中国医科大学学报  2024, Vol. 53 Issue (3): 240-245

文章信息

陈国华, 张浩宇, 韩升波, 何建鑫
CHEN Guohua, ZHANG Haoyu, HAN Shengbo, HE Jianxin
基于血清CA50、TSGF、TPA的中晚期乳腺癌放化疗敏感性列线图预测模型的建立与评价
Establishment and evaluation of the model for predicting the sensitivity to radiochemotherapy in patients with middle- and advanced-stage breast cancer based on serum CA50, TSGF, and TPA
中国医科大学学报, 2024, 53(3): 240-245
Journal of China Medical University, 2024, 53(3): 240-245

文章历史

收稿日期:2023-06-13
网络出版时间:2024-03-04 16:54:50
基于血清CA50、TSGF、TPA的中晚期乳腺癌放化疗敏感性列线图预测模型的建立与评价
郑州市中医院检验科, 郑州 450000
摘要目的 探究血清糖类抗原50(CA50)、肿瘤特异性生长因子(TSGF)、组织多肽抗原(TPA)表达对中晚期乳腺癌放化疗敏感性的预测价值,通过建立列线图模型进行验证与评价。方法 选取82例中晚期乳腺癌患者为研究对象,所有患者均接受紫杉醇化疗联合放疗,根据实体瘤疗效评价标准分为敏感组(n = 57)和不敏感组(n = 25)。统计2组患者的一般资料,治疗前后血清CA50、TSGF、TPA表达水平及差值,构建列线图模型,绘制校准曲线、受试者操作特征(ROC)曲线、决策曲线,评估列线图模型的预测效能和临床效用。结果 2组比较,肿瘤直径、脉管侵犯、TNM分期、淋巴结转移、分化程度的差异有统计学意义(P < 0.05)。治疗后不敏感组血清CA50、TSGF、TPA表达水平高于敏感组,CA50、TSGF、TPA差值低于敏感组(P < 0.05)。将上述有统计学差异的因素纳入LASSO回归,选出3个预测变量,分别为CA50、TSGF、TPA差值。logistic回归显示,CA50、TSGF、TPA差值是中晚期乳腺癌放化疗敏感性的影响因素(P < 0.05);基于CA50、TSGF、TPA差值构建列线图模型,校准曲线、ROC曲线、决策曲线显示该模型具有良好的预测精准度和临床效用。结论 中晚期乳腺癌放化疗不敏感的患者治疗后血清CA50、TSGF、TPA表达水平较高,且三者差值是影响放化疗敏感性的因素。基于血清CA50、TSGF、TPA构建的列线图预测模型具有良好的预测价值和临床效用。
Establishment and evaluation of the model for predicting the sensitivity to radiochemotherapy in patients with middle- and advanced-stage breast cancer based on serum CA50, TSGF, and TPA
Laboratory Department, Zhengzhou Traditional Chinese Medicine Hospital, Zhengzhou 450000, China
Abstract: Objective To evaluate the predictive value of serum carbohydrate antigen 50 (CA50), tumor specific growth factor (TSGF), and tissue polypeptide antigen (TPA) levels for sensitivity to radiochemotherapy in patients with middle- and advanced-stage breast cancer using a nomogram model. Methods Eighty-two patients with middle- and advanced-stage breast cancer were selected as the study subjects. All patients received paclitaxel chemotherapy combined with radiotherapy and were divided into sensitive (n = 57) and insensitive (n = 25) groups according to the Response Evaluation Criteria in Solid Tumors. The general information of the patients, serum expression of CA50, TSGF, and TPA, and their differences before and after treatment were recorded. A nomogram model was constructed, and calibration curves, receiver operating characteristic (ROC) curves, and decision curves were used to evaluate the predictive power and clinical utility of the nomogram model. Results Significant differences were observed in tumor diameter, vascular invasion, TNM stage, lymph node metastasis, and degree of differentiation between the two groups (P < 0.05). Compared to those in the sensitive group, the serum expression of CA50, TSGF, and TPA after treatment was higher, and the difference in CA50, TSGF, and TPA was smaller in the insensitive group (P < 0.05). Three predictive variables were identified in the LASSO regression: differences in CA50, TSGF, and TPA. The logistic regression results showed that differences in CA50, TSGF, and TPA influenced sensitivity to radiochemotherapy in middle- and advanced-stage breast cancer (P < 0.05). A nomogram model was constructed using differences in CA50, TSGF, and TPA. Calibration, ROC, and decision curves showed the model's good predictive accuracy and clinical utility. Conclusion Serum expression of CA50, TSGF, and TPA is high in patients with middle- and advanced-stage breast cancer who are insensitive to radiochemotherapy, and differences in CA50, TSGF, and TPA affect their sensitivity to radiochemotherapy. The nomogram model had good predictive value and clinical utility.

