【作者】 王勇；

【导师】 袁双虎；

【作者基本信息】 山东大学 ， 肿瘤学（专业学位）， 2022， 博士

【摘要】 第一章小细胞肺癌脑转移的空间分布及危险区域的识别研究背景:预防性全脑照射(PCI)已被广泛应用于小细胞肺癌(SCLC)患者预防脑转移的发生。尽管有强有力的证据支持对SCLC进行PCI治疗,但神经毒性风险增加,这可能导致认知能力的下降。保护海马(HS)的全脑放射治疗(WBRT)已被证实可减少神经认知能力下降的不良反应。SCLC脑转移的PCI是否也需要海马保护,海马免除照射后其转移的风险有多大,除了保护海马外其他部位是否也需要保护,首先需要了解SCLC脑转移的空间分布,了解各个脑区的转移风险,识别风险区域。然而检索国内外文献尚未发现有针对SCLC脑转移空间分布的研究。研究目的:探究SCLC的脑转移(BMs)的空间分布,并确定脑区域的转移风险水平。研究方法:本研究回顾性分析了来自中国三家医疗机构的SCLC患者的T1增强磁共振成像(MRI)。所有图像均配准到蒙特利尔神经学研究所(MNI)152数据库提供的标准大脑模板中,然后将所有脑转移瘤的位置转换到标准大脑空间。然后使用MNI结构图谱(10个脑区)、解剖自动标记(AAL)图谱(1 18个脑区)和改良后的AAL图谱(38个脑区)来分别识别解剖脑区,并使用双尾比例假设检验比较脑转移的观察率和预期率。图像标准化后,分析脑病变的位置和大小。识别出高风险和低风险的脑区。研究结果:通过MRI筛查出215例符合条件的SCLC患者,共1033个病灶,其中男性157例(73%),女性58例(27%)。关键结构的发生率为:海马0.68%,海马旁0.97%,脑干2.05%,尾状核0.68%,壳核0.68%,苍白球0.2%,丘脑1.36%。在杏仁核、脑下垂体或松果体中没有发现脑转移瘤。重要结构的累积频率为6.62%。基于MNI脑标准模板的10分区分割发现:小脑和脑深部结构(深部白质和脑干作为一个整体)被认定为高风险脑区(P=9.80×10-15,9.04×10-6),而颞叶是低风险脑区(P=1.65×10-4),将脑深部白质和脑干分开,单独分析脑干部位时发现其差异未见显著性。更详细的AAL图谱显示脑转移的低危险区域为额下回(P=6.971×10-4),而脑转移的高风险区域为小脑半球(P=1.177×10-9)。结论:在SCLC患者中有很多重要的脑功能区为低风险区域,特别是脑内重要结构区包括海马、海马旁、脑下垂体和丘脑等许多关键结构的脑转移频率较低。本研究为研究HA-WBRT和非均匀剂量PCI在SCLC患者中的临床可行性提供了帮助。第二章小细胞肺癌与非小细胞肺癌脑转移瘤空间分布的差异性分析研究背景:目前研究脑转移瘤空间分布差异的文献十分有限,而且有些研究结果之间还常存在矛盾。这些研究几乎都集中在分析乳腺癌和肺癌之间的差异。众所周知,小细胞肺癌(SCLC)和非小细胞肺癌(NSCLC)具有不同的基因突变模式和肿瘤生物学行为,因此我们假设他们的空间分布可能存在差异。而且目前小细胞肺癌和非小细胞肺癌脑转移的放疗策略也是不同的,因此研究其两者之间脑转移瘤空间分布的差异也显得十分有必要的。研究其空间分布不仅有利于预测脑转移部位,而且有利于优化全脑放疗中重要结构的保护。研究目的:探讨SCLC和NSCLC脑转移的空间分布差异,并确定脑区域的转移风险。研究方法:回顾性分析中国三家独立医疗机构728例符合条件的SCLC和NSCLC患者的2997个脑转移灶的T1增强MR图像。在3名放射科医生确认脑转移瘤轮廓后,在MITK软件上逐层勾画转移灶位置,将所有MRI图像变形配准到标准脑空间模板中。基于合并的AAL图谱(41个脑区),将所有脑区映射到配准后的图像中进行自动标记识别。采用卡方检验和Wilcoxon秩和检验比较SCLC和NSCLC患者的临床特征。采用双尾比例假设检验将脑转移的观察率与根据区域体积的预期率进行比较,最终确定转移危险区域。SCLC与NSCLC之间脑转移频率差异统计,将进一步采用更加严苛的Bonferroni校正多重比较,以1.219 × 10-3水平为有统计学意义。研究结果:在SCLC和NSCLC中,脑转移瘤均好发于深部白质(分别为22.51%和17.96%)、小脑半球(分别为9.84%和7.46%)和额中回(分别为6.72%和7.97%)。SCLC与NSCLC之间脑转移频率存在差异的脑区有:海马旁回(P=0.017),苍白球(P=0.050),小脑半球(P=0.024),深部脑白质(P=0.003)。然而采用更加严苛的Bonferroni校正多重比较后其差异未达到显著性要求。研究还发现在SCLC中脑转移在脑干、双侧丘脑、海马、海马旁回、杏仁核和颞极的累积频率为6.23%,在NSCLC中为3.77%。在这些关键结构中脑转移的累积频率在SCLC与NSCLC中有显著性差异(P=0.002)。危险区域分析发现:小脑半球(P=1.18×10-9)是SCLC的高危脑区。中央前回(P=1.75×10-15)、额中回(P=1.15×10-5)、中央旁小叶(P=4.49×10-4)和小脑半球(P=1.87×10-4)是NSCLC脑转移的高危部位。额下回(P=6.97×10-4)在SCLC中为低危脑区,颞极(P=6.77×10-4)在NSCLC中为低危脑区。在SCLC和NSCLC中,均未在杏仁核、脑下垂体或松果体中发现脑转移灶。结论:SCLC与NSCLC之间各脑区转移频率间存在细微的差异,然而更加严苛的多重检验矫正分析发现其间差异未见显著性。SCLC和NSCLC患者在小脑半球均为高风险区域。NSCLC较SCLC存在更多的风险区域。相比之下,脑干、双侧丘脑、海马、海马旁回、杏仁核、颞极等几个关键脑区在SCLC和NSCLC中出现脑转移的频率均相对较低,在充分评估的情况下可考虑给予低风险区域低剂量的放射分布。第三章常见癌种脑转移瘤空间分布图谱的构建、量化和影响因素探索研究背景:检索国内外文献发现罕有研究提供颅内常见癌种的脑转移瘤图谱及各个脑区的转移率的定量分析数据,且由于转移瘤位置分析技术的局限,很多既往研究多是研究人员仅通过观察影像资料就记录了肿瘤的解剖位置,缺乏准确性,很难精确的定位到具体的脑功能区。图像配准和三维结构变形算法的出现,将单个的脑图像变形和配准到平均的标准脑图像上成为可能,甚至可以检测出空间分布的微小差异。通过不同解剖自动标记图谱可以将脑区精细到我们想要的程度。定量分析的数据有利于临床医师查阅并评估不同脑区的转移风险。目前对于影响脑转移瘤空间分布的因素尚无统一的结论,既往研究几乎均是围绕某一单一因素进行分析,缺乏整体性。肿瘤栓塞的机械捕获理论认为血管口径突然改变,血流变慢,导致肿瘤栓塞的发生,是影响脑转移瘤分布的主要原因。而佩吉特的“种子和土壤”的理论认为某些肿瘤细胞(种子)对某些器官的环境(土壤)有一种特殊的亲和力,并且只有当种子和土壤是相容的,转移才会发生。许多文献也支持原发灶的病理差别对脑转移瘤的空间分布是有影响的。然而除此之外其他因素也是否有影响?因受影像分析技术和样本量的影响尚无系统的分析。