The impact of diabetes mellitus on short and long term outcomes in patients with gastric cancer following radical surgery: a retrospective cohort study with propensity score matching Authors: Zhiyuan Yu, Chen Liang, Rui Li, Qixuan Xu, Jingwang Gao, Peiyu Li, Sixin Zhou, Xudong Zhao, Manman Xu, Wenquan Liang Published: 2024 ABSTRACT AbstractBackgroundThe objective of this study was to analyze the impact of diabetes mellitus (DM) on short and long term outcomes following radical gastrectomy using propensity score matching (PSM) method, as well as to further investigate the factors influencing patient survival post-radical gastrectomy.MethodsThe patients who underwent radical gastrectomy for gastric cancer (GC) between January 2011 and December 2015 were selected as study population, and PSM procedure was conducted with a matching ratio of 1:3 between the DM and non-DM groups. The short-term recovery and long-term survival outcomes were compared between the DM and non-DM group. Subsequently, univariate and multivariate COX regression analyses were conducted to further explore the factors that influence survival outcomes.ResultsAmong the cohort of 2508 GC patients who underwent radical surgery prior to PSM, a total of 295 (11.8%) individuals were diagnosed with DM. After conducting 1:3 PSM, 293 DM patients and 821 control patients were enrolled. The results of comparative analysis revealed that the DM group exhibited a significantly higher rate of overall complications (P < 0.001), grade III-V complications (P = 0.010), and prolonged hospitalization (P = 0.001) compared to the control group. However, no statistical difference was observed in survival outcomes between the two groups. Besides, age > 65years (P < 0.001), BMI < 18.5 kg/m2 (P = 0.001), tumor size > 5 cm (P < 0.001), higher T (P = 0.012) and N tumor stage (P < 0.001), and the occurrence of severe complications (P = 0.007) were identified as independent risk factors for overall survival (OS).ConclusionsThe presence of combined DM in GC patients can heighten the susceptibility to postoperative complications and protract the duration of postoperative recovery, while exerting no significant impact on OS. This study also explored the predictors for OS, thereby providing valuable guidance for the perioperative management and postoperative monitoring.Supplementary InformationThe online version contains supplementary material available at 10.1186/s12885-024-13232-3.Keywords: Gastric cancer, Diabetes mellitus, Propensity score matching, Postoperative recovery, Survival outcomes Introduction The prevalence of gastric cancer (GC) in western populations has been progressively declining over the years due to modifications in dietary patterns, decreased tobacco consumption, and a reduced incidence of Helicobacter pylori. Despite this, GC still ranks fifth in terms of incidence (5.6%) and fourth in terms of mortality (7.7%) among all malignancies worldwide [1]. According to the latest cancer statistics, GC is the second most frequently diagnosed cancer and the second leading cause of cancer-related deaths in China [2]. Many of the risk factors for GC and diabetes mellitus (DM) overlap, including smoking, consumption of smoked, baked, and cured foods, etc [3, 4]. Consequently, over 11% of GC patients presented with diabetes at the time of diagnosis, with type 2 diabetes being the predominant form, accounting for more than 90% [2, 5]. The performance of surgical procedures on DM patients, including radical gastric surgery, poses significant challenges and risks due to complications involving the cardiovascular system, kidneys, skin, and other organs [6, 7]. Currently, the impact of DM on gastrointestinal complications and survival outcomes following radical gastrectomy remains unclear, with a dearth of relevant literature evidence. A systematic review and meta-analysis of cohort studies indicated that both type 1 DM (T1DM) and type 2 DM (T2DM) were associated with an elevated risk of GC development, and this risk could be influenced by the specific time intervals following the onset of DM [8]. The studies conducted by both He and Chen have demonstrated that DM constituted a significant risk factor for the development of surgical site infection in patients with GC [9, 10]. Han et al. discovered that the patients with GC and hypertension, DM, or dyslipidemia diagnosed with cardiovascular diseases within 1 year before their cancer diagnosis had a higher mortality risk [11]. Tseng’s study also revealed that diabetic patients have a higher risk of GC mortality and this was most remarkable in the youngest age group of 25–64 years [2]. The presence of DM is associated with age, body mass index (BMI), cardiovascular and cerebrovascular diseases, which also impacts the clinical symptoms related to GC as well as the complexity of surgical procedures [7, 12]. The pathological characteristics, size, and tumor stage also exert an influence on both short and long term surgical outcomes. Consequently, this retrospective study collected case data from one of the largest medical centers in China and employed propensity score matching (PSM) to mitigate the influence of confounding variables mentioned above, with the aim of analyzing the impact of DM on short and long term outcomes following