乳腺癌是全球范围常见的女性肿瘤,最新数据显示,2020年新发乳腺癌226万例,这也是全球女性癌症死亡的主要原因[1]。乳腺癌早期无特异症状,临床症状明显时患者多处于中晚期,此时已错过最佳治疗时期,只能采用综合治疗。放化疗是治疗中晚期乳腺癌的首选方法,可提高患者生存期,降低组织转移和复发风险,但部分患者对放化疗不敏感,导致治疗效果不佳[2]。近年来研究[3-4]发现,血清糖类抗原50(carbohydrate antigen 50,CA50)、肿瘤特异性生长因子(tumor specific growth factor,TSGF)、组织多肽抗原(tissue polypeptide antigen,TPA)等肿瘤标志物可用于乳腺癌等肿瘤的早期诊断、预后评估、疾病转归等,但关于在中晚期乳腺癌患者中CA50、TSGF、TPA联合应用的临床研究少见。因此,本研究分析了中晚期乳腺癌放化疗不敏感和敏感患者中血清CA50、TSGF、TPA的表达情况,构建列线图预测模型,并对该模型进行验证与评价。

1 材料与方法 1.1 研究对象

选取2020年10月至2022年10月我院收治的82例中晚期乳腺癌患者,年龄42~72岁,平均(56.65±5.21)岁;体质量指数(body mass index,BMI)18.7~25.4 kg/m2,平均(21.43±1.34)kg/m2;月经状态:绝经期52例,非绝经期30例;病理类型:导管内原位癌61例,浸润性导管癌4例,浸润性小叶癌17例。

纳入标准:符合《中国抗癌协会乳腺癌诊治指南与规范(2017年版)》[5]诊断标准,经病理检查确诊;为中晚期乳腺癌;预计生存期≥3个月;入组前3个月内未接受过放化疗等治疗。排除标准:妊娠期或哺乳期女性;患有其他恶性肿瘤;患有急慢性感染性疾病,如艾滋病、麻疹、病毒性肝炎等;患有自身免疫性疾病、内分泌疾病、血液系统疾病;严重心肝肾等器官功能障碍;意识或语言障碍;凝血功能障碍。本研究获得我院伦理委员会批准(批号:20200510005)。

1.2 治疗方法

化疗方法:135 mg/m2紫杉醇(美国Bristol-Myers Squibb公司,国药准字HJ20171227,规格5 mL:30 mg)静脉滴注1次。放疗方法:胸壁和锁骨区域用6和9 MV电子线照射,照射16次后,在胸壁处加盖0.5 cm组织等效膜,2 Gy/次,1次/d,5次/周,总量50 Gy。21 d为1个疗程,持续2个疗程。

根据实体瘤疗效评价标准[6],将82例患者分为敏感组(n = 57)和不敏感组(n = 25)。

1.3 观察指标

1.3.1 一般资料

包括年龄、肿瘤直径、BMI、月经状态(绝经、未绝经)、脉管侵犯(有、无)、TNM分期(Ⅲa期、Ⅲb期、Ⅳ期)、淋巴结转移(有、无)、淋巴结转移数量、孕激素受体(阳性、阴性)、雌激素受体(阳性、阴性)、病理类型(导管内原位癌、浸润性导管癌、浸润性小叶癌)、分化程度(中分化、低分化)、肿瘤位置(外侧象限、中央或内侧象限)。