研究目的:构建常见癌种脑转移瘤空间分布的图谱,定量分析不同脑功能区的转移频率。探讨影响脑转移瘤空间分布的相关因素。研究方法:在前期研究基础上继续扩大样本量,入组条件不再限于肺癌,增加乳腺癌、食管癌、肠癌、肾癌、及其他少见肿瘤的脑转移患者。按照前期方法人工勾画脑转移灶、图像配准、质心计算、构建可视化频率热图。基于MNI152脑图谱(10分区)、ALL脑图谱(118分区)和改良合并后的ALL图谱(38分区)利用计算机自动分割出各个脑功能区的转移灶数目。利用MITK软件在MNI标准脑模板中手动勾画了脑垂体、松果体、脑干、小脑蚓部和小脑半球(人工定义的谱图)。对于每个脑区,计算不同瘤种中脑转移瘤的数量及频率。将可能影响脑转移瘤空间分布的因素,分为内部因素和外部因素。内部因素包括:脑血管和血流因素。用“分水岭区”模式图构建脑血管边缘区模板,对配准后的图谱进行自动分割,得出各分水岭区脑转移瘤空间分布特点。外部因素包括:病理因素,治疗干预,基因突变状态,原发灶的位置。根据因素不同进行分组比较。使用内部软件自动计算脑表面到每个病灶中心的距离(深度)。分析不同因素下脑转移深度的差异。研究结果:共纳入914例脑转移患者,共4014个脑转移灶。其中肺癌患者728例,共2997个转移灶,乳腺癌患者59例,共计461个转移灶。食管癌患者21例,共计149个转移灶。肠癌患者13例,共计31个转移灶。肾癌患者1 1例,其他癌种82例。从可视化的热图中观察发现,脑功能不同区域的亮度并不均匀。部分额叶、双侧小脑、脑干区域突出,脑转移发生率高;双侧颞叶及部分枕叶亮度较低,脑转移发生率较低。而不同病理来源的转移瘤其空间分布在热图上也可看到不同的分布特点。各脑区脑转移瘤发生频率分别为:额叶1209(17.34%),顶叶739(13.95%),小脑 711(9.26%),颞叶 454(10.04%),枕叶 415(7.26%),深部白质和脑干 292(7.27%),丘脑 53(1.30%),壳核 39(0.86%),岛叶 48(0.83%),尾状核54(0.70%)。不同瘤种的各个脑区转移频率被列表展示。合并后的ALL图谱(38分区)分析发现转移率位于前十位的各脑区及转移灶数目(转移率)分别为:深部白质和脑干865(21.55%),小脑半球333(8.28%)额中回 291(7.24%)额上回 254(6.34%)中央前回 207(5.16%)小脑脚206(5.13%)枕回161(4.00%)中央后回159(3.96%)楔前叶125(3.11%)颞中回123(3.06%)。特别关注的部分脑功能区的转移率分别为:海马0.52%,海马旁回0.54%,苍白球0.07%,胼胝体裂隙和周围皮质2.04%,颞极0.96%。内部血管、血流因素分析发现无论是在分水岭前部(P=1.404×10-35)、后部(P=9.302×10-12),和小脑的分水岭(P=9.364×10-34)转移率均明显高于预期转移率。然而在大脑深部的灰白质交界区其转移率低于预期转移率(P=0.004)。病理组织来源对患者空间分布存在影响,肺癌在小脑半球高危(P=2.59×10-15)、颞叶低危(P=5.18×10-11),而乳腺癌尽管在小脑高危(P=4.92×10-7),但颞叶低危则未见显著性。肠癌最常发生转移的脑区均位于颅后窝:脑干和小脑半球。而肾癌好发转移部位为深部白质和脑干,额中回,颞下回。食管癌常见转移部位分别为脑深部白质和脑干,小脑半球,额中回。EGFR突变的NSCLC脑转移患者小脑半球为高危区域(P=0.007),颞叶为低危区域(P=0.029)。而EGFR无突变的NSCLC患者未发现危险区域,其所有MNI分区(10个脑区)均P>0.05。基于合并后的ALL分区(38个脑区)发现EGFR突变的患者在额中回(P=5.88×10-4)、中央前回为高危脑区(P=0.007),颞极为低危脑区(P=0.005),而EGFR未突变的患者未发现这一表现(P>0.05)。从转移深度角度研究发现EGFR突变的患者转移深度要大于无突变患者(11.823mm VS 11.463mm),但差异无显著性。转移前是否接受过化疗对MNI图谱(10分区)中的脑区危险度无明显影响,均表现为小脑高危和颞叶低危。基于合并后ALL图谱分析发现化疗后的患者其分布表现相对不均匀性增加,表现为比未化疗患者增加三个低危区域:缘上回(P=0.003×10-1),颞上回(P=0.003),颞极(P=0.009)。肺癌患者原发灶左右侧位置来源对转移瘤分布无影响(所有脑区均P>0.05)。结论:我们通过多中心的大样本研究构建了常见癌种脑转移瘤空间分布的图谱,用三种分割图谱定量分析了不同脑功能区的转移频率,识别出了一系列高风险和低分险的脑功能区。研究发现脑转移瘤的空间分布是不均一的,内部血管、血流分布因素及外部病理组织来源、基因突变状态、脑转移前治疗方式均会影响脑转移瘤的分空间布,且原发灶的左右位置对脑转移瘤空间分布并无影响。通过与预期频率的比较筛选出了常见瘤种的一系列危险区域并制作成图谱,这或将对脑转移瘤的放疗治疗的策略提供必要的帮助。更多还原

【Abstract】 Chapter one Spatial distribution of brain metastases in small cell lung cancer and identification of risk areasBackground:Prophylactic cranial irradiation(PCI)has been widely used to prevent brain metastases in patients with small cell lung cancer(SCLC).Although there are strong evidences supporting PCI for SCLC,there is an increased risk of neurotoxicity,which may lead to cognitive decline.Whole brain radiation therapy(WBRT)that protects the hippocampus(HS)has been shown to reduce the adverse effects of neurocognitive decline.Whether PCI of SCLC brain metastases also requires hippocampal protection,what is the risk of hippocampal metastasis after exemption from irradiation,and whether other parts besides hippocampal protection also need to be protected?Firstly,it is necessary to understand the spatial distribution of SCLC brain