Risk factors for diabetes mellitus in chronic pancreatitis

GASTROENTEROLOGY 2000;119:1324 –1332

Risk Factors for Diabetes Mellitus in Chronic Pancreatitis DAVID MALKA,* PASCAL HAMMEL,* ALAIN SAUVANET,* PIERRE RUFAT,‡ DERMOT O’TOOLE,* PASCAL BARDET,* JACQUES BELGHITI,* PIERRE BERNADES,* PHILIPPE RUSZNIEWSKI,* and PHILIPPE LE´VY* *Fe´de´ration Me´dico-Chirurgicale d’He´pato-Gastro-Ente´rologie and ‡Cellule M.S.I., Ho ˆpital Beaujon, Assistance Publique-Ho ˆpitaux de Paris, Clichy, France

Background & Aims: The influence of disease progression and pancreatic surgery on the appearance of diabetes mellitus in patients with chronic pancreatitis is unknown. Methods: A prospective cohort study of 500 consecutive patients with chronic pancreatitis (alcoholics, 85%) followed up over a mean period of 7.0 ⴞ 6.8 years in a medical-surgical institution between 1973 and 1996 was performed. Multivariate analysis of risk factors for diabetes mellitus was performed after exclusion of 47 patients. Patients who underwent elective pancreatic surgery (n ⴝ 231, 51%) were compared with patients who never underwent surgery (n ⴝ 222, 49%). Results: The cumulative rate of diabetes mellitus was 83% ⴞ 4% 25 years after the clinical onset of chronic pancreatitis (insulin requirement, 54% ⴞ 6%). The prevalence of diabetes mellitus did not increase in the surgical group overall but was higher 5 years after distal pancreatectomy (57% ⴞ 8%) than after pancreaticoduodenectomy (36% ⴞ 18%), pancreatic drainage (36% ⴞ 13%), or cystic, biliary, or digestive drainage (24% ⴞ 7%) (P ⴝ 0.005), without difference in the latter ones. Pancreatic drainage did not prevent the onset of diabetes mellitus. Distal pancreatectomy (risk ratio, 2.4; 95% confidence interval [CI ], 1.6 –3.8; P < 0.0001) and early onset of pancreatic calcifications (risk ratio, 3.2; CI, 2.2– 4.7; P < 0.0001) were the only independent risk factors for diabetes mellitus. Conclusions: The risk of diabetes mellitus is not influenced by elective pancreatic surgical procedures other than distal pancreatectomy in patients with chronic pancreatitis. This risk seems to be largely caused by progression of the disease because it increased by more than 3-fold after the onset of pancreatic calcifications.

iabetes mellitus is the major late sequelae of chronic pancreatitis (CP).1–3 It is an independent risk factor for mortality in patients with CP, whether they have undergone surgery1,4 –7 or not,1,7,8 and it affects the quality of life.8 Contrary to exocrine pancreatic insufficiency, endocrine pancreatic insufficiency may lead to life-threatening complications such as severe hypoglycemia9 or to chronic microangiopathic and macroangiopathic compli-

D

cations, which are as frequent in CP patients with diabetes as in other diabetic patients.10 –12 The prevalence of diabetes mellitus in large series of patients with CP has varied between 41% and 86% depending on the diagnostic criteria for CP and disturbances in glucose metabolism, definition of the onset of CP, duration of follow-up, and pattern of recruitment of the patients (medical, surgical, or mixed).1–3,9,10,13–16 Loss of functioning pancreatic tissue in patients with CP may result from disease progression or surgical resection. The effect of pancreatic drainage in slowing the development of diabetes mellitus is controversial. It is not clear whether disease progression or surgery prominently influences the risk of diabetes mellitus development; it is impossible to predict the effects of the different surgical procedures in patients with CP on the postoperative appearance of diabetes mellitus because most surgical series dealt with only a single surgical procedure or included patients with CP and those with other pancreatic diseases, or lacked a nonsurgical control group. This hampers the assessment of the real influence of pancreatic surgery on the postoperative onset of diabetes mellitus. To our knowledge, there is no large medical-surgical study assessing risk factors for diabetes mellitus in patients with CP treated surgically compared with conservative management. This study is based on a prospective, medical-surgical cohort of 500 patients followed up more than 35 years from the onset of CP. We aimed to (1) assess the risk of diabetes mellitus in patients with CP undergoing surgical or conservative treatment, (2) compare the influence of each surgical procedure on this risk, and (3) determine the risk factors for diabetes mellitus in patients with CP, whether they have undergone surgery or not. Abbreviations used in this paper: CI, confidence interval; CP, chronic pancreatitis. © 2000 by the American Gastroenterological Association 0016-5085/00/$10.00 doi:10.1053/gast.2000.19286

November 2000

Patients and Methods Patients All consecutive patients with suspected CP attending the medical and surgical gastroenterology units of our institution between 1973 and 1996, fulfilling the criteria of CP cited below during follow-up, were prospectively studied. Exclusion criteria were as follows: surgical procedures different from elective surgery for CP, such as gastrectomy or vagotomy for peptic ulcer disease (because of an unpredictable effect on glycoregulation); total pancreatectomy; pancreatic drainage procedure followed by pancreatic resection more than 1 year later; insulin-dependent diabetes mellitus diagnosed more than 2 years before the clinical onset of CP; and family history of insulin-dependent diabetes mellitus.

