exocrine pancreatic insufficiency. Although fecal elastase-1 is the most readily available clinical test for detection of pancreatic exocrine insufficiency, its sensitivity and specificity are low. Pancreatic enzyme replacement therapy should be initiated routinely after pancreatoduodenectomy and in patients with locally advanced disease and continued for at least 6 months after surgery, because untreated pancreatic exocrine insufficiency may result in severe nutritional derangement.

CONCLUSION

The importance of this position paper is the consensus reached on the topic. Concentrating on nutritional support and therapy is of utmost value in pancreatic surgery for both short- and long-term outcomes.

A Cohort Study for Derivation and Validation of Early Detection of Pancreatic Fistula After Pancreaticoduodenectomy

BACKGROUND

Pancreatic fistula (PF) remains the most important morbidity after pancreaticoduodenectomy (PD). Early drain removal was recently recommended. However, this is not applicable to all cases because the development of severe PF may not be obvious until a later postoperative day (POD). This study aimed to discover ways to detect clinically relevant PF early during the postoperative stage after PD.

METHODS

We studied 120 patients who underwent PD. Grades B/C PF classified according to the International Study Group of Pancreatic Surgery guidelines were defined as clinically relevant PF. Logistic regression was used to identify detection factors for clinically relevant PF. Receiver operating characteristic curves were used to identify the optimal cutoff value for clinically relevant PF, and the k-fold cross-validation model to validate the cutoff value.

RESULTS

Drain amylase on POD 1 and C-reactive protein (CPR) on POD 2 were independent factors for clinically relevant PF. Drain amylase >1300 IU/l on POD 1 and CRP >12.8 g/dl on POD 2 were the best cutoff values for clinically relevant PF detection and were confirmed by k-fold cross-validation. The sensitivity and specificity values were 79 and 81 %, respectively.

CONCLUSIONS

Values of drain amylase and CRP combined were useful to distinguish clinically relevant PF.

[Possibilities of treatment of external pancreatic fistula]

Evaluation of the efficacy of sekretolitičeskoj therapy with synthetic analogue of somatostatin, a short-acting oktreotid (group 1) and extended oktreotid-depo (group 2) in 24 patients with external pancreatic fistulas after destructive pancreatitis. Results of clinical studies have shown that against the backdrop of an analogue of somatostatin-depo true healing and purulent-necrotic pancreatic external fistula occurs in less time: average 19 ± 1.8, and 16.2 ± 1.2 day observations, respectively.

Effect of SMS 201-995 on exocrine pancreatic secretion in a patient with external pancreatic fistula

The effect of SMS 201-995 on pancreatic exocrine function was studied. The SMS 201-995 was administered to a patient with an artificial external pancreatic fistula following pancreaticoduodenectomy. Variations in pancreatic exocrine function were assessed by determining the volume and components of the fistula fluid during the following periods: 5 d prior to SMS 201-995 administration, for 5 d during actual administration, and for 5 d after it had been discontinued. The SMS 201-995 was administered by subcutaneous injection of 100 micrograms every 12 h for the first 2 d and then 100 micrograms every 6 h for 3 d. This experiment demonstrated that SMS 201-995 has a strong inhibitory effect on pancreatic exocrine function, markedly reducing the amount of fistula fluid and the production of amylase, total protein, and bicarbonate.

Pancreatic excretion of dimethadione and trimethadione by repeated oral administration of trimethadione in dogs

To examine pancreatic excretion of dimethadione (DMO), a weak organic acid, as well as of its precursor trimethadione (TMO), TMO was given orally to dogs with pancreatic fistulae at a dose of 10-160 mg/kg/day over a period of 14 days. Blood samples were taken once a day during the administration of TMO and for 7 days after discontinuation of the drug. On the 15th day, pancreatic juice was collected under stimulation by secretin (2 Crick-Haper-Raper units/kg/h). DMO concentration in plasma reached a maximal plateau around the 10th day after starting TMO administration, and depended directly on the dose of TMO. Pancreatic excretion of DMO at a steady state closely depended on both the dose of TMO and the DMO concentration in plasma. The pancreatic juice/plasma concentration ratio for DMO exceeded 1.0 at a steady rate and decreased with the increased flow rate. Pancreatic DMO clearance (DMO output/DMO concentration in plasma) increased, depending on the flow rate, the bicarbonate concentration, and pH of pancreatic juice. Pancreatic excretion of TMO was zero or extremely low.

