|
|
| |||||||||||||||||||||||||||||||||||||||
Background Diabetes and pancreatic cancer are known to be associated, but the cause of the association and whether diabetes is a risk factor for pancreatic cancer remain controversial.
Methods A total of 720 patients with pancreatic cancer and 720 control patients from 14 Italian centers were enrolled in the study. All subjects were interviewed personally and in detail about their clinical history. The diagnosis of diabetes was based on criteria recommended by the American Diabetes Association.
Results One hundred sixty-four patients with pancreatic cancer (22.8 percent) and 60 controls (8.3 percent) had diabetes. In the majority of the patients with pancreatic cancer (56.1 percent), diabetes was diagnosed either concomitantly with the cancer (in 40.2 percent), or within two years before the diagnosis of cancer (in 15.9 percent). The association between the two conditions was significant (odds ratio, 3.04; 95 percent confidence interval, 2.21 to 4.17). However, when only patients with diabetes of three or more years' duration were considered, the association was no longer significant (odds ratio, 1.43; 95 percent confidence interval, 0.98 to 2.07). All the patients with pancreatic cancer whose diabetes had been diagnosed before the cancer had non-insulin-dependent diabetes; all but one of the control patients with diabetes had the non-insulin-dependent form of the disease.
Conclusions Diabetes in patients with pancreatic cancer is frequently of recent onset and is presumably caused by the tumor. Diabetes is not a risk factor for pancreatic cancer.
Methods
This work, part of a multicenter study of the risk factors for pancreatic cancer in Italy, involved 14 Italian university and community hospitals with experience in the management of pancreatic disease29. All study subjects were patients who were hospitalized in 13 of the study centers between January 1987 and December 1989 and in the 14th center, in Bologna, between January 1987 and December 1992. Of the 76,144 patients hospitalized at the 14 centers during the study, 738 had pancreatic cancer. We studied 720 of them (415 men [57.6 percent] and 305 women [42.4 percent]); the mean age was 62.6 years (range, 22 to 79). The diagnosis of pancreatic cancer was based on the clinical history. It was confirmed by histologic findings in 512 of the patients (71.1 percent) and by surgery in 48 (6.7 percent); in the other 160 patients (22.2 percent), the diagnosis was confirmed by typical findings on at least three imaging procedures (ultrasonography, computed tomography, endoscopic retrograde cholangiopancreatography, or angiography), as well as by the clinical course.
For each patient, a control subject matched for sex, age (±5 years), social class, and geographic region was selected at random from the patients hospitalized at the same time in the same facility for acute nonmalignant disorders. The most frequent diagnoses in the control group were fractures, minor trauma, abdominal hernias, and disorders of the skin, throat, nose, or ear.
All the study subjects were interviewed personally about their medical history; special attention was paid to the presence of diabetes, the patient's age at the time of diagnosis, the type of diabetes, and the treatment received for it. Members of the immediate family were also interviewed in most cases. When possible, the results of previous blood tests, including the glucose concentration, were noted. Patients who were not able to provide detailed information, either because of the severity of their illness or for other reasons, were not included in the study; of the 738 patients with pancreatic cancer who were initially eligible for the study, 18 (2.4 percent) were excluded on this account. Of the 728 eligible controls, 8 (1.1 percent) were excluded because of insufficient cooperation.
The diagnosis of diabetes was based on the criteria recommended by the American Diabetes Association30: an unequivocal elevation of the plasma glucose concentration ( 
200 mg per deciliter [11.1 mmol per liter]) along with the classic symptoms of diabetes such as polydipsia, polyuria, polyphagia, and weight loss; and a fasting plasma glucose concentration greater than or equal to 140 mg per deciliter (7.8 mmol per liter) on more than one occasion.
Statistical Analysis
The chi-square test and the Mann-Whitney U test were used to compare the clinical characteristics of the case patients and control patients; Yates' correction was applied to two-by-two tables. The clinical duration of diabetes in the two groups of patients was compared by means of a log-linear model. A conditional logistic model for matched case-control studies was used in the analysis of the association between diabetes and pancreatic cancer. All statistical evaluations were performed with BMDP statistical software31 on a personal computer.
Results
One hundred sixty-four patients (96 men and 68 women) with pancreatic cancer had diabetes (22.8 percent), as did 60 patients (30 men and 30 women) without pancreatic cancer (8.3 percent). In the group of 164 case patients, the mean age was 64.4 years (range, 36 to 79) when pancreatic cancer was diagnosed and 59.4 years (range, 35 to 79) when diabetes was diagnosed. In the 60 controls, the mean age at the time of the diagnosis of diabetes was 56.1 years (range, 25 to 78). There was a family history of diabetes in 67 of the 164 case patients with diabetes (40.9 percent) and in 24 of the 60 control patients with diabetes (40.0 percent). None of these differences were significant. Among the case patients with diabetes, 127 (77.4 percent) had cancer in the head of the pancreas, and 31 (18.9 percent) in the body or tail of the pancreas (or both); the cancer was diffuse in the other 6 case patients (3.7 percent). The tumor was resectable in 31 case patients (18.9 percent).
