第一部分 第一章 ALK TKI真实世界序贯疗效及安全性特征探索 中文摘要 背景 洛拉替尼是第三代间变性淋巴瘤激酶（anaplastic lymphoma kinase, ALK）酪氨酸激酶抑制剂(tyrosine kinase inhibitors, TKI)，在提高ALK阳性非小细胞肺癌(non-small cell lung cancer, NSCLC)患者疗效方面表现卓越，目前尚不清楚洛拉替尼一线治疗相较于其他序贯治疗方案的疗效获益。本研究基于真实世界数据，系统评估洛拉替尼一线及序贯治疗的疗效与安全性特征，为优化ALK阳性NSCLC的个体化治疗策略提供依据。 方法 采用单中心真实世界回顾性结合前瞻性队列研究。回顾性（2020年2月–2023年4月）与前瞻性（2023年4月–2024年9月）纳入在中国医学科学院肿瘤医院接受洛拉替尼治疗的晚期ALK重排阳性非小细胞肺癌患者。按照治疗策略分为一线洛拉替尼组（30例）和二代ALK TKI后序贯洛拉替尼组（21例，既往接受过二代ALK TKI后进展序贯洛拉替尼）。安全性分析覆盖所有接受洛拉替尼的患者，未入队列的洛拉替尼治疗者同样纳入。收集患者临床特征、疗效和不良反应资料。主要疗效指标为客观缓解率（objective response rate，ORR）和无进展生存率/期（progression free survival，PFS），次要指标包括疾病控制率（disease control rate，DCR）、颅内客观缓解率（intracranial ORR，iORR）、颅内无进展生存期（intracranial PFS，iPFS）及总生存期（overall survival，OS）。统计分析采用R 4.2.2软件，计量资料正态分布时用t检验或单因素方差分析，非常态分布时用非参数检验，计数资料用χ2检验或Fisher精确检验；生存分析采用Kaplan-Meier法并经Log-rank检验比较组间差异。 结果 本研究共纳入51例ALK基因重排阳性晚期NSCLC患者（回顾性6例，前瞻性45例），队列1：一线洛拉替尼组30例，队列2：二代ALK TKI后序贯洛拉替尼组21例，两组基线特征相近。一线洛拉替尼组24例可评估病灶患者中ORR为91.7%（95%CI：73.0–99.0%），DCR 100.0%；中位PFS未达，12个月PFS率95.0%；序贯组二代ALK-TKI阶段16例可评估病灶中ORR为87.5%（95%CI：61.7–98.4%），DCR 100.0%；中位PFS 24.5个月（95%CI：13.9–31.3），12个月PFS率71.4%；序贯三代洛拉替尼阶段18例可评估病灶患者的ORR为22.2%（95%CI：6.4–47.6%），DCR 77.8%，中位PFS 19.5个月（95%CI：5.7–NA）。两组ORR、DCR差异均无统计学意义（P=1.00），12个月PFS率差异临界（P=0.05）；12、24个月OS率分别为92.3% vs. 83.5%，差异无统计学意义（P=0.84）。 安全性分析纳入所有接受洛拉替尼的ALK阳性NSCLC患者（n=63）进行不良事件统计，所有患者均在治疗早期出现高胆固醇血症（100%）和高甘油三酯血症（98.4%），经降脂药积极干预后血脂水平逐步回落，无一例因血脂异常停药。伴脑转移患者发生3级及以上高胆固醇血症的比例显著高于无脑转移组（37.0% vs. 12.5%，P=0.03）。其他常见不良反应包括外周水肿（63.5%）、中枢神经系统反应（57.1%）、体重增加（23.8%）、高血压（22.2%）、外周神经病变（22.2%）和关节疼痛/僵硬（20.6%）。1例既往有肾积水病史的患者因3级蛋白尿而停用洛拉替尼。 结论 洛拉替尼一线治疗ALK阳性晚期非小细胞肺癌显示出极高的肿瘤缓解率和良好的短期无进展生存获益，初步证明其一线应用具有显著且持久的抗肿瘤活性。二代ALK TKI序贯洛拉替尼策略同样在总体上实现了长期疾病控制，然而在序贯至第三代ALK TKI阶段疗效明显降低，提示前线使用二代TKI可能降低肿瘤对后续洛拉替尼的敏感性。洛拉替尼治疗的安全性可接受，其引发的高脂血症十分普遍，应在治疗早期密切监测血脂并及时干预。伴脑转移患者更易出现严重高脂血症现象，这一发现提示需针对该人群加强代谢监控和支持治疗。综合而言，本研究为洛拉替尼一线应用提供了初步有效性和安全性依据，也强调了高脂血症等不良反应的监测管理重要性。 关键词 非小细胞肺癌；ALK融合基因；高酯血症；靶向治疗；序贯治疗 第一部分 第二章 三代ALK TKI血脂异常特征及代谢机制探索 中文摘要 背景 尽管洛拉替尼在间变性淋巴瘤激酶（anaplastic lymphoma kinase, ALK）阳性非小细胞肺癌（non-small cell lung cancer, NSCLC）患者中展现出卓越疗效，其独特的不良反应谱——尤其是高脂血症的发生机制尚不清楚。本研究基于临床观察和代谢组学分析，探讨洛拉替尼所致高脂血症的代谢机制及管理策略。 方法 本研究为前瞻性单中心真实世界研究。共纳入22例ALK基因重排阳性的晚期NSCLC患者，2023年4月–2024年9月间于在中国医学科学院肿瘤医院接受洛拉替尼治疗。所有患者均口服洛拉替尼，纳入签署知情同意且符合条件的患者，于治疗时开启血脂相关指标监测，将其按洛拉替尼治疗过程中出现的主要血脂异常类型分为三组：甘油三酯明显升高为主（A组）、胆固醇明显升高为主（B组）和甘油三酯与胆固醇升高程度相同（C组）。在洛拉替尼治疗前（基线）及给药后2周和1个月分别采集空腹血清样本，动态监测血脂指标，包括甘油三酯（triglyceride, TG）、胆固醇（cholesterol，CHOL）、低密度脂蛋白胆固醇（low-density lipoprotein cholesterol, LDL-C）、高密度脂蛋白胆固醇（high-density lipoprotein cholesterol, HDL-C）、脂蛋白a（Lp(a)）和脂质代谢相关酶活性，包括肝脂酶（hepatic lipase, HL）、脂蛋白脂肪酶（lipoprotein lipase, LPL）、脂肪酶（lipase pancreatic, LPS）、乙酰辅酶A羧化酶（acetyl-CoA carboxylase, ACC）、ATP柠檬酸裂解酶（ATP citrate lyase, ACL）和脂肪酸合成酶（fatty acid synthase, FAS）。并选取各组部分患者（各亚组各3例，总计9例）于上述时间点采集血清行非靶向代谢组学分析，结合主成分分析、正交偏最小二乘判别分析及代谢通路网络分析，挖掘差异代谢物及相关通路。 