radical gastrectomy and identifying potential factors associated with overall survival (OS). Materials and methods Study participants The patients who underwent radical gastrectomy at our medical center between January 2011 and December 2015 were selected as study population. Patients were further assessed for eligibility based on predetermined inclusion and exclusion criteria. The inclusion criteria were as follows: (1) patients diagnosed with GC and underwent a radical resection procedure, which involved the complete or partial removal of the stomach, irrespective of the operative approach employed; (2) the tumors did not infiltrate adjacent organs and did not develop distant metastasis; (3) the outcome indicators requiring analysis were available. The exclusion criteria included: (1) gastroesophageal junction cancer (GEJC) patients undergoing thoracotomy; (2) patients underwent palliative operation, re-operation or extensive radical surgery, and the procedure of extensive radical surgery entails the excision of an organ other than the stomach, such as the gallbladder, colorectal region, or a segment of the liver; (3) non adenocarcinomas. The study was reviewed and approved by the Medical Ethics Committee and was performed in accordance with the Declaration of Helsinki. The perioperative management strategies for DM patients were as follows: (1) preoperative patient education, complemented by dietary counseling, for the prevention of postprandial hyperglycemia; (2) the blood glucose levels of fasting and three meals were monitored daily prior to the surgery; (3) for patients with inadequate glycemic control, administer insulin promptly and monitor blood glucose levels seven times a day in order to achieve stable blood glucose levels 72 h prior to surgery; (3) the blood glucose level was closely monitored within 3 days post-operation, and the insulin dosage of parenteral nutrition was promptly adjusted based on the test results; (4) the patients were administered prophylactic antibiotics, and blood routine and biochemical tests were conducted for three consecutive days post-surgery, with timely upgrading of antibiotics in case of suspected infection. Data collection and outcome evaluation The relevant clinicopathological data was retrieved and collected from electronic medical records for analysis in this study. Among the retrieved cases, approximately 200 patients underwent neoadjuvant chemotherapy, predominantly with SOX and XELOX regimens. Our hospital administered postoperative adjuvant chemotherapy to over 700 patients, with the main chemotherapy regimens being XELOX, SOX, and FOLFOX. The remaining patients did not receive adjuvant chemotherapy or chemotherapy regimens were unknown. In addition, the postoperative radiotherapy or immunotherapy was administered to over 60 patients in our hospital. Subsequently, the enrolled patients were categorized into DM and non-DM group based on the occurrence of DM, and a PSM analysis was performed between these two groups. The subsequent data, including patients’ demographic characteristics, surgical details, and pathological features, were utilized as covariates in the PSM procedures: sex, age, body mass index (BMI), smoking and drinking history, presence of hypertension, diabetes, cardia-cerebrovascular disease, anaemia, hypoalbuminemia, neoadjuvant chemotherapy, pyloric obstruction, gastrointestinal hemorrhage, or abdominal operation history, operation start time, digestive tract reconstruction method, operative scope, approach, and duration, intraoperative blood loss, number of tumors and lymph node dissection (LND), tumor size, histological types, TNM stage, and presence of neuroendocrine differentiation (NED), lymphovascular invasion (LVI), and nerve invasion. The outcome measures to be evaluated between DM and non-DM groups including the utilization and quantity of jejunal nutrition tube and peritoneal drainage, the duration of gastrointestinal decompression, indentation catheterization, fasting, and hospitalization, overall and severe complication rate, as well as survival outcomes at 1-, 3-, and 5-year. The postoperative complications were evaluated in accordance with the Clavien-Dindo classification system for surgical complications (2009 version) [13]. For patients who experienced concurrent multiple complications, the grading was determined based on the severity of the most critical complication, and Clavien-Dindo grade III-Vcomplications were regarded as severe complications. The occurrence of delayed gastric emptying following subtotal gastrectomy was observed as an early complication in this study. Delayed gastric emptying did not occur after total gastrectomy. The tumor stages were assessed in this study using the TNM staging system (8th edition), which is based on the pathological findings obtained from the patient’s surgical specimens [14]. The serological examination was conducted for all admitted cases. For serum hemoglobin (HGB) examination, male ≤ 120 g/L or female ≤ 110 g/L was defined as preoperative anemia. Serum albumin (ALB) < 35 g/L was defined as preoperative hypoalbuminemia. Statistical analysis SPSS (version 22.0) and R (version 2.15.3) software were adopted for statistical analyses in this study. If the data follows a normal distribution, the