Objective:To develop and validate machine learning-based radiomics models using preoperative CT images for individualized prediction of mitotic index (MI) in patients with gastrointestinal stromal tumors (GIST).Methods:The study was a case-control study. The data of 348 GIST patients confirmed by pathology were retrospectively collected from two independent medical centers: the Affiliated Hospital of Qingdao University (center 1) and Shandong Provincial Hospital Affiliated to Shandong First Medical University (center 2), covering the period from January 2013 to June 2018. Patients from center 1 were divided into a training cohort (176 cases) and an internal validation cohort (75 cases) at a ratio of 7∶3 using random sampling. Patients from center 2 served as an independent external validation cohort (97 cases). The primary endpoint was MI, categorized into high MI (145 cases) and low MI (203 cases) groups. Radiomic features were extracted from the portal venous phase images of preoperative contrast-enhanced CT scans. Five machine learning algorithms, including logistic regression, support vector machine, random forest, decision tree, and extreme gradient boosting (XGBoost),were employed to construct MI prediction models. The optimal model was identified using receiver operating characteristic curves. An individualized prediction model was developed by integrating the the optimal machine learning model combined with selected independent clinical factors, and the importance of features was visualized using Shapley Additive Explanation (SHAP) analysis. Patients were followed up, and Kaplan-Meier curves along with log-rank tests were used to evaluate recurrence-free survival (RFS) differences between the predicted high MI and low MI groups.Results:Among the five constructed machine learning models, the XGBoost model demonstrated the best predictive performance, with area under the curve (AUC) of 0.809 (95% CI 0.738-0.872), 0.693 (95% CI 0.571-0.809), and 0.718 (95% CI 0.605-0.822) in the training cohort, internal validation cohort, and external validation cohort, respectively. An individualized prediction model combining the XGBoost model with independent clinical factors (tumor location and tumor size) was developed. The model achieved AUC of 0.843 (95% CI 0.785-0.899), 0.791 (95% CI 0.680-0.894), and 0.777 (95% CI 0.678-0.861) in the training cohort, internal validation cohort, and external validation cohort, respectively. SHAP analysis indicated that radiomic features had the highest predictive impact. In both the training cohort and internal validation cohort, the RFS of patients predicted to be in the high MI group was lower than that of the low MI group, with statistically significant differences ( χ2=14.58, 9.52, both P<0.001). However, there was no statistically significant difference in RFS in the external validation set ( χ2=6.18, P=0.080). Conclusions:The optimal XGBoost model based on radiomic features extracted from preoperative portal venous phase CT images, when combined with clinical factors, can effectively predict the MI of GIST patients.

Objective:To investigate the diagnostic value and influencing factors of endoscopic ultrasonography (EUS) for detecting rectal neuroendocrine neoplasms (R-NENs).Methods:A retrospective case-control study was performed on data of patients with suspected R-NENs by white light endoscopy who underwent endoscopic diagnosis and treatment or surgical operation and obtained pathological diagnosis at the Affiliated Hospital of Qingdao University from March 2016 to June 2023. Clinical data, EUS characteristics and pathological results were statistically analyzed, and the diagnostic accuracy of EUS for R-NENs were obtained by comparing the EUS results with the pathological results. Influencing factors affecting accuracy were analyzed by using the binary logistic regression model.Results:A total of 317 patients were included. The sensitivity, the specificity, the positive predictive value and the negative predictive value of EUS in diagnosing R-NENs were 98.03% (249/254), 34.92% (22/63), 85.86% (249/290) and 81.48% (22/27) respectively. The accuracy was 85.49% (271/317) and the Jorden index was 0.33. Tumor size ≤5 mm ( P=0.002, OR=2.892, 95% CI: 1.464-5.713), absence of surface vascular dilation ( P=0.019, OR=2.613, 95% CI: 1.170-5.837), normal tumor coloration ( P=0.001, OR=3.460, 95% CI: 1.645-7.279) and erythematous surface appearance ( P=0.048, OR=7.242, 95% CI: 1.015-51.680) were independent risk factors affecting the accuracy of R-NENs diagnosis by EUS. Depth assessment accuracy of EUS was 76.77% (195/254), with echo heterogeneity ( P<0.001, OR=4.008, 95% CI: 1.980-8.113) and surface depression ( P=0.035, OR=2.664, 95% CI: 1.073-6.615) emerging as significant factors affecting invasion depth evaluation. Conclusion:EUS demonstrates substantial clinical utility for R-NENs assessment, with diagnostic performance being significantly associated with tumor morphology and sonographic features. Macroscopic characteristics including tumor size, vascular patterns, and chromatic features influence diagnostic accuracy, while echo-textural heterogeneity and surface depression affect invasion depth precision. These findings underscore the clinical relevance of comprehensive EUS evaluation in R-NENs management.