Definitions The diagnosis of CP was based on 1 or more of 3 criteria: (1) pancreatic calcifications as evidenced by plain radiography of the pancreatic area in 3 projections, computed tomography, ultrasonography, or endoscopic ultrasonography; (2) moderate-to-marked pancreatic ductal lesions on pancreatography (Cambridge criteria)17; and (3) typical histology of an adequate surgical pancreatic specimen. Diabetes mellitus was not considered as a sufficient diagnostic criterion of CP.18 CP was considered to be caused by alcohol when alcohol intake exceeded 60 g/day for at least 2 years in the absence of other causes; CP was considered idiopathic when no cause was found.18 Diabetes mellitus was diagnosed by at least 1 of the following criteria: (1) fasting blood glucose ⱖ6.7 mmol/L (whole venous blood) at at least 2 determinations; and (2) blood glucose ⱖ10.0 mmol/L (whole venous blood) 2 hours after ingestion of 75 g glucose.19 In cases of bouts of acute pancreatitis, these laboratory tests were postponed for at least 3 weeks after the end of the attack. Diabetes mellitus was considered to be caused by CP when it was diagnosed after the onset of CP or ⬍2 years before the diagnosis of CP in cases of painless CP. Insulin requirement was established when oral drugs and adequate diet were unable to prevent hyperglycemic symptoms and mean blood glucose throughout the day (at least 3 measurements) was ⬎10.0 mmol/L (whole venous blood). In this study, elective pancreatic surgery denotes all elective surgical procedures performed for CP, except total pancreatectomy. Patients who never underwent surgery served as controls. Surgical procedures were categorized in 4 subgroups. 1. Pancreatic drainage procedures (longitudinal pancreaticojejunostomy). 2. Cystic, digestive, or biliary drainage procedures. The time of surgery in patients who underwent several nonpancreatic drainage procedures was taken as that of the first intervention. Patients who underwent pancreatic drainage and other drainage procedures (whatever their type) simultaneously or consecutively were included in the pancreatic drainage subgroup.

RISK OF DIABETES MELLITUS IN CHRONIC PANCREATITIS

1325

3. Pancreaticoduodenectomy (in which either a pancreaticojejunostomy or a pancreaticogastrostomy was performed). 4. Distal pancreatectomy, in which 50%–70% of the distal pancreas was resected.20 Patients who underwent a pancreatic resection and subsequent drainage were included in the resection group. Attacks of acute pancreatitis were defined by acute abdominal pain requiring hospitalization associated with increased pancreatic enzyme levels above 3 times the upper limit of normal values and radiologic, intraoperative, or pathologic findings.21 Pancreatic pseudocysts were diagnosed on radiologic, intraoperative, or pathologic data.18 Splenic or portal vein occlusion was systematically searched for using ultrasonography and then confirmed by angiography, contrastenhanced computerized tomography, or surgery.22 Symptomatic duodenal stenosis was confirmed by a barium series.23 The diagnosis of biliary obstructive liver disease24 and/or alcoholic liver disease, with or without cirrhosis, was histologically proven or established using concordant clinical, laboratory, and radiologic data.

Study Design and Data Collection Patients were seen as outpatients at least once a year, or unscheduled (as outpatients or hospitalized) when symptomatic or when their disease was complicated. Duration of CP was defined as the time between the onset of CP (first symptom, sign, or complication clearly related to the disease) and either the cutoff date of the study, the last personal contact, or death. The course of the disease from onset to the first attendance in our institution was analyzed from hospital records or contacts with referring physicians. The time of onset of pancreatic calcifications, which do not result in symptoms, was considered as the time elapsed from the diagnosis of CP to the time of their first demonstration. The time of onset of all other time-dependent covariables was considered as the time elapsed from the onset of CP to their onset. At each attendance, patients underwent clinical assessment, abdominal ultrasonography or computerized tomography, and routine laboratory tests including fasting blood glucose and a 2-hour glucose tolerance test. Data were then registered prospectively on a database. Pancreatic exocrine insufficiency was searched for only in case of diarrhea or weight loss. Alcoholic abstinence was not assessed in this study.

Statistical Analysis General characteristics are expressed as mean ⫾ SD. The differences were statistically analyzed by using the Student t test or Mann–Whitney U test as necessary for continuous data and by using the ␹2 test or the Fisher exact test as necessary for categorical data. Cumulative rates of diabetes mellitus and insulin requirement since the onset of CP within the surgical and nonsurgical groups, or since the time of surgery within the different

1326

MALKA ET AL.

GASTROENTEROLOGY Vol. 119, No. 5

Figure 1. Patient subgroups. In the distal pancreatectomy subgroup, 8 patients also underwent nonpancreatic drainage procedures and 3 patients also underwent pancreatic drainage procedures. In the pancreaticoduodenectomy subgroup, 6 patients also underwent nonpancreatic (biliary, cystic, or digestive) drainage procedures. In the pancreatic drainage subgroup, 35 patients also underwent nonpancreatic (biliary, cystic, or digestive) drainage procedures.