1.3.2 治疗前后血清CA50、TSGF、TPA表达水平

取2组患者治疗前后静脉血5 mL,1 000 g离心10 min,收集上清液,采用化学发光法检测CA50、TSGF、TPA水平,试剂盒购自上海雅吉生物。

1.4 统计学分析

采用SPSS 25.0软件处理数据。计数资料用率(%)表示,采用χ2检验进行比较。计量资料行Shapiro-Wilk正态性检验和Levene法方差齐性检验,方差齐性且近似服从呈正态分布的计量资料以x±s表示,采用t检验进行比较。采用LASSO回归、logistic回归分析中晚期乳腺癌放化疗敏感性的影响因素,运用R语言rms包构建列线图模型,采用校准曲线、受试者操作特征(receiver operating characteristic,ROC)曲线、决策曲线评估列线图模型的预测效能和临床效用。双侧P < 0.05为差异有统计学意义。

2 结果 2.1 2组一般资料的比较

2组比较,肿瘤直径、脉管侵犯、TNM分期、淋巴结转移、分化程度的差异有统计学意义(P < 0.05),年龄、月经状态、BMI、淋巴结转移数量、孕激素受体、雌激素受体、病理类型、肿瘤位置的差异无统计学意义(P > 0.05)。见表 1

表 1 2组一般资料的比较 Tab.1 Comparison of general information between the two groups
Clinical data Insensitive group(n = 25) Sensitive group(n = 57) t2 P
Age(year) 55.82±5.46 57.01±4.37 1.050 0.297
Tumor diameter(cm) 4.45±1.03 3.85±1.26 2.092 0.040
BMI(kg/m2 21.52±1.33 21.39±1.42 0.389 0.698
Menstrual status [n(%)] 0.326 0.568
  Menopause period 17(68.00) 35(61.40)
  Non-menopausal period 8(32.00) 22(38.60)
Vascular invasion [n(%)] 5.722 0.017
  Yes 10(40.00) 9(15.79)
  No 15(60.00) 48(84.21)
TNM stage [n(%)] 2.817 0.005
  Ⅲa 5(20.00) 29(50.88)
  Ⅲb 8(32.00) 17(29.82)
  Ⅳ 12(48.00) 11(19.30)
Lymph node metastasis [n(%)] 4.661 0.031
  Yes 17(68.00) 24(42.11)
  No 8(32.00) 33(57.89)
Number of metastatic lymph node [n(%)] 0.120 0.729
  ≤3 13(52.00) 32(56.14)
  >3 12(48.00) 25(43.86)
Progesterone receptor [n(%)] 1.590 0.207
  Positive 12(48.00) 19(33.33)
  Negative 13(52.00) 38(66.67)
Estrogen receptor [n(%)] 1.645 0.200
  Positive 13(52.00) 21(36.84)
  Negative 12(48.00) 36(63.16)
Pathological type [n(%)] 0.284 0.777
  Ductal carcinoma in situ 18(72.00) 43(75.44)
  Infiltrating ductal carcinoma 1(4.00) 3(5.26)
  Infiltrating lobular carcinoma 6(24.00) 11(19.30)
Degree of differentiation [n(%)] 4.355 0.037
  Moderately 11(44.00) 39(68.42)
  Poorly 14(56.00) 18(31.58)
Tumor location [n(%)] 0.106 0.745
  Outer quadrant 15(60.00) 32(56.14)
  Central or inner quadrant 10(40.00) 25(43.86)

2.2 2组治疗前后血清CA50、TSGF、TPA表达水平的比较

治疗前2组比较,血清CA50、TSGF、TPA表达水平的差异无统计学意义(P > 0.05);治疗后不敏感组血清CA50、TSGF、TPA表达水平高于敏感组,三者差值低于敏感组(P < 0.05)。见表 2

表 2 2组治疗前后血清CA50、TSGF、TPA表达水平的比较 Tab.2 Comparison of serum levels of CA50, TSGF, and TPA before and after treatment in the two groups
Group n CA50(U/mL) TSGF(U/L) TPA(U/L)
Before treatment After treatment Difference* Before treatment After treatment Difference* Before treatment After treatment Difference*
Insensitive 25 44.96±5.52 35.52±4.08 9.44±3.05 90.02±8.44 70.72±6.62 19.30±3.44 72.90±6.46 65.48±5.95 7.42±1.33
Sensitive 57 45.78±4.69 19.91±3.32 25.87±4.44 88.96±9.56 40.35±4.42 48.61±5.89 74.11±5.02 50.52±4.77 23.59±3.56
t 0.690 18.253 16.816 0.478 24.445 23.159 0.918 12.104 21.983
P 0.492 <0.001 <0.001 0.634 <0.001 <0.001 0.361 <0.001 <0.001
* The difference is considered an absolute value.