Studies on neurotensin. II. Release of neurotensin

The objective of these experiments was to confirm the localization of neurotensin (NT) in gut endocrine cells of the canine small intestine using immunohistochemistry. In addition, the release of NT from the canine small intestine in response to selective perfusion of a fatty acid (oleate), triglyceride (Lipomul) or products of fat digestion into various segments of the small intestine was studied. In the immunohistochemical study, NT was found to be primarily localized in true endocrine cells of the ileal mucosa. In addition, NT was not found or only negligible numbers of cells were seen outside the lower small intestine. This observation supports previous results based on radioimmunoassay and immunohistochemistry studies. Based on these morphological findings, NT would be released by luminal secretagogues, of which fat appears to be the most potent. In the selective perfusion studies, perfusion of oleic acid into the jejunum of the chronic dog caused NT release, whereas perfusion of the ileum in which NT cells were most abundant was ineffective. This observation suggests that a neural or endocrine message is released to the ileal NT cell from the jejunum, causing NT release. This series of studies was carried out to elucidate the mechanism of NT release and to find the direct luminal stimulants of NT by using both chronic and acute experimental models. These studies suggest that NT is not significantly released under anesthesia and that undigested fat, like triglyceride, does not release NT in either the upper or lower small intestine. Furthermore, digested fat, like oleate or digestive juices in the lower small intestine, is not a direct stimulant of NT release.

The penetration of antibiotics into human pancreas

In order to analyse the penetration of two antibiotics (mezlocillin and metronidazole) which cover the spectrum of microorganisms involved in pancreatic infection, we determined their concentration in pancreatic tissue, juice and cyst fluid in 16 patients undergoing pancreatic surgery. In addition, the external pancreatic fistula fluid of one patient was analysed for antibiotic concentration and bacterial counts during a seven-day treatment with mezlocillin, metronidazole and netilmicin (i.v.). Antibiotic concentrations were determined by HPLC between 16 and 210 (median 74) min after i.v. administration of 4 g mezlocillin and 500 mg metronidazole, respectively. The median concentration of mezlocillin was 23.2 (range: 3.1-37.4) mg/kg, 15.9 (range: 4.2-55.0) mg/l and 9.9 (range: 5.2-14.8) mg/l in pancreatic tissue, juice and cyst fluid, respectively. The median concentration of metronidazole was 5.1 (range: 1.8-13.0) mg/kg, 8.5 (range: 3.6-16.2) mg/l and 1.2 (0.9-1.4) mg/l in pancreatic tissue, juice and cyst fluid, respectively. From the fistula patient, seven different bacteria were cultured (five aerobic and two anaerobic isolates); their concentration in fistula fluid ranged from 10(5) to 10(7) CFU/ml. The bacteria sensitive for mezlocillin and metronidazole disappeared after four days of i.v. treatment, whereas the two isolates sensitive for netilmicin showed continuous growth seven days after i.v. treatment. The peak concentrations for mezlocillin, metronidazole and netilmicin in the fistula fluid were 6.8 mg/l, 5.6 mg/l and less than 0.1 mg/l, respectively.

