第一部分 老年乳腺癌患者衰弱现状及生活质量报告: 一项前瞻性横断面调研

【背景】衰弱是指机体脆弱性增加及维持稳态能力下降的一种临床状态。衰弱状态会影响肿瘤患者的治疗决策。中国老年乳腺癌患者的衰弱状态尚未被充分重视, 也未见相关的研究报道。本研究旨在调查我国老年乳腺癌患者的衰弱患病率、相关因素及其对患者生活质量的影响。

【方法】本研究是一项前瞻性、横断面、注册临床研究。纳入年龄≥65岁病理确诊为乳腺癌的患者, 对其进行衰弱筛查、可能相关因素的分析和生活质量评估。使用国际公认的FRAIL衰弱筛查量表(包含五个项目: 疲劳, 行走, 爬楼, 合并症和不明原因体重下降, 每项1分)进行评分, 将患者分为非衰弱(FRAIL评分=0分), 衰弱前期(FRAIL评分=1-2分)和衰弱组(FRAIL评分=3-5分); 使用医院焦虑抑郁量表(Hospital Anxiety and Depression Scale, HADS)评估患者的情绪状态; 使用欧洲癌症研究与治疗组织生命质量核心量表(European Organization for Research and Treatment of Cancer Quality of Life questionnaire Core 30, EORTC QLQ-C30)评估患者的生活质量。同时收集入组患者的临床病理资料。应用卡方检验、单因素方差分析、logistic回归分析和多元线性回归模型进行数据的统计学分析。

【结果】研究最终纳入2021年10月至2023年7月期间在中国医学科学院肿瘤医院和北京市三环肿瘤医院就诊的老年乳腺癌患者946例。中位年龄为69(范围65-96)岁; 73.6%的患者诊断为早期(原位癌和I-III期)乳腺癌; 接受手术、化疗、放疗、内分泌治疗的比例分别为89.3%、56.7%、29.0% 和69.7%。83例患者(8.8%)被评估为衰弱, 541(57.2%)例患者评估为衰弱前期。研究显示, 随着年龄的增加, 衰弱的比例显著增加(P<0.001); 75岁及以上的患者中, 衰弱的患病率达22.3%; 晚期肿瘤患者衰弱和衰弱前期的比例均显著升高, 患病率分别为15.6%和63.6% (P<0.001); 在焦虑和抑郁的患者中, 衰弱的比例亦显著升高, 患病率分别为31.3% 和 29.3% (均P<0.001)。Logistic多因素回归分析显示, 年龄增长、晚期肿瘤、焦虑和抑郁情绪均是衰弱的独立相关危险因素(均P<0.05)。生活质量方面, 不论是功能维度还是症状维度的评估, 衰弱及衰弱前期患者的评分均劣于非衰弱患者(均P<0.001)。且校正了其他因素后, 衰弱状态仍然对生活质量有着独立的负面影响(均P<0.001)。 

【结论】我国老年乳腺癌患者中, 衰弱患病率为8.8%, 其在高龄、肿瘤晚期阶段及情绪障碍的人群中更常见(15.6-31.3%)。衰弱前期的患者达半数以上(57.2%)。衰弱及衰弱前期状态均明显降低患者的生活质量。在制定抗肿瘤治疗方案时, 有必要仔细评估患者的衰弱状态, 并对可控的因素(如心理健康、治疗相关症状等因素)进行干预, 以全面改善老年患者对肿瘤治疗的耐受性和依从性。

第二部分 老年衰弱评分(GVS)在老年晚期乳腺癌中的预后价值

【背景】老年综合评估(Comprehensive Geriatric Assessment, CGA)是国际公认的对老年人进行全面的生理、心理和功能等多维度进行评估的手段, 可以识别出“衰弱”的患者, 有预测老年肿瘤患者预后的价值。但由于其涉及的量表众多, 耗时长, 临床应用受限。前述部分的FRAIL量表作为衰弱的筛查工具不能替代CGA, 因此需要开发简化的老年评估方法以便于临床应用及推广。本研究拟针对老年晚期乳腺癌患者建立老年衰弱评分(Geriatric Vulnerability Score, GVS), 评估其与预后的关系, 并进一步探究GVS评分在晚期乳腺癌局部治疗回顾性研究中的作用。