metastases,understand the risk of metastasis in each brain region,and identify risk areas.However,no study on spatial distribution of SCLC brain metastases has been found by searching domestic and foreign literature.Objectives:This study aimed to explore the spatial distribution of brain metastases(BMs)in SCLC and determine the level of metastasis risk in brain regions.Methods:T1-enhanced magnetic resonance imaging(MRI)from SCLC patients were retrospectively reviewed from three medical institutions in China.All images were registered to the standard brain template provided by the Montreal Neurological Institute(MNI)152 database,followed by transformation of the location of all BMs to the space of standard brain.The MNI structural atlas and Anatomical Automatic Labeling(AAL)atlas and modified AAL atlas(38 brain regions)were used to identify anatomic brain regions,respectively,and the observed and expected rates of BMs were compared using the two-tail proportionality hypothesis test.After the images were standardized,the location and size of the lesions were analyzed and the high-risk and low-risk brain region were identified.Results:A total of 215 eligible patients with 1033 lesions were screened by MRI,including 157(73%)males and 58(27%)females.The incidence of crucial structures were as follows:hippocampus 0.68%,parahippocampal 0.97%,brainstem 2.05%,cauate 0.68%,putamen 0.68%,pallidum 0.2%,thalamus 1.36%.No BMs were found in the amygdala,pituitary gland,or pineal gland.The cumulative frequency of the important structures was 6.62%.Based on the results of MNI structural atlas,the cerebellum,deep white matter and brainstem was identified as a higher-risk region than expected for BMs(P=9.80×10-15,9.04×10-6),whereas temporal lobe were low-risk regions(P=1.65×10-4).More detailed AAL atlas revealed that the lowrisk regions for BMs was inferior frontal gyrus(P=6.971×10-4),while the high-risk regions for BMs was cerebellar hemispheres(P=1.177×10-9).Conclusion:Many crucial structures including the hippocampus,parahippocampus,pituitary gland and thalamus had low frequency of brain metastases in a population of SCLC patients.This study provided the help to investigate the clinical feasibility of HA-WBRT and non-uniform dose of PCI in a population of SCLC patients.Chapter two Differences in spatial distribution of brain metastases between small cell lung cancer and non-small cell lung cancerBackground:At present,the literatures on the difference of spatial distribution of BMs are very limited,and the results of some studies are often contradictory.Almost all of these studies have focused on analyzing differences between breast and lung cancer.It is well known that small cell lung cancer(SCLC)and non-small cell lung cancer(NSCLC)patients have different patterns of gene mutation and tumor biological behavior,so we hypothesized that their spatial distribution may differ.In addition,the current radiotherapy strategies for brain metastases of