(LF, 0.04 to 0.15 Hz) in normalized units [LF% � LF/(HF � LF) � 100%].HF was transformed by natural logarithm to correct the skewness ofdistribution (Am J Physiol 1999;277:H2233–H2239). Disease severity wasassessed by Ranson and Imrie scores.Results: Patients with acute pancreatitis had significantly lower In(HF) andLF%, compared with the age matched healthy controls, indicating vagaland sympathetic dysfunction respectively. The In(HF), but not the LF%,significantly correlated with Ranson scores in the patients with acutepancreatitis (r � 0.6, p � 0.01). This indicates that the vagal impairmentis related to disease severity.Conclusions: The results in this study indicate that patients with acutepancreatitis are deprived of parasympathetic and sympathetic activity. Theparasympathetic impairment is disease severity dependent. Correlatingwith Ranson scores, the high frequency power in heart rate variabilityanalysis may predict severity of disease in acute pancreatitis.

277

Comparison of five algorithms used to predict common bile ductstonesVaman S Jakribettuu MD 1 and Benoit C Pineau MD 1*. 1Division ofGastroenterology, Wake Forest University School of Medicine, Winston-Salem, NC-27157, United States.

Purpose: To prospectively compare five algorithms used to predict com-mon bile duct stones (CBDS).Methods: Seventy-seven consecutive patients referred for management ofsuspected CBDS were included in the study. Patients underwent routineblood work and imaging studies (US and/or CT). The algorithms were thenused to classify patients into low (�10%), moderate (10–55%) or high-risk(�55%) groups before a gold standard (ERCP) confirmed or excludedCBDS. Algorithm-I (Onken et al., 1996) is a nomogram that predicts thelikelihood of CBDS in patients with symptomatic cholelithiasis using liverenzymes, CBD size and total bilirubin. Algorithm-II (Barkun et al., 1994)predicts CBDS before laparoscopic cholecystectomy using CBD size, bil-irubin and age. Algorithm-III is a modification of an algorithm by Canto etal. (1998) that has been used to classify patients into different risk groupsusing clinical criteria, CBD size, bilirubin and liver enzyme levels. Algo-rithm-IV (Hauer-Jensen et al., 1984) and V (Lacaine et al., 1980) predictsCBDS based on predefined criteria (clinical findings, CBD size, bilirubinand liver enzymes) in patients with gallstones.Results: Results are presented in the table. Twenty-nine of the 77 patientshad common bile duct stones at the time of ERCP. Algorithms-I and Vcould not be used in a majority (�60%) of patients because of strictinclusion criteria. Algorithm-II did not allow classification into a low riskgroup. Algorithm-III accurately classified patients into high, moderate andlow probability groups, however, it could not be applied in 27% of patients.CBDS were found in 1 (9%) of 11 low risk, 9 (39%) of 23 moderate riskand 13 (59%) of 22 high-risk patients by this algorithm. Algorithm-IV wasnot accurate in identifying low risk patients.Conclusions: The above clinical algorithms have significant limitations.Some algorithms (I, III, V) exclude a large proportion of patients becauseof stringent inclusion criteria. Among the others (II, IV), none of thealgorithms accurately identify patients who are at low risk of CBDS.Therefore, accurate algorithms that can be used universally are needed forthe optimal management of patients with suspected CBDS.

N � 77Algorithm

N excludedfrom

algorithm

Highrisk,

CBDS�

Highrisk,

CBDS�

Moderaterisk,

CBDS�

Moderaterisk,

CBDS�

Lowrisk,

CBDS�

Lowrisk,

CBDS�

I 55 (71%) 0 0 1 (7%) 13 (93%) 1 (13%) 7 (87%)II 0 13 (72%) 5 (28%) 16 (27%) 43 (73%) 0 0III 21 (27%) 13 (59%) 9 (41%) 9 (39%) 14 (61%) 1 (9%) 10 (91%)IV 0 5 (63%) 3 (37%) 17 (40%) 25 (60%) 7 (26%) 20 (74%)V 47 (61%) 6 (55%) 5 (45%) 0 1 (100%) 4 (22%) 14 (78%)

278

Urinary trypsinogen activation peptide is more accurate thanhematocrit in determining severity in patients with acutepancreatitis. A prospective studyKhan Z, Vlodov J, Horovitz J, Jose RM, Iswara K, Smotkin J, Brown A,Tenner SM. Maimonides Medical Center, State University of New York-Health Sciences Center.