surgical subgroups, were calculated using the Kaplan–Meier method.25 Comparisons between the different surgical subgroups were performed using the score test in the Cox proportional hazards regression model.26 The time-dependent pattern of elective pancreatic surgery (surgery performed after the onset of CP) precluded statistical comparisons between the surgical and nonsurgical groups. For information, we have given the cumulative rates of diabetes mellitus and insulin requirement of patients undergoing, or not, elective pancreatic surgery since the clinical onset of CP. For multivariate analysis, we performed the Cox proportional hazards regression model to determine the independent prognostic factors.26 Covariables entered in the model were chosen according to the best score criterion strategy. Explanatory covariables were considered either as time independent (e.g., sex, age of onset of CP, cause of CP) or time dependent (e.g., pancreatic calcifications, liver cirrhosis, elective pancreatic surgery). Complications of CP, reversible and often recurrent, were coded as categorical, time-independent covariables (yes/no). The data were analyzed with the SAS 6.12 statistical software for Windows (SAS Institute Inc., Cary, NC). All

statistical tests were 2-sided. The critical level of statistical significance was P ⬍ 0.05.

Results Characteristics of the Patients The prospective cohort consisted of 500 patients with proven CP, 47 (9%) of whom were excluded from the present analysis. Of the remaining 453 patients, 231 (51%) underwent elective pancreatic surgery and 222 (49%) never underwent surgery (Figure 1 and Table 1). Alcohol was the cause of CP in 385 patients (85%). Besides idiopathic CP (n ⫽ 54; 12%), the causes of nonalcoholic CP were hereditary CP (n ⫽ 7), abdominal radiotherapy (n ⫽ 4), inflammatory bowel disease (n ⫽ 2), and Sjo¨gren’s syndrome (n ⫽ 1). The mean follow-up was of 7.0 ⫾ 6.8 years (range, 0 –35). Fifty-eight patients (13%) died within the follow-up at an average of 10.1 ⫾ 6.7 years after the onset of CP; 4 (0.9%) of them died of a pancreatic adenocarcinoma,

Table 1. Main Indications for Surgery

Indicationsa Pseudocysts Chronic pain Biliary stricture Duodenal stenosis Fistulas, pleural or peritoneal effusion Hemorrhage aMay

Distal pancreatectomy (n ⫽ 62)

Pancreaticoduodenectomy (n ⫽ 37)

Nonpancreatic drainages (n ⫽ 86)

Pancreatic drainage (n ⫽ 46)b

35 10 — — 26 4

18 11c 12 11 2 1

47 — 42 18 4 —

9 30 — — — —

be multiple for each type of surgical procedure. 16 patients in whom a pancreatic drainage was performed prophylactically during a nonpancreatic drainage procedure. cIncluding 1 case for suspicion of pancreatic cancer. bIncluding

November 2000

RISK OF DIABETES MELLITUS IN CHRONIC PANCREATITIS

Table 2. Clinical Characteristics of the Patients in the Surgical and Nonsurgical Groups

Age ( yr) at onset of CPa Age ( yr) at diagnosis of CPa Sex (M/F) Duration of follow-up ( yr)a Duration of CP before surgery ( yr)a Postoperative followup ( yr)a Death (%) Alcoholic CP (%) Smoking (%) Pancreatic calcifications (%) Acute pancreatitis (%) Pseudocysts (%) Duodenal stenosis (%) Splenic or portal vein occlusion (%) Liver disease (%) Biliary (%) Alcoholic (%) Other (%) Cirrhosis (%)

Surgical group (n ⫽ 231)

Nonsurgical group (n ⫽ 222)

39.3 ⫾ 12.1

43.1 ⫾ 13.5

⬍0.01

41.5 ⫾ 12.3 203/28

44.6 ⫾ 14.1 184/38

⬍0.05 NS

9.2 ⫾ 7.2

4.8 ⫾ 5.3

⬍0.0001

4.6 ⫾ 5.5

—

—

4.5 ⫾ 5.5 43 (19) 199 (86) 174 (75)

— 15 (7) 186 (84) 158 (71)

— ⬍0.001 NS NS

182 (79) 131 (57) 134 (58) 57 (25)

176 (79) 114 (51) 67 (30) 13 (6)

NS NS ⬍0.0001 ⬍0.0001

58 (25) 100 (43) 51 (22)b 39 (17)b 15 (6) 22 (10)

23 (10) 55 (25) 2 (1) 35 (16) 18 (8) 23 (10)

⬍0.0001 ⬍0.0001 ⬍0.0001 NS NS NS

P

NS, not significant. are mean ⫾ SD. bMixed alcoholic and biliary liver disease in 5 patients. aValues

in one case 22 months after the onset of insulindependent diabetes mellitus. Patients in the surgical group differed from the conservatively treated patients with respect to: younger age at onset and at diagnosis of CP; longer duration of follow-up; greater prevalence of pseudocysts, duodenal

Figure 2. Cumulative rates of diabetes mellitus (thick line) and insulin requirement (thin line) in 453 patients with CP since the onset of the disease.

