The Density, Contour, and Thickness of the Pancreas in Diabetics: CT Findings in 57 Patients

527 The Density, Contour, and Thickness of the Pancreas in Diabetics: CT Findings in 57 Patients Jean-Paul Gilbeau1 Insulin has a trophic effect on pancreatic acinar tissue, so the pancreas might be Vincent Poncelet1 expected to atrophy in persons who have diabetes. Accordingly, we analyzed the Ernest Libon1 density, contour (smooth or lobulated), and thickness of the pancreas on CT scans of Genevieve Derue2 diabetic patients and compared the results with those in control subjects. The prevalence Francis R. HeIIer of pancreatic diabetic and lobulation (incisurae deeper than control subjects were determined. 2 mm) and its The thickness correlation with age in of the pancreas was measured at three levels (head, body, tail). Three groups of diabetic patients were examined: 20 insulin-dependent patients, 25 patients not treated with nor dependent on insulin, and 12 patients treated with but not dependent on insulin. A control group included 57 nondiabetic patients. The ages of the control subjects were similar to those of the diabetic patients. The statistical significance of the differences between groups of diabetic patients and control subjects was estimated by using Student s t test for the values of density and thickness and the x2-test for the prevalence of pancreatic lobulation. The density of the pancreas in diabetic patients and control subjects was not statistically different. Diabetic patients had increased lobulation of the pancreas. All parts of the pancreas tended to be smaller in diabetic patients, but the degree of reduction varied. It was modest in the patients not treated with insulin, pronounced in insulin-dependent patients, and intermediate in non-insulin-dependent, insulin-treated patients. Moreover, the size of the body was significantly reduced in all three groups, whereas the size of the pancreatic head was preserved in patients not treated with insulin. In conclusion, CT of the pancreas shows that although density in diabetic patients is normal, lobulation is increased. Reduction in size involves the body of the pancreas more than other parts of the gland and is more pronounced in insulin-treated diabetic patients. CT of the pancreas might be useful to predict which diabetic patients will require insulin therapy. AJR 159:527-531, September 1992 The pancreas contains endocrine cells in discrete clusters (islets) within the exocrine parenchyma (acinar tissue). A functional relationship between endocrine and exocrine tissue is suggested by the high prevalence of exocrine pancreatic dysfunction in insulin-dependent diabetic patients [1 ]. As the islets are only 1-2% of the pancreatic mass, atrophy of the exocrine component of the gland might be expected in patients with diabetes, with consequent reduction in the size of the gland. Despite the high prevalence and the severity of diabetes, relatively few Received November 7, 1991 ; accepted after re- morphologic studies of the pancreas in diabetic patients have been done. Until vision March 9, 1992. recently, the only material available came from autopsies of diabetic patients [2]. 1 Department of Radiology, H#{244}pital de Jolimont, To our knowledge, except for a sonographic study [3] and one CT study in insulin- Rue Ferrer, 7100 Haine Saint Paul, Belgium. Ad-.... dress reprint requests to J-P. Gilbeau. dependent diabetic patients [4], no studies have been done of the pancreas in 2 Department of Intemal Medicine, H#{244}pitalde living subjects who have diabetes. Jolimont, 7100 Haine Saint Paul, Belgium. On clinical grounds, two types of diabetes mellitus are distinguished: insulin- 0361-803X/92/1593-0527 dependent diabetes mellitus, which affects mostly young persons and is charac- American Roentgen Aay Society terized by low or no secretion of insulin, and non-insulin-dependent diabetes

