龈沟液LTB4与hBD-3表达联合随机森林模型在种植体周围炎风险预测中的价值

doi:10.19439/j.sjos.2026.03.014

摘要/Abstract

摘要： 目的: 评价龈沟液（gingival crevicular fluid,GCF）中白三烯B4（leukotriene B4,LTB4）、人β防御素3（human beta-defensin-3,hBD-3）表达联合随机森林模型在种植修复患者维护期种植体周围炎（peri-implantitis,PI）风险预测中的价值。方法: 采取前瞻性研究,选择2023年3月—2025年3月收治的种植修复患者362例,根据维护期间发生种植体周围炎情况分为PI组（n=61）和非PI组（n=301）,比较两组基线资料,采用二元logistic回归分析种植修复患者维护期间发生PI的影响因素,应用R（R4.1.0）软件包及randomForest程序包构建随机森林模型。绘制受试者工作特征（receiver operating characteristic,ROC）曲线,计算曲线下面积（area under the curve,AUC）,检验随机森林模型对种植修复患者维护期间发生PI的预测价值。结果: PI组半胱天冬酶1（caspase-1）、白细胞介素1β（interleukin-1β,IL-1β）及LTB4水平显著高于非PI组,hBD-3水平显著低于非PI组（P<0.05）。二元logistic回归分析结果显示,Caspase-1、IL-1β、LTB4是种植修复患者维护期间发生PI的危险因素（OR>1,P<0.05）,hBD-3是保护因素（OR<1,P<0.05）。采用增加均方误差百分比（increase in mean squared error,%IncMse）打分并进行特征重要性排序,重要性前三位分别为LTB4、Caspase-1及IL-1β,%IncMse分别为29.372、27.488和23.123。随机森林模型P=0.01,362例患者中,317例预测正确,正确率为87.57%,随机森林在数量≥100时达到性能稳定期。基于测试集,以Caspase-1、IL-1β、LTB4和hBD-3建立随机森林模型,预测种植修复患者维护期间发生PI的AUC为0.841,敏感度为0.833,特异度为0.769,约登指数为0.603,95%CI为0.736~0.923,预测效能较好。结论: Caspase-1、IL-1β及LTB4水平是种植修复患者维护期间发生PI的危险因素,hBD-3水平是保护因素。随机森林模型分析表明,LTB4是预测PI的关键驱动因素,基于此构建的模型显示出良好的预测效能。

关键词: 种植修复, 龈沟液, 白三烯B4, 人β防御素3, 种植体周围炎

Abstract: PURPOSE: To evaluate the value of combining leukotriene B4 (LTB4) and human beta-defensin-3 (hBD-3) expression in gingival crevicular fluid (GCF) with a random forest model for predicting the risk of peri-implantitis (PI) during the maintenance period of implant restoration patients. METHODS: A prospective study was conducted, totally 362 patients who underwent implant restoration from March 2023 to March 2025 were selected. They were divided into PI group(n=61) and non PI group (n=301) according to the occurrence of peri-implantitis during the maintenance period. Baseline data were compared between the two groups, binary logistic regression was used to analyze the influencing factors of peri-implantitis during the maintenance of implant repair patients. The random forest model was constructed using R (R 4.1.0) software package and the randomForest package. The receiver operating characteristic (ROC) curve was drawn, and the area under the curve (AUC) was calculated to test the predictive value of the random forest model for peri-implantitis during the maintenance of patients with implant restoration. RESULTS: The levels of Caspase-1, IL-1β and LTB4 in the PI group were significantly higher than those in the non PI group, and the level of hBD-3 was significantly lower than that in the non PI group (OR>1, P<0.05). hBD-3 was a protective factor for PI during maintenance (OR<1, P<0.05). The variable importance measure (%IncMse) was used to score and rank the feature importance. The top three importance indicators were LTB4, Caspase-1 and IL-1β, respectively. The %IncMse was 29.372, 27.488 and 23.123, respectively. The random forest model was P=0.01. Among the 362 patients, totally 317 were correctly predicted, and the correct rate was 87.57%. The random forest reached a stable performance period when the number was≥100. Based on the test set, a random forest model was established with Caspase-1, IL-1β, LTB4 and hBD-3 to predict the occurrence of peri-implantitis during the maintenance of implant restoration patients. The AUC was 0.841, the sensitivity was 0.833, the specificity was 0.769, the Youden index was 0.603, and the 95%CI was 0.736-0.923, which had good predictive efficacy. CONCLUSIONS: Caspase-1, IL-1βand LTB4 levels are risk factors for peri-implantitis during the maintenance period in patients undergoing implant restoration, while hBD-3 level is a protective factor. Random forest model analysis indicates that LTB4 is a key driver in predicting PI, and the model constructed based on these results demonstrates good predictive performance.