1327

stenosis, splenic or portal vein occlusion, biliary complications, and higher mortality rate. The percentage of patients who experienced attacks of acute pancreatitis was similar in both groups (Table 2). The mean alcohol consumption of patients who admitted to drinking alcohol was of 165 g/day, and was similar in both surgical and nonsurgical groups. The proportion of patients who had never drunk or had drunk ⬍20 g alcohol daily at the time of diagnosis of CP was 2 times higher in the nonsurgical than the surgical group (47 [25%] vs. 24 [13%]; P ⫽ 0.03). The mean cumulative tobacco consumption of smokers at the time of diagnosis of CP was 30 pack-years in both groups. The proportion of patients who had never smoked or had smoked ⬍10 pack-years was 2 times higher in the nonsurgical than the surgical group (51 [24%] vs. 26 [12%]; P ⫽ 0.001). Onset of Diabetes Mellitus Diabetes mellitus and insulin requirement occurred at a mean of 4.5 ⫾ 5.9 years and 6.9 ⫾ 7.0 years after the onset of CP, respectively. They were present at the onset of CP in 10% and 2% of patients, respectively. The cumulative rates of appearance of diabetes mellitus calculated since the onset of CP were 50% ⫾ 3% and 83% ⫾ 4% at 10 and 25 years, respectively. The cumulative rates of insulin requirement were 26% ⫾ 3% and 53% ⫾ 6% at 10 and 25 years, respectively. Annual rates of diabetes mellitus and insulin requirement, estimated by linear regression, were 3.5% and 2.2%, respectively (Figure 2). Influence of Elective Pancreatic Surgery on the Onset of Diabetes Mellitus The cumulative rates of appearance of diabetes mellitus and insulin requirement within the surgical

1328

MALKA ET AL.

GASTROENTEROLOGY Vol. 119, No. 5

Figure 3. Cumulative rates of diabetes mellitus and insulin requirement in patients with CP who did (n ⫽ 231) or did not (n ⫽ 222) undergo elective pancreatic surgery since the onset of the disease.

group were 79% ⫾ 5% and 53% ⫾ 5%, respectively, 20 years after the onset of CP, vs. 69% ⫾ 6% and 42% ⫾ 8%, respectively, within the nonsurgical group (Figure 3). Statistical comparisons within the different surgical subgroups since the time of surgery showed that distal pancreatectomy was associated with a higher rate of diabetes mellitus and insulin requirement compared with the other types of surgical procedures (score test

[regression analysis]: P ⫽ 0.005 and P ⫽ 0.03, respectively). Differences between the latter ones were not statistically significant (Figure 4). Risk Factors for Diabetes Mellitus Diabetic patients differed from nondiabetic patients with respect to: younger age at the onset of CP; longer follow-up (and longer postoperative follow-up in patients treated surgically); higher prevalence of surgery;

Figure 4. Cumulative rates of (A ) diabetes mellitus and (B) insulin requirement since the time of surgery in 231 patients with CP who underwent pancreatic drainage (}; n ⫽ 46), nonpancreatic (biliary, cystic, or digestive) drainage procedures ({; n ⫽ 86), pancreaticoduodenectomy (E; n ⫽ 37), or distal pancreatectomy (F; n ⫽ 62). Differences were significant between distal pancreatectomy and other types of surgical procedures (score test [regression analysis]: diabetes mellitus, P ⫽ 0.005; insulin requirement, P ⫽ 0.03).

November 2000

RISK OF DIABETES MELLITUS IN CHRONIC PANCREATITIS

Table 3. Clinical Characteristics of the Diabetic and Nondiabetic Patients

Age ( yr) at onset of CPa Age ( yr) at diagnosis of CPa Sex (M/F) Duration of follow-up ( yr)a Elective pancreatic surgery (%) Duration of CP before surgery ( yr)a Postoperative follow-up ( yr)a Death (%) Alcoholic CP (%) Smoking (%) Pancreatic calcifications (%) Acute pancreatitis (%) Pseudocysts (%) Duodenal stenosis (%) Splenic or portal vein occlusion (%) Liver disease (%) Biliary (%) Alcoholic (%) Other (%) Cirrhosis (%)

No diabetes (n ⫽ 274)

Diabetes (n ⫽ 179)

P

42.3 ⫾ 13.1

39.5 ⫾ 12.5

0.003

43.5 ⫾ 14.2 236/38

42.3 ⫾ 11.9 151/28

NS NS

5.6 ⫾ 5.8

9.2 ⫾ 7.3

⬍0.0001

128 (47)

103 (58)

0.03

4.2 ⫾ 5.5

5.1 ⫾ 5.4

NS

3.1 ⫾ 3.9 29 (11) 230 (84) 197 (72)

6.4 ⫾ 6.6 29 (16) 155 (87) 133 (74)

⬍0.0001 NS NS NS

202 (74) 150 (55) 125 (46) 43 (16)

154 (86) 95 (53) 76 (42) 27 (15)

0.003 NS NS NS

40 (15) 87 (32) 31 (11)b 35 (13)b 23 (8) 20 (7)

41 (23) 68 (38) 22 (12) 36 (20) 10 (6) 25 (14)

NS NS NS NS NS 0.03

NS, not significant. are mean ⫾ SD. bMixed alcoholic and biliary liver disease in 2 patients. a Values

and pancreatic calcifications and liver cirrhosis. No difference was found in tobacco or alcohol consumption, or the prevalence of complications of CP or noncirrhotic liver disease between diabetic and nondiabetic patients (Table 3). Univariate analysis identified 2 time-dependent covariables associated with an increased risk of diabetes mellitus or insulin requirement: onset of pancreatic calcifications (risk ratios [95% confidence interval (CI)], 3.2 [2.2– 4.5] and 3.2 [1.9 –5.5], respectively; P ⬍ 0.0001) and distal pancreatectomy (risk ratios [CI], 2.9 [1.9 – 4.5] and 3.1 [1.8 –5.4], respectively; P ⬍ 0.0001). Distal pancreatectomy was the only surgical procedure associated with an increased risk of diabetes mellitus (P ⫽ 0.0001 vs. other surgical procedures) or insulin requirement (P ⫽ 0.0001). No significant difference was found between pancreaticoduodenectomy, pancreatic drainage, and nonpancreatic drainage procedures for diabetes mellitus or insulin requirement. Splenic or portal vein occlusion was significantly associated with the risk of insulin requirement (risk ratios [CI ]: 2.0 [1.0 –3.8], P ⫽ 0.04; and 1.8 [1.1–2.9], P ⫽ 0.01, respectively). Other