2.3 预测因素初筛

应用R语言glmnet程序包,以中晚期乳腺癌放化疗敏感性(不敏感=1,敏感=0)为因变量,肿瘤直径(实测值)、脉管侵犯(有=1,无=0)、TNM分期(Ⅲa期=1,Ⅲb期=2,Ⅳ期=3)、淋巴结转移(有=1,无=0)、分化程度(低分化=1,中分化=2)、治疗后血清CA50、TSGF、TPA表达水平及差值(实测值)纳入LASSO回归分析。随着惩罚系数λ变化,模型初始纳入影响因素系数被压缩,达到最佳影响因素选择的效果图(图 1)。采用交叉验证法绘制均方误差随lgλ的变化图(图 2),寻找到可使模型性能优良且影响因素最少的最佳惩罚系数λ,即图中虚线对应的λ=3。依据此λ值,选出3个预测变量,分别为TSGF差值、CA50差值、TPA差值。

图 1 LASSO回归筛选变量动态过程图 Fig.1 Dynamic graph of screening variable using LASSO regression

图 2 交叉验证最佳参数λ的选择过程图 Fig.2 Selection process of the optimal parameter λ for cross-validation

2.4 中晚期乳腺癌放化疗敏感性影响因素的多因素logistic回归分析

以中晚期乳腺癌放化疗敏感性(不敏感=1,敏感=0)为因变量,以LASSO回归筛选指标作为自变量,纳入多因素logistic回归分析,结果显示,CA50、TSGF、TPA差值是中晚期乳腺癌放化疗敏感性的影响因素(P < 0.05),见表 3

表 3 中晚期乳腺癌放化疗敏感性影响因素的多因素logistic回归分析 Tab.3 Multivariate logistic regression analysis of factors influencing sensitivity to radiochemotherapy in middle- and advanced-stage breast cancer
Independent variable β SE Wald χ2 OR 95%CI P
Constant 12.456
Difference in CA50 -0.596 0.171 12.149 0.551 0.345-0.888 <0.001
Difference in TSGF -2.561 0.496 26.663 0.077 0.011-0.542 <0.001
Difference in TPA -0.718 0.165 18.955 0.488 0.346-0.687 <0.001

2.5 列线图模型的构建

基于多因素logistic回归分析结果(CA50、TSGF、TPA差值),运用R语言rms包构建中晚期乳腺癌放化疗敏感性列线图模型。见图 3

图 3 中晚期乳腺癌放化疗敏感性列线图模型 Fig.3 Nomogram model for predicting sensitivity to radiochemotherapy in middle- and advanced-stage breast cancers

2.6 列线图模型的验证与评价

采用Hosmer-Lemeshow检验检测列线图模型拟合度,结果显示,该列线图预测模型具有良好拟合度(P = 0.482)。采用Bootstrap自抽样法及计算预测模型区分度(C-index),展开内部验证,结果显示,该列线图模型C-index为0.839,校准曲线趋向于理想曲线,见图 4A。ROC曲线显示,列线图模型的AUC为0.839(95%CI:0.814~0.871),灵敏度和特异度分别为80.00%和75.74%,说明基于CA50、TSGF、TPA差值构建的列线图模型具有良好的预测精准度,见图 4B。决策曲线显示,在0.1~1.0范围内,该列线图模型预测净获益值较高,说明该列线图模型临床预测效能良好,见图 4C