Treatment of pancreatic cutaneous fistulas with a somatostatin analog

Five pancreatic cutaneous fistulas were treated by subcutaneous administration of a long-acting synthetic analog of somatostatin, SMS 201-995. Patients included four men and one woman who ranged in age from 52 to 77 years. The fistulas developed after drainage of a pancreatic abscess, biopsy of a pancreatic mass, splenectomies for idiopathic thrombocytopenic purpura and Felty's syndrome, and operative trauma, respectively. Fistula output consisted of 1,000 ml/day of amylase- and lipase-rich fluid in the patient with a pancreatic biopsy. The other four patients had low-output fistulas (100 to 250 ml/day) that had been draining for 1 to 12 months. Direct communication with the pancreatic duct was demonstrated by endoscopic retrograde cholangiopancreatography, sinography, or both in four of the five patients. Fistula output decreased from 340 +/- 376 ml/day to 63 +/- 36 ml/day on the first day of therapy with two daily doses of 0.05 mg SMS 201-995 (p less than 0.03) and to 13 +/- 19 ml/day on the seventh day of therapy (p less than 0.03). Two patients had prompt closure of their fistulas and one closed in 3 months. One patient with chronic pancreatitis and a duct stricture and one patient with recurring infection did not achieve permanent fistula closure with SMS 201-995. Because of its safety, ease of administration, and efficacy in decreasing fistula output, we believe somatostatin analog therapy is beneficial in hastening closure of pancreatic fistulas.

Ciprofloxacin penetration in pancreatic juice

The penetration of ciprofloxacin in pancreatic juice was investigated in 5 patients with pancreatic fistula. The drug was administered as a single oral dose of 500 mg after which serial samples of pancreatic juice and serum were collected for ciprofloxacin assay. The following pharmacokinetic parameters (mean +/- SD) were estimated from the serum level versus time curves: clearance 11.51 +/- 2.85 (ml/min/kg); Vd area 3.08 +/- 1.20 ml/kg; terminal half-life 3.10 +/- 0.92 h; mean residence time 5.64 +/- 1.40 h. Ciprofloxacin serum levels declined rapidly after the third hour, whereas concentrations in pancreatic juice remained elevated (above 1 mg/1) for nearly 12 h. The pancreatic juice/serum ciprofloxacin concentration ratio increased gradually fom 0.63 +/- 0.45 after 0.5 h to 6.18 +/- 4.59 after 12 h (mean +/- SD). Our data indicate that while the drug elimination half-life from the serum is short, the time-course of ciprofloxacin levels in the pancreatic juice conforms to a much slower disappearance rate. In particular, the ciprofloxacin levels achieved in pancreatic juice are constantly greater than the MICs of the bacteria generally responsible for pancreatic infections.

Secretin is released by digestive products of fat in dogs

We investigated the effect of fat or digestive products of fat on the release of endogenous secretin in 15 gastric fistula dogs with either pancreatic fistulas or duodenal fistulas. In 4 dogs with both gastric and duodenal cannulas, intraduodenal administration of corn oil (Lipomul) at a dose of 15 mmol resulted in a significant increase in plasma secretin concentration, whereas in another group of 4 dogs with complete pancreatic duct ligation, the same amount of triglyceride failed to increase the secretin concentration. When Lipomul incubated with pancreatic enzymes was administered in the duodenum, a marked increase in plasma secretin concentration occurred in the 4 dogs with pancreatic duct ligations. In the 7 dogs with chronic pancreatic fistulas, intraduodenal administration of Lipomul resulted in a significant increase in both plasma secretin concentration and pancreatic secretion of bicarbonate when the pancreatic juice was allowed to flow into the duodenum, whereas no increase in either the secretin concentration or bicarbonate output was apparent using the same amount of Lipomul when the pancreatic juice was diverted from the duodenum. In 4 of these 7 dogs so studied, intraduodenal administration of oleic acid emulsion, with pH adjusted to 5.0 in graded doses, resulted in a dose-related increase in the secretin concentration that paralleled pancreatic bicarbonate output. The increases in both secretin concentration and pancreatic bicarbonate secretion were completely abolished by intravenous infusion of a rabbit antisecretin serum in the 4 dogs. Thus we conclude that release of endogenous secretin plays an important role in the mechanism of exocrine pancreatic secretion stimulated by digestive products of fat in dogs.