1329

covariables studied did not significantly affect the risk of diabetes mellitus or insulin requirement. Time-dependent multivariate analysis showed that the onset of pancreatic calcifications and distal pancreatectomy were the only independent prognostic factors for the risk of diabetes mellitus or insulin requirement. Alcoholic cause of CP was significant for the risk of diabetes mellitus, but not for the risk of insulin requirement (Table 4).

Discussion The first important result of this study is that elective surgery for CP, except distal pancreatectomy, did not seem to influence the risk of diabetes mellitus or insulin requirement up to 25 years after the onset of the disease. This has been previously suggested by the Zurich series1; however, comparisons of cumulative rates of diabetes mellitus in patients with CP, whether they underwent pancreatic surgery or not, by statistical means such as the log-rank test are theoretically impossible, because the event surgery, which is used to define the 2 groups, occurs during the course of the disease.7 Timedependent multivariate analysis allowed us to circumvent this pitfall. The strengths of our study are (1) inclusion of all consecutive patients with proven CP seen in a single medical-surgical center over a period of 23 years; (2) prospective standardized collection of all epidemiologic and clinical data concerning CP and its treatment; (3) comparisons of all the different elective pancreatic surgical procedures; (4) multivariate analysis of risk factors for diabetes mellitus and insulin requirement taking into account the time-dependent pattern of several important variables; and (5) long duration of patient follow-up. Patients in the surgical group were younger than conservatively treated patients at the onset and diagnosis of CP, were more often heavy drinkers and smokers, and had a higher mortality. Obviously, they also had a greater prevalence of pseudocysts, symptomatic duodenal stenosis, splenic or portal vein occlusion, and biliary obstruction. These characteristics may indicate a more severe course of CP in these patients, underlining the minor role of surgery in the occurrence of diabetes mellitus. The cumulative rate of appearance of diabetes mellitus 25 years after the onset of CP was 83% ⫾ 4%. Diabetes mellitus occurred a mean of 4.5 ⫾ 5.9 years after the onset of CP. Corresponding figures in the Zurich series were ⬃ 90% and 5.7 years, respectively.1 Distal pancreatectomy was the only surgical procedure that increased the risk of diabetes mellitus. This has been previously suggested by several reports.6,27–32 However, most of these studies followed up small numbers of patients,

1330

MALKA ET AL.

GASTROENTEROLOGY Vol. 119, No. 5

Table 4. Multivariate Analysis of Risk Factors for Diabetes Mellitus and Insulin Requirement Diabetes mellitus Risk ratio (95% CI) Sex (M) Age at onset of CP Alcoholic cause of CP Smoking Pancreatic calcifications Acute pancreatitis Pseudocysts Duodenal stenosis Splenic or portal vein occlusion Biliary liver disease Liver fibrosis or cirrhosis Elective pancreatic surgery Distal pancreatectomy Pancreaticoduodenectomy Nonpancreatic drainagesa Pancreatic drainage

0.7 (0.4–1.1) 1.0 (0.9–1.1) 1.7 (1.0–2.7) 0.9 (0.6–1.3) 3.2 (2.2–4.7) 0.9 (0.6–1.3) 1.0 (0.7–1.4) 0.8 (0.5–1.2) 1.2 (0.8–1.7) 1.3 (0.7–1.3) 1.1 (0.6–2.1) 1.2 (0.8–1.8) 2.4 (1.6–3.8) 0.7 (0.3–1.7) 0.7 (0.4–1.3) 0.8 (0.4–1.7)

Insulin requirement

P NS NS 0.04 NS ⬍0.0001 NS NS NS NS NS NS NS ⬍0.0001 NS NS NS

Risk ratio (95% CI) 0.9 (0.5–1.7) 1.0 (0.9–1.1) 1.2 (0.6–2.4) 0.8 (0.5–1.5) 3.3 (1.9–5.9) 1.0 (0.6–1.5) 1.4 (0.8–2.1) 0.9 (0.5–1.6) 1.4 (0.7–2.2) 1.1 (0.6–1.6) 1.0 (0.4–2.6) 1.4 (0.8–2.4) 2.6 (1.4–4.7) 1.3 (0.5–3.3) 0.9 (0.4–1.9) 0.6 (0.2–2.1)

P NS NS NS NS ⬍0.0001 NS NS NS NS NS NS NS 0.001 NS NS NS

NS, not significant. aDenotes biliary, cystic, or digestive drainage procedures.