结果 本研究共纳入22例ALK阳性晚期NSCLC患者进行血脂与脂代谢酶动态监测，其中A组3例，B组8例，C组11例。治疗2周时TG与CHOL分别从基线1.43±0.62 mmol/L和5.32±1.07 mmol/L显著升高至3.48±1.99 mmol/L和8.76±2.06 mmol/L（均P<0.01），并于1个月时维持高于基线水平。LDL-C与HDL-C亦同步升高（LDL-C：3.49±0.90 升至最大5.43±1.72 mmol/L；HDL-C：1.35±0.29 mmol/L升至最大1.86±0.55 mmol/L，均P<0.01）。亚组分析显示，第2周时B组CHOL显著高于C组（P=0.006），HDL-C与LDL-C亦存在整体组间差异趋势（分别为P=0.039和P=0.018），但TG在各组间差异未达显著（P>0.05）。治疗1个月后各指标水平虽有轻度下降，但仍明显高于基线（均P<0.01）。脂蛋白(a)及空腹血糖未见显著变化。脂质代谢酶活性整体变化不显著：各组别的LPL、ACC、FAS、ACL活性在治疗后与基线相比差异无统计学意义。仅HL活性在治疗1个月时较基线有升高趋势（总体分析P<0.01），但酶活性与血脂水平无显著相关（Pearson r<0.3, P>0.05）。 代谢组学共鉴定1831种代谢物，治疗后，各亚组患者血清代谢物谱发生了广泛改变。其中以脂肪酸氧化中间体（如酰基肉碱类）、胆固醇合成前体和胆汁酸代谢产物显著升高为特征，提示患者脂质氧化受阻和胆固醇合成-胆汁酸转化失衡。此外，氨基酸代谢、三羧酸循环及戊糖磷酸途径等能量代谢通路也发生显著扰动。代谢通路富集分析和网络重构显示，洛拉替尼诱导的血脂异常涉及多条代谢通路的广泛重塑，其中脂肪酸代谢、类固醇激素合成、鞘脂信号通路和线粒体能量代谢通路的异常最为显著。其中A组（甘油三酯升高）：膜脂代谢紊乱，脂解减弱、脂质转运障碍；首选贝特类（非诺贝特），激活PPAR-α、提升LPL活性，促进甘油三酯水解。B组（胆固醇升高）：膜磷脂重塑、胆固醇转运与吸收异常、氧化应激增强；首选他汀类（阿托伐他汀/瑞舒伐他汀），抑制HMG-CoA还原酶，改善膜脂代谢并减轻氧化应激C组（甘油三酯与胆固醇升高）：甘油磷脂与氨基酸代谢双重异常，炎症及能量代谢亢进；建议他汀＋贝特类联合或降脂＋SGLT2抑制剂协同治疗，双重控制血脂并抑制炎症。 结论 洛拉替尼可快速引起以胆固醇和甘油三酯全面升高为特征的高脂血症，其机制涉及复杂的代谢网络重构而非传统单一酶途径，可能由核受体介导的脂代谢调控紊乱所致。及时的个体化血脂监测和干预能够有效控制该不良反应，使患者在接受洛拉替尼突破性抗肿瘤疗效的同时降低心血管风险。上述发现为ALK阳性肺癌患者的治疗提供了代谢管理依据，并为进一步研究酪氨酸激酶抑制剂（tyrosine kinase inhibitor，TKI）相关代谢毒性的机制及干预策略奠定了基础。 关键词 洛拉替尼；非小细胞肺癌；ALK融合基因；高酯血症；代谢组学 第二部分 EGFR TKI序贯治疗模式潜在获益人群探索 摘要 背景 表皮生长因子受体（epidermal growth factor receptor, EGFR）突变晚期非小细胞肺癌（non-small cell lung cancer, NSCLC）的靶向治疗显著改善患者预后。识别潜在获益患者实施序贯治疗，可延长其生存获益。但不同EGFR突变亚型对“二代EGFR酪氨酸激酶抑制剂（tyrosine kinase inhibitors, TKI）一线治疗+三代EGFR TKI序贯治疗”策略的疗效差异尚不明确。本研究评估了不同EGFR突变类型患者序贯治疗疗效特征与影响因素，为优化个体化治疗提供依据。 方法 本研究为单中心真实世界回顾性队列研究。入组2017年6月至2023年1月于中国医学科学院肿瘤医院初诊的109例EGFR敏感突变Ⅲ–Ⅳ期NSCLC患者。患者一线接受二代EGFR TKI（达可替尼或阿法替尼）治疗，进展后基因检测明确耐药突变状态，其中87例序贯三代EGFR TKI治疗。根据EGFR突变类型分为经典突变（19del或L858R）与非经典突变组。同时对EGFR突变类型进行亚组分析探索潜在获益人群：经典突变组分为19del或L858R两组。非经典突变患者分为单一罕见突变和复合突变组。此外，额外对非经典突变依据结构分组，分为了P环和αC螺旋压缩突变（P-loop and αC-helix Compressing，PACC）突变和非PACC突变亚组。主要终点包括一线及序贯治疗阶段无进展生存期（progression-free survival，PFS）及总生存期（overall survival, OS），采用Kaplan-Meier法及Cox回归分析生存结局。 结果 入组109位患者中女性占70.6%，中位年龄57岁。一线阶段经典突变组骨转移比例显著高于非经典组（58.4% vs 34.4%，P=0.02），且达可替尼初始应用更常见（71.4% vs 43.8%，P=0.01）。初治L858R突变患者伴发TP53突变的比例（46.9%）显著高于19del突变（7.1%）及非经典突变（25.0%，P<0.01）。疾病进展后，非经典突变患者脑膜转移的发生率显著高于经典突变（21.7% vs 3.1%，P=0.01），而其获得性T790M突变的阳性率（33.3%）则显著低于经典突变组（58.5%，P=0.03）。非经典突变内结构亚组分析显示，PACC组患者（n=14）的多器官（≥3个）转移比例（57.1%）显著高于非PACC组（n=9，11.1%，P=0.04），骨转移比例也呈现出升高趋势（64.3% vs 22.2%，P=0.09）。 经典突变亚组中，L858R突变患者（n=49）中91.8%使用达可替尼治疗，ORR为68.2%，与19del突变患者（n=28，ORR为84.0%）相比差异不显著（P=0.64）；两组中位PFS1分别为15.7个月（95% CI：11.3–16.8）和16.3个月（95% CI：15.0–25.5）（P=0.14）。非经典突变患者（n=23）一线使用二代TKI疗效良好，ORR1为65.2%，中位PFS1为12.4个月（95% CI：10.8–19.9），与经典突变患者（15.8个月，95% CI：14.5–17.1）相比无显著差异（P=0.20）。 共87人接受三代EGFR-TKI序贯治疗，非经典组（n=23）疗效相对较差，ORR2为5.9%，较经典组（20.0%）相比呈现减低趋势（P=0.27）。