mean ± standard deviation (SD) was calculated for inter-group comparison using independent-samples T test. Otherwise, the median and interquartile range (IQR) was calculated and the Mann-Whitney U test was applied. The PSM procedure was conducted using the PS Matching plug-in of SPSS software, with a matching ratio of 1:3 between the DM and non-DM groups, while setting the matching capacity to 0.2. Subsequently, the data from the two groups of matched cases were compared and analyzed. To further explored the impact of DM on OS, COX regression analyses were performed. About sample size estimation, rules of thumb are often adopted that the sample size should be 5, 10, or 20 times the number of variables. For 21 predictors, 315 outcome events were ideal [15]. The correlations between covariates and dependent variables were examined through univariate analyses, and the potential covariates (P < 0.1) would be included in a multivariate COX regression model to identify the independent predictors. In addition, the Bonferroni correction method was employed to address the issue of multiple comparisons in regression analysis. In all analyses, P < 0.05 indicated a statistical difference. Results Patients selection and characteristics Initially, a cohort of 2973 GC patients who underwent gastrectomy were retrieved. Of these, 287 GEJC patients undergoing thoracotomy, 56 underwent palliative surgery, 106 underwent reoperation or extensive radical surgery, and 16 without cancerous pathological reports. After rigorous screening, 2508 patients were selected for further analysis. All patients included in this study completed R0 resection of tumors and D1+/D2 LND, and the mean number of LNM and LND were 4.32 and 26.29 respectively. Among the cohort of 2508 patients who underwent radical surgery prior to PSM, a total of 295 (11.8%) individuals were diagnosed with DM (Fig. 1). The results of comparative analysis revealed significant differences in sex (P = 0.006), age (P < 0.001), BMI (P < 0.001), rate of hypertension (P < 0.001), cardia-cerebrovascular disease (P < 0.001), anaemia (P < 0.001), gastrointestinal hemorrhage (P = 0.046), intraoperative blood loss (P = 0.004), number of tumors (P = 0.040) and LND (P = 0.011) between the DM and non-DM groups prior to PSM (Supplementary materials 1). The flowchart of the study population. GEJC, gastroesophageal junction cancer; PSM, propensity score matching; DM, diabetes mellitus Comparative analysis between the DM and non-DM groups after PSM After conducting 1:3 PSM, 293 DM patients and 821 control patients who underwent radical gastrectomy for GC were enrolled. The process of PSM can be found in Supplementary materials 2. The baseline data of the two groups were compared again after PSM, and no statistically significant differences were observed in any control variables (Table 1). Subsequently, the short-term recovery and long-term survival outcomes were compared between the DM and non-DM groups. The results of comparative analysis revealed that the DM group exhibited a significantly higher rate of overall complications (P < 0.001), grade III-V complications (P = 0.010), and prolonged hospitalization (P = 0.001) compared to the control group. Besides, a more comprehensive analysis of complications showed that the DM group exhibited a significantly elevated risk of delayed gastric emptying (P = 0.005), anastomotic leakage (P = 0.005), and wound infection (P = 0.027). However, no statistical difference was observed in 1-, 2-, or 3-year survival outcomes between the two groups (Table 2). Table 1Clinical characteristics of patients after 1:3 propensity score matchingTotal(n = 1114)DM group(n = 293)Non-DM group(n = 821)P valueSex, n (%)0.889 Male898 (80.6)237 (80.9)661 (80.5) Female216 (19.4)56 (19.1)160 (19.5)Age a, years63 (57–70)64 (57–70)63 (57–70)0.695BMI a, kg/m224.68 (22.60-27.12)24.80 (22.60-27.43)24.61 (22.59–27.02)0.444Smoking history, n (%)0.836 Absence797 (71.5)211 (72.0)586 (71.4) Presence317 (28.5)82 (28.0)235 (28.6)Drinking history, n (%)0.360 Absence711 (63.8)190 (64.8)521 (63.5) Presence403 (36.2)103 (35.2)300 (36.5)Hypertension, n (%)0.307 Absence678 (60.9)171 (58.4)507 (61.8) Presence436 (39.1)122 (41.6)314 (38.2)Cardia-cerebrovascular disease, n (%)0.360 Absence946 (84.9)244 (83.3)702 (85.5) Presence168 (15.1)49 (16.7)119 (14.5)Anaemia, n (%)0.237 Absence757 (68.0)191 (65.2)566 (68.9) Presence357 (32.0)102 (34.8)255 (31.1)Hypoalbuminemia, n (%)0.866 Absence974 (87.4)257 (87.7)717 (87.3) Presence140 (12.6)36 (12.3)104 (12.7)Neoadjuvant chemotherapy, n (%)0.683 Absence1032 (92.6)273 (93.2)759 (92.4) Presence82 (7.4)20 (6.8)62 (7.6)Pyloric obstruction, n (%)0.880 Absence982 (88.2)259 (88.4)723 (88.1) Presence132 (11.8)34 (11.6)98 (11.9)Gastrointestinal hemorrhage, n (%)0.615 Absence897 (80.5)233 (79.5)664 (80.9) Presence217 (19.5)60 (20.5)157 (19.1)Abdominal operation history, n (%)0.824 Absence921 (82.7)241 (82.3)680 (82.8) Presence193 (17.3)52 (17.7)141 (17.2)Operation start time, n (%)0.882 Morning608 (54.6)161 (54.9)447 (54.4) Afternoon506 (45.4)132 (45.1)374 (45.6)Digestive tract reconstruction, n (%)1.000 Esophagogastric anastomosis291 (26.1)77 (26.3)214 (26.1) Esophagojejunal Roux-en-Y221 (19.8)58 (19.8)163 (19.9) Esophagojejunal Billroth II23 (2.1)6 (2.0)17 (2.1) Gastroduodenostomy Billroth I301 (27.0)81 (27.6)220 (26.8) Gastrojejunum Roux-en-Y78 (7.0)20 (6.8)58 (7.1) Gastrojejunum Billroth II200 (18.0)51 (17.4)149 (18.1)Operative scope, n (%)0.997 Proximal stomach291 (26.1)77 (26.3)214 (26.1) Entire stomach243 (21.8)64 (21.8)179 (21.8) Distal stomach580 (52.1)152 (51.9)428 (52.1)Operative approach, n (%)0.887 Open721 (64.7)193 (65.9)528 (64.3) Laparoscopic321 (28.8)82 (28.0)239 (29.1) Robot-assisted72 (6.5)18 (6.1)54 (6.6)Operative duration a, min200 (170–243)199 (170–241)202 (170–243)0.575Intraoperative blood loss a, ml200 (100–300)200 (100–300)200 (100–300)0.639Number of LND a23 (17–31)22 (17–31)23 (17–31)0.737Tumor size a, cm4.0 (2.5-6.0)4.0 (2.5-6.0)4.0 (2.5-6.0)0.612Number of tumors, n (%)0.974 Single1099 (98.7)289 (98.6)810 (98.7) Multiple (≥ 2)15 (1.3)4 (1.4)11 (1.3)NED, n (%)0.900 Absence1066 (95.7)280 (95.6)786 (95.7) Presence48 (4.3)13 (4.4)35 (4.3)Histological types, n (%)0.888 Differentiated236 (21.2)65 (22.2)171 (20.8) Mixed362 (32.5)94 (32.1)268 (32.6) Undifferentiated516 (46.3)134 (45.7)382 (46.5)T stage, n (%)0.969 T1297 (26.7)77 (26.3)220 (26.8) T2154 (13.8)43 (14.7)111 (13.5) T3286 (25.7)75 (25.6)211 (25.7) T4377 (33.8)98 (33.4)279 (34.0)N stage, n (%)0.871 N0498 (44.7)134 (45.7)364 (44.3) N1195 (17.5)48 (16.4)147 (17.9) N2195 (17.5)54 (18.4)141 (17.2) N3226 (20.3)57 (19.5)169 (20.6)Tumor stage (pathological), n (%)0.948 I372 (33.4)100 (34.1)272 (33.1) II290 (26.0)76 (25.9)214 (26.1) III452 (40.6)117 (39.9)335 (40.8)LVI, n (%)0.770 Absence851 (76.4)222 (75.8)629 (76.6) Presence263 (23.6)71 (24.2)192 (23.4)Nerve invasion, n (%)0.741 Absence940 (84.4)249 (85.0)691 (84.2) Presence174 (15.6)44 (15.0)130 (15.8)aValues are presented as median (Q1-Q3)DM, diabetes mellitus; LND, lymph node dissection; BMI, body mass index; NED, neuroendocrine differentiation; LVI, lymphovascular invasion Clinical characteristics of patients after 1:3 propensity score matching aValues are presented as median (Q1-Q3) DM, diabetes mellitus; LND, lymph node dissection; BMI, body mass index; NED, neuroendocrine differentiation; LVI, lymphovascular invasion Table 2Comparison of postoperative recovery and survival outcomes between DM and Non-DM groupsDM group(n = 293)Non-DM group(n = 821)P valuePlacement of jejunal nutrient tube, n (%)0.078 Absence235 (80.2)695 (84.7) Presence58 (19.8)126 (15.3)Number of abdominal drainage tubes, n (%)0.221 Single132 (45.1)404 (49.2) Multiple (≥ 2)161 (54.9)417 (50.8)Indwelling gastric tube a, days6 (5–7)6 (5–7)0.162Indwelling catheter a, days4 (3–6)4 (3–6)0.486Time to the first flatus a, days7 (6–8)6 (6–8)0.186Hospital stays a, days19 (15–24)17 (14–22) 0.001 Total complications, n (%)62 (21.2)98 (11.9) < 0.001 Delayed gastric emptying, n (%)15 (5.1)16 (1.9) 0.005 Anastomotic leakage, n (%)16 (5.5)18 (2.2) 0.005 Anastomotic hemorrhage, n (%)2 (0.7)8 (1.0)0.925 Pancreatic leakage, n (%)4 (1.4)8 (1.0)0.821 Intestinal obstruction, n (%)4 (1.4)10 (1.2)0.846 Intra-abdominal infection, n (%)5 (1.7)7 (0.9)0.376 Pulmonary infection, n (%)8 (2.7)13 (1.6)0.215 Wound infection, n (%)7 (2.4)5 (0.6) 0.027 Other complications, n (%)5 (1.7)13 (1.6)1.000Severe complication (grade III-V) rate, n (%)21 (7.2)29 (3.5) 0.010 Reoperation, n (%)4 (1.4)9 (1.1)0.959Mortality, n (%)1 (0.3)2 (0.2)1.000OS b, months73.31 ± 2.2873.47 ± 1.330.9761-year survival outcome, n (%)0.459 Alive259 (88.4)746 (90.9) Died27 (9.2)58 (7.1) Unknown7 (2.4)17 (2.1)3-year survival outcome, n (%)0.937 Alive206 (70.3)570 (69.4) Died74 (25.3)216 (26.3) Unknown13 (4.4)35 (4.3)5-year survival outcome, n (%)0.778 Alive111 (37.9)314 (38.2) Died90 (30.7)266 (32.4) Unknown92 (31.4)241 (29.4)aValues are presented as median (Q1-Q3)bValues are presented as mean ± standard deviationDM, diabetes mellitus; OS, overall survival Comparison of postoperative recovery and survival outcomes between DM and Non-DM groups aValues are presented as median (Q1-Q3) bValues are presented as mean ± standard deviation DM, diabetes mellitus; OS, overall survival COX regression analysis of predictors for OS To account for potential confounding factors on survival outcomes, we employed COX regression analysis with OS as the dependent variable and included DM along with other relevant variables. This study included a total of 379 patients who experienced an outcome event (death), indicating that the sample size was adequate for detecting clinically significant differences. Subsequently, univariate and multivariate COX regression analyses were conducted between 293 DM and 821 non-DM patients. Univariate analysis showed that age > 65years [hazard ratio (HR) = 2.159, P < 0.001], BMI < 18.5 kg/m2 (HR = 2.536, P < 0.001), presence of pyloric obstruction (HR = 1.719, P < 0.001), total gastrectomy (HR = 1.665, P < 0.001), mixed (HR = 1.825, P < 0.001) and undifferentiated (HR = 2.104, P < 0.001) histological types, tumor size > 5 cm (HR = 2.817, P < 0.001), higher T (P < 0.001) and N tumor stage (P < 0.001), and the presence of NED (HR = 1.551, P = 0.041), LVI (HR = 1.793, P < 0.001), nerve invasion (HR = 1.877, P < 0.001), total (HR = 1.426, P = 0.008) and severe postoperative complications (HR = 2.005, P < 0.001) could influence the OS. The aforementioned potential covariates (P < 0.1) were incorporated into a multivariate COX