周華 李慧國 李君 晉志偉 項(xiàng)歡
[摘要]目的:探討腮腺良性腫瘤切除術(shù)后唾液瘺形成的危險(xiǎn)因素,并構(gòu)建預(yù)測模型計(jì)算預(yù)測模型的預(yù)測效能。方法:選取2015年1月-2019年12月于筆者醫(yī)院就診的80例腮腺良性腫瘤患者為研究對象,所有患者均擬行腮腺良性腫瘤切除術(shù),術(shù)后隨訪4周,根據(jù)患者術(shù)后是否發(fā)生唾液瘺分為唾液瘺組和對照組,比較兩組患者臨床資料,采用多因素Logistic回歸方程分析腮腺良性腫瘤切除術(shù)后唾液瘺形成的危險(xiǎn)因素;構(gòu)建邏輯回歸預(yù)測模型,分析該模型預(yù)測腮腺良性腫瘤切除術(shù)后唾液瘺形成風(fēng)險(xiǎn)的預(yù)測價(jià)值。結(jié)果:納入的80例患者未出現(xiàn)失訪病例,均完成隨訪。其中25例患者出現(xiàn)唾液瘺為唾液瘺組,剩余55例未發(fā)生唾液瘺的患者為對照組。對比兩組患者臨床資料后顯示,兩組患者在切除范圍、術(shù)后壓迫時(shí)間、傷口淀粉酶水平方面差異有統(tǒng)計(jì)學(xué)意義(P<0.05);多因素Logistic回歸分析顯示,傷口淀粉酶水平(OR=0.999)、切除范圍(OR=3.790)、術(shù)后壓迫時(shí)間(OR=1.450)是腮腺良性腫瘤切除術(shù)后唾液瘺形成的危險(xiǎn)因素;使用切除范圍、術(shù)后壓迫時(shí)間及傷口淀粉酶水平構(gòu)建邏輯回歸預(yù)測模型,將預(yù)測模型帶入ROC曲線分析后顯示:AUC為0.921,敏感性為90.91%,特異度為84.00%(P<0.05)。結(jié)論:切除范圍、術(shù)后壓迫時(shí)間、傷口淀粉酶水平是腮腺良性腫瘤切除術(shù)后唾液瘺形成的危險(xiǎn)因素,建立了預(yù)測腮腺良性腫瘤切除術(shù)后唾液瘺形成風(fēng)險(xiǎn)的模型,并進(jìn)行了預(yù)測分析,得出了該邏輯回歸模型能夠早期預(yù)測腮腺良性腫瘤切除術(shù)后唾液瘺風(fēng)險(xiǎn)的結(jié)論,這為臨床腮腺良性腫瘤切除術(shù)后唾液瘺的預(yù)防提供了很好的參考。
[關(guān)鍵詞]腮腺良性腫瘤;切除術(shù);唾液瘺;危險(xiǎn)因素;預(yù)測模型
[中圖分類號]R739.81? ? [文獻(xiàn)標(biāo)志碼]A? ? [文章編號]1008-6455(2022)05-0110-04
Predictive Model Analysis of the Risk of Salivary Fistula after Resection of Benign Parotid Tumor
ZHOU Hua, LI Huiguo, LI Jun, JIN Zhiwei, XIANG Huan
[Department of Stomatology, Wuhu Hospital Affiliated to East China Normal University (Wuhu Second People's Hospital), Wuhu 241000, Anhui, China]
Abstract: Objective? To investigate the risk factors of salivary fistula after resection of benign parotid tumors and construct a predictive model to calculate the predictive performance of the predictive model. Methods? A total of 80 patients with benign parotid gland tumors who were treated in our hospital from January 2015 to December 2019 were selected as the research subjects. All patients were planned to undergo resection of benign parotid tumors. The postoperative