Key words: Implant restoration, Gingival crevicular fluid, Leukotriene B4, Human β-defensin-3, Peri-implantitis

中图分类号:

R781.4

吴燕, 杨建新, 唐曌隆. 龈沟液LTB4与hBD-3表达联合随机森林模型在种植体周围炎风险预测中的价值[J]. 上海口腔医学, 2026, 35(3): 305-311.

Wu Yan, Yang Jianxin, Tang Zhaolong. Value of LTB4 and hBD-3 expression in gingival crevicular fluid combined with random forest model for predicting the risk of peri-implantitis during the maintenance period of implant restoration patients[J]. Shanghai Journal of Stomatology, 2026, 35(3): 305-311.

参考文献

[1] 张云霞, 杨娜, 姚莉, 等. TSLP, HIF-1α, RANKL在义齿修复后种植体周围炎患者龈沟液中的表达及意义[J]. 临床和实验医学杂志, 2024, 23(15): 1656-1659.
[2] Darby I.Risk factors for periodontitis & peri-implantitis[J]. Periodontol 2000, 2022, 90(1): 9-12.
[3] Fatima T, Khurshid Z, Rehman A, et al.Gingival crevicular fluid (GCF): a diagnostic tool for the detection of periodontal health and diseases[J]. Molecules, 2021, 26(5): 1208.
[4] Li KX, Zheng LX, Guo HQ, et al.LTB4-induced anti-apoptosis and infiltration of neutrophils in rheumatoid arthritis[J]. Clin Exp Rheumatol, 2020, 38(3): 543-551.
[5] Gürsoy UK, Salli K, Söderling E, et al.Regulation of hBD-2, hBD-3, hCAP18/LL37, and proinflammatory cytokine secretion by human milk oligosaccharides in an organotypic oral mucosal model[J]. Pathogens, 2021, 10(6): 739.
[6] Englezos E, Cosyn J, Koole S, et al.Resective treatment of peri-implantitis: clinical and radiographic outcomes after 2 years[J]. Int J Periodontics Restorative Dent, 2018, 38(5): 729-735.
[7] Roccuzzo M, Mirra D, Roccuzzo A.Surgical treatment of peri-implantitis[J]. Br Dent J, 2024, 236(10): 803-808.
[8] 谢雪梅, 张佳园, 张野, 等. 龈沟液中sICAM-1, IL-1β和HIF-1α水平与义齿种植修复并发种植体周围炎的关系[J]. 上海口腔医学, 2023, 32(1): 47-51.
[9] Tian L, Wu W, Yu T.Graph random forest: a graph embedded algorithm for identifying highly connected important features[J]. Biomolecules, 2023, 13(7): 1153.
[10] Chaar EE, Rutkowski J.Periodontal and peri-implant oral diseases as a public health crisis[J]. J Oral Implantol, 2024, 50(4): 297-298.
[11] Jervøe-Storm PM, Bunke J, Worthington HV, et al. Adjunctive antimicrobial photodynamic therapy for treating periodontal and peri-implant diseases[J]. Cochrane Database Syst Rev, 2024, 7(7): CD011778.
[12] Padial-Molina M, Montalvo-Acosta S, Martín-Morales N, et al.Correlation between inflammasomes and microbiota in peri-implantitis[J]. Int J Mol Sci, 2024, 25(2): 961.
[13] Du X, Li B, Cai Q, et al.D-aspartic acid protects against gingival fibroblasts inflammation by suppressing pyroptosis[J]. Mol Biol Rep, 2022, 49(7): 5821-5829.