regression model. Age > 65years (HR = 2.029, P < 0.001), BMI < 18.5 kg/m2 (HR = 2.143, P = 0.001), tumor size > 5 cm (HR = 1.519, P < 0.001), higher T (P = 0.012) and N tumor stage (P < 0.001), and the occurrence of severe complications (HR = 1.965, P = 0.007) were identified as independent risk factors for OS (Table 3). The P-values of multiple statistical tests in this analysis were all below 0.005 before correction, and even after correction, they remained below 0.05, thereby eliminating the possibility of false positives. Besides, Kaplan-Meier curves of OS for age (Fig. 2A), tumor size (Fig. 2B), TNM tumor stage (Fig. 2C), and severe complications (Fig. 2D) were both drawn using the data of 1114 GC cases. Table 3Univariate and multivariate Cox regression analysis of variables influencing overall survivalUnivariate analysisMultivariate analysisHR (95% CI)P valueHR (95% CI)P valueSex0.241 MaleReference Female0.853 (0.654, 1.112)Age (years) < 0.001 < 0.001 ≤ 65ReferenceReference > 652.159 (1.760, 2.648)2.029 (1.645, 2.502)BMI (kg/m2) < 0.001 < 0.001 18.5–28ReferenceReference < 18.52.536 (1.645, 3.909) < 0.001 2.143 (1.363, 3.370) 0.001 > 280.531 (0.384, 0.736) < 0.001 0.643 (0.462, 0.895) 0.009 Neoadjuvant chemotherapy0.645 AbsenceReference Presence1.098 (0.737, 1.636)Hypertension0.987 AbsenceReference Presence0.998 (0.812, 1.227)Diabetes0.976 AbsenceReference Presence0.996 (0.792, 1.254)Cardia-cerebrovascular disease0.505 AbsenceReference Presence1.098 (0.835, 1.444)Pyloric obstruction < 0.001 0.858 AbsenceReferenceReference Presence1.719 (1.314, 2.249)1.026 (0.775, 1.358)Gastrointestinal hemorrhage0.207 AbsenceReference Presence0.843 (0.646, 1.099)Operative scope < 0.001 0.765 Proximal stomachReferenceReference Entire stomach1.665 (1.270, 2.182) < 0.001 1.005 (0.754, 1.339)0.974 Distal stomach0.857 (0.669, 1.098)0.2220.925 (0.716, 1.195)0.550Operative approach0.232 OpenReference Laparoscopic0.870 (0.692, 1.094)0.233 Robot-assisted0.709 (0.434, 1.159)0.170Number of tumors0.543 SingleReference Multiple (≥ 2)1.285 (0.573, 2.878)NED 0.041 0.269 AbsenceReferenceReference Presence1.551 (1.017, 2.365)1.272 (0.830, 1.951)Histological types < 0.001 0.183 DifferentiatedReferenceReference Mixed1.825 (1.320, 2.524) < 0.001 1.305 (0.933, 1.825)0.120 Undifferentiated2.104 (1.547, 2.863) < 0.001 1.349 (0.978, 1.860)0.068Tumor size (cm) < 0.001 < 0.001 ≤ 5ReferenceReference > 52.817 (2.301, 3.450)1.519 (1.202, 1.919)T stage < 0.001 0.012 T1ReferenceReference T22.615 (1.670, 4.096) < 0.001 1.987 (1.249, 3.162) 0.004 T34.302 (2.940, 6.293) < 0.001 1.859 (1.206, 2.867) 0.005 T45.814 (4.053, 8.341) < 0.001 1.977 (1.279, 3.054) 0.002 N stage < 0.001 < 0.001 N0ReferenceReference N12.450 (1.762, 3.408) < 0.001 1.688 (1.184, 2.406) 0.004 N23.714 (2.732, 5.050) < 0.001 2.364 (1.681, 3.324) < 0.001 N37.367 (5.576, 9.732) < 0.001 4.741 (3.364, 6.683) < 0.001 LVI < 0.001 0.459 AbsenceReferenceReference Presence1.793 (1.444, 2.226)0.915 (0.724, 1.157)Nerve invasion < 0.001 0.888 AbsenceReferenceReference Presence1.877 (1.477, 2.385)0.982 (0.760, 1.269)Postoperative complications 0.008 0.773 AbsenceReferenceReference Presence1.426 (1.097, 1.854)0.952 (0.684, 1.326)Severe complications < 0.001 0.007 AbsenceReferenceReference Presence2.005 (1.355, 2.965)1.965 (1.200, 3.217)HR, hazard ratio; CI, confidence interval; BMI, body mass index; NED, neuroendocrine differentiation; LVI, lymphovascular invasion Univariate and multivariate Cox regression analysis of variables influencing overall survival HR, hazard ratio; CI, confidence interval; BMI, body mass index; NED, neuroendocrine differentiation; LVI, lymphovascular invasion Kaplan-Meier curve illustrating OS for (A) age, (B) tumor size, (C) TNM tumor stage, and (D) severe complications in patients following radical gastrectomy To validate the aforementioned analysis findings, we constructed several COX regression models incorporating diverse variables: six independent predictors, aforementioned six predictors + variables that did not exhibit statistical significance in univariate analysis, and all variables involved in univariate analysis (Table 4). The analysis results of the three COX regression models exhibit no discernible disparities, indicating a higher level of reliability in the findings. In addition, we stratified the study population into different subgroups based on DM prevalence and performed COX regression analysis to examine the association between each independent predictor with OS (Fig. 3). The remaining five independent predictors were adjusted as covariates during each analysis. The analyses of individuals without DM revealed significant correlations between age (Fig. 3A), BMI (Fig. 3B), tumor size (Fig. 3C), T stage (Fig. 3D), N stage (Fig. 3E), and severe complications (Fig. 3F) with OS. Owing to the limited sample size of DM patients, the analysis of certain predictors exhibited a correlation trend without achieving statistical significance. Table 4Multi-model COX regression analysis of variables influencing overall survivalModel 1Model 2Model 3HR (95% CI)P valueHR (95% CI)P valueHR (95% CI)P valueAge (years)< 0.001< 0.001< 0.001≤ 65ReferenceReferenceReference> 652.009 (1.631, 2.473)2.018 (1.627, 2.504)2.032 (1.636, 2.524)BMI (kg/m2)< 0.001< 0.001< 0.00118.5–28ReferenceReferenceReference< 