Table 1 shows the distribution of case and control patients according to the clinical duration of diabetes. In 92 of the 164 case patients (56.1 percent), diabetes was diagnosed either at the same time as the pancreatic cancer (40.2 percent) or within two years before the diagnosis of cancer (15.9 percent). The diagnosis of diabetes preceded the diagnosis of pancreatic cancer by 3 to 4 years in 11 case patients (6.7 percent), by 5 to 9 years in 27 (16.5 percent), and by 10 or more years in 34 (20.7 percent). The frequency of recent-onset diabetes (zero to two years' duration) was significantly higher in the case patients than in the control patients (56.0 percent vs. 13.3 percent, P<0.001), whereas the frequency of long-standing diabetes (10 or more years' duration) was significantly higher in the control patients (53.3 percent vs. 20.7 percent, P<0.001).
|
|
|
Discussion
We found a significant association between diabetes and pancreatic cancer. However, we found no association between diabetes and pancreatic cancer among patients with long-standing diabetes. The increased prevalence of diabetes in patients with pancreatic cancer is apparently due largely to diabetes of recent onset, presumably caused by the tumor.
Many studies6,7,8,9,10,11,12,13,14,15,16,17,18 have concluded that diabetes is a risk factor for pancreatic cancer. However, the duration of the diabetes before the diagnosis of pancreatic cancer was not ascertained in these investigations. Other methodologic differences between our study and earlier work include the small numbers of patients, the use of mailed questionnaires and death certificates, and possibly unrepresentative control groups in previous studies.
In our patients with pancreatic cancer who had diabetes that was diagnosed at the same time as or shortly before the tumor, it is likely that the diabetes was a consequence of the tumor. Pancreatic cancer can cause diabetes by destroying islet cells2,3,32 or by causing peripheral resistance to insulin33. Such insulin resistance frequently occurs early in the course of the disease,19,20,34 which may explain why diabetes can appear before the symptoms of the pancreatic tumor.
As compared with the case patients who did not have diabetes, those with diabetes that was presumably secondary to pancreatic cancer were significantly more likely to have a family history of diabetes and an unresectable tumor, suggesting that a hereditary predisposition to diabetes and a larger tumor size contribute to the development of diabetes. The absence of a relation between the location of the cancer and the frequency of diabetes would be compatible with a dependence of tumor-related diabetes on humoral factors. Green et al.3 found no relation between the location of the pancreatic tumor and the frequency of recent-onset diabetes, but they also reported no relation between a family history of diabetes or the size of the tumor and the incidence of diabetes.
In our study, all the patients with pancreatic cancer in whom the diagnosis of diabetes preceded the diagnosis of cancer were considered to have non-insulin-dependent diabetes. All the control patients with diabetes also had the non-insulin-dependent form of the disease except one, who had insulin-dependent diabetes. Most of the case patients (90 percent) and control patients (92 percent) were treated with oral antidiabetic agents or dietary modifications.
The prevalence of diabetes in our control subjects (8.3 percent) was similar to that reported in the Italian population of a corresponding age (7 percent)35. We did not use the glucose-tolerance test to diagnose diabetes; for this reason, we may have underestimated the percentages of patients with diabetes. However, in patients with cancer, the glucose-tolerance test frequently yields false positive results36.
Supported by grants (880716 and 892616) from the Centro Nazionale delle Ricerche.
We are indebted to Miss Janice Capan for assistance in translating the manuscript.
Source Information
From the Institute of Medicine and Gastroenterology, University of Bologna, St. Orsola Hospital, Bologna, Italy. The members of the Italian Pancreatic Cancer Study Group include Patrizia Priori, M.D., Orazio Campione, M.D., Riccardo Casadei, M.D., Maria Brambati, M.D., Carlo Lesi, M.D., and Antonio Frena, M.D. (Bologna); Alberto Malesci, M.D., and Alessandro Zerbi, M.D. (Milan); Angelo Andriulli, M.D., and Patrizio Acquadro, M.D. (Turin); Alessandro D'Ambrosi, M.D., and Vittorio Alvisi, M.D. (Ferrara); Giuseppe Montalto, M.D., and Antonio Carroccio, M.D. (Palermo); Valerio De Conca, M.D., Alberto Mornese, M.D., and Carlo Mansi, M.D. (Genoa); Carlo Battistini, M.D. (Parma); Cosimo Sperti, M.D., and Claudio Pasquali, M.D. (Padua); Luigi Gaeta, M.D., and Mario Mazzeo, M.D. (Naples); and Martina Felder, M.D. (Bolzano).
Address reprint requests to Professor Gullo at the Istituto di Clinica Medica e Gastroenterologia, Ospedale S. Orsola, Via Massarenti, 9, 40138 Bologna, Italy.
References
| |||||||||||||||||||||||||||||||||||||||
Related Letters:
Pancreatic Cancer and Diabetes
Balkau B., Barrett-Connor E., Eschwege E., Jones S.C., Alberti K.G.M.M., O'Shea R. S., Strom B. L., Berlin J. A., Gullo L., Pezzilli R., Morselli-Labate A. M.
Extract |
Full Text
N Engl J Med 1994;
331:1526-1528, Dec 1, 1994.
Correspondence
This article has been cited by other articles:
HOME | SUBSCRIBE | SEARCH | CURRENT ISSUE | PAST ISSUES | COLLECTIONS | PRIVACY | HELP | beta.nejm.org Comments and questions? Please contact us. The New England Journal of Medicine is owned, published, and copyrighted © 2008 Massachusetts Medical Society. All rights reserved. |
Previous