often for a short time, did not compare distal pancreatectomy with other elective pancreatic surgical procedures, or lacked a control group. Seven studies including a total of 199 patients with CP followed up for 3.6 –7.0 years after a 40%– 80% distal pancreatectomy have shown that the prevalence of insulin-dependent diabetes mellitus increased from 0%–10% preoperatively to 19%– 48% in the late postoperative period.28 –34 Increasing the extent of distal pancreatectomy to 80%–95% led to a postoperative rate of diabetes mellitus of 58%–100% and increased late mortality.6,29,30,35–37 The influence of pancreaticoduodenectomy on the risk of postoperative diabetes mellitus is unclear.4,27–32,38 – 41 It has been shown that pancreaticoduodenectomy results in decreased insulin secretion after surgery.42 Eight series including a total of 325 patients with CP followed up between 25 months and 6.2 years after pancreaticoduodenectomy found an increase in the rate of insulin-dependent diabetes mellitus from 0%–16%

preoperatively to 11%– 68% in the late postoperative follow-up.4,28,30,36,38,39,41,43 In contrast with these studies, we found that pancreaticoduodenectomy did not influence the risk of diabetes mellitus compared with the natural history of CP in patients treated conservatively. The lack of a statistically significant difference between pancreaticoduodenectomy and nonpancreatic drainage procedures (which may be considered as positive surgical controls) with respect to the risk of diabetes mellitus is consistent with this finding. This result is in accordance with the heterogeneous distribution of Langerhans islets along the pancreatic gland, which are prominently localized to the tail of the pancreas.44 Similar to the present study, a recent report has shown that diabetes mellitus appearing after pancreaticoduodenectomy performed for CP was almost always delayed for a minimum of 1 year after surgery, suggesting that disease progression prevails in the risk of diabetes mellitus.4 The postoperative appearance of diabetes mellitus has been noted less fre-

November 2000

quently when pancreatectomy was performed for focal disease than for diffuse CP.41 Thus, pancreaticoduodenectomy should preserve sufficient endocrine pancreatic tissue to ensure normal glycemic regulation in most patients. Whether pancreatic drainage preserves or even improves pancreatic function by releasing obstructed pancreatic juice is controversial.5,27–29,31,45–52 We found that pancreatic drainage did not prevent the eventual appearance of diabetes mellitus. The first suggestion of a significant delay in pancreatic dysfunction by surgical pancreatic drainage was provided by Nealon et al.51 In a partially randomized prospective study including 143 patients followed up for a mean of 47.3 months, they showed that 13% of patients who underwent pancreatic drainage had progressive loss of their functional status compared with 78% of those conservatively treated.52 However, the principal criterion to determine candidacy for duct drainage in this series was the presence of a dilated pancreatic duct. Thus, what the study actually reports is the outcome of pancreatic drainage in patients with dilated ducts vs. the natural history of patients with CP and no ductal dilation. It is therefore uncertain whether the slower deterioration of pancreatic endocrine function in patients who underwent surgery is actually caused by the surgical drainage procedure. Other series did not report long-term favorable results of pancreatic drainage procedures despite radiologic evidence of anastomotic patency.5,29,45,46,53,54 A recent study showed that duct drainage achieved after endoscopic therapy and extracorporeal shockwave lithotripsy did not correlate with an improvement in pancreatic endocrine function.55 Our study suggests that pancreatic duct obstruction plays a minor role in the appearance of diabetes mellitus compared with parenchymal destruction and does not support a policy of early surgery in patients with CP before significant loss of pancreatic endocrine function.51,52 However, we cannot exclude a lack of statistical power because of the small number of patients in this surgical subgroup. The onset of pancreatic calcifications was the only clinical factor significantly associated with the risk of diabetes mellitus; once they appeared, the risk of diabetes mellitus and insulin requirement increased by more than 3-fold. Because we took into account the time-dependent pattern of the onset of pancreatic calcifications, we can conclude that the earlier they appear in the course of CP, the greater the risk of diabetes mellitus and insulin requirement. A tight correlation between the onset of pancreatic calcifications and pancreatic endocrine (and exocrine) insufficiency has been shown.1 This fact may attest that the onset of diabetes mellitus in patients with CP is mainly caused by progression of the disease, rather than by elective pancreatic surgery.

RISK OF DIABETES MELLITUS IN CHRONIC PANCREATITIS

1331

In conclusion, distal pancreatectomy was the only elective surgical procedure that modified the risk of diabetes mellitus compared with the natural course of CP. Efforts should be made to avoid distal pancreatectomy when other surgical (or endoscopic) procedures are feasible. Pancreatic drainage did not prevent the onset of diabetes mellitus. This finding does not support prophylactic pancreatic drainage procedures performed early in the course of CP as a means of preventing the progression of endocrine pancreatic insufficiency. Progression of CP seems to be the main determinant of diabetes mellitus in both surgical and nonsurgical patients.