small cell lung cancer and nonsmall cell lung cancer are also different,so it is necessary to explore the differences in spatial distribution of brain metastases between them.The study of its spatial distribution is not only helpful to predict the location of BMs,but also to optimize the protection of important structures in whole brain radiotherapy.Objectives:The aims of our study were to investigate the difference of spatial distribution of brain metastases(BMs)between small-cell lung cancer(SCLC)and non-small-cell lung cancer(NSCLC),and to identify the metastatic risk in brain regions.Methods:T1-enhanced MR images of 2997 BMs from 728 eligible SCLC and NSCLC patients from three independent medical institutions in China were retrospectively analyzed.After the contour of BMs was confirmed by 3 radiologists,the location of metastases was delineated layer by layer on MITK software,and all images were spatially normalized to the standard spatial template.The brain regions in the normalized image were identified based on the combined AAL map(41 brain regions),and all the brain regions were mapped to the brain regions in the normalized image.Chi-square test and Wilcoxon rank-sum test were used to compare the clinical characteristics of patients with SCLC and NSCLC.The observed rate of BMs was compared with the expected rate based on area volume using a two-tailed proportionality hypothesis test to determine the metastatic risk area.The frequency difference of BMs between SCLC and NSCLC was statistically significant at the level of 1.219×10-3 by multiple comparison with Bonferroni correction.Results:In both SCLC and NSCLC,BMs were biased to the deep white matter(22.51%and 17.96%,respectively),cerebellar hemisphere(9.84%and 7.46%,respectively),and middle frontal gyrus(6.72%and 7.97%,respectively).The cerebellar hemisphere(P=1.18x10-9)was a high-risk brain region in SCLC.The precentral gyrus(P=1.75×1015),middle frontal gyrus(P=1.15×10-5),paracentral lobule(P=4.49×10-4),and cerebellar hemisphere(P=1.87×10-4)were at high risk of BM in NSCLC.The inferior frontal gyrus(P=6.97×10-4)was a low-risk brain region in SCLC and the temporal pole(P=6.77×10-4)was at low risk of BM in NSCLC.In both SCLC and NSCLC,no BM was found in the amygdala,pituitary gland,or pineal gland.Conclusion:There were slight differences in the frequency of metastasis in various brain regions between SCLC and NSCLC,and no significant differences were found after more rigorous multiple test correction analysis.The cerebellar hemisphere in patients with SCLC or NSCLC showed a high risk of developing BMs.In comparison,several critical brain regions,including the brainstem,bilateral thalamus,hippocampus,parahippocampal gyrus,amygdala,and temporal pole had relatively low frequency of BMs from both SCLC and NSCLC,where the low-dose distribution of radiation may be given under the condition of adequate preoperative evaluations.Chapter three Construction,quantification and influencing