Glucose Intolerance After Pancreatectomy Was Associated With Preoperative Hemoglobin A1c, Insulin Resistance, and Histological Pancreatic Fatty Infiltration.

To the Editor:

The incidence of new-onset diabetes mellitus has been reported to be 18% to 39% after pancreaticoduodenectomy and 5% to 42% after distal pancreatectomy.[1] Although factors such as high body mass index (BMI),[2-4] high resected pancreatic volume,[2,3,5] operative procedure (distal pancreatectomy),[3,6] presence of chronic pancreatitis,[1] glucose tolerance,[5,6] and age[3] are reported to contribute to glucose intolerance after pancreatectomy in preoperative nondiabetic patients, some of these factors are controversial, partly due to inaccurate or poor assessment of preoperative glucose tolerance. Moreover, the relationship between postoperative glucose intolerance and underlying conditions of glucose tolerance, such as insulin secretory capacity and insulin resistance or sensitivity, as well as histological characteristics of normal pancreatic tissue, have never been studied. We aimed to identify predictive factors of glucose intolerance after pancreatectomy, including histological factors, in preoperative nondiabetic patients whose glucose tolerance was examined in detail.

Fifty-six nondiabetic patients who underwent pancreatectomy in the Department of Gastroenterological Surgery, Osaka University Hospital, between April 2007 and September 2013, provided written comprehensive informed consent, and were followed up for at least 1 year were enrolled in our study. Preoperative diabetes was defined by (a) fasting plasma glucose level of 126 mg/dL or greater, (b) plasma glucose level at 120 minutes in a 75-g oral glucose tolerance test of 200 mg/dL or greater, (c) casual plasma glucose level of 200 mg/dL or greater, (d) hemoglobin A1c (HbA1c) of 6.5% or greater, or (e) presence of history of diabetes or use of antidiabetic drugs. Patients who either had relapses of the primary diseases or other active diseases; were diagnosed as having neuroendocrine tumors with positive immunostaining for either insulin or glucagon; were treated with drugs affecting glucose tolerance, with underlying chronic pancreatitis, anemia, liver cirrhosis, or renal dysfunction; and were without preoperative data on HbA1c were excluded from the study. The maximum value of HbA1c (= postoperative MAX HbA1c) within 1 year after surgery was evaluated to be an index of glucose intolerance. Among the 56 patients, histological analyses of β-cell area and pancreatic fatty infiltration were performed in 36 patients whose normal pancreatic sections were isolated from near the resected margins and judged to be suitable for analyses, as determined by rejecting cancer elements and fibrosis changes. The ratio of β-cell area to the entire pancreatic section (excluding fat tissue) was defined as the relative β-cell area (%), and the ratio of the sum of the interlobular and intralobular fat-cell areas to the entire pancreatic section (including fat tissue) was defined as “fat-cell area” (%). Representative views of hematoxylin and eosin (HE)–stained pancreatic sections with the minimum (0.033%) and maximum (53%) values of fat-cell area are shown in Figures 1A and B, respectively, whereas the median value of fat-cell area in 36 subjects' specimens was 0.94%. Among the subjects of histological analyses, fat-cell area had the strongest correlation with postoperative MAX HbA1c (r = 0.90, P < 0.0001) of all of the preoperative clinical parameters that were correlated with postoperative MAX HbA1c, such as C-peptide index (r = 0.51, P = 0.018), HbA1c (r = 0.52, P = 0.0012), fasting C-peptide (r = 0.59, P = 0.0046), BMI (r = 0.66, P < 0.0001), and homeostasis model assessment of insulin resistance (HOMA-R; r = 0.68, P = 0.0001). Multiple regression analyses revealed that fat-cell area, HOMA-R, and HbA1c were independently associated with postoperative MAX HbA1c and that fat-cell area had the strongest contribution (Table 1). Receiver operating characteristic curve analysis revealed that the cutoff value of fat-cell area for identifying subjects whose postoperative HbA1c deteriorated to diabetic state (HbA1c ≥6.5%, one of the criteria for diabetes in Japan) was 3.7%.

FIGURE 1

Views of HE-stained pancreatic sections with the minimum fat-cell area (0.033%; A) and maximum fat-cell area (53%; B). Fat-cell area (black arrowhead) was found scattered or spread in normal pancreatic structure of islets (black arrow) and exocrine tissue. Fat-cell area was distinguished from lymph ducts or vessels (white arrowhead) by the absence of surrounding connective tissue and endothelial cells. A, A representative low-magnification view of a HE-stained pancreatic section with the minimum fat-cell area (0.033%) from a 55-year-old woman, whose BMI and HOMA-R were 22.3 kg/m2 and 1.1, respectively. Her postoperative MAX HbA1c was 6.2%. B, A representative low-magnification view of a HE-stained pancreatic section with the maximum fat-cell area (53%) from a 29-year-old woman, whose BMI and HOMA-R were 34.3 kg/m2 and 3.8, respectively. Her postoperative MAX HbA1c was 13.3%.

TABLE 1

Multiple Regression Analyses for Postoperative MAX HbA1c (%)

As for the reasons why fat-cell area was associated with postoperative glucose intolerance, the following explanations are given. First, fat-cell area in the pancreas may reflect insulin resistance. Indeed, fat-cell area had positive correlations with HOMA-R (r = 0.56, P = 0.0031) and BMI (r = 0.64, P < 0.0001) in this study, in line with previous studies.[7,8] Second, fat-cell area may be related to another factor of glucose tolerance impairment other than insulin resistance because fat-cell area contributed to postoperative MAX HbA1c more strongly than HOMA-R. A previous study in humans revealed that the degree of pancreatic fatty infiltration was negatively correlated with indices of insulin secretory capacity,[9] although this was not shown in this study. Further studies are needed to confirm the association of fatty infiltration and deterioration of β-cell function.

Unexpectedly, postoperative MAX HbA1c was not associated with relative β-cell area, an index reflecting insulin secretory capacity.[10] This result may suggest that insulin secretory capacity contributed less to postoperative glucose tolerance within 1 year than insulin resistance in preoperative nondiabetic patients. Longer follow-up studies might clarify the contribution of relative β-cell area to postoperative glucose tolerance.

In conclusion, glucose intolerance within 1 year of pancreatectomy could be predicted by preoperative HbA1c, HOMA-R, and especially pancreatic fat-cell area in preoperative nondiabetic patients.