References 1. Ammann RW, Akovbiantz A, Largiader F, Schueler G. Course and outcome of chronic pancreatitis. Longitudinal study of a mixed medical-chirurgical series of 245 patients. Gastroenterology 1984;86:820 – 828. 2. Bernades P, Belghiti J, Athouel M, Mallardo N, Breil P, Fe´ke´te´ F. Histoire naturelle de la pancre´atite chronique: e´tude de 120 cas. Gastroenterol Clin Biol 1983;7:8 –13. 3. Ammann RW, Muellhaupt B, Zurich Pancreatitis Study Group. The natural history of pain in alcoholic chronic pancreatitis. Gastroenterology 1999;116:1132–1140. 4. Traverso LW, Kozarek RA. Pancreatoduodenectomy for chronic pancreatitis. Anatomic selection criteria and subsequent longterm outcome analysis. Ann Surg 1997;226:429 – 438. 5. Sato T, Miyashita E, Yamauchi H, Matsuno S. The role of surgical treatment for chronic pancreatitis. Ann Surg 1986;203:266 – 271. 6. Eckhauser FE, Strodel WE, Knol JA, Harper M, Turcotte JG. Neartotal pancreatectomy for chronic pancreatitis. Surgery 1984;96: 599 – 607. 7. Le´vy P, Milan C, Pignon JP, Baetz A, Bernades P. Mortality factors associated with chronic pancreatitis. Unidimensional and multidimensional analysis of a medical-surgical series of 240 patients. Gastroenterology 1989;96:1165–1172. 8. Petrin P, Andreoli A, Antoniutti M, Zaramella D, DaLio C, Bonadimani B, Garbin L, Pedrazzoli S. Surgery for chronic pancreatitis: what quality of life ahead? World J Surg 1995;19:398 – 402. 9. Linde J, Nilsson LH, Barany FR. Diabetes and hypoglycemia in chronic pancreatitis. Scand J Gastroenterol 1977;12:369 –373. 10. Koizumi M, Yoshida Y, Abe N, Shimosegawa T, Toyota T. Pancreatic diabetes in Japan. Pancreas 1998;16:385–391. 11. Gullo L, Parenti M, Monti L, Pezzilli R, Barbara L. Diabetic retinopathy in chronic pancreatitis. Gastroenterology 1990;98: 1577–1581. 12. Ziegler O, Candiloros H, Guerci B, Got I, Crea T, Drouin P. Lowerextremity arterial disease in diabetes mellitus due to chronic pancreatitis. Diabetes Metab 1994;20:540 –545. 13. James O, Agnew JE, Bouchier IAD. Chronic pancreatitis in England: a changing picture? BMJ 1974;2:34 –38. 14. Lankisch PG, Otto J, Erkelenz I, Lembcke B. Pancreatic calcifications: no indicator of severe exocrine pancreatic insufficiency. Gastroenterology 1986;90:617– 621. 15. Lankisch PG, Lohr-Happe A, Otto J, Creutzfeldt W. Natural course in chronic pancreatitis. Pain, exocrine and endocrine pancreatic insufficiency and prognosis of the disease. Digestion 1993;54: 148 –155. 16. Cavallini G, Frulloni L, Pederzoli P, Talamini G, Bovo P, Bassi C, Di Francesco V, Vaona B, Falconi M, Sartori N, Angelini G, Brunori MP, Filippini M. Long-term follow-up of patients with chronic pancreatitis in Italy. Scand J Gastroenterol 1998;33:880 – 889.

1332

MALKA ET AL.

17. Axon ATR, Classen M, Cotton PB, Cremer M, Freeny PC, Lees WR. Pancreatography in chronic pancreatitis: international definitions. Gut 1984;25:1107–1112. 18. Ammann RW. A clinically based classification system for alcoholic chronic pancreatitis: summary of an international workshop on chronic pancreatitis. Pancreas 1997;14:215–221. 19. Anonymous. Classification and diagnosis of diabetes mellitus and other categories of glucose tolerance. National Diabetes Data Group. Diabetes 1979;28:1039 –1057. 20. Rossi RL, Heiss FW, Braasch JW. Surgical management of chronic pancreatitis. Surg Clin North Am 1985;65:79 –101. 21. Steinberg W, Tenner S. Acute pancreatitis. N Engl J Med 1994; 330:1198 –1210. 22. Bernades P, Baetz A, Le´vy P, Belghiti J, Menu Y, Fe´ke´te´ F. Splenic and portal venous obstruction in chronic pancreatitis. A prospective longitudinal study of a medical-surgical series of 266 patients. Dig Dis Sci 1992;37:340 –346. 23. Le´vy P, Lesur G, Belghiti J, Fe´ke´te´ F, Bernades P. Symptomatic duodenal stenosis in chronic pancreatitis: a study of 17 cases in a medical-surgical series of 306 patients. Pancreas 1993;8: 563–567. 24. Lesur G, Le´vy P, Fle´jou JF, Belghiti J, Fe´ke´te´ F, Bernades P. Factors predictive of liver histopathological appearance in chronic alcoholic pancreatitis with common bile duct stenosis and increased serum alkaline phosphatase. Hepatology 1993; 18:1078 –1081. 25. Kaplan EL, Meier P. Nonparametric estimation from incomplete observations. J Am Stat Assoc 1958;53:457– 481. 26. Cox DR. Regression models and life-tables. J R Soc Stat [B] 1972;34:187–220. 27. Traverso LW, Tompkins RK, Urrea PT, Longmire WP Jr. Surgical treatment of chronic pancreatitis. Twenty-two years’ experience. Ann Surg 1979;190:312–319. 28. Jalleh RP, Williamson RCN. Pancreatic exocrine and endocrine function after operations for chronic pancreatitis. Ann Surg 1992; 216:656 – 662. 29. Morrow CE, Cohen JI, Sutherland DER, Najarian JS. Chronic pancreatitis: long-term surgical results of pancreatic duct drainage, pancreatic resection, and near-total pancreatectomy and islet autotransplantation. Surgery 1984;96:608 – 616. 30. Frey CF, Child III CG, Fry W. Pancreatectomy for chronic pancreatitis. Ann Surg 1976;184:403– 413. 31. Taylor RH, Bagley FH, Braasch JW, Warren KW. Ductal drainage or resection for chronic pancreatitis. Am J Surg 1981;14:28 –33. 32. Williamson RCN, Cooper MJ. Resections in chronic pancreatitis. Br J Surg 1987;74:807– 812. 33. Warshaw AL. Conservation of pancreatic tissue by combined gastric, biliary, and pancreatic duct drainage for pain from chronic pancreatitis. Am J Surg 1985;149:563–569. 34. Sawyer R, Frey CF. Is there still a role for distal pancreatectomy in surgery for chronic pancreatitis? Am J Surg 1994;168:6 –9. 35. Keith RG, Saibil FG, Sheppard RH. Treatment of chronic alcoholic pancreatitis by pancreatic resection. Am J Surg 1989;157:156 – 162. 36. Gall FP, Muhe E, Gebhardt C. Etude comparative de 117 duode´no-pancre´atectomies subtotales et totales pour pancre´atite chronique. Chirurgie 1979;105:187–192. 37. White TT, Slavotinek AH. Results of surgical treatment of chronic pancreatitis. Report of 142 cases. Ann Surg 1979;189:217–224. 38. Rossi RL, Rothschild J, Braasch JW, Munson JL, Remine SG.