The management of patients with acute pancreatitis is complicated by theinability to distinguish mild from severe disease during the early stages. Ithas been previously shown that Urinary Trypsinogen Activation Peptide(TAP) and the Hematocrit (HCT) may serve as early predictors of severityin patients with acute pancreatitis. In order to determine which marker ismore accurate in the determination of severity in acute pancreatitis, aprospective study was performed. All patients admitted with pain consis-tent with acute pancreatitis and an amylase of greater than 3 times the upperlimit of normal were included. The admission and 24 hour HCT wasobtained. A urine sample was obtained within 12 hours of admission. TheUrinary TAP was determined using TAPKIT (Biotrin, Dublin, Ireland).Severity was defined by the Atlanta Symposium as the presence of organfailure and/or pancreatic necrosis. Fiftyeight consecutive patients withacute pancreatitis participated. There were 33 male, 25 female patients witha mean age of 60 � 19. The etiology of the acute pancreatitis varied: 12patients with alcohol, 20 with gallstones, 9 with postoperative, 17 withother causes, including hypertriglyceridemia, and idiopathic. Thirty-ninepatients had mild disease; 19 had severe disease. The Urinary TAP waselevated in 26 patients. All patients with severe acute pancreatitis werecorrectly identified as having severe disease by an elevated Urinary TAP(Sensitivity 100 percent, Specificity 73 percent). The admission HCT washigher than 47 in only 3 patients, all with mild disease. Of the patients witha rise in HCT, 8 had mild disease, only 1 had severe disease. There was noassociation between a rise in the HCT and severity of disease (r � 0.51).In comparison to admission HCT, Urinary TAP was more accurate indetermining severity in patients with acute pancreatitis by Atlanta, ApacheII, and Ranson Criteria (p � 0.001). We conclude that Urinary TAP is amore accurate predictor of severity in patients with acute pancreatitiscompared to HCT.

279

Incidental gall bladder cancerVK Kapoor, GD Wagholikar, A Behari, A Kumar, SS Sikora, R Saxena.Sanjay Gandhi Post-graduate Institute of Medical Sciences, Lucknow,India.

Purpose: Some patients who are operated for gall stones (GS) are found tohave gall bladder cancer (GBC) on histological examination-incidentalGBC.Methods: From 1989 to 1999, about 2,600 cholecystectomies were per-formed for GS-23 (1%) were found to have incidental GBC. These con-stituted 6.5% of 356 patients operated for GBC during the period.Results: The median age of these patients (55 years) was more than that ofGS patients (40 years); only 2 patients were less than 35 years old. Themedian duration of symptoms (4.5 months) was less than that in GSpatients (12 months); as many as 12 patients had symptoms for less than 6months. Choledocholithiais was suspeted clinically in 9 and on US in 11patients; 14 patients had evidence of biliary obstruction (history of jaun-dice, elevated serum bilirubin or alkaline phosphotase, dilated CBD onUS). Cholecystectomy was performed in all patients-12 received adjuvantradiotherapy/chemotherapy. TNM stages were I � 8, II � 5 and III � 10.Median survival was 59 months and actuarial survival was 45% at 5 years.Conclusions: GBC should be suspected in patients with GS who are old,have short duration of symptoms or evidence of biliary obstruction. Inci-dental GBC is in early stages, can be resected and has better prognosis.

S88 Abstracts AJG – Vol. 96, No. 9, Suppl., 2001