【方法】研究纳入2012年2月至2023年9月在中国医学科学院肿瘤医院住院治疗年龄≥65岁的晚期乳腺癌患者。患者于治疗前进行老年评估, 评估内容包括合并症(查尔森共患病指数Charlson Comorbidity Index, CCI), 身体功能(日常生活活动能力量表Activity of Daily Living, ADL), 免疫功能储备(外周血淋巴细胞绝对值)及营养功能储备(外周血白蛋白水平) 4项。基于上述4项评估建立GVS评分。研究还进一步纳入了SEER数据库初诊IV期乳腺癌患者的数据。采用cox多因素回归方法进行预后的危险因素分析; 采用t检验或单因素方差分析比较GVS评分与患者临床特征之间的关系; Kaplan-Meier曲线对比组间总生存(overall survival, OS)或肿瘤特异性生存率(cause-specific survival, CSS); Harrell’s 一致性检验(C指数)计算预后模型的性能。

【结果】233例中国医学科学院肿瘤医院住院治疗的患者纳入最终分析, 中位年龄为69(65-82)岁。159(66.8%)例患者为晚期初治状态(未接受任何晚期抗肿瘤治疗); HR-/HER2-占比 27.5%; 骨、肺、肝、脑转移比例分别为50.2%, 48.1%, 26.2%, 6.0%。CCI中位值为0(0-4)分; 33(14.2%)例患者ADL分数<90分; 25(10.7%)例患者外周淋巴细胞绝对值≤1×10⁹/升; 17(7.3%)例患者血白蛋白≤35克/升。多因素分析显示, HR-/HER2-, 肝或脑转移, ADL分数<90分, 淋巴细胞绝对值≤1×10⁹/升, 白蛋白≤35克/升是患者总生存(overall survival, OS)独立不良预后因素(均P<0.05)。CCI不是老年晚期乳腺癌患者的独立预后因素(P>0.05)。将上述有显著统计学意义的老年评估项目进行计分(ADL分数<90分、淋巴细胞绝对值≤1×10⁹/升、白蛋白≤35克/升, 每项计1分)以建立GVS评分。GVS评分越高, 患者OS越差(HR: 2.228, 95%CI: 1.640-3.027, P<0.001)。该评分在159例晚期初治患者组中也得到了内部验证(HR: 3.215, 95%CI: 1.975-5.232, P<0.001)。无论是所有患者还是晚期初治患者, 在肿瘤特征(含分子分型和转移部位)的基础上加入GVS评分均能够提高预后模型的准确性(C指数, 全人群: 从0.670提高到0.709; 初治患者: 从0.694提高到0.742)。68例初诊IV期患者的KM曲线分析发现接受了原发灶手术的患者的OS似乎略高于未接受原发灶手术的患者, 但差别无明显统计学意义(P=0.101)。其中56例GVS评分为0的患者KM曲线分析发现, 无论是接受原发灶手术还是未接受原发灶手术, 患者的OS相当(P=0.660)。然而在纳入分析的SEER数据库的3194例初诊IV期患者的多因素cox分析发现, 局部治疗方式是影响患者CSS的独立预后因素( P<0.001)。在化疗的基础上接受手术或接受手术联合放疗的患者预后均显著优于仅接受放疗的患者(放疗5年CSS＝16％, 手术5年CSS＝43%, 手术联合放疗=47%, P<0.001)。

【结论】基于ADL评分、淋巴细胞绝对值、白蛋白构建的GVS评分与老年晚期乳腺癌患者的预后密切相关, GVS评分越高, 预后越差。提示在老年晚期患者的治疗过程中, 需要特别注重老年衰弱评分每个领域的干预, 包括身体功能的增强、免疫功能的保护和营养的支持。此外, GVS评分可能有助于消除回顾性研究中患者的身体功能储备因素而引起的偏移, 有利于得出更加准确的结论。该评分方法客观且简便易行, 值得在更大的样本中验证其临床应用价值。