Physiological role and localization of cholecystokinin release in dogs

In dogs with pancreatic fistulas, meat feeding and intestinal perfusion with a sodium oleate or amino acid mixture increased pancreatic protein secretion to approximately 110, 100, and 50%, respectively, of the response to cholecystokinin (CCK) at a dose of 85 pmol X kg-1 X h-1. Plasma CCK response increased in these studies to approximately 100, 180, and 40%, respectively, of the value obtained with exogenous CCK, suggesting that, in addition to CCK, other neurohormonal factors contribute to pancreatic enzyme secretion in response to endogenous stimulants. Feeding and duodenal oleate or amino acids also stimulate the release of pancreatic polypeptide (PP), which may be involved in the control of pancreatic secretion in response to endogenous stimulants, including CCK. Perfusion of the intact intestine with graded amounts of oleate (0.5-16 mmol/h) produced dose-dependent increments in plasma CCK and pancreatic protein similar to those obtained with intravenous infusion of graded doses of CCK (0.85-255 pmol X kg-1 X h-1). Oleate perfusion of isolated Thiry loops (30 cm long) made of duodenojejunal (D-J) and ileal (I) segments also stimulated protein secretion but elevated plasma CCK only after perfusion of the D-J but not of the I loop. We conclude that 1) the endogenous CCK released by various luminal stimulants drives the pancreatic protein secretion; 2) the release of CCK is confined to the foregut; and 3) PP concomitantly released by various intestinal stimulants may contribute to the control of pancreatic secretion induced by endogenous CCK.

A comparison of the effects of two somatostatin analogues in a patient with an external pancreatic fistula

In a 30-year-old man, a total external pancreatic fistula developed after enucleation of an insulinoma of the head of the gland. The output of the fistula was reduced to some extent by total parenteral nutrition. Much greater reductions of volume were noted during short-term administration of two somatostatin preparations, SMS 201-995 and Somatofalk, for 1 week consecutively. Although the former showed several advantages over the latter drug, such as absence of an escape phenomenon and of a rebound effect, neither drug caused a closure of the fistula. Spontaneous closure of the fistula occurred after 3 1/2 months.

Loperamide-induced inhibition of pancreatic secretion in rats

The effects of loperamide on exocrine pancreatic secretion were studied in rats fitted with chronic or acute fistulas. Intraduodenal injection of loperamide in conscious rats resulted in a dose-dependent inhibition of basal pancreatic secretion involving volume and bicarbonate and protein output with an ED50 of about 0.5 mg/kg. The maximal inhibition observed was about 60% for volume and bicarbonate output and 90% for protein output. Loperamide induced an inhibition of pancreatic secretion in conscious rats that was naloxone-sensitive and persisted in cimetidine-treated rats. Thus, it did not depend on modifications of gastric secretion. In anaesthetized rats, loperamide did not inhibit the pancreatic secretion evoked by agents acting directly on the pancreatic cells (acetylcholine, secretin, CCK) but it inhibited by 100% the pancreatic secretion induced by vagal electrical stimulation (VES) and by 80-100% that induced by 5 thio-glucose, a centrally acting vagal stimulatory agent. Loperamide inhibition of VES-induced pancreatic secretion was different from that obtained with morphine or methadone since these opiate drugs could only inhibit by 50-60% maximally the VES-stimulated pancreatic secretion. The loperamide inhibition of VES-induced secretion was naloxone-insensitive, while loperamide inhibition of 5 thio-glucose-induced secretion was in part naloxone-sensitive. These results suggest that loperamide exerts a potent inhibition of pancreatic secretion by acting on the nerve supply to the pancreas through both opiate and non-opiate mechanisms.