follow-up was 4 weeks, according to whether the patients occurred after surgery. Salivary fistulas were divided into salivary fistula group and control group. The clinical data of the two groups were compared. The multivariate logistic regression equation was used to analyze the risk factors of salivary fistula after parotid benign tumor resection; a logistic regression prediction model was constructed and analyzed to predict the benign parotid gland. The predictive value of the risk of salivary fistula formation after tumor resection. Results The 80 included cases that did not appear to be lost to follow-up were followed up. Among them, 25 patients with salivary fistula were in the salivary fistula group, and the remaining 55 patients without salivary fistula were in the control group. Comparing the clinical data of the two groups of patients showed that there were statistical differences in the range of resection, postoperative compression time, and wound amylase levels between the two groups (P<0.05), multivariate Logistic regression analysis showed that wound amylase levels (OR=0.999), resection range (OR=3.790), postoperative compression time (OR=1.450) are the results of parotid benign tumor resection Risk factors for the formation of salivary fistula; use the range of resection, postoperative compression time, and wound amylase level to construct a logistic regression prediction model. After the prediction model is incorporated into the ROC curve analysis, it shows that the AUC is 0.921, the sensitivity is 90.91%, and the specificity is 84.00% (P<0.05). Conclusion Resection range, postoperative compression time, and wound amylase levels are risk factors for salivary fistula after parotid benign tumor resection. A model for predicting the risk of salivary fistula after parotid benign tumor resection was established and analyzed. It is concluded that the logistic regression model can predict the risk of salivary fistula after parotid benign tumor resection early, which provides a good reference for the prevention of salivary fistula after clinical parotid benign tumor resection.