A, calibration curve; B, ROC curve; C, decision curve. 图 4 中晚期乳腺癌放化疗敏感性列线图模型的校准曲线、ROC曲线和决策曲线 Fig.4 Calibration, ROC, and decision curves of the nomogram model for predicting sensitivity to radiochemotherapy in middle- and advanced-stage breast cancer

3 讨论

研究[7]显示,中晚期乳腺癌患者采取同步放化疗治疗,可提高临床缓解率,缩小肿瘤体积,并改善生活质量和免疫功能。但部分患者未获得临床缓解,导致预后不佳,还易出现复发、转移等。因此,寻找一些特异度和灵敏度高的无创分子标志物,对预测中晚期乳腺癌患者放化疗敏感性具有重要作用。

肿瘤标志物主要以激素、抗原、酶等形式存在于肿瘤细胞和宿主体液中,可准确反映肿瘤细胞的生化性质和代谢情况,临床主要用于鉴别良恶性肿瘤、筛选肿瘤高危人群、判断肿瘤进展、评估肿瘤预后等[8]。血清肿瘤标志物检测是一种非介入性检查措施,操作简单、检测快捷。目前临床发现,血清CA153、癌胚抗原、CA125等在乳腺癌患者组织和血清中差异表达,且与远处转移、分化程度、血管形成等密切相关,可用于乳腺癌的早期诊断、预后评估等[9-10]。CA50、TSGF、TPA是临床常见肿瘤标志物。CA50是一种非特异性肿瘤抗原,与癌抗原19-9具有交叉性。CA50在人正常组织中含量较低,细胞癌变后可激活糖基化酶,引起细胞膜表层糖基结构改变,导致大量CA50进入血液或组织中[11]。已有研究[12]显示,CA50在多种恶性肿瘤中高表达,肿瘤分期越高、分化程度越低、直径越小,CA50阳性率越高。TSGF是国际公认的与恶性肿瘤有关的广谱肿瘤标志物,是由多种小分子代谢产物、氨基酸和糖类物质形成的一种多肽因子,TSGF分泌增多可刺激恶性肿瘤细胞新血管形成,促进肿瘤恶性生长分化、转移和毛细血管增生;TSGF在多种肿瘤中表达升高,具有较高灵敏度和特异度,可用于良恶性肿瘤鉴别、预后评估等[13]。TPA是一种与角蛋白8、18、19有关的一种多肽物质,主要在多种恶性肿瘤组织和胎盘组织中表达;研究[14]发现,TPA水平升高与恶性肿瘤直径、分化程度、浸润程度等有关。本研究发现,治疗后不敏感组血清CA50、TSGF、TPA表达水平高于敏感组,CA50、TSGF、TPA差值低于敏感组,提示血清CA50、TSGF、TPA表达水平与中晚期乳腺癌患者放化疗敏感性有关。笔者推测:患者肿瘤直径越大、脉管发生侵犯、TNM分期越高、淋巴结发生转移可促进肿瘤细胞恶性生长,进而促进新血管形成;放化疗治疗后,约30.49%的患者不敏感,影响治疗效果,故不敏感组患者治疗前后血清CA50、TSGF、TPA差值变化较小。

本研究使用LASSO回归筛选的有统计学差异的11个变量,发现最佳惩罚系数λ=3,筛选出3个预测变量为治疗前后TSGF、CA50、TPA差值。logistic回归也证实三者差值是中晚期乳腺癌放化疗敏感性的影响因素,进一步说明血清CA50、TSGF、TPA与中晚期乳腺癌放化疗敏感性有关。根据logistic回归结果构建中晚期乳腺癌放化疗敏感性列线图模型,提示三者差值越小,中晚期乳腺癌放化疗不敏感风险越高。采用C-index对该模型进行内部验证,发现C-index为0.839,具有较高的准确度,且校准曲线趋向于理想曲线;ROC曲线和决策曲线显示,该模型具有良好的预测精准度和临床预测效能。

综上所述,在中晚期乳腺癌放化疗不敏感患者中血清CA50、TSGF、TPA表达水平较高,且血清CA50、TSGF、TPA差值是患者放化疗敏感性的影响因素,基于以上因素构建的列线图预测模型,具有较好的参考价值和临床适用性。