第三部分  老年乳腺癌化疗相关认知障碍的临床影响因素和血浆代谢组学研究

【背景】乳腺癌患者在化疗期间或之后常有的认知功能下降的主诉。老年患者更易出现化疗相关认知障碍(chemotherapy induced cognitive impairment, CICI)。在该人群中, 与CICI有关的危险因素仍存在争议, 且缺乏相关的生物标志物。代谢组学可以更加直接反映生理状态的动态变化, 其与CICI之间的关系鲜有研究。本研究旨在分析老年乳腺癌患者CICI的临床影响因素, 并基于代谢组学方法, 分析化疗前后患者血浆代谢物的变化, 以评估其与CICI的关系。

【方法】本研究是一项前瞻性、非干预、单中心临床研究。纳入年龄≥60岁、经病理确诊的、初次接受化疗的乳腺癌患者。在化疗前后进行主观认知功能和肿瘤相关症状的评估, 使用的4份量表包括: 癌症患者认知功能量表(Functional Assessment of Cancer Therapy-Cognitive Function Version 3, FACT-cog V3), 医院焦虑抑郁量表(Hospital Anxiety and Depression Scale, HADS), 匹兹堡睡眠质量指数量表(Pittsburgh sleep quality index, PSQI), 简易疲劳量表( Brief Fatigue Index, BFI)。收集同期治疗前后的外周血标本进行基于液相色谱-质谱联用技术(Liquid Chromatography-Mass Spectrometry, LC-MS)的非靶向代谢组学检测分析。通过线性混合模型确定与FACT-cog总分独立相关的因素; 使用多元变量统计分析方法进行化疗前后外周血差异代谢物的筛选; 对差异代谢物进行最小绝对收缩和选择算子(least absolute shrinkage and selection operator, LASSO)回归, 建立危险评分(Risk Score); 对Risk Score进行相关性分析及多因素线性混合模型分析。 

【结果】研究于2022年8月至2024年3月在中国医学科学院肿瘤医院内科病房开展。共筛选了177例患者, 最终纳入分析的患者98例。中位年龄为64(范围60-77)岁; 72.5%患者拥有高中/中专及以上学历; 高血压和糖尿病患者的比例为54.1%和21.4%。大部分(95.9%)患者为早期乳腺癌, 仅4例(4.1%)为晚期患者; 61例(62.2%)接受了紫杉类方案的化疗, 34例(34.7%)患者接受了蒽环联合紫杉类方案化疗, 3例(3.1%) 患者接受了蒽环方案的化疗。相比于化疗前(T0), 化疗后(T1)FACT-cog总分显著下降(T0: 120.5 vs T1: 112.0, P<0.0001)。量表中各个子量表的评分亦普遍下降, 在知觉的认知功能障碍(T0: 65.8 vs T1: 61.1, P<0.001)、其他人的评语(T0: 14.6 vs T1: 14.0, P<0.05)、知觉的认知能力(T0: 27.5vs T1: 25.3, P<0.001)领域下降程度有统计学意义。经过线性混合分析调整其他可能影响认知的因素后, 化疗仍然是FACT-cog总分下降的独立相关因素(P=0.013)。焦虑和抑郁也是FACT-cog总分下降的重要危险因素(分别 P=0.008和0.007)。61例患者在化疗前后均留取了合格的外周血标本, 共122份。在LC-MS鉴定出的1258种代谢物中, 相比于T0, 发现有42种代谢物的表达值在T1时间点上调或下调, 且均有显著统计学意义(P<0.05)。通过LASSO回归我们最终筛选出5个与FACT-Cog总分相关的差异代谢物,包括CerP(d18:0/16:0)和LPC(O-22:1)的上调, 硫酸雄酮(Androsterone sulfate), 谷氨酸-精氨酸二肽(Glu-Arg)和γ-谷氨酰谷氨酰胺(gamma-Glutamylglutamine)下调。人类代谢组数据库(Human Metabolome Database, HMDB)显示, 这5种代谢物涉及到鞘脂代谢、甘油磷脂胆碱代谢、雄激素代谢和谷氨酸-谷氨酰胺循环等途径。我们进一步将这5个代谢物与其系数建立Risk Score, 经过线性混合模型分析发现, 在调整了其他因素(包括焦虑、抑郁和PSQI评分)后, Risk Score仍然是FACT-cog总分变化的独立危险因素(P=0.022)。

