Changes in basal insulin secretion capacity after pancreatectomy: a retrospective study

This study evaluated the changes in basal insulin secretion capacity (BISC) after pancreatectomy across two surgical procedures. We also investigated the association between decreased BISC and the introduction of postoperative insulin therapy. The data on 110 patients who underwent pancreatectomy during October 2018-February 2022 at our hospital were analyzed retrospectively. We focused on the C-peptide index (CPI) as an index for BISC. A decreased postoperative BISC was defined as a postoperative CPI (post-CPI) <1.0, which was in turn defined as the event occurrence in our study. The receiver operating characteristic curve for the event occurrence was plotted for factors related to preoperative glucose tolerance. Of the preoperative non-diabetic 73 patients, 44 and 29 who underwent pancreaticoduodenectomy (PD) and distal pancreatectomy (DP), respectively, were examined. A post-CPI of <1.0 was observed in 29 patients (39%). Although the proportion of remnant pancreatic volume was significantly smaller in patients with PD than in patients with DP (38% vs. 68%, p <0.0001), no significant difference was observed in the event rates (39% vs. 41%). In patients with PD, preoperative CPI (area under the curve: 0.75) was used for predicting post-CPI <1.0. Multivariate analysis revealed that preoperative CPI <1.65 (odds ratio: 7.69, 95% confidence interval: 1.87?31.5) was an independent predictor of decreased postoperative BISC. However, in patients with DP, no significant predictors were identified. Induction of insulin therapy was significantly lower in preoperative non-diabetic patients (n=73) after undergoing a pancreatectomy (1.4% vs. 37.5%, p<0.0001) than in preoperative medically treated patients (n=16); there was no significant difference in event (post-CPI <1.0) rates (39.7% vs. 56.2%, p=0.23). Although pancreatectomy reduces BISC after surgery, the coexistence of insulin resistance, which is a pathophysiology of type 2 diabetes mellitus, may play a role in whether postoperative glucose tolerance is reduced to the point where insulin therapy is necessary.

CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted June 16, 2022. ;https://doi.org/10.1101https://doi.org/10. /2022 4 58 Introduction 59 Pancreatectomy is essential in several neoplastic diseases, such as cholangiocarcinoma, 60 cancer of the duodenal papilla, intraductal papillary mucinous tumors, and neuroendocrine 61 tumors, besides pancreatic cancer. Although some postoperative complications, such as 62 pancreatic fistulas, remain to be resolved, surgical techniques and perioperative management 63 have improved the short-term outcomes, such as perioperative mortality [1,2]. Medium-to-64 long-term complications after pancreatectomy include disease-free survival and deterioration 65 of exocrine and endocrine functions.

66
Pancreatectomy has an impact on pancreatic endocrine functions; it reduces the 67 capacity for insulin secretion and may lead to the development of new-onset diabetes mellitus 68 (NODM) [3]. Insulin is secreted by β-cells of the islets of Langerhans in the pancreas. Many 69 studies have indicated that the islets of Langerhans are predominantly distributed in the 70 pancreatic tail [4][5][6]. While some studies suggest that pancreatic tail resection is likely to cause 71 NODM [7], others have reported that the incidence of NODM does not differ between 72 pancreaticoduodenectomy (PD) and distal pancreatectomy (DP) [8,9]. As noted above, some 73 studies examining factors contributing to the occurrence of NODM after pancreatectomy have 74 reported contradictory findings [7][8][9]. However, only a few studies have evaluated the changes 75 in endocrine function, particularly insulin secretion capacity, before and after pancreatectomy 76 [10,11].
. CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

77
There are two known types of insulin secretion: basal insulin secretion, which remains 78 constant regardless of food intake, and additional insulin secretion, which is in response to 79 rising blood glucose levels post food intake. Decreased basal insulin secretion reduces the 80 blood glucose uptake by the muscles and other tissues and weakens the ability of insulin to 81 inhibit gluconeogenesis in the liver during fasting, leading to increased fasting blood glucose 82 levels, which is a diagnostic criterion for diabetes mellitus (DM is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted June 16, 2022. ;https://doi.org/10.1101https://doi.org/10. /2022 6 96 resection [n=2]). Moreover, patients with missing data (inadequate preoperative glucose 97 tolerance assessments) were also excluded (n=7). Patients on preoperative insulin therapy were 98 also excluded (n=10). Finally, 89 patients were enrolled (Fig 1). The following main analyses 99 were performed on preoperative non-diabetic patients (n=73), while preoperative type 2 100 diabetic patients on medication therapy (n=16) were primarily evaluated for the presence of 101 postoperative induction of insulin therapy. 104 CPI, C-peptide index 105 106 Index of basal insulin secretion 107 Among the various indices of BISC, we used the C-peptide index (CPI), a useful 108 indicator for requirement of insulin therapy [12]. The CPI can be calculated using a simple 109 formula: fasting C-peptide immunoreactivity (ng/mL)/fasting plasma glucose (mg/dL)×100.
110 For normal glucose tolerance, the CPI is approximately 1.60 ± 0.57 [12]. A CPI of <0.8 -1.0 111 is indicated as the criterion for insulin therapy requirement for type II DM [12,13]. We defined 112 a decrease in BISC as a postoperative CPI (post-CPI) of <1.0, and the changes in CPI pre-and 113 post-surgery were compared according to the operative procedure.

114
. CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted June 16, 2022. ; https://doi.org/10.1101/2022.06.14.22276406 doi: medRxiv preprint 7 115 Perioperative factors 116 The surgical procedures of PD and DP were analyzed. Data on operative time,

132
In Japan, prior to liver surgery, hepatic function is estimated based on the K 133 indocyanine green (K-ICG) value, which is obtained by calculating the extinction rate of . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted June 16, 2022. ; https://doi.org/10.1101/2022.06.14.22276406 doi: medRxiv preprint 8 134 intravenously injected ICG. The future remnant hepatic function is estimated by multiplying 135 the K-ICG value by the expected proportion of remnant liver volume [16,17]. However, we 136 introduced a new index, CPV, calculated using the following formula: preoperative CPI (pre-137 CPI)×PRPV measured by CT volumetry.

138
Similarly, using the homeostasis model assessment of β-cell function (HOMA-β), that 139 is one index of basal insulin secretion capacity, we generated a new index of HPV, obtained 140 by multiplying preoperative HOMA-β by PRPV, as a predictor of event occurrence (post-CPI 141 <1.0).

142
Blood tests were performed in the fasting state, and the values were measured 143 preoperatively and within 1 week to 1 month postoperatively. CT volumetry was performed 144 using ZIOstation2 (Ziosoft, Inc., Tokyo, Japan); the pancreatic parenchyma contour was 145 plotted on a 1-mm-width slice from the preoperative and first postoperative CT. The plot area 146 of each slice was calculated, and the pancreatic volume was obtained by integrating the areas.

147
148 Perioperative blood glucose control 149 In cases of patients with poor preoperative blood glucose control, surgery was 150 performed after inpatient fortified insulin therapy (these patients were excluded in this study).
151 After surgery, a sliding scale was used, and fasting blood glucose level was maintained at <200 152 mg/dL. When fasting blood glucose level was found to be over 200 mg/dL after food intake . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted June 16, 2022. ; https://doi.org/10.1101/2022.06.14.22276406 doi: medRxiv preprint 9 153 was stabilized, oral hypoglycemic agents or self-injection of insulin was introduced under the 154 guidance of diabetes specialists.

155
156 Statistical analysis 157 To evaluate the predictive power of each factor for event occurrence (post-CPI <1.0), 158 receiver operator characteristic (ROC) curves were generated for each factor, and the area 159 under the ROC curve (AUC) was calculated and compared for each factor. Thereafter, the 160 ROC-AUC variable suggestions found to be the most useful were added to the logistic 161 regression model for event occurrence that was adjusted for age, sex, and surgical procedure 162 to verify whether they remained significant after these adjustments.