528 GILBEAU ET AL. AJA:159, September 1992 mellitus, which occurs in older obese persons and is characterized by insulin resistance of the target organs and normal or increased secretion of insulin. Data on the morphology of the pancreas in these two types of diabetes are scanty. The aim of the present work was to study the density, contour, and thickness of the pancreas of diabetic patients with CT and to compare the results in three different groups of diabetic patients: non-insulin-dependent diabetic patients who need diet or diet and hypoglycemic drugs for their treatment, non-insulin-dependent diabetic patients who require insulin for good glycemic control, and insulin-dependent diabetic patients. Subjects and Methods Fifty-seven diabetic patients were included in the study. They belonged to one of three groups. The 20 insulin-dependent patients included 1 0 men and 10 women 23-72 years old (mean ± standard error of the mean [SEM], 48.0 ± 3.9 years). Duration of diabetes was 9.4 ± 2.9 years. The 25 non-insulin-dependent diabetic patients whose treatment consisted of adherence to an appropriate diet or adherence to a diet and use of oral hypoglycemic agents included nine men and 1 6 women 44-86 years old (mean ± SEM, 61.7 ± 2.2 years). Duration of diabetes was 6.6 ± 1.8 years. The 12 non-insulindependent diabetic patients who required insulin treatment included three men and nine women 51-83 years old (mean ± SEM, 63.5 ± 9.9 years). Duration of diabetes was 1 1.2 ± 3.4 years. Patients with diabetes caused by or associated with chronic pancreatitis, malignant neoplasm, or other endocrinopathies were not included. Fifty-seven patients (27 men and 30 women) were used as control subjects. They were 27-83 years old (mean ± SEM, 60.5 ± 20 years). They were not obese (body mass index less than 25 kg/me) and were not taking any drugs known to be diabetogenic (e.g., glucocorticoids, thiazides). They were scheduled to undergo abdominal CT as part of a medical workup for evaluation of unexplained abdominal pain or hepatomegaly. Results of an oral glucose tolerance test were normal in these patients, and examination of their clinical records confirmed the absence of exocrine or endocrine pancreatic dysfunction. Abdominal CT was performed with a CR-Siemens unit (Erlangen, Germany). Serial transverse scans with 8-mm-thick contiguous slices were obtained. The patient was supine and was asked to take a deep breath and hold it. lodinated contrast medium (Micropaque scanner, Nicholas, Brussels) was given orally for better delineation of the outline of the pancreatic head. Determinations of the size of the pancreas may be more accurate when IV contrast material is used because the vessels are defined better. However, we did not use IV contrast material because ofthe risk of renalfailure in diabetic patients who receive it. The density of the pancreas was measured at the level of the tail, thus avoiding the beam-hardening effect, caused by the presence of oral contrast medium in the proximal duodenum, that occurs when measurements are made in the body or in the head of the pancreas. The contour of the pancreas was defined as smooth or lobulated on the basis of the presence of incisurae deeper than 2 mm at the surface of the gland. Analysis of the pancreatic contour was done by following the technique described by Heuck et al. [5] for normal subjects. The thickness of the pancreas was measured at three levels (head, body, and tail), also according to the method of Heuck et al. Thickness was measured on the right side of the mesenteric vessels for the head, in the extension of a line drawn from the left border ofthe vertebral body for the body, and from the internal border of the left kidney for the tail. The interpreters for the CT scans were not aware of the clinical diagnosis. In order to look for an eventual effect of age or duration of diabetes on pancreatic size, correlations were calculated between these parameters and the thickness of the different parts of the pancreas in control subjects and diabetic patients. Levels of C peptide were analyzed to evaluate residual insulin secretion in diabetic patients. Measurement was made by radioimmunoassay of serum samples obtained after the patient had fasted overnight. In normal subjects, basallevels of C peptide are 1.0-3.5 ng/ml. The statistical significance of a difference between the means of the different groups was estimated by using Student s t test or the x2-test, and the correlation between parameters was analyzed by calculating Spearman s correlation Results coefficient. The density of the pancreas did not differ significantly between control subjects (40 ± 2.0 H [mean ± SEM]) and the different groups of diabetic patients: insulin-dependent, 37 ± 4.6 H; non-insulin-dependent, non-insulin-treated, 36 ± 3.2 H; and non-insulin-dependent, insulin-treated, 41 ± 3.8 H. Heuck at al. [5] showed that lobulation of the pancreas increases progressively in healthy persons 20-80 years old. In our study (Table 1), the percentage of pancreatic lobulation was higher in control subjects more than 40 years old than in those less than 40 years old. For the age ranges 41 to 60 years and 61 to 90 years, diabetic patients had a significantly higher lobulation of the pancreas than did control subjects. When compared with control subjects, non-insulin-treated diabetic patients had a significant reduction in the size of the body of the pancreas (p =.035). The sizes of the head and tail were not significantly smaller. All parts of the pancreas were smaller in insulin-treated diabetic patients than in control subjects. The difference was significant in insulin-dependent patients (head, p <.03; body, p <.001 ; and tail, p <.005) and in non-insulin-dependent diabetic patients (head, p <.0001 ; body, p <.0005; and tail, p <.02). When compared with non-insulin-dependent diabetic patients who received no insulin, insulin-treated patients had a reduction in the size of the head of the pancreas (p <.02), whereas the sizes of body and tail were not significantly different (Figs. 1 and 2). TABLE 1: Prevalence of Pancreatic Lobulation by Age Range in Control Subjects and Diabetic Patients Group No. with Lo bulation/tota Group (%) I No. in Age 20-40 Yr 41-60 Yr 61-90 Yr Control subjects 0/8 (0) 7/21 (33) 9/28 (32) Insulin-dependent diabetic patients 1 /1 0 (1 0) 4/5 (80)8 4/5 (80) Non-insulin-treated, noninsulin-dependent diabetic patients 0/0 8/1 1 (73)b 1 1 /1 4 (79)C Insulin-treated, non-insuin-dependent diabetic patients 0/0 4/5 (80) 5/7 (71)8 a Statistically different from control subjects in the same age range at a level ofp=.05. b Statistically different from control subjects in the same age range at a level ofp=.0013. C Statistically different from control subjects in the same age range at a level ofp=.0045.

AJR:159, September1992 CT OF PANCREAS IN DIABETES 529 (-) 18- : 12 10 tad SIOY TAIL Fig. 1.-Graph gives mean (plus or minus standard error of mean) thickness of pancreas at level of head, body, and tail in different groups of patients: C control subjects, 1 = insulin-dependent diabetic patients, 2 = non-insulin-treated, non-insulin-dependent diabetic patients, and 3 = Insulin-treated, non-lnsulln-dependentdiabetic patients.asterlsk indicates that values are significantly different from those in control subjects. Fig. 2.-CT scans of pancreas in control subjects and three different groups of diabetic patients. A, Control subject. B, Non-insulin-treated, non-insulindependent diabetic patient Lobulation and atrophy of body of pancreas (arrow) are clearly seen. C, Insulin-treated, non-insulin-dcpendent diabetic patient. Marked afrophy of body of pancreas (arrow) is seen. D, Insulin-dependent diabetic patient. Marked atrophy of body and tail of pancreas (arrows) is seen. The size of the tail of the pancreas correlated significantly with the duration of diabetes in the whole group of diabetic patients (r =.27, p =.03). The size of the body of the pancreas correlated significantly with age in non-insulin-dependent, non-insulin-treated patients (r = -.43, p =.01) and in non-insulin-dependent, insulin-treated patients (r = -.51, p =.04). Age was not correlated with pancreatic size in control subjects. Basal serum levels of C peptide were significantly lower in insulin-dependent diabetic patients (0.83 ± 0.35 ng/dl [mean ± SEM]) than in non-insulin-dependent, non-insulin treated diabetic patients (4.1 2 ± 0.61 ng/dl, p <.0001) and in non-insulin-dependent, insulin-treated diabetic patients (2.86 ± 0.46 ng/di, p <.003). Non-insulin-dependent, insulin-treated patients had lower concentrations of C peptide than did non-insulin-dependent, non-insulin-treated patients, but the difference was not significant (p =.27). Discussion From autopsy data, it is known that the pancreas in patients with insulin-dependent diabetes is smaller than the pancreas