Key words: benign tumor of parotid gland; resection; salivary fistula; risk factors; prediction model
在口腔頜面部腫瘤中腮腺腫瘤的發(fā)生率約80%,通常發(fā)生在腮腺淺葉,多為良性[1]。表現(xiàn)為緩慢生長的無痛性包塊,與周圍組織界限清楚,治療以手術(shù)徹底切除為原則。但部分患者術(shù)后出現(xiàn)涎液流至面頰部,內(nèi)涎瘺涎液流入口腔的唾液瘺,臨床主要表現(xiàn)為腮腺區(qū)皮膚可見點(diǎn)狀瘺孔,伴少量透明液體流出,進(jìn)食時(shí)增多,當(dāng)管瘺靠前在頰肌部或靠后位于嚼肌部時(shí)可有透明或混濁液體自瘺孔外流至面頰[2]。隨著唾液瘺的發(fā)生發(fā)展,可增加傷口感染及瘢痕的風(fēng)險(xiǎn),還可引發(fā)面部自限性腫脹影響進(jìn)食、說話和容貌美觀。腮腺良性腫瘤切除術(shù)后唾液瘺形成的高危因素諸多,有研究證實(shí)[3]腮腺良性腫瘤切除術(shù)后唾液瘺形成與切除范圍相關(guān),可作為腮腺良性腫瘤切除術(shù)后唾液瘺形成的預(yù)測因子,術(shù)后可通過加壓壓迫預(yù)防唾液瘺的發(fā)生,但長期臨床調(diào)查顯示[4],腮腺良性腫瘤切除術(shù)后唾液瘺的發(fā)生率卻一直在升高。推測還存在其他腮腺良性腫瘤切除術(shù)后唾液瘺形成的高危因素,然而當(dāng)前關(guān)于腮腺良性腫瘤切除術(shù)后唾液瘺形成的高險(xiǎn)因素尚缺乏有力數(shù)據(jù)支持。鑒于此,本研究通過Logistic回歸分析其發(fā)生的危險(xiǎn)因素,并建立預(yù)測模型評估其預(yù)測效能。
1? 資料和方法
1.1 研究對象:選取2015年1月-2019年12月筆者醫(yī)院就診的80例腮腺良性腫瘤患者為研究對象,以隨訪是否發(fā)生唾液瘺進(jìn)行分組,其中25例患者出現(xiàn)唾液瘺為唾液瘺組,剩余55例未發(fā)生唾液瘺的患者為對照組。
1.2 納入標(biāo)準(zhǔn):①年齡21~72歲;②經(jīng)病理組織學(xué)檢查證實(shí)為良性病變[5];③首次行腮腺腫瘤切除術(shù);④簽署知情同意書。
1.3 排除標(biāo)準(zhǔn):①近期接受過激素、輸血等治療者;②合并肝、腎等嚴(yán)重疾病或其他惡性腫瘤、全身感染者;③合并胰腺炎、胰腺腫瘤等疾病者;④臨床資料不全者。
1.4 臨床資料收集:于患者入院時(shí)收集患者人口學(xué)資料,包括性別、年齡、體重及身高等,病理學(xué)指標(biāo)包括腫瘤類型、位置及大小,記錄手術(shù)切除范圍,翻瓣方式及術(shù)后壓迫時(shí)間,使用酶比色法采用羅氏Cobas c 701/702全自動生化分析儀測量患者傷口淀粉酶水平。
1.5 隨訪:兩組患者術(shù)后隨訪4周,記錄兩組發(fā)生唾液瘺的情況。將唾液瘺定義為涎液自腺體外滲或涎液自主導(dǎo)管處外滲。
1.6 預(yù)測變量篩選:將收集的變量采用單變量方差分析,篩選出具統(tǒng)計(jì)學(xué)意義的變量,再使用Logistic回歸分析有統(tǒng)計(jì)學(xué)意義的變量進(jìn)行變量篩選。
1.7 邏輯回歸預(yù)測模型的構(gòu)建:本研究限于樣本量,不按比例單獨(dú)劃分訓(xùn)練集和測試集,而采用十折交叉驗(yàn)證方法建立模型,即每次將數(shù)據(jù)隨機(jī)分為10份,其中9份作為訓(xùn)練集,1份作為測試集,該過程重復(fù)10次,以防止“過擬合”現(xiàn)象的出現(xiàn)。根據(jù)多因素Logistic回歸分析得出腮腺良性腫瘤切除術(shù)后唾液瘺形成的獨(dú)立危險(xiǎn)因素,通過建立邏輯回歸預(yù)測模型,使用Medcalc軟件計(jì)算預(yù)測模型數(shù)據(jù),帶入ROC曲線進(jìn)行效能分析。
