Abstract
目的
传统的宫颈癌筛查手段——高危型人乳头瘤病毒(high-risk human papillomavirus,hrHPV)和宫颈脱落细胞学检测存在局限性。本研究通过检测宫颈脱落细胞中连接黏附分子3(junctional adhesion molecule 3,JAM3)/配对盒基因1(paired box gene 1,PAX1)高甲基化水平,并将其与液基薄层细胞学(liquid based cytology,LBC)进行比较,评价JAM3/PAX1甲基化对于宫颈高级别病变的诊断能力,以探索新的宫颈高级别病变的诊断模式,实现宫颈高级别病变“精准筛查”的目的。
方法
回顾性收集2021年6月至2022年6月在中南大学湘雅三医院妇科阴道镜门诊接受检查的患者共136例,包括宫颈非高级别病变122例,高级别病变14例。研究的变量包括:基本临床信息(年龄、体重指数、是否绝经)、LBC、hrHPV、宫颈组织病理、阴道微生态结果、阴道镜结果(宫颈转化区类型)、JAM3(ΔCtJ)和PAX1(ΔCtP)基因甲基化的ΔCt值。首先通过logistic回归分析筛选影响宫颈高级别病变的影响因素,并进行相关性分析,然后用差异有统计学意义的变量构建条件推断树模型。
结果
Logistic回归分析提示高级别宫颈病变组与非高级别宫颈病变组的PAX1与JAM3基因甲基化ΔCt值以及LBC检测结果差异均有统计学意义(均P<0.05)。相关性分析发现,宫颈病理结果与ΔCtP(r=-0.360,P<0.001)、ΔCtJ(r=-0.448,P<0.001)、LBC(r=-0.305,P<0.001)、菌群多样性(r=-0.183,P=0.037)呈负相关。条件推断树显示:当ΔCtJ>10.13时,全部为宫颈非高级别病变的患者;当ΔCtP>6.22时,非高级别病变的患者有117例,占97.5%,高级别病变者仅3例,占2.5%。ΔCtJ>8.61且LBC为不明确的非鳞状上皮细胞(atypical squamous cell of undetermined significance,ASC-US)或未见上皮内病变细胞时,105例(99.1%)为宫颈非高级别病变的患者,仅1例(0.9%)检出为高级别病变;当LBC结果为高级别病变时,仅9例患者检出高级别病变,3例检出非高级别病变;而当LBC提示低级别病变、ASC-US、未见上皮内病变细胞,且ΔCtP>6.22时,有117例(97.5%)的患者检出非高级别病变。
结论
在人乳头瘤病毒感染妇女中,JAM3/PAX1双基因甲基化检测可独立应用于宫颈高级别病变/非高级别病变的分层诊断,且不依赖于宫颈脱落细胞学检测结果;亦可与LBC联合使用以弥补LBC敏感度低的缺点。此外,未来将甲基化试剂盒应用于大规模宫颈癌筛查,有利于发现更多宫颈高级别病变患者,并达到早筛、早治宫颈病变/癌的目的。
Keywords: 宫颈病变, DNA甲基化, 连接黏附分子3, 配对盒基因1, 条件推断树
Abstract
Objective
Currently, traditional cervical cancer screening methods, such as high-risk human papillomavirus testing and liquid based cytology (LBC), still possess limitations. This study aims to identify new diagnostic biomarkers to achieve the goal of “precision screening” via exploring the clinical value of DNA methylation [ΔCtP: paired box gene 1 (PAX1)and ΔCtJ: junctional adhesion molecule 3 (JAM3)] detection in cervical exfoliated cells for the diagnosis of high-grade cervical lesions.
Methods
A total of 136 patients who underwent gynecological examinations in the vaginal room of the Department of Gynecology at the Third Xiangya Hospital of Central South University from June 2021 to June 2022 were retrospectively studied. Among them, 122 patients had non-high-grade cervical lesions, and 14 patients had high-grade cervical lesions. The variables included general information (age, body mass index, and menopause status), LBC, high-risk human papillomavirus, cervical tissue pathology, vaginal examination results, and the ΔCt values of JAM3 and PAX1 gene methylation. Logistic regression analysis was used to identify the factors affecting the diagnosis of high-grade cervical lesions, followed by correlation analysis and construction of a conditional inference tree model.
Results
Logistic regression analysis showed that the methylation ΔCt values of PAX1 and JAM3 genes and LBC detection results were statistically significant between the high-grade cervical lesions group and the non-high-grade cervical lesions group (all P<0.05). Correlation analysis revealed a negative correlation between cervical pathological changes and ΔCtP (r=-0.36, P<0.001), ΔCtJ (r=-0.448, P<0.001), LBC (r=-0.305, P<0.001), or bacterial diversity (r=-0.183, P=0.037). The conditional inference tree showed that when ΔCtJ>10.13, all of patients had non-high-grade cervical lesions, while ΔCtP>6.22, the number of non-high-grade lesions accounted for 97.5% (117/120), and high-grade lesions accounted for only 2.5% (3/120). When ΔCtJ>8.61 and LBC were atypical squamous cell of undetermined significance or negative for intraepithelial lesions or malignancy (NILM), 105 (99.1%) patients were non-high-grade cervical lesions, only 1 (0.9%) patient was high-grade lesion. When the results of LBC were high-grade lesions, only 9 patients’ histopathological examination was the high-grade lesions and 3 non-high-grade lesions. When LBC indicated low-grade lesions, atypical squamous cell of undetermined significance, no intraepithelial lesions, and ΔCtP>6.22, 117 (97.5%) of patients’ histopathological examination was the non-high-grade lesions.