GASTROENTEROLOGY Vol. 119, No. 5

39.

40.

41.

42.

43.

44. 45.

46.

47. 48. 49.

50.

51.

52.

53. 54. 55.

Pancreatoduodenectomy in the management of chronic pancreatitis. Arch Surg 1987;122:416 – 420. Stone WM, Sarr MG, Nagorney DM, McIlrath DC. Chronic pancreatitis: results of Whipple’s resection and total pancreatectomy. Arch Surg 1988;123:815– 819. Gall FP, Zirngibl H, Gebhardt C, Schneider MU. Duodenal pancreatectomy with occlusion of the pancreatic duct. Hepatogastroenterology 1990;37:290 –294. Sato N, Yamaguchi K, Yokohata K, Shimizu S, Noshiro H, Mizumoto K, Chijiiwa K, Tanaka M. Changes in pancreatic function after pancreatoduodenectomy. Am J Surg 1998;176:59 – 61. Miyata M, Yamamoto T, Hamaji M, Izukura M, Nakamura M, Taketani H, Nakao K, Kawashima Y. Pancreatic endocrine functions in long-term survivors after pancreatoduodenectomy: special reference to reversibility of insulin and glucagon secretion. World J Surg 1988;12:651– 657. Traverso LW, Kozarek RA. The Whipple procedure for severe complications of chronic pancreatitis. Arch Surg 1993;128: 1047–1053. Wittingen J, Frey CF. Islet concentration in the head, body, tail and uncinate process of the pancreas. Ann Surg 1974;179:412– 414. Adams DB, Ford MC, Anderson MC. Outcome after lateral pancreaticojejunostomy for chronic pancreatitis. Ann Surg 1994; 219:481– 489. Warshaw AL, Popp JW, Schapiro RH. Long term patency, pancreatic function, and pain relief after lateral pancreaticojejunostomy for chronic pancreatitis. Gastroenterology 1980;79:289 –292. Bradley EL III, Nasrallah SM. Fat absorption after longitudinal pancreaticojejunostomy. Surgery 1984;95:640 – 643. Prinz RA, Greenlee HB. Pancreatic duct drainage in 100 patients with chronic pancreatitis. Ann Surg 1981;194:313–320. Jordan GL, Strug BS, Crowder WE. Current status of pancreaticojejunostomy in the management of chronic pancreatitis. Am J Surg 1977;133:46 –51. Sato T, Saitoh Y, Noto W, Matsuno K. Appraisal of operative treatment for chronic pancreatitis with special reference to side to side pancreaticojejunostomy. Am J Surg 1975;129:621– 627. Nealon WH, Townsend CM Jr, Thompson JC. Operative drainage of the pancreatic duct delays functional impairment in patients with chronic pancreatitis: a prospective analysis. Ann Surg 1988; 208:321–329. Nealon WH, Thompson JC. Progressive loss of pancreatic function in chronic pancreatitis is delayed by main pancreatic duct decompression. A longitudinal prospective analysis of the modified Puestow procedure. Ann Surg 1993;217:458 – 468. Ebbehoj N, Klaaborg KE, Kronborg O, Madsen P. Pancreaticogastrostomy for chronic pancreatitis. Am J Surg 1989;157:315–317. Bradley EL III. Long-term results of pancreatojejunostomy in patients with chronic pancreatitis. Am J Surg 1987;153:207–213. Adamek HE, Jakobs R, Buttmann A, Adamek MU, Schneider AR, Riemann JF. Long term follow up of patients with chronic pancreatitis and pancreatic stones treated with extracorporeal shock wave lithotripsy. Gut 1999;45:402– 405.

Received February 29, 2000. Accepted June 21, 2000. Address requests for reprints to: David Malka, M.D., Service de GastroEnte ´rologie, Ho ˆpital Beaujon, 100, Bld du Gal Leclerc, 92110 Clichy, France. e-mail: [email protected]; fax: (33) 1-42-70-37-84.