成人胶质瘤早期死亡影响因素分析及风险预测模型构建

摘要/Abstract

摘要：

目的探讨成人胶质瘤发生早期死亡（3个月）的影响因素，并构建风险预测模型。方法回顾性分析第九〇九医院（厦门大学附属东南医院）2020年6月至2024年6月收治的228例成人胶质瘤患者的临床资料。根据患者3个月内是否发生死亡分为死亡组（n=32）和存活组（n=196），比较两组临床资料。采用多因素logistic回归分析患者3个月内死亡的影响因素，并构建logistic回归预测模型，绘制受试者操作特征（ROC）曲线分析该模型的预测价值。结果两组患者年龄、性别、高血压、糖尿病、肿瘤部位、肿瘤累及范围、神经功能受损、肿瘤最大径、化疗、放疗差异均无统计学意义（均P>0.05），死亡组患者脑疝（χ²=20.74，P<0.001）、入院Karnofsky功能状态（KPS）评分≤70分（χ²=26.66，P<0.001）、肿瘤分级Ⅲ~Ⅳ级（χ²=28.70，P<0.001）、MGMT启动子未甲基化（χ²=10.25，P=0.001）、IDH野生型（χ²=6.18，P=0.013）、肿瘤未全切除（χ²=10.37，P=0.001）比例高于存活组。多因素分析显示，脑疝（OR=19.78，95%CI为5.33~73.41，P<0.001）、入院KPS评分≤70分（OR=19.64，95%CI为5.54~69.59，P<0.001）、肿瘤分级Ⅲ~Ⅳ级（OR=9.40，95%CI为3.02~29.27，P<0.001）、MGMT启动子未甲基化（OR=4.28，95%CI为1.18~15.54，P=0.027）、肿瘤未全切除（OR=9.50，95%CI为2.72~33.23，P<0.001）是胶质瘤患者发生早期死亡的独立危险因素。基于上述指标构建胶质瘤患者早期死亡风险预测模型为logit（P）=-18.04+2.96×脑疝（有=1、无=0）+2.98×入院KPS评分（≤70分=1、>70分=0）+2.24×肿瘤分级（Ⅲ~Ⅳ=1、Ⅰ~Ⅱ=0）+1.45×MGMT启动子甲基化（否=1、是=0）+2.25×肿瘤全切除（否=1、是=0）。ROC曲线分析显示，该模型对胶质瘤患者早期死亡具有预测价值，曲线下面积为0.920，95%CI为0.868~0.972，敏感性为0.842，特异性为0.906。结论脑疝、入院KPS评分≤70分、肿瘤分级Ⅲ~Ⅳ级、MGMT启动子未甲基化、肿瘤未全切除是成人胶质瘤患者发生早期死亡的独立危险因素，基于以上指标构建的风险预测模型具有较好的预测价值。

关键词:神经胶质瘤,死亡,预后,脑膨出

Abstract:

ObjectiveTo explore the influencing factors of early death （within 3 months） in adult glioma patients， and to construct a risk prediction model.MethodsRetrospective analysis was performed on the clinical data of 228 adult glioma patients admitted to the 909th Hospital （Dongnan Hospital of Xiamen University） from June 2020 to June 2024. Patients were divided into a death group （n=32） and a survival group （n=196） based on whether death occurred within 3 months， and the clinical data between the two groups were compared. Multivariate logistic regression was used to analyze the influencing factors of death within 3 months， a logistic regression prediction model was constructed， and receiver operator characteristic （ROC） curve was plotted to analyze the predictive value of the model.ResultsThere were no statistically significant differences between the two groups in age， gender， hypertension， diabetes， tumor location， tumor involvement， neurological impairment， maximum tumor diameter， chemotherapy， or radiotherapy （allP>0.05）. The death group showed higher proportions of cerebral herniation （χ²=20.74，P<0.001）， hospital admission Karnofsky performance status （KPS） score ≤70 （χ²=26.66，P<0.001）， tumor grade Ⅲ-Ⅳ （χ²=28.70，P<0.001）， MGMT promoter unmethylation （χ²=10.25，P=0.001）， IDH wild-type （χ²=6.18，P=0.013）， and incomplete tumor resection （χ²=10.37，P=0.001） compared with the survival group. Multivariate analysis revealed that cerebral herniation （OR=19.78， 95%CI： 5.33-73.41，P<0.001）， hospital admission KPS score ≤70 （OR=19.64， 95%CI： 5.54-69.59，P<0.001）， tumor grade Ⅲ-Ⅳ （OR=9.40， 95%CI： 3.02-29.27，P<0.001）， MGMT promoter unmethylation （OR=4.28， 95%CI： 1.18-15.54，P=0.027）， and incomplete tumor resection （OR=9.50， 95%CI： 2.72-33.23，P<0.001） were independent risk factors for early death in glioma patients. The risk prediction model for early death in glioma patients constructed based on these indicators was logit（P）=-18.04+2.96×cerebral herniation （with=1， without=0）+2.98×hospital admission KPS score （≤70=1， >70=0）+2.24×tumor grade （Ⅲ-Ⅳ=1， Ⅰ-Ⅱ=0）+1.45×MGMT promoter methylation （no=1， yes=0）+2.25×complete tumor resection （no=1， yes=0）. ROC curve analysis demonstrated that this model had predictive value for early death in glioma patients， with an area under the curve of 0.920 （95%CI： 0.868-0.972）， a sensitivity of 0.842， and a specificity of 0.906.ConclusionsCerebral herniation， hospital admission KPS score ≤70， tumor grade Ⅲ-Ⅳ， MGMT promoter unmethylation， and incomplete tumor resection are independent risk factors for early death in adult glioma patients. The risk prediction model constructed based on these indicators has good predictive value.