Conclusion
The JAM3/PAX1 gene methylation test can be used independently for the stratified diagnosis of high-grade/non-high-grade cervical lesions in women with high-risk human papillomavirus infection, independent of the cytological results of cervical excision. The JAM3/PAX1 gene methylation test can also be used in combination with LBC to make up for the shortcomings of low sensitivity of LBC. In addition, the application of methylation kit in large-scale cervical cancer screening in the future will be good to the detection of more patients with high-grade cervical lesions, and achieve early screening and early treatment for cervical lesions/cancer.
Keywords: cervical lesions, DNA methylation, junctional adhesion molecule 3, paired box gene 1, conditional inference tree
宫颈癌是全球常见的女性恶性肿瘤之一,在发展中国家更是位居女性生殖道肿瘤之首[1],由于发展中国家和不发达国家筛查率和疫苗接种率均低,宫颈癌的病死率呈现上升趋势[2],对妇女健康造成严重影响[3]。2018年,世界卫生组织(World Health Organization,WHO)提出了于2030年实现“消除宫颈癌”的目标,并且对于全球70%的30~45岁女性,在2030年达到“精准筛查”[4]。因此,高效与精确的检测手段对早期精准发现宫颈高级别鳞状上皮内病变(high-grade squamous intraepithelial lesion,HSIL),及时进行干预,降低宫颈浸润癌发生率并提高患者生存率具有重要意义。
高危型人乳头瘤病毒(high-risk human papillomavius,hrHPV)持续感染是导致宫颈上皮内瘤变(cervical intraepithelial neoplasia,CIN)和宫颈癌发生、发展的重要原因[5]。hrHPV DNA检测敏感度高但特异度较差,绝大部分女性为“一过性”感染且有自愈倾向,因此,hrHPV DNA检测结果的假阳性会导致过度治疗,从而带给患者及其家属心理负担[6-7]。液基薄层细胞学(liquid based cytology,LBC)检查,又称液基薄层细胞学检测(thinprep cytologic test,TCT),是宫颈癌筛查的重要手段,敏感度为50%~80%[8-9],其对意义不明确的非典型鳞状细胞(atypical squamous cells of undefined significance,ASC-US)的判读易受细胞病理医师主观因素与经验的影响,可能导致转诊阴道镜检查和治疗费用增加等问题[10]。因此,在目前相关指南推荐hrHPV DNA与LBC作为筛查条件下需要寻找能够准确且有效检测出宫颈高级别病变的新型生物标志物,单独或联合现有方法应用于宫颈癌筛查[11]。
近年来,有研究[12-13]指出,宫颈癌的发生、发展与表观遗传学改变具有密切关联性,而DNA甲基化属于癌症常见表观遗传现象——癌症细胞中抑癌基因启动子高度甲基化造成抑癌基因表达异常甚至基因失活,出现细胞恶性生长,是目前癌症发生的重要理论基础。宫颈癌甲基化基因检测可尽早发现宫颈高级别病变,指导临床进行干预,将成为未来宫颈癌防治的新手段[14]。连接黏附分子3(junctional adhesion molecule 3,JAM3)参与白细胞游走、血管形成及肿瘤转移等过程[15]。研究[11]发现,对于hrHPV阳性的患者,JAM3甲基化水平相对于LBC显示出更高的灵敏度,认为JAM3甲基化标志物可能成为hrHPV阳性患者的一种指导分级诊疗的新型生物标志物。配对盒基因1(paired box gene 1,PAX1)被认为是多种癌症的抑癌基因[16]。研究[17-18]表明PAX1基因高甲基化程度随着宫颈病变程度加重而加强,比单独的 HPV分型检测具有更高的准确性[16]。
传统的宫颈癌筛查手段——hrHPV和LBC都存在临床局限性,使得检测DNA甲基化程度作为宫颈病变新的诊断方式成为可能。然而,JAM3/PAX1甲基化对于宫颈病变的分类诊断能力如何,其判断能力与LBC相比,谁更准确尚不得而知。因此,本研究采用机器学习的方法,通过单因素分析初步筛选研究变量,构建具有较高分类预测准确率的条件推断树模型[19],以探讨JAM3/PAX1高甲基化在宫颈病变分类诊断中的应用价值。
1. 对象与方法
1.1. 对象
回顾性收集2021年6月至2022年6月在中南大学湘雅三医院妇科阴道镜门诊接受检查的患者共136例。其中,经病理确诊的宫颈非高级别病变患者122例,高级别病变患者14例。所有患者均签署知情同意书,研究获得中南大学湘雅三医院伦理委员会批准(审批号:23137)。
纳入标准:1)宫颈组织样本经病理学明确诊断,符合诊断标准者;2)严格遵循宫颈癌筛查流程,均接受LBC、hrHPV及阴道镜检查者;3)有性生活史者;4)3 d内无性生活、无阴道冲洗和阴道用药者;5)近6个月内未接受抗人乳头瘤病毒(human papillomavirus,HPV)治疗者;6)配合检查、资料完整并完成知情同意书签署的患者。