【结论】化疗和(或)焦虑抑郁的情绪显著降低老年乳腺癌患者的主观认知功能。化疗后血浆中5种代谢成分的上调或下调与患者的主观认知功能的下降显著相关, 涉及到鞘脂代谢、甘油磷脂胆碱代谢, 雄激素代谢和谷氨酸-谷氨酰胺循环途径。这些代谢物或其涉及的途径可能是未来改善CICI的潜在靶点。

PART I  Prevalence of Frailty and Its Impact on Quality of Life in Older Breast Cancer: A Prospective Cross-Sectional Study

Background Frailty refers to a clinical state characterized by increased vulnerability of the body and a decline in the ability to maintain homeostasis. Frailty status influences treatment decisions of older breast cancer patients. The frailty status of older breast cancer patients in China has not been fully considered and reported. This study aims to evaluate the prevalence of frailty, relative factors, and its impact on the quality of life(QoL) of older breast cancer patients in China.

Methods This is a prospective cross-sectional registry study. Pathologically confirmed breast cancer patients aged 65 years and older at the time of survey were included. They were screened for frailty and assessed for related factors and QoL. The internationally recognized FRAIL frailty screening scale (including five items: fatigue, resistance, ambulation, illnesses, and loss of weight; 1 point for each item) was used for scoring. Patients were categorized into robust (FRAIL score = 0), pre-frail (FRAIL score = 1-2), and frail (FRAIL score = 3-5) groups. Emotional functioning was evaluated using the Hospital Anxiety and Depression Scale (HADS), and QoL was assessed using the European Organization for Research and Treatment of Cancer Quality of Life Questionnaire-Core 30 (EORTC QLQ-C30). Patients clinicopathologic data were also collected. Statistical analysis of the data was performed using chi-squared test, one-way ANOVA, logistic regression analysis and multivariate linear regression model.

Results The study enrolled 946 older breast cancer patients who visited the Cancer Hospital of the Chinese Academy of Medical Sciences or Beijing Sanhuan Hospital from October 2021 to July 2023. The median age was 69 years (range 65-96). 73.6% of patients were diagnosed with early stage (0-III) breast cancer. The proportions of patients undergoing surgery, chemotherapy, radiotherapy or endocrine therapy were 89.3%, 56.7%, 29.0%, and 69.7%, respectively. 83 patients (8.8%) were classified as frail and 541 (57.2%) were classified as pre-frail. The study revealed that the prevalence of frailty increased significantly with age (P <0.001). In patients aged 75 years and older, the  proportions of frailty reached 22.3%. The prevalence of frailty and pre-frailty increased significantly in patients with advanced tumors, reaching 15.6% and 63.6%, respectively (P <0.001). The prevalence of frailty also increased significantly in patients with anxiety and depression (31.3% and 29.3%, respectively, both P <0.001). Logistic multivariate regression analysis showed that older age, advanced tumors, anxiety, and depression were independent risk factors for frailty (all P <0.05). Regarding QoL, both functional and symptomatic domain scores of frail and pre-frail patients were inferior to those of robust patients (all P <0.001). Even after adjustment for other factors, frailty still had an independent negative impact on QoL (P <0.001).

Conclusions In Chinese older breast cancer patients, the prevalence of frailty is 8.8%, and it is more common (15.6-31.3%) in people with advanced age, advanced cancer stages, and emotional disorders. More than half (57.2%) of the patients were in the pre-frailty stage. Both frailty and pre-frailty states significantly reduce patients' quality of life. When formulating treatment plans, it is necessary to carefully evaluate patients frailty status and intervene in modifiable factors (such as mental health, treatment-related symptoms, etc.) to comprehensively improve patients tolerance and compliance with tumor treatment.

PART II Prognostic Value of the Geriatric Vulnerability Score(GVS) in Older Patients with Metastatic Breast Cancer

Background Comprehensive Geriatric Assessment (CGA) is an internationally recognized method for various aspects including physiological, psychological and functional assessment of the older. It can identify "frail" patients and has value in predicting the prognosis of older cancer patients. However, due to the large number of scales involved, it is time-consuming and has limited clinical application. The aforementioned FRAIL scale cannot replace CGA as a frailty screening tool. Therefore, it is necessary to develop a simplified geriatric assessment method for clinical application and promotion. This study intends to establish a Geriatric Vulnerability Score (GVS) for older patients with advanced breast cancer and evaluate its relationship with prognosis, and further explore the role of GVS in retrospective study of local treatment of advanced breast cancer.