1.8 統(tǒng)計(jì)學(xué)分析:采用SPSS 20.0軟件進(jìn)行數(shù)據(jù)處理,符合正態(tài)分布的計(jì)量資料采用均數(shù)±標(biāo)準(zhǔn)差表示,組間比較采用t檢驗(yàn);不符合正態(tài)分布的計(jì)量資料采用中位數(shù)(取值范圍)表示,組間比較采用秩和檢驗(yàn);計(jì)數(shù)資料采用例/百分比表示,組間比較采用卡方檢驗(yàn);組外驗(yàn)證使用 k 折交叉驗(yàn)證,k=10;多因素分析采用Logistic回歸方程;通過邏輯回歸建立預(yù)測模型,繪制受試者工作特征曲線,分析邏輯回歸預(yù)測模型的診斷效能,以P<0.05表示差異有統(tǒng)計(jì)學(xué)意義。
2? 結(jié)果
2.1 兩組患者臨床資料比較:對比兩組患者臨床資料后顯示,兩組患者在切除范圍、術(shù)后壓迫時(shí)間及傷口淀粉酶水平方面比較,差異有統(tǒng)計(jì)學(xué)意義(P<0.05),見表1。
2.2 腮腺良性腫瘤切除術(shù)后唾液瘺形成的危險(xiǎn)因素分析:以是否發(fā)生唾液瘺為因變量,傷口淀粉酶水平(OR=0.999,95%CI=0.999~1.000)、切除范圍(OR=3.790,95%CI=1.139~12.616)及術(shù)后壓迫時(shí)間(OR=1.450,95%CI=1.040~2.020)是腮腺良性腫瘤切除術(shù)后唾液瘺形成的危險(xiǎn)因素,見表2。
2.3 邏輯回歸預(yù)測模型在預(yù)測腮腺良性腫瘤切除術(shù)后唾液瘺形成的ROC曲線與診斷效能:將邏輯回歸預(yù)測模型帶入患者工作特征曲線(ROC)進(jìn)行分析后顯示邏輯回歸預(yù)測模型在預(yù)測腮腺良性腫瘤切除術(shù)后唾液瘺形成的AUC為0.921,敏感性為90.91%,特異度為84.00%。見圖1,表3。
3? 討論
腮腺良性腫瘤多為混合瘤,多發(fā)生在淺表腮腺葉中,有一定惡變風(fēng)險(xiǎn),手術(shù)切除病變組織是治療的首選,主要包括部分或完全的腮腺淺表切除術(shù),但術(shù)后部分患者并發(fā)唾液瘺。唾液瘺通常被認(rèn)為是唾液產(chǎn)生的薄壁組織不斷分泌液體的結(jié)果,絕大多數(shù)在術(shù)后1個月內(nèi)出現(xiàn),常在進(jìn)食期間發(fā)生,傷口完全愈合后,部分瘺管可持續(xù)較長時(shí)間,給患者帶來功能和美學(xué)不適[6-7]。有效掌握腮腺良性腫瘤切除術(shù)后唾液瘺形成的高危因素,早期預(yù)測唾液瘺的發(fā)生,并采取早期防治措施是預(yù)防腮腺良性腫瘤切除術(shù)后唾液瘺形成的關(guān)鍵。有學(xué)者認(rèn)為[8]手術(shù)切除范圍是腮腺良性腫瘤切除術(shù)后唾液瘺形成的高危因素,但有文獻(xiàn)報(bào)道[9],部分和完全的腮腺淺表切除術(shù)術(shù)后患者唾液瘺發(fā)生率無顯著差異。表明腮腺良性腫瘤切除術(shù)后還存在其他唾液瘺形成的高危因素。據(jù)報(bào)道[10],淀粉酶濃度是胰腺切除術(shù)后胰瘺發(fā)展的有效預(yù)測指標(biāo)。唾液中主要的有機(jī)物為唾液淀粉酶等,同時(shí)也廣泛存在于胰腺中,因此,推測淀粉酶濃度與唾液瘺管形成之間存在關(guān)聯(lián)。鑒于此,本研究納入傷口淀粉酶水平建立邏輯回歸預(yù)測模型并探討其預(yù)測效能。