18.52.029 (1.302, 3.163)0.0022.019 (1.290, 3.161)0.0022.124 (1.346, 3.353)0.001> 280.639 (0.460, 0.887)0.0080.669 (0.480, 0.931)0.0170.672 (0.482, 0.938)0.019Tumor size (cm)< 0.001< 0.001< 0.001≤ 5ReferenceReferenceReference> 51.518 (1.214, 1.899)1.536 (1.225, 1.925)1.530 (1.210, 1.935)T stage0.0040.0050.013T1ReferenceReferenceReferenceT22.113 (1.335, 3.345)0.0012.114 (1.330, 3.358)0.0021.995 (1.250, 3.184)0.004T31.939 (1.268, 2.966)0.0021.903 (1.241, 2.918)0.0031.832 (1.186, 2.829)0.006T42.085 (1.365, 3.186)0.0012.075 (1.351, 3.186)0.0011.978 (1.275, 3.067)0.002N stage< 0.001< 0.001< 0.001N0ReferenceReferenceReferenceN11.695 (1.190, 2.414)0.0031.727 (1.210, 2.464)0.0031.715 (1.200, 2.450)0.003N22.406 (1.717, 3.373)< 0.0012.474 (1.762, 3.474)< 0.0012.414 (1.714, 3.401)< 0.001N34.845 (3.492, 6.723)< 0.0014.948 (3.562, 6.874)< 0.0014.830 (3.420, 6.819)< 0.001Severe complications0.0030.0030.009AbsenceReferenceReferenceReferencePresence1.806 (1.215, 2.683)1.842 (1.235, 2.747)1.941 (1.182, 3.187)HR, hazard ratio; CI, confidence interval; BMI, body mass indexModel 1: CrudeModel 2: Adjust: sex, neoadjuvant chemotherapy, hypertension, diabetes, cardia cerebrovascular disease, gastrointestinal hemorrhage, operative approach, number of tumorsModel 3: Adjust: sex, neoadjuvant chemotherapy, hypertension, diabetes, cardia cerebrovascular disease, gastrointestinal hemorrhage, operative approach, number of tumors, pyloric obstruction, operative scope, NED, histological types, LVI, nerve invasion, postoperative complications Multi-model COX regression analysis of variables influencing overall survival HR, hazard ratio; CI, confidence interval; BMI, body mass index Model 1: Crude Model 2: Adjust: sex, neoadjuvant chemotherapy, hypertension, diabetes, cardia cerebrovascular disease, gastrointestinal hemorrhage, operative approach, number of tumors Model 3: Adjust: sex, neoadjuvant chemotherapy, hypertension, diabetes, cardia cerebrovascular disease, gastrointestinal hemorrhage, operative approach, number of tumors, pyloric obstruction, operative scope, NED, histological types, LVI, nerve invasion, postoperative complications With OS as the dependent variables, the forest plots present the subgroup analysis of (A) age, (B) BMI, (C) tumor size, (D) T stage, (E) N stage, and (F) severe complications based on diabetes prevalence Discussion Surgery remains an important component of curative treatment for GC. The presence of DM can impact the patient’s surgical and anesthetic tolerance, as well as exacerbate adverse reactions such as impaired phagocytic ability, reduced body resistance, and increased wound exudation caused by postoperative blood glucose elevation. The aforementioned factors contribute to an elevated susceptibility to postoperative complications [16, 17]. The occurrence of postoperative complications, such as anastomotic leakage, gastrointestinal dysfunction, and infection, can impact surgical outcomes by prolonging hospitalization and increasing medical expenses. In severe cases, these complications may even result in perioperative mortality. The presence of preexisting diabetes in GC patients may also be associated with decreased rate and shorter duration of postoperative survival. Therefore, the success of radical surgery for GC patients with DM relies heavily on effective perioperative nursing and the prevention of postoperative complications, and the impact of DM on survival outcome necessitates additional investigation [18–20]. Based on the analysis of 412 GC cases, Mukoyama et al. observed that the majority of delayed gastric emptying cases occurred between 7 and 14 days post-surgery, with a higher prevalence among female patients, those with DM, and individuals diagnosed with distal gastric tumors [18]. The digestive activity of the stomach is regulated by a complex series of interdependent neuro-hormonal mechanisms, in which the vagus nerve plays a major role. The elevation of blood glucose during the perioperative period can result in gastric nerve conduction dysfunction and abnormal gastric motility, ultimately leading to the development of gastroparesis syndrome [21, 22]. Besides, surgical procedure can also induce alterations in anatomical relationships, impede partial innervation, and disrupt the regulation of gastrointestinal neurohormones, thereby resulting in persistent postoperative dysfunction [23]. The impact of DM on the occurrence of anastomotic leakage has also been evidenced in published studies [24, 25]. The previous studies have indicated that diabetes can impact all stages of wound healing, encompassing cell migration, cell proliferation, angiogenesis and granulation tissue formation, as well as impede or diminish the regenerative capacity of body tissues. The study conducted by Onodera et al. revealed that hyperglycemia significantly augmented the non-enzymatic glycosylation process of proteins, including collagen. This heightened glycosylation, coupled with hindered cross-linking between molecules, may potentially compromise the structural integrity of newly formed collagen and consequently contribute to a higher incidence of anastomotic leakage [26, 27]. The findings of our study also demonstrated that GC patients with diabetes exhibited an elevated susceptibility to