Objective:To develop and validate machine learning-based radiomics models using preoperative CT images for individualized prediction of mitotic index (MI) in patients with gastrointestinal stromal tumors (GIST).Methods:The study was a case-control study. The data of 348 GIST patients confirmed by pathology were retrospectively collected from two independent medical centers: the Affiliated Hospital of Qingdao University (center 1) and Shandong Provincial Hospital Affiliated to Shandong First Medical University (center 2), covering the period from January 2013 to June 2018. Patients from center 1 were divided into a training cohort (176 cases) and an internal validation cohort (75 cases) at a ratio of 7∶3 using random sampling. Patients from center 2 served as an independent external validation cohort (97 cases). The primary endpoint was MI, categorized into high MI (145 cases) and low MI (203 cases) groups. Radiomic features were extracted from the portal venous phase images of preoperative contrast-enhanced CT scans. Five machine learning algorithms, including logistic regression, support vector machine, random forest, decision tree, and extreme gradient boosting (XGBoost),were employed to construct MI prediction models. The optimal model was identified using receiver operating characteristic curves. An individualized prediction model was developed by integrating the the optimal machine learning model combined with selected independent clinical factors, and the importance of features was visualized using Shapley Additive Explanation (SHAP) analysis. Patients were followed up, and Kaplan-Meier curves along with log-rank tests were used to evaluate recurrence-free survival (RFS) differences between the predicted high MI and low MI groups.Results:Among the five constructed machine learning models, the XGBoost model demonstrated the best predictive performance, with area under the curve (AUC) of 0.809 (95% CI 0.738-0.872), 0.693 (95% CI 0.571-0.809), and 0.718 (95% CI 0.605-0.822) in the training cohort, internal validation cohort, and external validation cohort, respectively. An individualized prediction model combining the XGBoost model with independent clinical factors (tumor location and tumor size) was developed. The model achieved AUC of 0.843 (95% CI 0.785-0.899), 0.791 (95% CI 0.680-0.894), and 0.777 (95% CI 0.678-0.861) in the training cohort, internal validation cohort, and external validation cohort, respectively. SHAP analysis indicated that radiomic features had the highest predictive impact. In both the training cohort and internal validation cohort, the RFS of patients predicted to be in the high MI group was lower than that of the low MI group, with statistically significant differences ( χ2=14.58, 9.52, both P<0.001). However, there was no statistically significant difference in RFS in the external validation set ( χ2=6.18, P=0.080). Conclusions:The optimal XGBoost model based on radiomic features extracted from preoperative portal venous phase CT images, when combined with clinical factors, can effectively predict the MI of GIST patients.