本研究結(jié)果顯示術(shù)后80例患者中,25例出現(xiàn)唾液瘺,對比出現(xiàn)唾液瘺患者和未出現(xiàn)唾液瘺患者的臨床資料顯示,切除范圍、術(shù)后壓迫時(shí)間及傷口淀粉酶水平均存在差異,多因素Logistic回歸分析顯示,切除范圍、術(shù)后壓迫時(shí)間及傷口淀粉酶水平是腮腺良性腫瘤切除術(shù)后唾液瘺形成的獨(dú)立危險(xiǎn)因素。分析認(rèn)為:①切除范圍:完全的腮腺淺表切除術(shù)需將腫瘤以及腮腺淺葉組織切除,手術(shù)切除范圍大,切口較長,手術(shù)解剖較為復(fù)雜,術(shù)中需解剖、暴露面神經(jīng),增大對面神經(jīng)的損傷;其次術(shù)中需結(jié)扎腮腺導(dǎo)管,增大術(shù)后出現(xiàn)唾液瘺的發(fā)生。部分腮腺淺表切除術(shù)在手術(shù)切除過程中僅將腮腺腫瘤及周圍0.5~1.0 cm的正常腺體切除,切除的正常腺體小,盡可能保留了腮腺淺葉組織、主導(dǎo)管及分支導(dǎo)管,較為完整地保留了腮腺功能,降低唾液瘺的發(fā)生。在Daniel[11]的研究中指出,區(qū)域性腮腺淺表切除術(shù)術(shù)后涎腺瘺的發(fā)生率顯著低于傳統(tǒng)完全切除術(shù);②術(shù)后壓迫時(shí)間:腮腺良性腫瘤切除術(shù)后唾液瘺發(fā)生的主要原因?yàn)橄岩和ㄟ^腮腺創(chuàng)面滲入切口下方形成,術(shù)后壓迫是一種通過正壓讓皮瓣緊貼腮腺創(chuàng)面,達(dá)到“不透水”的效果[12-13]。在壓迫過程中皮瓣和深層組織緊密貼附,使肌組織擠入死腔嚴(yán)密覆蓋腮腺創(chuàng)面,減少因積血、積液引起的感染。消滅創(chuàng)面死腔的同時(shí)可壓迫殘余腺體組織使腺體萎縮,降低分泌功能,減少唾液瘺的發(fā)生;③傷口淀粉酶水平:淀粉酶是一種由腮腺及胰腺分泌的消化酶,在血清中含量較少,在唾液中含量較多[14-15]。當(dāng)唾液腺組織被損傷時(shí),淀粉酶大量入血,血淀粉酶增高,在腮腺良性腫瘤切除術(shù)后的傷口中淀粉酶主要來源于血清。傷口淀粉酶濃度過高損傷新生組織的生成,抑制創(chuàng)口愈合。而唾液瘺漏出的唾液由于腮腺組織完整性破壞而導(dǎo)致液體滯留,尤其是主導(dǎo)管或葉間導(dǎo)管的不閉合,有研究證實(shí)[16],在早期瘺形成過程中淀粉濃度呈上升趨勢,證實(shí)血清淀粉酶水平升高表明腮腺或?qū)Ч軗p傷,增加術(shù)后唾液瘺形成的風(fēng)險(xiǎn)。
邏輯回歸模型是一種預(yù)測性的建模技術(shù),它研究的是因變量和自變量之間的關(guān)系,通常用于預(yù)測分析及發(fā)現(xiàn)變量之間的因果關(guān)系。本研究為探討腮腺良性腫瘤切除術(shù)后唾液瘺形成,將切除范圍、術(shù)后壓迫時(shí)間及傷口淀粉酶水平通過邏輯回歸方式建立早期預(yù)測模型,經(jīng)進(jìn)一步分析顯示AUC為0.921,敏感性為90.91%,特異度為84.00%,進(jìn)一步表明邏輯回歸預(yù)測模型在早期預(yù)測腮腺良性腫瘤切除術(shù)后唾液瘺形成方面具有良好的診斷效能。這為一些腮腺良性腫瘤切除術(shù)術(shù)前制定手術(shù)方案提供了很好參考。切除范圍越大對患者創(chuàng)傷越大、腮腺功能及結(jié)構(gòu)損傷越大,而傷口淀粉酶水平過高可影響傷口愈合,同時(shí)術(shù)后壓迫是減低患者術(shù)后出血及促進(jìn)恢復(fù)的手段,壓迫時(shí)間過短可能導(dǎo)致皮瓣與深層組織貼附不緊密,腺體萎縮不徹底,增加唾液瘺形成風(fēng)險(xiǎn)。三者聯(lián)合建立預(yù)測模型更具臨床優(yōu)勢,能為臨床腮腺良性腫瘤切除術(shù)后唾液瘺的預(yù)防提供很好參考。
綜上所述,本研究建立了預(yù)測腮腺良性腫瘤切除術(shù)后唾液瘺形成風(fēng)險(xiǎn)的模型并進(jìn)行了預(yù)測分析,得出該邏輯回歸模型能夠早期預(yù)測腮腺良性腫瘤切除術(shù)后唾液瘺風(fēng)險(xiǎn)的結(jié)論,這為臨床腮腺良性腫瘤切除術(shù)后唾液瘺的預(yù)防提供了很好參考;但限于本研究樣本量較少,科學(xué)性有限,也未納入諸如炎癥因子、血管生長因子等進(jìn)行探討,研究結(jié)果可能存在偏差,后期還需增大樣本量,納入炎癥因子等進(jìn)一步驗(yàn)證該結(jié)論。
[參考文獻(xiàn)]
[1]余和東.腮腺腫瘤手術(shù)的演變及術(shù)后并發(fā)癥預(yù)防的研究進(jìn)展[J].西部醫(yī)學(xué),2018,30(9):1401-1404.