[14] Zhang C, Jiao B, Qian D.Protective effect of resveratrol against peri-implant inflammation and oxidative stress: an in vivo study[J]. Discov Med, 2025, 37(196): 862-870.
[15] Zhao S, Ge C, Li Y, et al.Desferrioxamine alleviates UHMWPE particle-induced osteoclastic osteolysis by inhibiting caspase-1-dependent pyroptosis in osteocytes[J]. J Biol Eng, 2022, 16(1): 34.
[16] Syngouna SA, Mitsikostas P, Sioutis S, et al.Diagnostic value of IL-1β, IL-17A, and IL-17F serum levels in patients with upper extremity infections[J]. J Long Term Eff Med Implants, 2021, 31(4): 39-44.
[17] Mehrabi A, Negahdari R, Parnia F, et al.Effect of tetracycline on IL-1β and IL-6 levels of the peri-implant sulcular fluid[J]. J Adv Periodontol Implant Dent, 2021, 13(2): 56-60.
[18] Guarnieri R, Reda R, Di Nardo D, et al.Expression of IL-1β, IL-6, TNF-α, and a-MMP-8 in sites with healthy conditions and with periodontal and peri-implant diseases: a case-control study[J]. J Dent Res Dent Clin Dent Prospects, 2024, 18(2): 135-142.
[19] Lashkarizadeh N, Foroudisefat M, Abyari S, et al.Is it safe to reuse healing abutments? an experimental study on IL-1β and TNF-α cytokine levels in peri-implant crevicular fluid[J]. J Prosthodont, 2022, 31(5): 399-404.
[20] Wei C, Xu Y, Zheng Y, et al.The LTB4-BLT1 axis attenuates influenza-induced lung inflammation by suppressing NLRP3 activation[J]. Cell Death Discov, 2025, 11(1): 148-160.
[21] Liu M, Yan M, He J, et al.Macrophage MST1/2 disruption impairs post-infarction cardiac repair via LTB4[J]. Circ Res, 2021, 129(10): 909-926.
[22] 辜义陆, 缪克红, 杨凤, 等. 血清CX3CL1, LTB4水平与牙列缺失患者口腔种植修复预后的关系[J]. 检验医学与临床, 2024, 21(16): 2396-2399.
[23] Yokomizo T, Shimizu T.The leukotriene B4 receptors BLT1 and BLT2 as potential therapeutic targets[J]. Immunol Rev, 2023, 317(1): 30-41.
[24] Chessa C, Bodet C, Jousselin C, et al.Antiviral effect of hBD-3 and ll-37 during human primary keratinocyte infection with west nile virus[J]. Viruses, 2022, 14(7): 1552.
[25] Özdemir M, Caglayan F, Bikker FJ, et al.Gingival tissue human beta-defensin levels in relation to infection and inflammation[J]. J Clin Periodontol, 2020, 47(3): 309-318.
[26] Peng G, Tsukamoto S, Ikutama R, et al.Human β-defensin-3 attenuates atopic dermatitis-like inflammation through autophagy activation and the aryl hydrocarbon receptor signaling pathway[J]. J Clin Invest, 2022, 132(17): e156501.
[27] Gürsoy UK, Salli K, Söderling E, et al.Regulation of hBD-2, hBD-3, hCAP18/LL37, and proinflammatory cytokine secretion by human milk oligosaccharides in an organotypic oral mucosal model[J]. Pathogens, 2021, 10(6): 739.