Objective:To investigate the diagnostic value and influencing factors of endoscopic ultrasonography (EUS) for detecting rectal neuroendocrine neoplasms (R-NENs).Methods:A retrospective case-control study was performed on data of patients with suspected R-NENs by white light endoscopy who underwent endoscopic diagnosis and treatment or surgical operation and obtained pathological diagnosis at the Affiliated Hospital of Qingdao University from March 2016 to June 2023. Clinical data, EUS characteristics and pathological results were statistically analyzed, and the diagnostic accuracy of EUS for R-NENs were obtained by comparing the EUS results with the pathological results. Influencing factors affecting accuracy were analyzed by using the binary logistic regression model.Results:A total of 317 patients were included. The sensitivity, the specificity, the positive predictive value and the negative predictive value of EUS in diagnosing R-NENs were 98.03% (249/254), 34.92% (22/63), 85.86% (249/290) and 81.48% (22/27) respectively. The accuracy was 85.49% (271/317) and the Jorden index was 0.33. Tumor size ≤5 mm ( P=0.002, OR=2.892, 95% CI: 1.464-5.713), absence of surface vascular dilation ( P=0.019, OR=2.613, 95% CI: 1.170-5.837), normal tumor coloration ( P=0.001, OR=3.460, 95% CI: 1.645-7.279) and erythematous surface appearance ( P=0.048, OR=7.242, 95% CI: 1.015-51.680) were independent risk factors affecting the accuracy of R-NENs diagnosis by EUS. Depth assessment accuracy of EUS was 76.77% (195/254), with echo heterogeneity ( P<0.001, OR=4.008, 95% CI: 1.980-8.113) and surface depression ( P=0.035, OR=2.664, 95% CI: 1.073-6.615) emerging as significant factors affecting invasion depth evaluation. Conclusion:EUS demonstrates substantial clinical utility for R-NENs assessment, with diagnostic performance being significantly associated with tumor morphology and sonographic features. Macroscopic characteristics including tumor size, vascular patterns, and chromatic features influence diagnostic accuracy, while echo-textural heterogeneity and surface depression affect invasion depth precision. These findings underscore the clinical relevance of comprehensive EUS evaluation in R-NENs management.

Objective: To explore the effect of high mobility group box-1 protein (HMGB1) on the balance of Th17/Treg in patients with immune thrombocytopenia (ITP). Methods: A total of 30 patients who were first diagnosed as ITP in the Fifth People's Hospital of Datong from July 2017 to April 2018 were selected as the case group, and another 30 healthy volunteers in the corresponding period were taken as the control group. The proportion of Th17 and Treg cells was detected by using flow cytometry, and the concentration of HMGB1, interleukin (IL)-17 and transforming growth factor β (TGF-β) in plasma was tested by using enzyme-linked immunosorbent assay (ELISA). Isolated peripheral blood mononuclear cells (PBMC) were cultured in vitro. After the treatment with recombinant human HMGB1 (rhHMGB1), real-time polymerase chain reaction (RT-PCR) was applied to detect the mRNA expression changes in Treg cell transcription factor intracellular forkhead helix transcription factor 3 (Foxp3) and Th17 cell transcription factor retinoid related orphan receptor γt (RORγt). The differences of indicators in Treg cell transcription factor peripheral blood between the case group and the control group were compared, and the balance correlation between HMGB1 and Th17/Treg was analyzed. Results: Compared with the healthy control group, the proportion of Th17 cells and the expression level of HMGB1 and IL-17 in peripheral blood of ITP patients were increased (all P < 0.01), while the proportion of Treg cells and the level of TGF-β were decreased (all P < 0.01). The proportion of Th17 cells and the expression level of IL-17 and HMGB1 in peripheral blood of ITP patients were positively correlated with the concentration of HMGB1 (all P < 0.01); the proportion of Treg cells and the level of TGF-β were negatively correlated with the expression level of HMGB1 (all P < 0.01). In vitro experiments, the expression of intracellular RORγt mRNA was increased compared with the negative control group (1.50±0.24 vs. 0.93±0.22, t = 9.612, P < 0.01), and the expression of Foxp3 mRNA was decreased compared with the negative control group after the stimulation of PBMC by rhHMGB1 (0.72±0.19 vs. 1.08±0.18, t = 7.387, P < 0.01). Conclusion The high level of HMGB1 in the peripheral blood of ITP patients induces Th17/Treg imbalance and aggravates inflammatory reactions, which may be an important cause of ITP.