[2]Costa M G E S T D,Maranh?o-Filho P A,Santos I C,et al.Parotidectomy-related facial nerve lesions: proposal for a modified sunnybrook facial grading system[J].Arq Neuropsiquiatr,2019,77(7):460-469.
[3]張敏,賈志宇,劉樹妍,等.腮腺良性腫瘤淺葉部分切除術(shù)與淺葉切除術(shù)的循證醫(yī)學(xué)分析[J].臨床耳鼻咽喉頭頸外科雜志,2019,33(9):875-882.
[4]Bonavolontà P,Dell'Aversana Orabona G,Maglitto F,et al.Postoperative complications after removal of pleomorphic adenoma from the parotid gland: A long-term follow up of 297 patients from 2002 to 2016 and a review of publications[J].Br J Oral Maxillofac Surg,2019,57(10):998-1002.
[5]Britt CJ,Stein A P,Gessert T,et al.Factors influencing sialocele or salivary fistula formation postparotidectomy[J].Head Neck,2017,39(2):?387-391.
[6]Fiacchini G,Cerchiai N, Tricò D, et al.Frey syndrome, first bite syndrome,great auricular nerve morbidity,and quality of life following parotidectomy[J].Eur Arch Otorhinolaryngol,2018,275(7):?1893-1902.
[7]李浩,吳坡,蔣佶,等.面神經(jīng)中段解剖法在區(qū)域性腮腺切除術(shù)中的應(yīng)用[J].華西口腔醫(yī)學(xué)雜志,2020,38(1):37-41.
[8]Mantsopoulos K,Goncalves M,Iro H.Transdermal scopolamine for the prevention of a salivary fistula after parotidectomy[J].Br J Oral Maxillofac Surg,2018,56(3):212-215.
[9]Zou H W,Li W G,Huang S Y,et al.New method to prevent salivary fistula after parotidectomy[J].Br J Oral Maxillofac Surg,2019,57(8):801-802.
[10]朱軍利,夏琪,馬浩.69例腮腺腫瘤術(shù)后并發(fā)癥的危險(xiǎn)因素分析[J].實(shí)用癌癥雜志,2019,34(7):1144-1146.
[11]Daniel F,Tamim H,Hosni M,et al.Validation of day 1 drain fluid amylase level for prediction of clinically relevant fistula after distal pancreatectomy using the NSQIP database[J].Surgery,2019,165(2):315-322.
[12]鄒弘駒,段瑞.彈力套包扎在腮腺腫瘤術(shù)后患者康復(fù)中的應(yīng)用價(jià)值[J].中國腫瘤臨床與康復(fù),2018,25(3):306-308.
[13]唐正琪.不同切口下腮腺淺葉腫瘤切除術(shù)對患者圍術(shù)期臨床指標(biāo)、美觀性及術(shù)后并發(fā)癥的影響[J].中國眼耳鼻喉科雜志,2019,19(1):35-38.
[14]Infante-Cossio P,Gonzalez-Cardero E,Garcia-Perla-Garcia A,et al.Complications after superficial parotidectomy for pleomorphic adenoma[J].Med Oral Patol Oral Cir Bucal,2018,23(4):e485-e492.
[15]Lu Y,Zhang S,Peng C,et al.Drain fluid amylase as a predictor of postoperative salivary fistula in cases with benign parotid tumours[J].BMC Oral Health,2020,20(1):184.
[16]Wong W K,Shetty S.The extent of surgery for benign parotid pathology and its influence on complications: A prospective cohort analysis[J].Am J Otolaryngol,2018,39(2):162-166.
[收稿日期]2020-12-02
本文引用格式:周華,李慧國,李君,等.腮腺良性腫瘤切除術(shù)后唾液瘺形成風(fēng)險(xiǎn)的預(yù)測模型分析[J].中國美容醫(yī)學(xué),2022,31(5):110-113.