530 GILBEAU ET AL. AJR:159, September1992 in persons without diabetes [2]. The pancreas appears to be slightly smaller in non-insulin-dependent diabetic patients than in persons without diabetes [6] and larger in obese patients (Ibrahim MM, Rahier J, unpublished data), whether or not the patients are diabetic [7]. The results of these studies, particularly those done in patients dying of diseases other than diabetes, may have been influenced by the effects of the underlying disease on pancreatic function and morphology. To our knowledge, only two studies have analyzed the pancreas in living diabetic patients. Fonseca et al. [3] used sonography to study the head and body of the pancreas in diabetic patients and showed that both parts of the pancreas were smaller in patients with insulin-dependent, non-insulindependent, and insulin-requiring diabetes mellitus than in control subjects. The pancreas was the smallest in the first group of diabetic patients, largest in the second group, and intermediate in the third group. Compared with sonography, CT has the advantage of accurately showing all parts of the pancreas, including the tail. In our study, all parts of the pancreas tended to be smaller in diabetic patients than in the control subjects. The reduction was most prominent in insulindependent patients and in insulin-treated, non-insulin-dependant patients and least pronounced in non-insulin-treated, non-insulin-dependent patients. The body of the pancreas was significantly smaller in the three groups of diabetic patients. The head and tail of the pancreas were significantly smaller in insulin-treated diabetic patients only, whether they had insulin-dependent or non-insulin-dependent diabetes. In a CT study of insulin-dependent diabetic patients by Brunova et al. [4], only the sizes of the pancreatic body and tail were below the lower limit of normal. Although Fonseca et al. [3] found no correlation between the size of the pancreas and duration of diabetes, our results suggest that the sizes of the tail and body of the pancreas tend to decrease with duration of diabetes or age. This process seems to be related to the disease itself more than to the patient s age because no correlation was found between the size of the pancreas and age in control subjects. Similarly, Brunova at al. [4] found that insulin-dependent diabetic patients who had had diabetes longer than 1 0 years had a smaller pancreatic body than did patients who had been diabetic fewer than 10 years. It is unclear why the reduction in size affects the body more than the other parts of the pancreas in diabetic patients. However, Rahier et al. [8] have shown that the decrease in the weight of the pancreas in patients with long-term insulin-dependent and noninsulin-dependent diabetes is not caused by diffuse atrophy. Indeed, the pancreatic polypeptide-nch lower part of the head region of the pancreas (derived from the embryologic ventral primordium) is entirely preserved, and the atrophy is limited to the pancreatic polypeptide-poor lobe derived from the dorsal primordium [8]. This could explain why the size of head of the pancreas in diabetic patients is reduced less than the size of the body is. Rahier et al. [8] have speculated that atrophy of the exocrine pancreas due to the disappearance of the trophic local effect of insulin on acinar tissue does not occur in the pancreatic polypeptide-rich lobe of the pancreas, as pancreatic polypeptide also could have a trophic effect on acinar cells. This is in accordance with our findings on the serum levels of C peptide which reflect insulin secretion in the different groups of diabetic patients. Insulin-treated patients who had insulin-dependent or non-insulin-dependent diabetes had both the most prominent reduction in pancreatic size and the lowest concentrations of C peptide, whereas noninsulin-treated patients with non-insulin-dependent diabetes had the least pronounced reduction in pancreatic size and normal concentrations of C peptide. These data also agree with those of Brunova at al. [4], who found that the only patients who had a smaller pancreas were those with insulindependent diabetes who had the lowest serum levels of C peptide. Some non-insulin-dependent diabetic patients eventually require insulin for equilibration because their disease is not adequately controlled by oral hypoglycemic agents alone. This has been attributed to a loss of insulin secretion. On the average, beta-cell mass has not been shown to be reduced in