Pancreatic secretory responses to intravenous hyperalimentation and intraduodenal elemental and full liquid diets

Intravenous hyperalimentation and enteral elemental diets have both been advocated for the nutritional support of patients with acute pancreatitis, pancreatic fistula, and proximal small bowel fistula. The exocrine pancreatic responses to these nutrients compared to one another and to full liquid diet have been inadequately studied. Therefore, pancreatic protein, volume, and bicarbonate responses to graded doses of (1) intravenous hyperalimentation, (2) intraduodenal elemental diet, and (3) intraduodenal full liquid diet were compared in duplicate experiments in five dogs with chronic pancreatic fistulas. Both intraduodenal elemental and full liquid diets caused comparable and significant dose-related increases in pancreatic protein, volume, and bicarbonate outputs over basal levels (p less than 0.05). In contrast, there was no stimulation of pancreatic secretion by intravenous hyperalimentation. It therefore appears that of the methods studied, only intravenous hyperalimentation can provide full nutritional support while maintaining the pancreas at rest.

Interaction between stimulants of exocrine pancreatic secretion in dogs

We measured bicarbonate and protein secretory responses to graded doses of intravenous caerulein and bethanechol and intraduodenal L-phenylalanine alone or with background secretin; graded doses of secretin alone or with background caerulein or L-phenylalanine; and background secretin plus graded doses of caerulein or L-phenylalanine plus background atropine sulfate. Potentiation (more-than-additive response) occurred for bicarbonate secretion between secretin and caerulein, between secretin and L-phenylalanine, but not between secretin and bethanechol. The only potentiating interaction for protein secretion was between secretin and low doses of caerulein. Atropine abolished the potentiated bicarbonate response to secretin plus L-phenylalanine but had no effect on the response to secretin plus caerulein. Potentiation between secretin and cholinergic mechanisms and cholecystokinin for pancreatic bicarbonate secretion may be an important regulatory mechanism, while potentiation of protein secretion with these stimulants does not appear to be important in dogs. A cholinergic mechanism mediates much of the bicarbonate potentiation between secretin and intestinal L-phenylalanine.

[Influence of dermorphin on pancreatic secretion in the rat with gastric fistula]

The influence of the opioid peptide dermorphin on 2DG-stimulated pancreatic secretion of the rat was studied. Experiments were performed in conscious rats with pancreatic fistulae (8 animals), or with pancreatic and gastric fistulae (8 rats) to allow diversion of gastric juice. In animals with gastric fistulae the peak of pancreatic protein output in response to 2DG was about 50% lower than in animals with intact stomachs. Dermorphin almost completely inhibited the stimulant effect of 2DG, independently by different experimental conditions. It is concluded that the increase of pancreatic secretion produced by 2DG is, at least in part, independent from the entrance of gastric acid into the duodenum. As a consequence, the inhibiting effect of dermorphin on 2DG-stimulated pancreatic secretion may be attributed to a decrease of vagal stimulation of the pancreas, at central and/or peripheral sites.

Gastric inhibitory polypeptide (GIP) is not the primary mediator of the enterogastrone action of fat in the dog

We compared the inhibition of food-stimulated gastric acid secretion and changes in serum concentrations of immunoreactive gastric inhibitory polypeptide and gastrin caused by: (a) duodenal perfusion of oleic acid, and (b) intravenous infusion of pure, natural, porcine gastric inhibitory, polypeptide in dogs with gastric fistula and pancreatic fistula. A rate of duodenal perfusion of oleic acid (12 ml/hr) which gave near maximal pancreatic protein response was chosen. This dose of oleic acid caused complete suppression of acid response to a meal of liver extract (300 ml of a 15% solution) while elevating serum immunoreactive gastric inhibitory polypeptide modestly. By contrast, intravenous administration of gastric inhibitory polypeptide that raised serum immunoreactive gastric inhibitory polypeptide several fold caused only 40% inhibition of acid response to the same meal. Other effects of duodenal perfusion of oleic acid were exaggeration of pancreatic protein secretion and significant inhibition of gastrin release in response to the meal. Exogenous administration of gastric inhibitory polypeptide, on the other hand, was without significant effect on these responses. These results suggest that, in the innervated dog stomach, the enterogastrone action of fat is not primarily mediated by gastric inhibitory polypeptide.