Methods This study included breast cancer patients aged 65 years and above who underwent inpatient treatment and geriatric assessment at the Cancer Hospital of the Chinese Academy of Medical Sciences from February 2012 to September 2023. Geriatric assessment included Charlson Comorbidity Index (CCI), activities of daily living (ADL) assessment, peripheral blood lymphocyte absolute count, and albumin level. The GVS was determined based on these assessment items. The study further included data from the SEER database on patients diagnosed with primary stage IV breast cancer. Cox multivariate regression analysis was used to identify prognostic factors. The relationship between GVS score and clinical characteristics was compared using student’s t-tests or one-way ANOVA. Kaplan-Meier curves were used to compare overall survival (OS) or cause-specific survival (CSS) between groups. Harrell's concordance statistic (C‐statistic) was used to calculate property of the prognostic model.

Results A total of 233 patients hospitalized at the Cancer Hospital of the Chinese Academy of Medical Sciences were included in this study with a median age was 69 (65-82) years. Of these, 97 (41.6%) were HR+/HER2-, 61 (26.2%) were HR+/HER2+, and 64 (27.5%) were HR-/HER2-. 159 (66.8%) patients received no treatment in metastatic setting at the time of assessment(treatment-naive group). The proportions of bone, lung, liver, and brain metastases were 50.2%, 48.1%, 26.2%, and 6.0%, respectively. The median CCI was 0 (0-4). 33 (14.2%) patients had ADL <90; 25 (10.7%) patients had peripheral lymphocyte absolute counts ≤1×10⁹/L, and 17 (7.3%) patients had peripheral blood albumin levels ≤35g/L. Multivariate analysis showed that triple-negative subtype, liver and brain metastases, ADL<90, lymphocyte counts ≤1×10⁹/L, and albumin ≤35 g/L were independent adverse factors for OS in patients (all P <0.05), whereas CCI was not an independent prognostic factor for older metastatic breast cancer(P >0.05). When the significant geriatric assessment items (ADL <90, absolute lymphocyte count ≤1×10⁹/L, albumin ≤35 g/L, each scored 1 point) were scored to determine the GVS, it was found that the higher the GVS, the worse the OS of patients (HR: 2.228, 95% CI: 1.640-3.027, P <0.001). Furthermore, this score was validated in the treatment-naive group (HR: 3.215, 95% CI: 1.975-5.232, P <0.001). Harrell's concordance statistic showed that adding the GVS to tumor characteristics (including molecular subtype and metastatic site) improved the accuracy of prognostic prediction (C-index, all patients: from 0.670 to 0.709; treatment-naive group: from 0.694 to 0.742). However, a multivariate Cox analysis of the 3,194 patients with primary stage IV breast cancer included in the SEER database found that the type of local treatment was an independent prognostic factor affecting patients' CSS (P<0.001). Patients who received surgery or surgery combined with radiotherapy in addition to chemotherapy had significantly better prognoses compared to those who only received radiotherapy (5-year CSS for radiotherapy = 16%, 5-year CSS for surgery = 43%, 5-year CSS for surgery combined with radiotherapy = 47%, P<0.001).

Conclusion The GVS composed of ADL score, lymphocyte absolute value, and albumin is closely related to the prognosis of older patients with advanced breast cancer. The higher the GVS score, the worse the prognosis. It is suggested that during the treatment of advanced patients, special attention needs to be paid to intervention in each area of the geriatric frailty score, including enhancement of physical function, protection of immune function, and nutritional support. Additionally, the GVS may help eliminate the bias caused by patients' physical functional reserve factors in retrospective studies, leading to more accurate conclusions. This scoring method is objective, simple and feasible, and is worthy of verifying its clinical application value in a larger sample.

Part III Clinical Impact Factors of Chemotherapy-Induced Cognitive Impairment in Older Breast Cancer and Plasma Metabolomics Study

Background Breast cancer patients often complain of cognitive decline during or after chemotherapy. Older patients are more susceptible to chemotherapy-induced cognitive impairment (CICI). In this population, the risk factors associated with CICI remain controversial and blood biomarkers are lacking. Metabolomics provides a more direct way to detect physiological states, but its relationship with CICI has rarely been studied. This study aims to investigate the clinical factors influencing CICI in older breast cancer patients and to evaluate the relationship between changes in plasma metabolites before and after chemotherapy using metabolomics analysis.