参考文献
[1]
WILKINSON L, GATHANI T. Understanding breast cancer as a global health concern[J]. Br J Radiol, 2022, 95(1130): 20211033. DOI:10.1259/bjr.20211033
[2]
WANG X, WANG SS, HUANG H, et al. Effect of capecitabine maintenance therapy using lower dosage and higher frequency vs observation on disease-free survival among patients with early-stage triple-negative breast cancer who had received standard treatment: the SYSUCC-001 randomized clinical trial[J]. JAMA, 2021, 325(1): 50-58. DOI:10.1001/jama.2020.23370
[3]
张雁, 刘松岭, 周爱凤, 等. 乳腺癌患者血清因子水平与疾病的相关性研究[J]. 中国肿瘤临床与康复, 2022, 29(4): 412-415. DOI:10.13455/j.cnki.cjcor.2022.04.07
[4]
陈鹏, 马德寿, 赵海宁. 保乳术治疗对乳腺癌患者疗效、血清肿瘤标志物的影响及预后分析[J]. 湖南师范大学学报(医学版), 2019, 16(3): 51-54. DOI:10.3969/j.issn.1673-016X.2019.03.016
[5]
中国抗癌协会乳腺癌专业委员会. 中国抗癌协会乳腺癌诊治指南与规范(2017年版)[J]. 中国癌症杂志, 2017, 27(9): 695-759. DOI:10.19401/j.cnki.1007-3639.2017.09.004
[6]
杨学宁, 吴一龙. 实体瘤治疗疗效评价标准——RECIST[J]. 循证医学, 2004, 4(2): 85-90. DOI:10.3969/j.issn.1671-5144.2004.02.012
[7]
张叶青, 孙成晖, 卢柳岑. 心理护理干预在多西他赛+表柔比星方案化疗中晚期乳腺癌患者中的应用效果[J]. 中国医药科学, 2020, 10(19): 143-146. DOI:10.3969/j.issn.2095-0616.2020.19.035
[8]
IWAMOTO T, KAJIWARA Y, ZHU Y, et al. Biomarkers of neoadjuvant/adjuvant chemotherapy for breast cancer[J]. Chin Clin Oncol, 2020, 9(3): 27. DOI:10.21037/cco.2020.01.06
[9]
LI H, WANG S, LI X, et al. Dual-channel detection of breast cancer biomarkers CA15-3 and CEA in human serum using dialysis-silicon nanowire field effect transistor[J]. Int J Nanomedicine, 2022, 17: 6289-6299. DOI:10.2147/IJN.S391234
[10]
LI J, LIU L, FENG Z, et al. Tumor markers CA15-3, CA125, CEA and breast cancer survival by molecular subtype: a cohort study[J]. Breast Cancer, 2020, 27(4): 621-630. DOI:10.1007/s12282-020-01058-3
[11]
LI H, LI L, SUN J, et al. Value of TCT combined with serum CA153 and CA50 in early diagnosis of cervical cancer and precancerous lesions[J]. Pak J Med Sci, 2022, 38(6): 1471-1476. DOI:10.12669/pjms.38.6.5503
[12]
ZHANG Q, DONG G, WANG F, et al. Correlation between the changes of serum COX-2, APE1, VEGF, TGF-β and TSGF levels and prognosis in patients with osteosarcoma before and after treatment[J]. J Cancer Res Ther, 2020, 16(2): 335-342. DOI:10.4103/jcrt.JCRT_11_20
[13]
YU C, SUN C. Diagnostic value of multislice spiral computed tomography combined with serum AFP, TSGF, and GP73 assay in the diagnosis of primary liver cancer[J]. Evid Based Complement Alternat Med, 2022, 2022: 6581127. DOI:10.1155/2022/6581127
[14]
CHEN Z, LIU X, SHANG X, et al. The diagnostic value of the combination of carcinoembryonic antigen, squamous cell carcinoma-related antigen, CYFRA 21-1, neuron-specific enolase, tissue polypeptide antigen, and progastrin-releasing peptide in small cell lung cancer discrimination[J]. Int J Biol Markers, 2021, 36(4): 36-44. DOI:10.1177/17246008211049446