factors of spatial distribution map of common cancer brain metastasesBackground:A search of domestic and foreign literature showed that few studies provided maps of brain metastases of common intracranial cancers and quantitative analysis data of metastasis rates in each brain region.In addition,due to the limitations of metastatic tumor location analysis technology,many previous studies mostly recorded the anatomical location of tumors only by observing image data,which lacked accuracy and made it difficult to accurately locate specific brain functional areas.With the advent of image registration and 3D structure deformation algorithms,it is possible to transform and register a single brain image onto an average standard brain image,and even detect small differences in spatial distribution.We can fine-tune brain regions to the extent that we want to through different anatomical autolabeling atlas.Quantitative analysis of the data allows clinicians to review and assess the risk of metastasis in different brain regions.At present,there is no unified conclusion on the factors affecting the spatial distribution of BMs,and almost all previous studies focused on a single factor,lacking integrity.The mechanical capture theory of tumor embolism holds that the sudden change of vascular diameter and slow blood flow lead to the occurrence of tumor embolism,which is the main reason affecting the distribution of BMs.Paget’s "seed and soil"theory suggests that certain tumor cells(seeds)have a special affinity for the environment of certain organs(soil),and that metastasis can only occur if the seeds and soil are compatible.Many literatures also support the influence of primary pathological differences on the spatial distribution of BMs.But are there other factors also affect?There is no systematic analysis due to the influence of image analysis technology and sample size.Objectives:The spatial distribution map of brain metastases of common cancer species was constructed to quantitatively analyze the frequency of metastasis in different brain functional areas.To explore the factors influencing spatial distribution of brain metastases.Methods:On the basis of previous studies,the sample size was further expanded,and the inclusion conditions were no longer limited to lung cancer,but increased the number of patients with brain metastasis from breast cancer,esophageal cancer,colon cancer,kidney cancer,and other rare tumors.According to the previous method,brain metastases were delineated manually,image registration,centroid calculation and visual frequency heat map were constructed.Based on MNI152 brain atlas(10 regions),ALL brain atlas(118 regions)and modified merged ALL atlas(38 regions),the observed metastasis frequency of each brain functional region was segmented automatically by computer.The pituitary gland,pineal gland,brain stem,cerebellar vermis and cerebellar hemisphere were delineated manually in MNI standard brain template by MITK