排除标准:1)月经期者;2)已知生殖道恶性肿瘤或其他肿瘤史者;3)有自身免疫性疾病史,或正在服用免疫抑制剂者;4)妊娠期或哺乳期者;5)宫颈癌疫苗接种史者。
1.2. 方法
1.2.1. 临床信息收集
提取患者基本信息,包括:1)一般信息[门诊号、姓名、年龄、身高、体重、体重指数(body mass index,BMI)、是否绝经];2)LBC结果,分为低级别鳞状上皮内病变(low-grade squamous intraepithelial lesions,LSIL)、高级别鳞状上皮内病变(high-grade squamous intraepithelial lesions,HSIL)、ASC-US及未见上皮内病变细胞(negative for intraepithelial lesions or malignancy,NILM);3)hrHPV DNA结果,分为16/18型HPV、非16/18高危型HPV、阴性(无hrHPV型感染);4)阴道镜下宫颈转化区类型,分为I型、II型、III型;5)宫颈组织病理,分为HSIL、LSIL及慢性宫颈炎;6)常规阴道微生态结果,包括优势菌(革兰氏阳性大杆菌G+b、革兰氏阳性短杆菌G+b、革兰氏阳性球菌G+c、革兰氏阴性短杆菌G-b)、白细胞酯酶(+,-)、唾液酸苷酶(+,-)、乙酰氨基葡萄糖苷酶(+,-)、菌群密集度(-,+,++,+++)、菌群多样性(+,-)、乳酸杆菌(未检出,少量,中量,大量)、Nugent评分(0~3分,4~6分,≥7分)、白细胞计数[<5,≥5~<10,≥10~<15,≥15~<20,≥20个/高倍视野(high power field,HP)]、酸碱度(pondus hydrogenii,pH)(≤4.6,>4.6)、真菌感染(+,-)、细菌性阴道炎(bacterial vaginosis,BV)(+,-)、过氧化氢(+,-)。以上数据由2名研究助理仔细记录并复核。
1.2.2. LBC、HPV及DNA甲基化检测
暴露宫颈后,用无菌大头棉签轻轻擦拭宫颈表面分泌物,用宫颈脱落细胞取材刷于宫颈表面及宫颈管内顺时针旋转5~10周,获得宫颈脱落细胞,并将刷头放入细胞保存液,贴好检测条形码。1)取5~10 mL标本经处理后,将脱落细胞制片、染色。由病理医师阅片,并依照2014年修订的宫颈细胞学结果分级系统(the Bethesda system,TBS)[20]进行描述并诊断。2)取5~10 mL标本采用聚合酶链反应对HPV DNA进行扩增,用人乳头瘤病毒(21型)核酸分型检测试剂盒(广州凯普生物科技有限公司生产)进行检测。可以检测出21种HPV类型,包括14种高危型HPV(16、18、31、33、35、39、45、51、52、56、58、59、66和68)、5种低危型HPV(6、11、42、43和44)和2种未知风险型HPV(53和CP8304)[21]。3)取5~10 mL标本进行DNA提取后,进行DNA浓度和质量测定,并以200 ng DNA进行重亚硫酸盐转化。重亚硫酸盐转化后DNA依照人PAX1和JAM3基因甲基化检测试剂盒(PCR-荧光探针法)(北京起源聚禾生物科技公司生产)说明书,采用ABI 7500实时荧光定量PCR仪进行多重实时荧光定量PCR(real-time PCR)检测,记录样本目标基因ΔCt值,JAM3结果以ΔCtJ表示,PAX1结果以ΔCtP表示,ΔCt值越低提示目标基因甲基化程度越高。
1.2.3. 阴道镜检查及宫颈活检
患者取截石位,放置阴道镜,充分暴露宫颈,用生理盐水把宫颈的表面黏液擦干净,观察宫颈的形态、颜色等自然状态。将涂有5%冰醋酸的湿润棉球涂到宫颈上,停留在宫颈上120 s以后取掉棉球,肉眼直接观察宫颈颜色变化,观察宫颈转化区类型、醋白反应、有无镶嵌点状血管、有无异型血管等病变。鳞柱交界区可见、部分可见、不可见分别对应I型、II型、III型转化区[22]。用活检钳在阴道镜下提示可疑病变区域取组织,取样组织经10%甲醛固定。宫颈组织固定后立即送至病理科予以制片,采用苏木精-伊红染色。由1名病理医师进行判读,另1名病理科医师审核结果。依照2020年WHO关于宫颈病变的分类进行诊断:正常/慢性宫颈炎、LSIL、HSIL、宫颈癌。
1.3. 统计学处理
先用Shapiro-Wilk检验进行正态性检验。符合正态分布的数据用均数±标准差( ±s)描述,不确定分布状态的数据采用95%置信区间提供对参数估计的范围和不确定性的度量。计量资料使用t检验或者Wilcoxon秩和检验;计数资料用例(%)表示,使用χ 2检验或Fisher’s确切概率法。先通过SAS 9.4软件进行统计描述和单因素logistic回归分析,采用Pearson相关分析进行相关性分析。再用R4.1.3(party)构建条件推断树(conditional inference tree)建立宫颈病变分类诊断的预测模型,探讨宫颈高级别病变的影响因素。双侧P<0.05为差异有统计学意义。
2. 结 果
2.1. 单因素分析
单因素分析显示:HSIL组较非HSIL组ΔCtJ值和ΔCtP值均降低(均P<0.001);LBC的阳性预测值为66.67%,灵敏度仅为42.9%。
与非HSIL组比较,HSIL组一般人口学特征差异均无统计学意义(均P>0.05),年龄稍高(P=0.244)。2组均以未绝经状态为主,绝经状态2组之间差异无统计学意义(P=0.143)。并且,2组队列hrHPV感染情况、宫颈转化区类型、阴道微生态各指标差异均无统计学意义(均P>0.05):92.65%的患者感染hrHPV,且44.12%存在HPV16/18型感染,仅7.35%未检测出hrHPV感染(P=0.569);宫颈III型转化区占60.29%;74.26%的患者阴道清洁度为III度;发生BV感染的患者均在非HSIL组(表1)。
表1.