OBJECTIVETo explore the feasibility and application value of the preservation of vegetative nervous functions in radical resection for right-sided colon cancer.

METHODSClinical data of 55 cases with right-sided colon cancer undergoing laparoscopic D3+ complete mesocolic excision (CME) radical resection from January 2016 to July 2017 at Department of Gastrointestinal Surgery of Guangdong Province Hospital of Traditional Chinese Medicine were retrospectively analyzed. Exclusion criteria included emergency surgery for various reasons, intestinal obstruction or perforation, distant metastasis or locally advanced cancer, previous history of abdominal surgery and preoperative neoadjuvant chemoradiotherapy. Twenty-nine cases underwent lymphadenectomy with intrathecal dissection of superior mesenteric artery (SMA) and part of superior mesenteric plexus was resected (nerve partial resection group, NPR group). Twenty-six cases received lymphadenectomy with the clearance of lymphatic adipose tissue on the right side of SMA by sharp or obtuse method outside the sheath; the sheath of superior mesenteric vein (SMV) was entered at the junction of SMA and SMV; the SMV was naked in the sheath; the third station lymph node dissection was completed with preservation of superior mesenteric plexus (nerve preserved group, NP group). Intra-operative and postoperative complications were compared between two groups.

RESULTSThe baseline data were not significantly different between two groups (all P>0.05). The operation time in NP group was significantly shorter than that in NPR group [(164.0±19.8) minutes vs. (176.0±19.7) minutes, t=2.249, P=0.029]. No significant differences in operative blood loss, operative vessel damage, postoperative time to flatus, postoperative hospital stay and abdominal pain were observed between two groups(all P>0.05). The number of harvested lymph node in two groups was 28.5±7.8 and 27.6±6.5 respectively without significant difference(P>0.05). As compared to NPR group, NP group had lower incidence of chylous leakage[3.8%(1/26) vs. 37.9%(11/29), χ²=9.337, P=0.002] and postoperative diarrhea [15.4%(4/26) vs. 41.4%(12/29), χ²=4.491, P=0.034].

CONCLUSIONAutonomic nerve-preserving D3+ CME radical resection for right-sided colon cancer is safe and feasible, and can prevent the postoperative gastrointestinal dysfunction caused by nerve injury and decrease the risk of chylous leakage.

Autonomic Pathways ; surgery ; Colonic Neoplasms ; surgery ; Humans ; Laparoscopes ; Laparoscopy ; methods ; Lymph Node Excision ; Mesocolon ; surgery ; Retrospective Studies

OBJECTIVETo observe the clinical effects differences and partial mechanism for chronic nonbacterial prostatitis (CNP) among drug oil moxibustion, simple moxibustion, and conventional western medicine.

METHODSA total of 120 patients who met the criteria of inclusion were randomly assigned into a drug oil moxibustion group, a moxibustion group and a western medication group, 40 cases in each one. Moxibustion was used at Guanyuan (CV 4), Zhongji (CV 3), Qihai (CV 6) and bilateral Yinlingquan (SP 9), Sanyinjiao (SP 6), Shenshu (BL 23), Mingmen (GV 4), Pangguangshu (BL 28), Ciliao (BL 32), and Zhibian (BL 54), etc. The same moxibustion was used at the same acupoints in the drug oil moxibustion group after external application of medicated oil. Thirty min treatment was used once a day in alternated abdomen and back. In the western medication group, oral tamsulosin hydrochloride capsules were applied once a day, one capsule at a time. All the treatment was given for 30 days. Chronic prostatitis symptom index from National Institutes for Health (NIH-CPSI), the contents of Zinc (Zn) and C-reactive protein (CRP), as well as the number of white blood cells (WBC) and density of lecithin bodies were observed before and after treatment and 1 month after treatment. The effects were evaluated after treatment.