T790M阳性的经典突变患者（n=41）表现出明确的长期生存优势，中位OS达到80.8个月（95% CI：47.4–NA），而非经典型组中位OS仅为44.0个月（95%CI：40.8–NA），差异未达显著（P=0.36）。所有经典突变患者中，L858R突变患者中位OS尚未达到，与19del患者中位OS（47.4个月，95% CI：35.8–NA）无显著差异（P=0.23）。而非经典突变组中，不同T790M状态对序贯治疗的PFS2和OS均未表现出明显影响：T790M阳性者（n=10）的中位PFS2为7.8个月（95% CI：3.0–NA），而阴性者（n=13）为8.1个月（95% CI：3.0–NA），差异不显著（P=0.14）；两者中位OS分别为44.0个月（95% CI：40.8–NA）和41.7个月（95% CI：31.3–NA），差异亦不显著（P=0.69）。COX多因素回归分析显示，进展后伴随TP53突变为OS的独立不良预后因素（HR=2.59，95% CI：1.25–5.36，P=0.01）。 结论 L858R突变患者在我们队列中主要接受达可替尼一线治疗选择下，尽管TP53伴随突变占比高，整体疗效与19del突变患者接近。不同突变亚型序贯治疗获益不同，经典EGFR突变患者序贯治疗后获得明显长期生存优势，尤其在伴随获得性T790M突变的患者中尤为突出；非经典突变患者序贯疗效相对较差，内部异质性较强，仅根据T790M状态进行治疗决策可能不足以获得显著的疗效改善。进展后伴随TP53突变是预后独立不良危险因素。需针对具体突变特征选择个体化治疗策略。

Part I, Chapter 1: Real-World Exploration of Sequential Efficacy and Safety of ALK TKIs Abstract Background Lorlatinib, a third-generation anaplastic lymphoma kinase (ALK) tyrosine kinase inhibitor (TKI), has shown remarkable efficacy in ALK-positive non-small-cell lung cancer (NSCLC). However, its benefit as first-line therapy compared with other sequential ALK-TKI regimens remains unclear. This real-world study evaluates the efficacy and safety of lorlatinib both as initial treatment and in a sequential setting, aiming to optimize individualized therapy for ALK-positive NSCLC. Methods We conducted a single-center observational cohort study at the Cancer Hospital of the Chinese Academy of Medical Sciences, retrospectively enrolling patients treated with lorlatinib from February 2020 to April 2023 and prospectively from April 2023 to September 2024. All participants had stage IIIB–IV ALK-rearranged NSCLC and were assigned by treatment sequence to either first-line lorlatinib (cohort 1) or lorlatinib following progression on a second-generation ALK TKI (cohort 2). Safety was assessed in all lorlatinib-treated patients (n = 63). Efficacy endpoints included objective response rate (ORR) and progression-free survival (PFS), with secondary endpoints of disease control rate (DCR), intracranial ORR (iORR), intracranial PFS (iPFS), overall survival (OS), and safety. Prospective patients underwent lipid (triglycerides, total cholesterol, LDL-C, HDL-C) and liver-function monitoring at baseline, two weeks, and one month, then at intervals guided by lipid stability. Statistical analyses were performed using R 4.2.2: continuous variables were tested for normality (Shapiro–Wilk) and described as mean ± SD or median (IQR), with comparisons by t-test/ANOVA or Mann–Whitney/Kruskal–Wallis; categorical variables were compared by Pearson’s χ2 or Fisher’s exact test; ORR, DCR, and time-point PFS/OS rates were calculated with 95% CIs via Clopper–Pearson; survival was analyzed by Kaplan–Meier with log-rank test and Cox modeling for OS prognostic factors. Two-sided P < 0.05 was considered significant. The study received ethics