The effect of intravenous fat emulsions in patients with pancreatic fistula

Three patients with pancreatic fistulae were given parenteral 10% fat emulsion (Intralipid) to study the effect of intravenous fat on pancreatic fistula output. Each patient received nutritional support with isovolumetric, isonitrogenous, and isocaloric solutions containing either hypertonic dextrose and amino acids, or hypertonic dextrose, amino acids, and a fat emulsion. Measurements of fistula volume, fistula amylase, lipase, and chloride concentrations, and fistula trypsin activity were performed. The patients were studied for periods of 10 to 26 days. No significant increases in any of the above parameters were noted during the periods when the fat emulsions were infused. In one patient the fistula closed spontaneously. We conclude that intravenous fat emulsions may be used to provide nutritional support for patients with pancreatic fistula without increasing pancreatic juice volume or enzyme content.

Effects of intraduodenal amylase, lipase, trypsin, and bile on pancreatic enzyme secretion in the rat

The effect of duodenal infusion of trypsin, amylase, lipase and bile on pancreatic enzyme secretion was studied in conscious rats surgically prepared with bile-pancreatic fistulae. Trypsin infusion resulted in a depression of the secretory volume and protein and trypsinogen output. All these effects were reversed after additional infusion of trypsin inhibitor. Infusion of amylase or lipase in doses comparable to those of trypsin did not influence the secretory volume, protein or enzyme output. Likewise, intraduodenal bile infusion to rats with diverted bile-pancreatic juice did not change these parameters; also in rats with trypsin reinfused into the duodenum, bile infusion was without effect.

Metabolic problems in the management of a traumatic pancreatic fistula

A case of persistent, high output, traumatic pancreatic fistula, complicated by the development of a gastric fistula secondary to a pancreatic abscess, is described. The problems associated with the metabolic and nutritional management are discussed.

[Effect of somatostatin on canine exocrine pancreas secretion stimulated by pancreozymin, secretin or test meal]

In 4 dogs with pancreatic and gastric fistulae the effect of somatostatin (SST) on the pancreas was investigated during stimulation by pancreozymin-secretin or by a test meal. The pancreozymin-secretin-induced enzyme output and bile flow, as well as the enzyme concentration and duodenal volume during test meal stimulation were significantly reduced by SST. SST caused a less marked decrease in bicarbonate concentration, while volume and bicarbonate output did not change.

External biliary-pancreatic fistulas

Twenty-seven patients treated for pancreatic and/or biliary-cutaneous fistulas have been reviewed. Four patients died mainly because of cardiopulmonary and septic complications. Spontaneous sealing of the fistula occurred in 81% of the conservatively treated cases (48% of all cases). All the LO fistulas but only 68% of the HO fistulas treated conservatively sealed spontaneously. Eleven patients were treated surgically. There were three deaths and three failures (reappearance of fistula). All the patients who died had been operated on within three months after the appearance of HO fistulas. There was no mortality among the patients with LO fistulas or among patients operated on at a later stage. We have reached the following conclusions: 1. There is a significant difference in prognosis between low output and high output fistulas. 2. In LO fistulas, there is no need for a surgical intervention aimed to close the fistula unless it persists for at least one year. 3. In HO fistulas, if a corrective operation is necessary, it should be withheld for at least three months whenever possible. 4. Roux-en-Y fistulojejunostomy is considered to be the procedure of choice. 5. Infection and premature colsure of the external part of the fistulous tract should be avoided by insertion of drains and repeated surgical drainage, where necessary. 6. High caloric feeding, elemental diet and intravenous hyperalimentation are very important factors that enhance recovery in the surgically and conservatively treated patients.

The relationship between flow and electrolyte secretion in the canine pancreas

The response of the pancreatic bicarbonate secretory mechanism to secretin and the relationship between bicarbonate concentration and flow rate in pancreatic juice have been re-examined following reports describing decreasing levels of bicarbonate concentration at high flow rates. Both anaesthetized and chronic fistula dogs were used. The results show that when high doses of secretin elicit high rates of flow of pancreatic juice the bicarbonate concentration rises to a peak which is constant over a wide range of stimulation and flow rates.