非高级病变组与高级病变组临床和病理特征的单因素logistic回归分析
Table 1 Univariate logistic analysis for the clinical and pathological characteristics of high-grade cervical lesions and non-high-grade cervical lesions
| 组别 | n | 年龄/岁 |
BMI/ (kg·m-2) |
绝经状态/[例(%)] | 甲基化检测 | ||
|---|---|---|---|---|---|---|---|
| 是 | 否 | JAM3(ΔCtJ) | PAX1(ΔCtP) | ||||
| 非高级病变组 | 122 | 36.8±11.8 | 21.6±3.4 | 21(80.77) | 101(91.82) | 13.63(10.85, 15.63) | 9.71(7.97, 14.30) |
| 高级病变组 | 14 | 40.9±10.3 | 22.2±3.1 | 5(19.23) | 9(8.18) | 7.94(6.04, 8.61) | 3.60(2.82, 6.22) |
| 合计 | 37.2±11.7 | 21.7±3.2 | 26(19.12) | 110(80.88) | 13.49(10.33, 15.28) | 9.44(7.35, 12.64) | |
| t/Z/χ 2 | 1.17 | 0.60 | — | 5.37 | 4.47 | ||
| P | 0.244 | 0.550 | 0.143 | <0.001 | <0.001 | ||
| 组别 | LBC检测/[例(%)] | HPV DNA检测/[例(%)] | 宫颈转化区类型/[例(%)] | |||||||
|---|---|---|---|---|---|---|---|---|---|---|
| HSIL | LSIL | ASC-US | NILM | HPV16/18感染 | 非16/18hrHPV感染 | 非hrHPV感染 | III | II | I | |
| 非高级病变组 | 3(33.33) | 8(80.00) | 54(96.43) | 57(93.44) | 54(90.00) | 60(90.91) | 8(80.00) | 75(91.46) | 11(78.57) | 36(90.00) |
| 高级病变组 | 6(66.67) | 2(20.00) | 2(3.57) | 4(6.56) | 6(10.00) | 6(9.09) | 2(20.00) | 7(8.54) | 3(21.43) | 4(10.00) |
| 合计 | 9(6.62) | 10(7.35) | 56(41.18) | 61(44.85) | 60(44.12) | 66(48.53) | 10(7.35) | 82(60.29) | 14(10.29) | 40(29.41) |
| t/Z/χ 2 | — | 1.13 | — | |||||||
| P | <0.001 | 0.570 | 0.270 | |||||||
| 组别 | 清洁度/[例(%)] | 优势菌/[例(%)] | 白细胞酯酶/[例(%)] | |||||
|---|---|---|---|---|---|---|---|---|
| III | II |
革兰氏阳性 大杆菌G+b |
革兰氏阳性 短杆菌G+b |
革兰氏阳性 球菌G+c |
革兰氏阴性 短杆菌G-b |
+ | - | |
| 非高级病变组 | 92(91.09) | 30(85.71) | 10(90.91) | 96(88.89) | 4(100) | 12(92.31) | 23(88.46) | 99(90.00) |
| 高级病变组 | 9(8.91) | 5(14.29) | 1(9.09) | 12(11.11) | 0(0) | 1(7.69) | 3(11.54) | 11(10.00) |
| 合计 | 101(74.26) | 35(25.74) | 11(8.09) | 108(79.41) | 4(2.94) | 13(9.56) | 26(19.12) | 110(80.88) |
| t/Z/χ 2 | — | — | — | |||||
| P | 0.350 | 1.000 | 0.731 | |||||
| 组别 | 唾液酸苷酶/[例(%)] |
乙酰氨基葡萄糖 苷酶/[例(%)] |
菌群密集度/[例(%)] | 菌群多样性/[例(%)] | ||||||
|---|---|---|---|---|---|---|---|---|---|---|
| + | - | + | - | - | + | ++ | +++ | - | + | |
| 非高级病变组 | 11(91.67) | 111(89.52) | 8(88.89) | 114(89.76) | 4(66.67) | 3(100) | 8(100) | 107(89.92) | 4(66.67) | 118(90.77) |
| 高级病变组 | 1(8.33) | 13(10.48) | 1(11.11) | 13(10.24) | 2(33.33) | 0(0) | 0(0) | 12(10.08) | 2(33.33) | 12(9.23) |
| 合计 | 12(8.82) | 124(91.18) | 9(6.62) | 127(93.38) | 6(4.41) | 3(2.21) | 8(5.88) | 119(87.5) | 6(4.41) | 130(95.59) |
| t/Z/χ 2 | — | — | — | — | ||||||
| P | 1.000 | 1.000 | 0.208 | 0.117 | ||||||
| 组别 | 乳杆菌/[例(%)] | 白细胞计数/[例(%)] | |||||||
|---|---|---|---|---|---|---|---|---|---|
| 未检出 | 少量 | 中量 | 大量 | <5个/HP | ≥5~<10个/HP | ≥10~<15个/HP |
≥15~ <20个/HP |
≥20个/HP | |
| 非高级病变组 | 64(88.89) | 23(95.83) | 3(100) | 32(86.49) | 87(92.55) | 12(75.00) | 10(90.91) | 7(100) | 6(75.00) |
| 高级病变组 | 8(11.11) | 1(4.17) | 0(0) | 5(13.51) | 7(7.45) | 4(25.00) | 1(9.09) | 0(0) | 2(25.00) |
| 合计 | 72(52.94) | 24(17.65) | 3(2.21) | 37(27.21) | 94(69.12) | 16(11.76) | 11(8.09) | 7(5.15) | 8(5.88) |