Key words:Glioma,Death,Prognosis,Encephalocele

代玉娟, 陈羡英, 黄巍, 陈大朝. 成人胶质瘤早期死亡影响因素分析及风险预测模型构建[J]. 国际肿瘤学杂志, 2025, 52(10): 609-613.

Dai Yujuan, Chen Xianying, Huang Wei, Chen Dachao. Analysis of influencing factors and construction of a risk prediction model for early death in adult glioma[J]. Journal of International Oncology, 2025, 52(10): 609-613.

图/表3

表1

两组成人胶质瘤患者临床资料对比（例）"

临床资料	死亡组（n=32）	存活组（n=196）	χ²值	P值	临床资料	死亡组（n=32）	存活组（n=196）	χ²值	P值
年龄（岁）					入院KPS评分（分）
<50	21	109	1.13	0.289	≤70	18	31	26.66	<0.001
≥50	11	87	1.13	0.289	>70	14	165	26.66	<0.001
性别					肿瘤最大径（cm）
男	21	116	0.48	0.490	<3	18	112	0.01	0.925
女	11	80	0.48	0.490	>3	14	84	0.01	0.925
高血压					肿瘤分级
有	8	25	2.42	0.120	Ⅰ~Ⅱ	11	156	28.70	<0.001
无	24	171	2.42	0.120	Ⅲ~Ⅳ	21	40	28.70	<0.001
糖尿病					MGMT启动子甲基化
有	8	26	2.13	0.144	是	20	168	10.25	0.001
无	24	170	2.13	0.144	否	12	28	10.25	0.001
肿瘤部位					IDH野生型
左侧	16	100			否	22	169	6.18	0.013
右侧	13	87	1.29	0.525	是	10	27	6.18	0.013
中线或双侧	3	9			肿瘤全切除
肿瘤累及范围					否	25	93	10.37	0.001
单叶	18	123	0.49	0.482	是	7	103	10.37	0.001
多叶	14	73	0.49	0.482	化疗
脑疝					无	5	19	0.49	0.482
无	18	173	20.74	<0.001	有	27	177	0.49	0.482
有	14	23	20.74	<0.001	放疗
神经功能受损					无	5	21	0.26	0.610
无	24	171	2.42	0.120	有	27	175	0.26	0.610
有	8	25	2.42	0.120

表1

因素	β值	SE值	Waldχ²值	OR值	95%CI	P值
脑疝（有/无）	2.96	0.67	19.91	19.78	5.33~73.41	<0.001
入院KPS评分（≤70分/>70分）	2.98	0.65	21.27	19.64	5.54~69.59	<0.001
肿瘤分级（Ⅲ~Ⅳ/Ⅰ~Ⅱ）	2.24	0.58	14.96	9.40	3.02~29.27	<0.001
MGMT启动子甲基化（否/是）	1.45	0.66	4.88	4.28	1.18~15.54	0.027
IDH野生型（是/否）	0.39	0.66	0.36	1.48	0.41~5.37	0.551
肿瘤全切除（否/是）	2.25	0.64	12.42	9.50	2.72~33.23	<0.001