postoperative anastomotic leakage. Patients with diabetes often experience metabolic disorders and malnutrition, leading to reduced immune function. In a state of surgical stress, the body’s energy utilization rate decreases, resulting in inadequate blood supply to the intestinal microcirculation. Additionally, some patients may also have complications such as microvascular sclerosis, which can delay tissue healing and increase the risk of anastomotic leakage [28, 29]. Several studies have suggested that there is an elevated susceptibility to surgical site infections following diabetes [9–19]. A meta-analysis including thirty-four eligible articles with 66,066 patients found that the presence of DM was a significant risk factors for surgical site infections [9]. The abdominal surgical incision is a frequently performed procedure in surgery, and it is also prone to incision infection, which often leads to prolonged hospitalization and increased patient discomfort [19]. Diabetes is widely acknowledged as a significant risk factor for wound infection. The immune response of diabetic patients is compromised, and the hyperglycemic environment facilitates tissue edema, thereby promoting bacterial growth and leading to incision dehiscence and infection. In addition, diabetes can impair fibroblast formation and impede the process of incision healing. Therefore, continued efforts are needed to improve surgical outcomes for diabetic patients [30]. In the perioperative management for GC patients complicated with DM, effective glycemic control is crucial. The first step is to ensure appropriate preoperative blood sugar control, followed by patient education and supplemented with dietary guidance to prevent blood sugar spikes. Due to the extensive nature of the procedure, the administration of insulin is necessary to ensure blood sugar stability in the 72 h preceding surgery [16, 31]. The surgery should be conducted when the underlying disease is effectively managed and nutrition is at an optimal level. Secondly, factors such as trauma and anesthesia during surgery can lead to an upregulation of glucocorticoid secretion in patients, consequently inducing insulin resistance and resulting in elevated blood glucose levels. The use of sympathetic agonists during surgery should be minimized to mitigate the risk of inducing insulin resistance in patients. While maintaining blood sugar control is crucial, it is imperative to prioritize incision protection during surgical procedures. Prior to opening the digestive tract, it is crucial to isolate the digestive tract from the abdominal cavity and incision, while also preparing the intestines for selective surgery [17, 32, 33]. Furthermore, the operation should be performed with utmost care to minimize potential damage to blood vessels and nerves, control bleeding effectively, and optimize the duration of the procedure. After the surgical procedure, it is necessary to implement fasting and parenteral nutrition in order to expedite the recovery of gastrointestinal function and promote healing of the anastomosis. The early use of diabetes-specific enteral nutrition preparations after surgery is recommended to ensure adequate nutrient supply while minimizing the risk of blood sugar fluctuations, thereby assisting patients in safely navigating the perioperative period. In addition, insulin should be continued within 3 days post-surgery while closely monitoring blood glucose levels. Rational administration of antibiotics, timely dressing replacement, and catheter removal are essential measures to prevent surgical and urinary tract infections [34, 35]. After the surgical procedure, it is imperative to minimize the retention of the peritoneal drainage tube, prioritize anti-infection treatment, facilitate tissue healing, and mitigate the risk of complications. It is also crucial to closely monitor the patient’s physical condition, abdominal drainage, and test results in order to promptly identify potential complications such as abdominal infection, bleeding, or leakage. Depending on the severity of these complications, appropriate measures including anti-infection treatment, abdominal irrigation, endoscopy, intervention procedures or surgical interventions should be taken [28, 31, 36]. It was determined that factors such as age > 65years, BMI < 18.5 kg/m2, tumor size > 5 cm, higher tumor stage, and the occurrence of severe complications significantly influenced OS. The physiological characteristics of the elderly, such as cellular and organ aging, muscle atrophy and loss, result in compromised compensatory function of various bodily organs, diminished immune function, and an increased susceptibility to sarcopenia and malnutrition. These factors contribute to a higher incidence of surgical complications, elevated risk of postoperative tumor recurrence, and shortened survival time [37, 38]. The presence of cardiovascular and cerebrovascular diseases, hypertension, diabetes, and other comorbidities in elderly patients also exert a certain impact on their postoperative quality of life and survival time. Malignant tumors are characterized by cachexia, and a low BMI often indicates a higher degree of malignancy, wider lesion range, longer duration, and advanced tumor stage. Consequently, the curative