| t/Z/χ 2 | — | — | |||||||
| P | 0.709 | 0.107 | |||||||
| 组别 | Nugent评分/[例(%)] | pH/[例(%)] | 真菌感染/[例(%)] | BV/[例(%)] | 过氧化氢/[例(%)] | ||||||
|---|---|---|---|---|---|---|---|---|---|---|---|
| ≥7 | 4~6 | 0~3 | ≤4.6 | >4.6 | + | - | + | - | + | - | |
| 非高级病变组 | 8(100) | 0(0) | 114(89.06) | 34(87.18) | 88(90.72) | 10(90.91) | 112(89.60) | 8(100) | 113(88.98) | 86(90.53) | 36(87.80) |
| 高级病变组 | 0(0) | 0(0) | 14(10.94) | 5(12.82) | 9(9.28) | 1(9.09) | 13(10.40) | 0(0) | 14(11.02) | 9(9.47) | 5(12.20) |
| 合计 | 8(5.88) | 0(0) | 128(94.12) | 39(28.68) | 97(71.32) | 11(8.09) | 125(91.91) | 8(5.93) | 127(94.07) | 95(69.85) | 41(30.15) |
| t/Z/χ 2 | — | — | — | — | — | ||||||
| P | 1.000 | 0.543 | 1.000 | 1.000 | 0.759 | ||||||
符合正态分布的数据用均数±标准差描述,不确定分布状态的数据采用95%置信区间提供对参数估计的范围和不确定性的度量。HSIL:高级别鳞状上皮内病变;LSIL:低级别鳞状上皮内病变;ASC-US:不明确的非鳞状上皮细胞;NILM:未见上皮内病变细胞;HPV:人乳头瘤病毒;LBC:液基薄层细胞学;HP:高倍视野;BMI:体重指数;BV:细菌性阴道炎。
2.2. 相关性分析结果
相关性分析结果显示:宫颈HSIL病理结果与ΔCtP(r=-0.360,P<0.001)、ΔCtJ(r=-0.448,P<0.001)、LBC(r=-0.305,P<0.001)、菌群多样性(r=-0.183, P=0.037)呈负相关;与年龄、BMI、HPV型别、是否绝经、宫颈转化区、清洁度、阴道微生态(Nugent评分、乳杆菌、清洁度等)无相关性(均P>0.05,图1)。
图 1.
临床特征相关性分析热图
Figure 1 Correlation heatmap of clinical parameters
*P<0.05. HPV: Human papillomavirus; BMI: Body mass index; BV: Bacterial vaginosis; TCT: Thinprep cytologic test.
2.3. 条件推断树模型
根据单因素logistic回归分析结果,本研究分别将ΔCtP、ΔCtJ、ΔCtP+LBC、ΔCtJ+LBC、ΔCtP+ΔCtJ、ΔCtP+ΔCtJ+LBC纳入条件推断树模型(因JAM3与PAX1间检测病变差异无统计学意义,故未建构JAM3与PAX1相关模型),推断树模型见图2。
图 2.
子宫颈高级别病变条件推断树模型(灰色为0,代表非高级别病变;黑色为1,代表高级别病变;右侧0~1代表高级别病变的概率)
Figure 2 Conditional inference tree for the hierarchical diagnosis of high-grade cervical lesions (Gray is 0, representing non-high-grade lesions; Black is 1, representing high-grade lesions; 0-1 on the right side represents the probability of high-grade lesions)
图2A显示:推断树共分2层,由2个特征变量得到3条规则。1)ΔCtJ≤8.61共20例,为第1类,该分类中宫颈非HSIL和HSIL人数相当,分别占9例和11例;2)ΔCtJ>8.61且≤10.13共12例,为第2类,该分类中宫颈非HSIL例数(9例)较第1类增多,而宫颈HSIL仅为3例,提示当ΔCtJ=10.13时,对2组人群有较好的分流能力;3)ΔCtJ>10.13共104例,为第3类,该分类全部为宫颈非HSIL患者。图2B显示:推断树共分2层,由2个特征变量得到3条规则。1)ΔCtJ≤8.61共20例,为第1类,可以将宫颈非HSIL和HSIL人数大致按照1꞉1划分,分别为9例和11例;2)ΔCtJ>8.61且 LBC提示HSIL或LSIL共10例为第2类,其中宫颈非HSIL有8例,占比与图2A中Node 4类似,宫颈HSIL仅2例;3)ΔCtJ>8.61且LBC提示ASC-US或NILM共106例,为第3类,该分类中105例(99.1%)为宫颈非HSIL的患者,仅1例(0.9%)检出HSIL病理结果。该模型表明ΔCtJ联合LBC能够较好地预测宫颈HSIL。图2C展示了仅用ΔCtP区分宫颈HSIL的能力:1)当ΔCtP≤6.22时共有16例,其中5例为非HSIL,11例为HSIL;2)当ΔCtP>6.22时,共有120例,检出117例非HSIL,占比明显增加,达到97.5%,而仅3例检出HSIL。结果显示当ΔCtP=6.22时对2组人群具有较好的鉴别能力,ΔCtP>6.22时发生HSIL的可能性极低。
图2D中的条件推断树由2个特征变量LBC和ΔCtP构成,结果显示:1)当LBC结果为HSIL时,9 例患者中6例检出病理高级别病变,另3例检出非病理高级别病变;2)当LBC提示LSIL、ASC-US与NILM,且ΔCtP≤6.22时,7例患者中有2例检出非HSIL,另5例检出HSIL;3)当LBC提示LSIL、ASC-US、NILM,且ΔCtP>6.22时,120例患者中仅有3例检出病理高级别病变,较ΔCtP≤6.22时明显降低,另有117例(97.5%)检出非病理高级别病变,与病理诊断的结果更趋于一致。
3. 讨 论
肿瘤细胞特定基因启动子高甲基化被公认为肿瘤表观遗传学机制之一[23]。基因启动子高甲基化主要发生在基因启动子区域CpG岛及其附近区域,在不影响蛋白质编码序列的基础上下调多种基因的表达水平,导致一系列基因沉默,使基因表达调控失常[23-24]。在宫颈癌的表观遗传学研究中特定基因的高甲基化被认为可成为宫颈癌分子诊断的新方法[25]。