approval. Results Fifty-one patients were evaluable for efficacy (six retrospective, 45 prospective): 30 in cohort 1 (median follow-up 9.4 months) and 21 in cohort 2 (median follow-up 45.3 months), with similar baseline characteristics. In cohort 1, ORR was 91.7% (22/24; 95% CI 73.0–99.0%), DCR was 100.0%, median PFS was not reached, and the 12-month PFS rate was 95.0%. In cohort 2 during the second-generation TKI phase, ORR was 87.5% (14/16; 95% CI 61.7–98.4%), DCR was 100.0%, median PFS was 24.5 months (95% CI 13.9–31.3), and the 12-month PFS rate was 71.4%. During the lorlatinib phase, ORR was 22.2% (4/18; 95% CI 6.4–47.6%), DCR was 77.8%, and median PFS was 19.5 months (95% CI 5.7–NA). There were no significant differences in ORR or DCR between cohorts (P = 1.00), and the 12-month PFS rate difference was marginal (P = 0.05). The 12- and 24-month OS rates were 92.3% versus 83.5% (P = 0.84), with median OS not reached in either group. Safety: Among all 63 lorlatinib-treated patients, hypercholesterolemia occurred in 100% and hypertriglyceridemia in 98.4% during early treatment; lipid-lowering interventions restored values toward baseline without any discontinuations. Grade ≥ 3 hypercholesterolemia was more frequent in patients with brain metastases (37.0% vs. 12.5%, P = 0.03). Other common adverse events included peripheral edema (63.5%), central nervous system effects (57.1%), weight gain (23.8%), hypertension (22.2%), peripheral neuropathy (22.2%), and arthralgia or stiffness (20.6%). One patient with preexisting hydronephrosis discontinued lorlatinib due to grade 3 proteinuria. Conclusion Lorlatinib achieves high response rates and durable PFS both as first-line and in sequential use after second-generation ALK TKIs in ALK-positive NSCLC. Although lipid abnormalities are ubiquitous, early monitoring and intervention effectively prevent treatment discontinuation. These findings support the use of lorlatinib in both settings and underscore the importance of proactive lipid management and tailored care for patients with brain metastases. Keywords Non-small cell lung cancer (NSCLC);ALK fusion gene;Hyperlipidemia;Targeted therapy; sequential therapy; Part I, Chapter 2: Exploration of Dyslipidemia Characteristics and Metabolic Mechanisms Induced by Third-generation ALK TKIs Abstract Background Although lorlatinib, a third-generation anaplastic lymphoma kinase (ALK) tyrosine kinase inhibitor (TKI), delivers outstanding antitumor activity in ALK-rearranged non-small-cell lung cancer (NSCLC), its unique adverse-event profile—particularly the mechanism of treatment-induced hyperlipidemia—remains unclear. This prospective study combines clinical observation with untargeted metabolomics to elucidate the metabolic drivers of lorlatinib-associated hyperlipidemia and