effect of radical surgery is suboptimal and the prognosis for these patients is unsatisfactory [39]. Additionally, advanced tumors exhibit larger volumes, deeper infiltration, more severe inflammatory responses, increased lymph node metastasis, greater surgical extent and complexity, as well as an elevated risk of postoperative complications [38, 40]. The present study also investigated the factors that influence survival outcomes following radical gastrectomy, thereby offering valuable insights for enhancing the postoperative prognosis of GC. The findings of this study enable surgeons to identify relevant risk groups, including patients with advanced age, weight loss, large tumors, high tumor stages, or severe postoperative complications. This allows for active perioperative intervention and postoperative follow-up and treatment in order to enhance the overall quality of life. The present study, despite its analysis of GC case data from one of China’s largest medical centers, still possesses certain limitations. Firstly, the level of evidence was constrained by administrative and technical disparities between retrospective cohort studies and cases originating from a single center, thereby potentially impeding the generalizability of the findings to other healthcare settings or populations. Secondly, the operational efficacy and prognosis were influenced to a certain extent by the variations in surgical skills and practices among different surgeons, despite their extensive experience as chief physicians in this study. Thirdly, the propensity score model includes a large number of variables relative to the number of events, which could lead to overfitting and potentially biased estimates. Besides, only approximately one-third of the patients included in this study exhibited a survival outcome, necessitating further long-term follow-up. Owing to these inherent limitations, the impact of DM on survival outcomes necessitates confirmation and supplementation through large-scale, prospective, multi-center randomized controlled clinical trials. The prognosis of GC patients with DM combined with postoperative complications requires further analysis. Additionally, future research may focus on comparing and evaluating the efficacy of various postoperative adjuvant treatment regimens in cases of GC complicated by DM. Conclusions The presence of combined DM in GC patients can heighten the susceptibility to overall and grade III-Vpostoperative complications, and protract the duration of postoperative recovery, while exerting no significant impact on survival outcomes. Besides, age > 65years, BMI < 18.5 kg/m2, tumor size > 5 cm, higher T and N tumor stage, and the occurrence of severe complications were identified as independent risk factors for OS. The present study examined the impact of DM on both short and long term outcomes following radical gastrectomy utilizing PSM method, as well as explored the factors influencing survival outcomes, thereby providing valuable guidance for the perioperative management and postoperative follow-up in GC patients with DM. Electronic supplementary material Below is the link to the electronic supplementary material. Supplementary Material 1 Supplementary Material 2 Acknowledgements We thank all the patients whose data were used for the study. Our greatest acknowledgement goes to the authors who made detailed data available for this study and to all our colleagues in this study for their hard work. Author contributions Zhiyuan Yu: Conceptualization, Data curation, Writing - original draft. Chen Liang: Formal analysis, Methodology, Writing - original draft. Rui Li: Data curation, Investigation, Visualization. Qixuan Xu: Data curation, Software, Visualization. Jingwang Gao: Data curation, Formal analysis. Peiyu Li: Conceptualization, Supervision. Sixin Zhou: Methodology, Supervision. Xudong Zhao: Conceptualization, Formal analysis, Writing-review & editing. Manman Xu: Methodology, Software, Writing-review & editing. Wenquan Liang: Data curation, Project administration, Writing-review & editing. All authors reviewed the manuscript. Funding This research did not receive any specific grants from funding agencies in the public, commercial, or not-for-profit sectors. Data availability The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request. Declarations Ethics approval and informed consent This study was approved by the Medical Ethics Committee of the First Medical Center of the Chinese PLA General Hospital (Approval No. S2022-323-01) and was performed in accordance with the Declaration of Helsinki. Informed consent was obtained from all participants during the follow-up period. Consent for publication Not applicable. Competing interests The authors declare no competing interests. Conflict of interest The authors have no conflicts of interest to declare. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Zhiyuan Yu, Chen Liang and Rui Li contributed equally to this work. Contributor Information Xudong Zhao, Email: 601489554@qq.com. Manman Xu, Email: xmm1903@ccmu.edu.cn. Wenquan Liang, Email: lwq0618@163.com. References Associated Data Supplementary Materials Supplementary Material 1 Supplementary Material 2 Data Availability Statement The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.