RESULTSAfter treatment, the total effective rate of the drug oil moxibustion group was 90.0% (36/40), which was significantly higher than 72.5% (29/40) of the moxibustion group and 62.5% (25/40) of the western medication group (both<0.05). After treatment and at follow-up in the three groups, the NIH-CPSI scores were lower than those before treatment (all<0.05), and those in the drug oil moxibustion group were lower than the results in the moxibustion group and the western medication group (all<0.05). The contents of Zn in the three groups were higher than those before treatment (all<0.05), with better results in the drug oil moxibustion group (all<0.05), and higher Zn contents in the moxibustion group compared with those in the western medication group (both<0.05). The CRP levels were lower than those before treatment (all<0.05), and those in the drug oil moxibustion group were better than those in the moxibustion group and western medication group (all<0.05). The CRP contents in the moxibustion group were lower than those in the western medication group (both<0.05). The number of WBC were lower than those before treatment (all<0.05), with better results in the drug oil moxibustion group (all<0.05). The concentrations of lecithin were higher than those before treatment (all<0.05), with better results in the drug oil moxibustion group (all<0.05).

CONCLUSIONSThe clinical effect of drug oil moxibustion is better than those of simple moxibustion and western medicine, which has advantages in improving clinical symptoms, Zn, the density of lecithin body and decreasing CRP content and the number of WBC.

Objective To evaluate the efficacy and safety of laparoscopic technique in the treatment of refractory adhesive intestinal obstruction. Methods In a retrospective matched?pair analysis, 68 patients of re?fractory adhesive intestinal obstruction whose relapsed more than 3 times were included in the research. All patients were treated with laparoscopic operation or continued follow?up observation after conservative treatment from Janu?ary 2011 to January 2016. The patients were divided into 2 groups, laparoscopic surgery (32 cases) and observation group (36 cases), to contrast the recurrence rate and safety of the two types of treatment. Results The recurrence rate of laparoscopic surgery group was significantly lower in the observation group (18.8%vs. 77.8%, P<0.05). Re?fractory adhesive intestinal obstruction is further divided into Mixed type, Patchy adhesion type and Cable Belt Com?pression type, in the laparoscopic surgery subgroup analysis, hybrid recurrence rate (66.7%, 4/6 cases) was signifi?cantly higher than that of patchy adhesion type (10.0%, 1/10 cases) and cord compression type (6.3%, 1/16 cases). Conclusion Laparoscopic operation is safe and feasible in the treatment of refractory adhesive intestinal obstruc?tion, and then it can effectively reduce the recurrence rate of refractory adhesive intestinal obstruction;the curative effect is closely related to the type of adhesion.

OBJECTIVETo investigate the safety and feasibility of laparoscopic radical right hemicolectomy using caudal-to-cranial approach (yellow-white borderline between right mesostenium and retroperitoneal is firstly cut as the entry to dissect the fusion fascial space between the visceral and parietal peritoneum, which is called caudal-to-cranial approach for right hemicolectomy).

METHODSFrom January 2014 to May 2015, 76 consecutive patients with right side colon cancer underwent laparoscopic radical right hemicolectomy using caudal-to-cranial approach. The baseline characteristics, intraoperative and postoperative outcomes were prospective collected and reviewed retrospectively.

RESULTSAll the 76 patients completed operations successfully, and one patient (1.3%) was converted to open surgery because of intraoperative bleeding due to unexpected injury of ileocolic artery. The mean operative time was (152.8±42.1) min with a mean estimated blood loss of (70.4±43.5) ml. The mean time of first flatus was (49.3±22.9) h and mean liquid oral intake was (58.5±17.6) h. The postoperative complications appeared in 7 patients (9.2%), including one (1.3%) of pulmonary infection, one(1.3%) of urinary system infection, two (2.6%) of wound infection, two (2.6%) of inflammatory bowel obstruction and one (1.3%) of lymphatic fistula, and they were all cured with conservative treatments. The postoperative hospital stay was (7.8±5.4) d. The mean number of harvested lymph node was 34.2±10.9, among which 4.1±2.8 was positive.

CONCLUSIONSLaparoscopic radical right hemicolectomy using caudal-to-cranial approach is safe and feasible.