inform management strategies. Methods Between April 2023 and September 2024, we enrolled 21 patients with stage IIIB–IV ALK-rearranged NSCLC receiving lorlatinib at the Cancer Hospital, Chinese Academy of Medical Sciences. All participants provided informed consent and underwent serial monitoring of fasting lipid panels at baseline, two weeks, and one month after treatment initiation. Patients were categorized by their dominant lipid abnormality during therapy into three groups: predominantly hypertriglyceridemia (Group A), predominantly hypercholesterolemia (Group B), and combined hypertriglyceridemia and hypercholesterolemia (Group C). Measurements included triglycerides (TG), total cholesterol (TC), low-density lipoprotein cholesterol (LDL-C), high-density lipoprotein cholesterol (HDL-C), lipoprotein(a) [Lp(a)] and activities of key lipid-metabolizing enzymes: hepatic lipase (HL), lipoprotein lipase (LPL), pancreatic lipase (LPS), acetyl-CoA carboxylase (ACC), ATP-citrate lyase (ACL) and fatty acid synthase (FAS). Untargeted metabolomic profiling using ultra-performance liquid chromatography–mass spectrometry (UPLC-MS) was performed on serum from nine representative patients (three per group) at the same time points. Lipid and enzyme changes over time were analyzed by Friedman test and Wilcoxon signed-rank test; correlations between enzyme activity and lipid levels by Pearson’s test; differential metabolites by principal component analysis (PCA) and orthogonal partial least squares discriminant analysis (OPLS-DA); and metabolic pathway network analysis. Results A total of 22 patients were enrolled for lipid and enzyme dynamic monitoring, including 3 patients in Group A, 8 in Group B, and 11 in Group C. At 2 weeks post-treatment, TG and CHOL significantly increased from baseline values of 1.43±0.62 mmol/L and 5.32±1.07 mmol/L to 3.48±1.99 mmol/L and 8.76±2.06 mmol/L, respectively (both P<0.01), remaining elevated at 1 month. LDL-C and HDL-C levels also increased synchronously (LDL-C: from 3.49±0.90 mmol/L to a maximum of 5.43±1.72 mmol/L; HDL-C: from 1.35±0.29 mmol/L to a maximum of 1.86±0.55 mmol/L; both P<0.01). Subgroup analysis revealed significant differences in CHOL between Group B and Group C at 2 weeks (P=0.006), while HDL-C and LDL-C showed significant overall subgroup differences (P=0.039 and P=0.018, respectively). TG did not differ significantly between groups (P>0.05). Although lipid levels slightly declined by 1 month, they remained significantly elevated compared to baseline (all P<0.01). Lp(a) and fasting glucose did not significantly change. Enzyme activity analysis revealed no significant overall changes in LPL, ACC, FAS, and ACL enzyme activities after treatment compared with baseline. Only HL showed a significant increasing trend at 