non-insulin-dependent diabetic patients. However, a high variability exists [6]. Our data are consistent with the hypothesis of progressive exhaustion of the endocrine pancreas in some non-insulin-dependent diabetic patients. Indeed, compared with those who did not require insulin for glycemic control, those who did had both a reduction in the size of the pancreas and low serum concentrations of C peptide. Interestingly, the best discriminant parameter between these two groups of non-insulin-dependent diabetic patients was the size of the head of the pancreas. Pathologic examination of the pancreas has shown that insular fibrosis together with atrophic islets is usual in patients with long-term insulin-dependent diabetes and that fat infiltration or increased fibrotic tissue may be observed in patients with non-insulin-dependent diabetes, but the latter findings are far from constant [2]. On CT scans, fat has a low coefficient of attenuation and fibrosis has a high one. Thus, it is not surprising that the density of the pancreas is not different in control subjects and in diabetic patients on one hand and in the different groups of diabetic patients on the other hand. Lobulation of the pancreas increased with age in our control subjects, a finding in accord with the work of Heuck at al. [5]. The cause of this age-related process is not known. When the age of the subjects is considered, diabetic patients have increased lobulation of the pancreas. This suggests that the process is accelerated by diabetes mellitus. CT examination of the pancreas in diabetic patients leads to several conclusions. First, the pancreas is smaller in diabetic patients than in control subjects, in accordance with previous reports of autopsy findings. The reduction in size is more pronounced in insulin-treated than in non-insulin-treated diabetic patients. This finding together with the results of the C peptide determinations suggests a loss of insulin secretion in insulin-treated patients. Second, the body of the pancreas appears to be the most affected part of this organ. Third, a requirement for insulin in non-insulin-dependent diabetic patients seems to be associated with reduction in size of the head of the pancreas. This finding suggests that in noninsulin-dependent diabetic patients, pancreatic atrophy is an evolving process affecting first the body and then the entire pancreas, with the head being the last involved. At this stage,

AJR:159, September 1992 CT OF PANCREAS IN DIABETES 531 insulin secretion appears to be extremely low, so exogenous insulin is required for glycemic control. However, only a Iongitudinal study of pancreatic size in these patients could confirm this hypothesis. CT examination of the pancreas could help detect those patients who are at risk for diabetes mellitus (atrophy of the body of the pancreas) and, among non-insulindependent diabetic patients, those who will soon require insulin therapy (atrophy of the head of the pancreas). REFERENCES 1. Frier BM, Faber OK, Binder C, Ellioth HL. The effect of residual secretion on exocrine pancreatic function in juvenile-onset diabetes mellitus. Diabetologia 1978:14:301-304 2. Rahier J. The diabetic pancreas: pathologist s view. In: Lefebvre PJ, Pipeleers DG, eds. The pathology of the endocrine pancreas in diabetes. New York: Springer-Verlag, 1988:17-40 3. Fonseca V, Berger LA, Beckett AG, Dandona P. Size of pancreas in diabetes rnellitus: a study based on ultrasound. BMJ 1985:291: 1240-1 241 4. Brunova E, Bruna J, Bartos V, et al. Pancreatic and renal size and density determined by computed tomography in patients with type I (insulindependent) diabetes. Radiol Diagn (Boil) 1986:27:737-744 5. Heuck A, Maubach PA, Reiser M, et al. Age-related morphology of the normal pancreas on computed tomography. Gastrointest Radiol 1987:12:18-22 6. Rahier J, Goebbels RM, Henquin JC. Cellular composition of the human diabetic pancreas. Diabetologia 1983;24:366-371 7. Kloppel G, Lohr M, Habich K, Oberholzer M, Heitz PU. Islet pathology and the pathogenesis of type I and type II diabetes mellitus revisited. Surv Synth Pathol Res 1985:4: 1 10-1 25 8. Aahier J, Wallon J, Loozen S, Lefevre A, Gepts W, Haot J. The pancreatic polypeptide cells in the human pancreas: the effects of age and diabetes. J Clin Endocrinol Motab 1983;56:441-444