163
Further, we conducted stratified four-fold cross-validation to evaluate the 164 generalizability of the factors selected in the above tests. Specifically, the data were divided 165 into two groups (training and testing groups) at a ratio of 3:1; the data were further distributed 166 such that the proportion of the event occurrence (post-CPI <1.0) was equal between the two 167 groups (training and testing groups). The same division was randomly performed 10 times, and 168 logistic regression was performed to predict the event occurrence (post-CPI <1.0) in the split 169 data sets: using the previously obtained event predictors for each surgery category as 170 explanatory variables-to predict the event occurrence (post-CPI <1.0) for each surgery.
171 Logistic regression was performed using the following formula: . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted June 16, 2022. ; https://doi.org/10.1101/2022.06.14.22276406 doi: medRxiv preprint The authors are accountable for all aspects of the work in ensuring that questions related 189 to the accuracy or integrity of any part of the work are appropriately investigated and resolved.
190 The study was conducted as per the Declaration of Helsinki (as revised in 2013) guidelines.
. CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted June 16, 2022. is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
. CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted June 16, 2022.  The changes in pre-and-postoperative CPI are shown in Fig 2. The CPI decreased 225 postoperatively in 85% of the preoperative non-diabetic patients; post-CPI was significantly 226 lower than pre-CPI in patients with these two operative procedures (Fig 2A, (Fig 2A). In preoperative non-238 diabetic patients who underwent DP, the median preoperative and postoperative CPIs were 239 1.63 (range: 1.01-3.79) and 1.03 (range: 0.40-2.72), respectively (Fig 2B). A significant 240 difference was noted among the two corresponding groups (pre-and post-operative CPI), with 241 p=2.7×10 -4 (Fig 2B).

242
. CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted June 16, 2022. ; https://doi.org/10.1101/2022.06.14.22276406 doi: medRxiv preprint 21 243 To assess the ability of each factor to predict the occurrence of post-CPI <1.0, we used 244 the ROC-AUC. Preoperative CPI (pre-CPI) was the best predictor of the occurrence of post-245 CPI <1.0 in PD patients (Table 3,  246 Similarly, pre-BMI was the most useful predictor in DP patients ( is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.  is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted June 16, 2022. is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
287 288 is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted June 16, 2022. ;https://doi.org/10.1101https://doi.org/10. /2022 25 300 induction rate was significantly lower in the preoperative non-diabetic patients ( 338 that can be measured using blood tests and is considered more versatile.

339
The results of this study suggest that in terms of predictors of postoperative decreased . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted June 16, 2022. ; https://doi.org/10.1101/2022.06.14.22276406 doi: medRxiv preprint 29 340 BISC, the factors that are significant in PD are not significant in DP, suggesting that the 341 predictors differ by surgical procedure. This difference might be due to the heterogeneous 342 distribution of the islets of Langerhans in the pancreas. Generally, the islets of Langerhans are 343 largely distributed in the pancreatic tail [4][5][6]. In PD, the pancreatic head, which has a large 344 volume, is removed, while the pancreatic tail, which is thought to contain many islets of 345 Langerhans, is preserved. Therefore, it was suggested that BISC after PD strongly reflects the 346 index of preoperative BISC, which is represented by preoperative CPI. Since the residual 347 pancreatic volume was large in DP cases, we evaluated the index considering the PRPV but 348 did not obtain significant results. In a previous report, PRPV was not a risk factor for NODM 349 after DP, and obese patients were significantly more likely to develop NODM [19].
350 Preoperative BMI was a candidate predictor of event occurrence in DP patients of this study, 351 but it was not significant on multivariate analysis.

352
Regulation of insulin secretion is through stimulation by hyperglycemia or impact of 353 various endocrine hormones. Incretin is a general term for gastrointestinal-derived hormones 354 that act on pancreatic beta-cells to promote insulin secretion [20,21]. Glucagon-like peptide-1 355 (GLP-1) and gastric inhibitory peptide (GIP) are known as incretins, and their mechanisms of 356 enhancing incretin effects and inhibiting metabolism have been elucidated and clinically 357 applied as therapeutic targets for DM [20,21]. Muscogiuri et al. showed that duodenal resection 358 in PD resulted in an increase in GLP-1 concentration, which was expected to increase insulin . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted June 16, 2022. ; https://doi.org/10.1101/2022.06.14.22276406 doi: medRxiv preprint