JAM是免疫超家族中一种特殊类型的跨膜蛋白家族,它位于极性上皮细胞和内皮细胞的紧密连接处。Boers等[26-27]应用定量的方法检测了JAM3基因甲基化水平在hrHPV阳性的妇女中诊断宫颈上皮内瘤变(cervical intraepithelial neoplasia,CIN)2及以上病变敏感度为68%,特异度为94%,诊断CIN3+病变的敏感度和特异度分别为80%和76%。本研究发现:单独应用JAM3(ΔCtJ)时,随着ΔCtJ值增加,高级别病变检出率逐渐降低;当ΔCtJ>10.13时,全部患者最终被证实为非高级别患者。结论与上述研究相符。
PAX1编码一个与转录因子相关的保守区域,在胚胎发育过程中扮演重要角色[28]。Liu等[29]研究发现PAX1基因甲基化常见于CIN3及浸润性宫颈癌患者的宫颈脱落细胞中。Xu等[30]研究发现在正常宫颈脱落细胞中PAX1基因并没有发生甲基化,而在CIN1中的甲基化率为9%,在CIN2/3中为44%,在浸润性宫颈癌中为100%。本研究发现:单独应用ΔCtP,且当ΔCtP>6.22时,在区分宫颈HSIL/非HSIL的准确性上,较LBC更接近病理诊断,检测出非HSIL的例数占比达到97.5%,仅2.5%为HSIL。提示ΔCtP与LBC联用较单独使用LBC,能大大提高对高级别病变的判断能力,降低LBC的假阴性率。然而,该模型与单用ΔCtP相比,高级别病变的检出率相当,2种方案联用在增加成本的同时并未显示出更明显的优势。
LBC检测的敏感度和特异度因各地区的医疗水平而异[31]。ASC-US被认为是一种细胞病理学医师难以重复的诊断结果,其诊断意义不大,但是又隐藏着一定比例的病理HSIL[32-34]。本研究发现:单用LBC时即使提示非HSIL,仍可有一定数量病理高级别病变存在,需要更可靠的手段对患者进行分流;将ΔCtJ/ΔCtP与LBC联用,可提高LBC对于高级别病变预测的准确性。有多项研究认为:对ASC-US病例中DNA甲基化的检测显示出比HPV检测更高的特异度,并且比HPV16/18基因分型具有更高的敏感度,甲基化标志比蛋白质标志物更敏感[35-37],有望成为宫颈癌筛查的新型方法[30]。
另外,大规模以LBC作为宫颈癌筛查时存在许多缺点:1)更容易受到阅片医师主观因素的影响;2)细胞形态学ASC-US存在漏诊的可能;3)对于病理科医师匮乏的地区较难实施细胞学检查;4)无法自采样检测。相比而言,DNA甲基化检测有以下优势:1)仪器检测与判读更具客观性;2)检测时间短,可批量检测,效率更高,节省人力和时间成本;3)妇女自我取样被验证可行并与医师取样时具有较高的一致率[27]。
本研究仍然存在一定的不足。虽然条件推断树的预测能力高,却不能量化变量的风险值[38]。另外,2个甲基化指标因无检测差异性无法在同一棵决策树中纳入这2个指标。此外,本研究中该类患者最终获得高级别病变病理结果的例数偏少,后续研究需扩大样本量,以便更好地研究PAX1/JAM3甲基化在分层诊断宫颈病变中的价值。
综上,本研究发现:1)单独应用ΔCtJ/ΔCtP,且当ΔCtJ>10.13或ΔCtP>6.22时,在区分宫颈高级别病变/非高级别病变的准确性上,较LBC更接近病理诊断;2)单用ΔCtJ/ΔCtP诊断宫颈高级别病变的能力不低于ΔCtJ/ΔCtP与LBC联用;3)ΔCtJ/ΔCtP与LBC联用可提高LBC对于高级别病变预测的准确性。结果提示ΔCtJ/ΔCtP可不依赖于LBC,有望独立应用于宫颈高级别病变/非高级别病变的诊断;亦可联合使用以弥补LBC敏感度不高的缺点。另外,将甲基化试剂盒应用于大规模宫颈癌筛查,特别对于经济欠发达地区,更有利于节省经济成本和人力资源成本,易于推广。
基金资助
湖南省临床医疗技术创新引导项目(2020SK53604)。
This work was supported by the Hunan Provincial Clinical Medical Technology Innovation Guiding Project, China (2020SK53604).
利益冲突声明
作者声称无任何利益冲突。
作者贡献
李翔 实验操作,论文撰写和修改;贺斯黎、赵行平、孙丹 数据采集和临床采样;吴思、徐大宝 数据分析,论文修改;李颖佳 研究设计,论文撰写。所有作者阅读并同意最终的文本。
Footnotes
http://dx.chinadoi.cn/10.11817/j.issn.1672-7347.2023.230175
原文网址
http://xbyxb.csu.edu.cn/xbwk/fileup/PDF/2023121820.pdf
参考文献
- 1. Allouch S, Malki A, Allouch A, et al. High-risk HPV oncoproteins and PD-1/PD-L1 interplay in human cervical cancer: recent evidence and future directions[J]. Front Oncol, 2020, 10: 914. 10.3389/fonc.2020.00914. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2. 杨雅兰. SLC5A8基因甲基化及与高危型HPV的相关性在宫颈病变中的研究[D]. 湖北十堰: 湖北医药学院, 2021. [Google Scholar]; YANG Yalan. Significance of methylation of SLC5A8 gene and its relationship with HPV in cervical lesions[D]. Shiyan, Hubei: Hubei University of Medicine, 2021. [Google Scholar]
- 3. McCormack M, Gaffney D, Tan D, et al. The Cervical Cancer Research Network (Gynecologic Cancer InterGroup) roadmap to expand research in low- and middle-income countries[J]. Int J Gynecol Cancer, 2021, 31(5): 775-778. 10.1136/ijgc-2021-002422. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4. World Health Organization . Draft global strategy towards the elimination of cervical cancer as a public health problem[EB/OL]. (2019-12-16)[2023-04-02]. https:// www.who.int/docs/default-source/cervical-cancer/cerv-cancer-elimn-strategy-16dec-12pm.pdf.
