血清CC趋化因子配体20、Akt激酶联合超声内镜和增强CT诊断胃黏膜下肿物及侵袭风险分级的应用价值

Application value of serum CCL20, Akt kinase combined with endoscopic ultrasound and enhanced CT for the diagnosis of submucosal gastric tumors and the classification of invasion risk

摘要:

目的 探讨血清CC趋化因子配体20（CCL20）、Akt激酶联合超声内镜和增强CT在胃黏膜下肿物（GSMT）诊断和侵袭危险度分级评估中的应用价值。方法 回顾性分析2022年3月－2024年6月在该院经白光内镜检查发现，并经术后病理检查证实的128例GSMT患者的临床资料。所有患者术前均进行了超声内镜、CT平扫和增强CT检查，并检测血清CCL20和Akt激酶。根据美国国立卫生研究院关于胃肠道间质瘤侵袭危险分级标准，将76例胃间质瘤患者纳入极低危组（36例）、低危组（19例）和中高危组（21例）。比较3组超声内镜和CT下表现、血清CCL20和Akt激酶水平，将差异有统计学意义的自变量纳入有序分类多因素Logistic回归模型，分析胃肠道间质瘤侵袭危险度分级的独立危险因素，并构建加权预测模型，通过受试者操作特征曲线（ROC curve），评估模型效能。结果 超声内镜诊断GSMT和胃间质瘤的符合率分别为79.69%和97.37%，明显高于增强CT的67.19%和84.21%，差异均有统计学意义（P < 0.05）。极低危组和低危组超声内镜特征中的回声均匀和边界清晰占比，以及增强CT特征中的瘤周脂肪间隙清晰、边界清晰、均匀强化和密度均匀占比，明显高于中高危组，差异均有统计学意义（P < 0.0167）；极低危组和低危组超声内镜特征中的表面破溃占比明显低于中高危组，差异有统计学意义（P < 0.0167）；极低危组超声内镜和增强CT特征中的边缘规则占比，明显高于中高危组，增强CT特征中的肿瘤血管显示占比明显低于中高危组，差异均有统计学意义（P < 0.0167）。胃间质瘤组血清CCL20和Akt激酶水平明显高于非胃间质瘤组，差异均有统计学意义（P < 0.05）。胃间质瘤中高危组血清CCL20和Akt激酶水平明显高于低危组和极低危组，低危组血清CCL20和Akt激酶水平明显高于极低危组，差异均有统计学意义（P < 0.05）。有序分类多因素Logistic回归分析显示，超声内镜回声不均匀、超声内镜表面破溃、增强CT强化不均匀、血清CCL20 > 30.59 pg/mL和Akt激酶 > 2.08 ng/mL是胃间质瘤侵袭危险度分级的独立危险因素（P < 0.05）。基于上述5个独立危险因素构建胃间质瘤侵袭风险概率预测模型：logit（P） = -5.215 + 超声内镜回声不均匀×1.140 + 超声内镜表面破溃×1.267 + 增强CT强化不均匀×1.056 + 血清CCL20 > 30.59 pg/mL×0.258 + 血清Akt > 2.08 ng/mL×1.396。ROC curve分析显示，该模型预测中高危胃间质瘤的曲线下面积（AUC）为0.943（95%CI：0.893~0.992），采用最大约登指数确定最佳临界值，当logit（P）≥-2.358时，预测为中高危胃间质瘤，此时的约登指数为0.873，敏感度为100.00%，特异度为87.27%。结论 超声内镜诊断GSMT的符合率高于增强CT；基于超声内镜回声不均匀、超声内镜表面破溃、增强CT强化不均匀、血清CCL20和Akt激酶构建的多因素Logistic回归预测模型，能够有效评估胃间质瘤的侵袭风险概率，对中高危胃间质瘤具有较高的预测价值，可为术前评估和治疗方案的制定，提供量化参考。

Abstract:

Objective To explore the application value of serum CC chemokine ligand 20 (CCL20), Akt kinase combined with endoscopic ultrasound and enhanced CT in the diagnosis of submucosal gastric tumors (GSMT) and the assessment of invasion risk classification.Methods A retrospective analysis was conducted on the clinical data of 128 patients with GSMT who were identified by white light endoscopy and confirmed by postoperative pathological examination from March 2022 to June 2024. All patients underwent endoscopic ultrasound examination, plain CT scan and enhanced CT before the operation, and serum CCL20 and Akt were detected. According to the risk classification criteria for invasion of gastrointestinal stromal tumors by the National Institutes of Health of the United States, 76 patients with gastric stromal tumors were included in the very low-risk group (36 cases), the low-risk group (19 cases), and the medium-high-risk group (21 cases). The endoscopic ultrasound and CT features, serum CCL20 and Akt levels of the three groups were compared. The independent variables with statistically significant differences were included in the ordered classification multivariate Logistic regression model. The independent risk factors for the invasion risk classification of gastrointestinal stromal tumors were analyzed, and a weighted prediction model was constructed. Through the receiver operating characteristic curve (ROC curve) , evaluate the effectiveness of the model.Results The consistency rates of endoscopic ultrasound in diagnosing GSMT and gastric stromal tumors were 79.69% and 97.37% respectively, which were significantly higher than 67.19% and 84.21% of enhanced CT, and the differences were statistically significant (P < 0.05). The proportions of uniform echo and clear boundary in endoscopic ultrasound, as well as the proportions of clear peritumoral fat space, clear boundary, uniform enhancement and uniform density in enhanced CT in the very low-risk group and the low-risk group were significantly higher than those in the medium-high-risk group, and the differences were statistically significant (P < 0.0167). The proportion of endoscopic ultrasound of surface ulceration in the very low-risk group and the low-risk group was significantly lower than that in the medium-high-risk group, and the difference was statistically significant (P < 0.0167). The proportion of regular margins of endoscopic ultrasound and CT in the very low-risk group was significantly higher than that in the medium-high-risk group, while the proportion of tumor vascular display in enhanced CT was significantly lower than that in the medium-high-risk group. The differences were statistically significant (P < 0.0167). The levels of serum CCL20 and Akt kinase in the gastric stromal tumor group were significantly higher than those in the non-gastric stromal tumor group, and the differences were statistically significant (P < 0.05). The levels of serum CCL20 and Akt kinase in the medium-high-risk group of gastric stromal tumors were significantly higher than those in the low-risk group and the very low-risk group. The levels of serum CCL20 and Akt kinase in the low-risk group were significantly higher than those in the very low-risk group, and the differences were statistically significant (P < 0.05). Ordered classification multivariate Logistic regression analysis showed that non-uniform echo of endoscopic ultrasound, surface ulceration of endoscopic ultrasound, non-uniform enhancement of enhanced CT, serum CCL20 > 30.59 pg/mL and Akt kinase > 2.08 ng/mL were independent risk factors for the risk classification of gastric stromal tumor invasion (P < 0.05). Based on the above five independent risk factors, the invasion risk probability prediction model of gastric stromal tumor was constructed: logit (P) = -5.215 + endoscopic ultrasound non-uniform × 1.140 + endoscopic ultrasound surface ulceration ×1.267 + enhanced CT non-uniform enhancement × 1.056 + serum CCL20 > 30.59 pg/mL × 0.258 + serum Akt > 2.08 ng/mL × 1.396. ROC curve analysis showed that the area under the curve (AUC) of this model for predicting medium and high-risk gastric stromal tumors was 0.943 (95%CI: 0.893 ~ 0.992), when logit (P) ≥ -2.358 was used to predict intermediate-high risk gastric stromal tumors, the Youden index was 0.873, the sensitivity was 100.00%, and the specificity was 87.27%.Conclusion The consistency rate of endoscopic ultrasound in diagnosing GSMT is higher than that of enhanced CT. The multivarate Logistic regression prediction model based on the non-uniform echo of endoscopic ultrasound, surface ulceration of endoscopic ultrasound, non-uniform enhancement of enhanced CT, serum CCL20 and Akt kinase can effectively evaluate the invasion risk probability of gastric stromal tumors, and has high predictive value for intermediate and high-risk gastric stromal tumors, and can provide quantitative references for preoperative assessment and treatment plan formulation.