表2

图1

参考文献25

[1]	Yin J, Liu G, Zhang Y, et al. Gender differences in gliomas: from epidemiological trends to changes at the hormonal and molecular levels[J].Cancer Lett,2024,598: 217114. DOI:10.1016/j.canlet.2024.217114.
[2]	Drexler R, Khatri R, Sauvigny T, et al. A prognostic neural epigenetic signature in high-grade glioma[J].Nat Med,2024,30(6): 1622-1635. DOI:10.1038/s41591-024-02969-w. pmid:38760585
[3]	国家卫生健康委员会医政医管局. 脑胶质瘤诊疗规范(2018年版)[J].中华神经外科杂志,2019,35(3): 217-239. DOI:10.3760/cma.j.issn.1001-2346.2019.03.001.
[4]	陈雯琳, 王月坤, 刘千舒, 等. 2022年度我国脑胶质瘤领域研究进展[J].协和医学杂志,2023,14(5): 983-990. DOI:10.12290/xhyxzz.2023-0321.
[5]	Huang F, Li S, Wang X, et al. Serum lipids concentration on prognosis of high-grade glioma[J].Cancer Causes Control,2023,34(9): 801-811. DOI:10.1007/s10552-023-01710-1.
[6]	Zhang D, Li H, Jia W. Exploration of the prognostic value of the resection of adult brainstem high-grade glioma based on competing risk model, propensity score matching, and conditional survival rate[J].Neurol Sci,2023,44(5): 1755-1764. DOI:10.1007/s10072-022-06557-z. pmid:36604383
[7]	Wu S, Wang C, Li N, et al. Analysis of prognostic factors and surgical management of elderly patients with low-grade gliomas[J].World Neurosurg,2023,176: e20-e31. DOI:10.1016/j.wneu.2023.02.099.
[8]	Lin Y, Li H, Ge Q, et al. Establishment and validation of a prognostic prediction model for glioma based on key genes and clinical factors[J].Transl Cancer Res,2025,14(1): 240-253. DOI:10.21037/tcr-24-1035. pmid:39974385
[9]	Hanif S, Yousaf I, Iqbal M, et al. Brain herniation and subsequent complications following partial resection of high-grade glioma: a case report[J].Clin Case Rep,2024,12(1): e8407. DOI:10.1002/ccr3.8407.
[10]	Lipková J, Menze B, Wiestler B, et al. Modelling glioma progression, mass effect and intracranial pressure in patient anatomy[J].J R Soc Interface,2022,19(188): 20210922. DOI:10.1098/rsif.2021.0922.
[11]	Pasqualetti F, Lombardi G, Gadducci G, et al. Brain stem glioma recurrence: exploring the therapeutic frontiers[J].J Pers Med,2024,14(9): 899. DOI:10.3390/jpm14090899.
[12]	Fuster-Garcia E, Thokle Hovden I, Fløgstad Svensson S, et al. Quantification of tissue compression identifies high-grade glioma patients with reduced survival[J].Cancers (Basel),2022,14(7): 1725. DOI:10.3390/cancers14071725.
[13]	Gunawan PY, Islam AA, July J, et al. Karnofsky performance scale and neurological assessment of neuro-oncology scale as early predictor in glioma[J].Asian Pac J Cancer Prev,2020,21(11): 3387-3392. DOI:10.31557/APJCP.2020.21.11.3387.
[14]	Hei B, Ouyang J, Gao Q, et al. Analysis of changes in KPS scores and imaging signs in deep brain gliomas by DWI combined with intraoperative ultrasound resection[J].Altern Ther Health Med,2024,30(5): 110-117. DOI:10.23736/S0390-5616.19.04735-0.
[15]	Shin I, Sim Y, Choi SH, et al. Revisiting prognostic factors of gliomatosis cerebri in adult-type diffuse gliomas[J].J Neurooncol,2024,168(2): 239-247. DOI:10.1007/s11060-024-04656-9.
[16]	Kivioja T, Posti JP, Sipilä J, et al. Motor dysfunction as a primary symptom predicts poor outcome: multicenter study of glioma symptoms[J].Front Oncol,2023,13: 1305725. DOI:10.3389/fonc.2023.1305725.
[17]	Decavèle M, Gatulle N, Weiss N, et al. One-year survival of patients with high-grade glioma discharged alive from the intensive care unit[J].J Neurol,2021,268(2): 516-525. DOI:10.1007/s00415-020-10191-0.
[18]	Wan Y, Li G, Deng J, et al. A gene signature predicting prognosis of patients with lower-grade gliomas receiving temozolomide therapy[J].Horm Cancer,2023,14(1): 202. DOI:10.1007/s12672-023-00818-9.
[19]	Yu S, Wu J, Jing Y, et al. Research trends in glioma chemoradiotherapy resistance: a bibliometric analysis (2003-2023)[J].Front Oncol,2025,15: 1539937. DOI:10.3389/fonc.2025.1539937.
[20]	Wetzel EA, Nohman AI, Hsieh AL, et al. A multi-center, clinical analysis of IDH-mutant gliomas, WHO grade 4: implications for prognosis and clinical trial design[J].J Neurooncol,2025,171(2): 373-381. DOI:10.1007/s11060-024-04852-7.
[21]	Zhang J, Qiu X, Feng J, et al. MGMT promoter methylation is a strong prognostic factor for survival after progression in high-grade gliomas[J].Chin Neurosurg J,2024,10(1): 24. DOI:10.1186/s41016-024-00375-2.
[22]	Singh A, Singh A, Agrawal S, et al. Prognostic significance and clinicopathological correlations of epigenetic MGMT gene silencing in high grade diffuse gliomas[J].Discoveries (Craiova),2023,11(3): e175. DOI:10.15190/d.2023.14.
[23]	朱一硕, 崔玉洁, 刘崎, 等. 脑胶质瘤患者术后早期复发危险因素分析及预测模型构建[J].国际肿瘤学杂志,2022,49(2): 79-83. DOI:10.3760/cma.j.cn371439-20210121-00012.
[24]	Fan ZC, Zhao WJ, Jiao Y, et al. Risk factors and predictive nomogram for survival in elderly patients with brain glioma[J].Curr Med Sci,2024,44(4): 759-770. DOI:10.1007/s11596-024-2880-4.
[25]	牛弘川, 梁剑峰. 脑胶质瘤预后因素分析及其列线图预测模型的建立[J].中华肿瘤防治杂志,2023,30(3): 140-146. DOI:10.16073/j.cnki.cjcpt.2023.03.03.