- 5. 李明蔚, 王建东. 阴道微生物与HPV感染及宫颈病变的相关性研究进展[J]. 医学综述, 2022, 28(11): 2145-2151. 10.3969/j.issn.1006-2084.2022.11.012. [DOI] [Google Scholar]; LI Mingwei, WANG Jiandong. Research progress in correlation of vaginal microbiota with HPV infection and cervical lesions[J]. Medical Recapitulate, 2022, 28(11): 2145-2151. 10.3969/j.issn.1006-2084.2022.11.012. [DOI] [Google Scholar]
- 6. Poirson J, Biquand E, Straub ML, et al. Mapping the interactome of HPV E6 and E7 oncoproteins with the ubiquitin-proteasome system[J]. FEBS J, 2017, 284(19): 3171-3201. 10.1111/febs.14193. [DOI] [PubMed] [Google Scholar]
- 7. Preventive Services Task Force US, Curry SJ, Krist AH, et al. Screening for cervical cancer: us preventive services task force recommendation statement[J]. JAMA, 2018, 320(7): 674-686. 10.1001/jama.2018.10897. [DOI] [PubMed] [Google Scholar]
- 8. Nkwabong E, Laure Bessi Badjan I, Sando Z. Pap smear accuracy for the diagnosis of cervical precancerous lesions[J]. Trop Doct, 2019, 49(1): 34-39. 10.1177/0049475518798532. [DOI] [PubMed] [Google Scholar]
- 9. Güneri SE, Şen S. Women’s experiences after abnormal Pap smear results: a qualitative study[J]. J Psychosom Obstet Gynaecol, 2020, 41(1): 22-29. 10.1080/0167482X.2019.1657090. [DOI] [PubMed] [Google Scholar]
- 10. 赵爽. 适宜中低资源国家和地区的宫颈癌筛查策略及推广实施研究[D]. 北京: 北京协和医学院, 2022. 10.27648/d.cnki.gzxhu.2022.000321. [DOI] [Google Scholar]; ZHAO Shuang. Study on cervical cancer screening strategy and its popularization and implementation suitable for countries and regions with low and medium resources[D]. Beijing: Peking Union Medical College, 2022. 10.27648/d.cnki.gzxhu.2022.000321. [DOI] [Google Scholar]
- 11. 梁婷婷, 杨勇霞, 侯丛哲, 等. PAX1基因甲基化与宫颈高级别上皮内病变及高危型HPV分型的关联性[J]. 山东大学学报(医学版), 2021, 59(11): 48-52. 10.6040/j.issn.1671-7554.0.2021.0795. [DOI] [Google Scholar]; LIANG Tingting, YANG Yongxia, HOU Congzhe, et al. Correlation among PAX1 methylation, high-grade cervical intraepithelial lesions and HR-HPV classification[J]. Journal of Shandong University. Health Science, 2021, 59(11): 48-52. 10.6040/j.issn.1671-7554.0.2021.0795. [DOI] [Google Scholar]
- 12. Zhu HM, Zhu H, Tian M, et al. DNA methylation and hydroxymethylation in cervical cancer: diagnosis, prognosis and treatment[J]. Front Genet, 2020, 11: 347. 10.3389/fgene.2020.00347. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13. van den Helder R, Steenbergen RDM, van Splunter AP, et al. HPV and DNA methylation testing in urine for cervical intraepithelial neoplasia and cervical cancer detection[J]. Clin Cancer Res, 2022, 28(10): 2061-2068. 10.1158/1078-0432.CCR-21-3710. [DOI] [PubMed] [Google Scholar]
- 14. Güzel C, van Sten-Van’t Hoff J, de Kok IMCM, et al. Molecular markers for cervical cancer screening[J]. Expert Rev Proteomics, 2021, 18(8): 675-691. 10.1080/14789450.2021.1980387. [DOI] [PubMed] [Google Scholar]
- 15. Yin AJ, Zhang Q, Kong XN, et al. JAM3 methylation status as a biomarker for diagnosis of preneoplastic and neoplastic lesions of the cervix[J]. Oncotarget, 2015, 6(42): 44373-44387. 10.18632/oncotarget.6250. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16. Fang C, Wang SY, Liou YL, et al. The promising role of PAX1 (aliases: HUP 48, OFC2) gene methylation in cancer screening[J/OL]. Mol Genet Genomic Med, 2019, 7(3): e506[2023-04-02]. 10.1002/mgg3.506. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17. Chen Y, Cui ZL, Xiao ZZ, et al. PAX1 and SOX1 methylation as an initial screening method for cervical cancer: a Meta-analysis of individual studies in Asians[J]. Ann Transl Med, 2016, 4(19): 365. 10.21037/atm.2016.09.30. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18. Nikolaidis C, Nena E, Panagopoulou M, et al. PAX1 methylation as an auxiliary biomarker for cervical cancer screening: a Meta-analysis[J]. Cancer Epidemiol, 2015, 39(5): 682-686. 10.1016/j.canep.2015.07.008. [DOI] [PubMed] [Google Scholar]
- 19. Obermeyer Z, Emanuel EJ. Predicting the future—big data, machine learning, and clinical medicine[J]. N Engl J Med, 2016, 375(13): 1216-1219. 10.1056/NEJMp1606181. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20. Nayar R, Wilbur DC. The pap test and Bethesda 2014[J]. Cancer Cytopathol, 2015, 123(5): 271-281. 10.1002/cncy.21521. [DOI] [PubMed] [Google Scholar]