Methods This is a prospective, non-interventional, single-center clinical trial. It includes breast cancer patients aged ≥60 years who have been pathologically diagnosed and are receiving chemotherapy for the first time. Subjective cognitive function and tumor-related symptoms were assessed before and after chemotherapy, and peripheral blood samples were collected simultaneously. The four assessment scales used were Functional Assessment of Cancer Therapy-Cognitive Function Version 3(FACT-cog V3), Hospital Anxiety and Depression Scale (HADS), Pittsburgh Sleep Quality Index (PSQI), and Brief Fatigue Index (BFI). Peripheral blood samples collected were subjected to non-targeted metabolomics analysis using liquid chromatography-mass spectrometry (LC-MS). Factors independently associated with FACT-cog scores were determined using linear mixed models. Least absolute shrinkage and selection operator (LASSO) regression was performed on differential metabolites to establish a Risk Score. Correlation analysis and multi-factor linear mixed-effects model analysis were conducted on the Risk Score.

Results The study was conducted from August 2022 to March 2024 in the Department of Internal Medicine, Cancer Hospital of the Chinese Academy of Medical Sciences. A total of 177 patients were screened, and 98 patients were finally included in the analysis. The median age was 64 years (range 60-77). 72.5% of the patients had a high school education or higher. The proportions of patients with hypertension and diabetes were 54.1% and 21.4%, respectively. Most of the patients (95.9%) patients had stage I-III tumors and only 4 cases (4.1%) were stage IV patients. 61 patients (62.2%) received taxane-based chemotherapy, 34 patients (34.7%) received anthracycline combined with taxane-based chemotherapy, and 3 patients (3.1%) received anthracycline-based chemotherapy. Compared to pre-chemotherapy (T0), there was a significant decrease in the total FACT-cog score after chemotherapy (T1: 120.5 vs. T0: 112.0, P<0.0001). Scores on each subscale of the scale also generally decreased, with statistically significant decreases in the areas of perceived cognitive impairments (T1: 61.1 vs. T0: 65.8, P<0.001), comments from others (T1: 14.0 vs. T0: 14.6, P<0.05), and perceived cognitive abilities (T1: 25.3 vs. T0: 27.5, P<0.001). After adjusting for other factors that may affect cognition using linear mixed-effects analysis, chemotherapy remained an independent factor associated with the decrease in FACT-cog score (P=0.013). Anxiety and depression were also important factors for the decrease in FACT-cog score (P=0.008 and 0.007, respectively). 61 patients had qualified peripheral blood samples collected before and after chemotherapy, for a total of 122 samples. Among the 1258 metabolites identified by LC-MS, the expression levels of 42 metabolites were found to be up- or down-regulated at T1 compared to T0, all with significant statistical significance (t-test P<0.05). By LASSO regression, we finally selected 5 differential metabolites related to FACT-Cog score, including down-regulation of Androsterone sulfate, Glu- Arg and gamma-Glutamylglutamine, up-regulation of CerP(d18:0/16:0) and LPC(O-22:1). According to the Human Metabolome Database (HMDB), these 5 metabolites are involved in pathways including metabolism of androgens and estrogens, the glutamine-glutamate cycle, sphingolipid metabolism, and glycerophospholipid choline. We further created a risk score by combining these 5 metabolites with their coefficients. After adjusting for other factors (including anxiety, depression, and PSQI) through linear mixed-effects model analysis, the risk score remained an independent risk factor for changes in FACT-cog score (P=0.022).

Conclusion Subjective cognitive decline in older breast cancer patients is associated with chemotherapy, anxiety and depression. The up-regulation or down-regulation of five metabolic components in plasma after chemotherapy is significantly related to the decline of patients' subjective cognitive functions, involving sphingolipid metabolism, glycerophospholipid choline metabolism, androgen metabolism and glutamate-glutamine cycle pathway. These metabolites or the pathways involved may be potential targets for improving CICI in the future.