software.For each brain region,the number and frequency of brain metastases in different tumor types were calculated.Factors that may affect the spatial distribution of BMs were divided into internal factors and external factors.Internal factors include cerebrovascular and blood flow factors.The "watershed region" model map was used to construct the cerebrovascular marginal region template,and the map after registration was automatically segmented to obtain the spatial distribution characteristics of BMs in each watershed region.External factors include pathological factors,therapeutic interventions,gene mutation status,and primary site.Group comparison was carried out according to different factors.Internal software is used to automatically calculate the distance(depth)from the brain surface to the center of each lesion.The differences of depth of brain metastasis under different factors were analyzed.Results:A total of 914 patients with brain metastases were included,including 4014 metastases.Among them,728 patients with lung cancer had 2997 metastases,and 59 patients with breast cancer had 461 metastases.There were 21 cases of esophageal cancer,13 cases of colon cancer,11 cases of kidney cancer and 82 cases of other cancers.The visualized heat map shows uneven brightness in different functional areas of the brain.Partial frontal lobe,bilateral cerebellum and brainstem regions were prominent,and the incidence of brain metastasis was high.The brightness of bilateral temporal lobe and part of occipital lobe was low,and the incidence of brain metastasis was low.The spatial distribution characteristics of metastases from different pathological sources can also be seen on the heat map.The frequency of BMs in each brain region was as follows:Frontal lobe 1209(17.34%),parietal lobe 739(13.95%),cerebellum 711(9.26%),temporal lobe 454(10.04%),occipital lobe 415(7.26%),deep white matter and brain stem 292(7.27%),thalamus 53(1.30%),putamen 39(0.86%),Insula 48(0.83%),caudate 54(0.70%).Analysis of the combined ALL atlas(38 regions)showed that the top ten brain regions with metastasis rate and the number of metastatic foci(metastasis rate)were as follows:Deep white matter and brain stem 865(21.55%),Cerebellar 333(8.28%),middle frontal gyrus 291(7.24%),superior frontal gyrus 254(6.34%),precentral gyrus 207(5.16%),cerebellar foot 206(5.13%),occipital gyrus 161(4.00%),posterior central gyrus 159(3.96%),precuneus 125(3.11%),middle temporal gyrus 123(3.06%).The metastasis rates of some functional areas of the brain deserving special attention including hippocampus,parahippocampal gyrus,globus pallidus,callosal fissure and peripheral cortex and temporal pole were 0.52%,0.54%,0.07%,2.04%and 0.96%,respectively.The analysis of internal vascular and blood flow factors showed that the rate of metastasis was significantly higher than expected in the anterior(P=1.404×10-35),posterior(P=9.302×10-12)and cerebellum watershed(P=9.364×10-34).However,the rate of