1 month (overall P<0.01); however, enzyme activity was not significantly correlated with lipid levels (Pearson r<0.3, P>0.05). Metabolomic profiling identified 1,831 metabolites in total, and after treatment the serum metabolic signatures of all subgroups underwent extensive remodeling. These changes were marked by pronounced elevations in fatty acid oxidation intermediates (e.g., acylcarnitines), cholesterol synthesis precursors, and bile acid metabolites, suggesting impaired lipid oxidation and disrupted cholesterol–bile acid conversion. Significant perturbations were also observed in energy metabolism pathways, including amino acid metabolism, the tricarboxylic acid (TCA) cycle, and the pentose phosphate pathway. Pathway enrichment analysis and network reconstruction revealed that lorlatinib-induced dyslipidemia involves widespread remodeling across multiple metabolic pathways, with the most prominent abnormalities in fatty acid metabolism, steroid hormone biosynthesis, sphingolipid signaling, and mitochondrial energy metabolism. Group A (elevated triglycerides): membrane lipid metabolism is dysregulated, lipolysis is reduced, and lipid transport is impaired. First-line therapy: fibrates (fenofibrate), which activate PPAR-α and increase lipoprotein lipase (LPL) activity to promote triglyceride hydrolysis. Group B (elevated cholesterol): membrane phospholipid remodeling, abnormal cholesterol transport and absorption, and enhanced oxidative stress. First-line therapy: statins (atorvastatin or rosuvastatin), which inhibit HMG-CoA reductase to improve membrane lipid metabolism and reduce oxidative stress. Group C (elevated triglycerides and cholesterol): concurrent abnormalities in glycerophospholipid and amino acid metabolism, increased inflammation, and hyperactive energy metabolism. Recommended therapy: combination of statins and fibrates, or combined lipid-lowering therapy with SGLT2 inhibitors, to simultaneously control triglycerides and cholesterol and suppress inflammation. Conclusion Lorlatinib rapidly induces hyperlipidemia characterized by comprehensive elevations in cholesterol and triglycerides. Its mechanism involves complex metabolic network remodeling beyond traditional single-enzyme pathways, likely mediated by nuclear receptor-driven lipid metabolic dysregulation. Prompt individualized lipid monitoring and intervention can effectively control this adverse reaction, enabling patients to receive the substantial antitumor benefits of lorlatinib while reducing cardiovascular risks. These findings provide a metabolic management framework for ALK-positive lung cancer patients and establish groundwork for future studies into the mechanisms and intervention strategies of TKI-related metabolic toxicities. Keywords Lorlatinib;Non-small cell lung cancer (NSCLC); ALK fusion gene;Hyperlipidemia;Metabolomics Exploration of Potential Beneficiaries of EGFR TKI Sequential Therapy Abstract Background Targeted therapy significantly improves outcomes for advanced non-small cell lung cancer (NSCLC) patients with epidermal growth factor receptor (EGFR) mutations. Identifying patients who potentially benefit from sequential therapy (second-generation followed by third-generation EGFR tyrosine kinase inhibitors [TKIs]) may extend survival. However, the efficacy of this sequential approach across different EGFR mutation subtypes remains unclear. This study evaluates the therapeutic effects and influencing factors of sequential EGFR TKI therapy in various EGFR mutation subtypes to optimize individualized treatment. Methods This single-center retrospective real-world cohort study enrolled 109 patients with EGFR-sensitive mutations (stages III–IV NSCLC), initially diagnosed at the Cancer Hospital of the Chinese Academy of Medical Sciences from June 2017 to January 2023. All patients received second-generation EGFR TKIs (dacomitinib or afatinib) as first-line therapy. After disease progression, resistance mutation status was determined by genetic testing, and 87 patients underwent sequential third-generation EGFR TKI therapy. Patients were grouped according to EGFR mutation type into classic mutations (19del or L858R) and uncommon mutations. Subgroup analyses were conducted to identify potential beneficiaries: classic mutations were subdivided into 19del and L858R groups, while uncommon mutations were classified into single uncommon mutations and compound mutations. Further, uncommon mutations were structurally grouped into P-loop and αC-helix Compressing (PACC) and non-PACC mutations. Primary endpoints included progression-free survival (PFS) and overall survival (OS) during first line and sequential therapy, analyzed via Kaplan-Meier methods and Cox regression. Results Of the 109 enrolled patients, 70.6% were female, with a median age of 57 years. The classic mutation group exhibited a significantly higher incidence of bone metastasis (58.4% vs 34.4%, P=0.02) and more frequent initial use of dacomitinib (71.4% vs 43.8%, P=0.01). The initial L858R mutation group had a significantly higher proportion of concurrent TP53 mutations (46.9%) compared to 19del mutations (7.1%) and non-classic mutations (25.0%, P<0.01). After progression, the incidence of leptomeningeal metastasis was significantly higher in non-classic mutations compared to classic mutations (21.7% vs 3.1%, P=0.01), while the acquired T790M mutation positivity rate was significantly lower (33.3% vs 58.5%, P=0.03). Structural subgroup analysis of non-classic mutations showed that patients in the PACC group (n=14) had a significantly higher proportion of multi-organ (≥3 organs) metastases (57.1%) compared to the non-PACC group (n=9, 11.1%, P=0.04), with a trend toward increased bone metastasis (64.3% vs 22.2%, P=0.09). In the classic mutation subgroup, 91.8% of L858R patients (n=49) received dacomitinib, achieving an ORR of 68.2%, like the 19del group (n=28, ORR 84.0%; P=0.64). Median PFS1 was 15.7 months (95% CI: 11.3–16.8) and 16.3 months (95% CI: 15.0–25.5) for L858R and 19del groups, respectively (P=0.14). Non-classic mutation patients (n=23) showed good efficacy with second-generation TKIs as first-line treatment (ORR1: 65.2%, median PFS1: 12.4 months, 95% CI: 10.8–19.9), with no significant difference compared to classic mutation patients (median PFS1: 15.8 months, 95% CI: 14.5–17.1; P=0.20). Among 87 patients receiving third-generation EGFR-TKI sequential therapy, non-classic mutation patients (n=23) had relatively poorer efficacy (ORR2: 5.9%) compared to classic mutation patients (20.0%), although the difference was not significant (P=0.27). Classic mutation patients positive for T790M (n=41) demonstrated clear long-term survival benefits, achieving a median OS of 80.8 months (95% CI: 47.4–NA), compared to 44.0 months (95% CI: 40.8–NA) in the non-classic group (P=0.36). Within classic mutation patients, the median OS for L858R mutation has not been reached, showing no significant difference compared to 19del patients (median OS: 47.4 months, 95% CI: 35.8–NA; P=0.23). For non-classic mutation patients, different T790M statuses did not significantly affect PFS2 and OS: median PFS2 for T790M-positive (n=10) and T790M-negative (n=13) were 7.8 months (95% CI: 3.0–NA) and 8.1 months (95% CI: 3.0–NA), respectively (P=0.14); median OS was 44.0 months (95% CI: 40.8–NA) and 41.7 months (95% CI: 31.3–NA), respectively (P=0.69). Cox multivariate regression analysis indicated concurrent TP53 mutations post-progression as an independent adverse prognostic factor for OS (HR=2.59, 95% CI: 1.25–5.36, P=0.01). Conclusion Despite a higher proportion of concurrent TP53 mutations, patients with L858R mutations treated primarily with dacomitinib achieved efficacy comparable to those with 19del mutations. The benefits of sequential therapy varied significantly across EGFR mutation subtypes. Classic EGFR mutation patients showed clear long-term survival advantages with sequential treatment, particularly among those acquiring T790M mutations. Non-classic mutation patients exhibited relatively poorer sequential therapy outcomes and greater internal heterogeneity, suggesting treatment decisions based solely on T790M status may not be sufficient. Concurrent TP53 mutation post-progression is an independent adverse prognostic factor, emphasizing the need for personalized treatment strategies based on specific mutation characteristics.