- 21. Tao PP, Zheng WP, Wang YG, et al. Sensitive HPV genotyping based on the flow-through hybridization and gene chip[J]. J Biomed Biotechnol, 2012, 2012: 938780. 10.1155/2012/938780. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22. Bornstein J, Bentley J, Bösze P, et al. 2011 colposcopic terminology of the international federation for cervical pathology and colposcopy[J]. Obstet Gynecol, 2012, 120(1): 166-172. 10.1097/AOG.0b013e318254f90c. [DOI] [PubMed] [Google Scholar]
- 23. 霍会蚕, 王如意, 李艳云, 等. 高危型HPV感染及CALCA基因启动子甲基化在宫颈病变检测中的临床价值[J]. 中华医院感染学杂志, 2021, 31(17): 2628-2632. 10.11816/cn.ni.2021-203679. 34229919 [DOI] [Google Scholar]; HUO Huican, WANG Ruyi, LI Yanyun, et al. Clinical value of high-risk HPV infection and CALCA promoter methylation in diagnosis of cervical lesions[J]. Chinese Journal of Nosocomiology, 2021, 31(17): 2628-2632. 10.11816/cn.ni.2021-203679. [DOI] [Google Scholar]
- 24. Nishiyama A, Nakanishi M. Navigating the DNA methylation landscape of cancer[J]. Trends Genet, 2021, 37(11): 1012-1027. 10.1016/j.tig.2021.05.002. [DOI] [PubMed] [Google Scholar]
- 25. Zhang LY, Tan WX, Yang HM, et al. Detection of host cell gene/HPV DNA methylation markers: a promising triage approach for cervical cancer[J]. Front Oncol, 2022, 12: 831949. 10.3389/fonc.2022.831949. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26. Boers A, Wang R, van Leeuwen RW, et al. Discovery of new methylation markers to improve screening for cervical intraepithelial neoplasia grade 2/3[J]. Clin Epigenetics, 2016, 8: 29. 10.1186/s13148-016-0196-3. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27. Boers A, Bosgraaf RP, van Leeuwen RW, et al. DNA methylation analysis in self-sampled brush material as a triage test in hrHPV-positive women[J]. Br J Cancer, 2014, 111(6): 1095-1101. 10.1038/bjc.2014.392. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28. Feederle R, Gerber JK, Middleton A, et al. Generation of Pax1/PAX1-specific monoclonal antibodies[J]. Monoclon Antibodies Immunodiagn Immunother, 2016, 35(5): 259-262. 10.1089/mab.2016.0029. [DOI] [PubMed] [Google Scholar]
- 29. Liu HF, Meng X, Wang JY. Real time quantitative methylation detection of PAX1 gene in cervical cancer screening[J]. Int J Gynecol Cancer, 2020, 30(10): 1488-1492. 10.1136/ijgc-2019-001144. [DOI] [PubMed] [Google Scholar]
- 30. Xu J, Xu L, Yang BH, et al. Assessing methylation status of PAX1 in cervical scrapings, as a novel diagnostic and predictive biomarker, was closely related to screen cervical cancer[J]. Int J Clin Exp Pathol, 2015, 8(2): 1674-1681. [PMC free article] [PubMed] [Google Scholar]
- 31. Goodman A. HPV testing as a screen for cervical cancer[J]. BMJ, 2015, 350: h2372. 10.1136/bmj.h2372. [DOI] [PubMed] [Google Scholar]
- 32. Wang M, Hou B, Wang XZ, et al. Diagnostic value of high-risk human papillomavirus viral load on cervical lesion assessment and ASCUS triage[J]. Cancer Med, 2021, 10(7): 2482-2488. 10.1002/cam4.3653. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33. Oranratanaphan S, Kobwitaya K, Termrungruanglert W, et al. Value of CCNA1 promoter methylation in triaging ASC-US cytology[J]. Asian Pac J Cancer Prev, 2020, 21(2): 473-477. 10.31557/apjcp.2020.21.2.473. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34. Kelly H, Benavente Y, Pavon MA, et al. Performance of DNA methylation assays for detection of high-grade cervical intraepithelial neoplasia (CIN2+): a systematic review and meta-analysis[J]. Br J Cancer, 2019, 121(11): 954-965. 10.1038/s41416-019-0593-4. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35. Brancaccio M, Natale F, Falco G, et al. Cell-free DNA methylation: the new frontiers of pancreatic cancer biomarkers’ discovery[J]. Genes, 2019, 11(1): 14. 10.3390/genes11010014. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36. Ciechomska M, Roszkowski L, Maslinski W. DNA methylation as a future therapeutic and diagnostic target in rheumatoid arthritis[J]. Cells, 2019, 8(9): 953. 10.3390/cells8090953. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37. Yanatatsaneejit P, Chalertpet K, Sukbhattee J, et al. Promoter methylation of tumor suppressor genes induced by human papillomavirus in cervical cancer[J]. Oncol Lett, 2020, 20(1): 955-961. 10.3892/ol.2020.11625. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38. Hothorn T, Hornik K, Zeileis A. Unbiased recursive partitioning: a conditional inference framework[J]. J Comput Graph Stat, 2006, 15(3): 651-674. 10.1198/106186006x133933. [DOI] [Google Scholar]