metastasis in the deep gray matter junction was lower than expected(P=0.004).The source of pathological tissue had a significant effect on the spatial distribution of patients.Lung cancer was at high risk in cerebellar hemisphere(P=2.59×10-15)and low risk in temporal lobe(P=5.18×10-11),while breast cancer was at high risk in cerebellar hemisphere(P=14.92×10-7),but low risk in temporal lobe was not significant.The most frequently metastasized brain regions of colorectal cancer are located in the posterior fossa:deep white matter,brainstem and cerebellar hemisphere.The most common metastatic sites of renal carcinoma were deep white matter and brain stem,middle frontal gyrus and inferior temporal gyrus.The most common metastatic sites of esophageal cancer were deep brain white matter and brain stem,cerebellum and middle frontal gyrus.Patients with EGFR mutation were at high-risk of cerebellum(P=0.007)and low temporal lobe(P=0.029).However,no risk regions were found in patients without EGFR mutations,P>0.05 in all MNI subdivision(10 brain regions).Based on the combined ALL subdivision,the middle frontal gyrus(P=0.006×10-1),precentral gyrus(P=0.007)were at high risk in patients with EGFR mutation and the The temporal pole were at low-risk area(P=0.005),while this phenomenon was not found in patients without EGFR mutation.From the perspective of metastasis depth,it was found that the depth of metastasis in patients with EGFR mutation was greater than that in patients without mutation,but with no statistical significance.Whether chemotherapy had been received before metastasis had no significant effect on the risk of brain region in MNI atlas(10 sub-regions),which both showed high risk of cerebellum and low risk of temporal lobe.Based on the analysis of the combined ALL atlas,it was found that the distribution of patients after chemotherapy shows an increase in relative unevenness.Three low-risk areas were increased compared with patients without chemotherapy:superior marginal gyrus 0.003×10-1,superior temporal gyrus 0.003,and temporal pole 0.009.The location of the left and right sides of the primary foci in lung cancer patients had no effect on the distribution of metastatic tumor(P>0.05 in all brain regions).Conclusion:We constructed the spatial distribution map of brain metastases of common cancer types through multi-center large sample study,and quantitatively analyzed the frequency of metastasis in different brain functional areas.The spatial distribution of BMs was found to be uneven.The distribution of BMs was influenced by the factors of internal blood vessel and blood flow distribution,source of external pathological tissue,EGFR mutation status and treatment before BMs,but the left and right location of primary foci had no influence on the spatial distribution of BMs.Compared with the expected frequency,a series of risk areas were screened out,which may provide necessary help for the strategy of radiotherapy for brain metastases.更多还原