背景

肺鳞癌（Squamous cell lung cancer, SCC）是肺癌常见的病理亚型，确诊时常为晚期，传统化疗效果差。近年来，免疫检查点抑制剂（Immune checkpoint inhibitors，ICIs）在临床试验中显示可改善晚期SCC患者的预后，但是反映治疗模式演变对其预后影响的真实世界数据仍较缺乏。此外，程序性死亡配体-1（Programmed death ligand-1，PD-L1）阴性的患者接受免疫治疗的获益也亟需真实世界数据支持。针对治疗反应存在较大差异的患者，其肿瘤微环境特征亦需要探索。

目的

本研究拟建立真实世界的晚期SCC结构化数据库，系统分析ICIs上市前后单中心晚期SCC治疗模式的演变过程对患者生存的影响和不良预后的关键危险因素。同时聚焦PD-L1阴性的晚期SCC患者，分析化疗联合帕博利珠单抗方案的生存获益。最后，通过数字空间组学方法，解析不同治疗反应的PD-L1阴性SCC患者的肿瘤微环境特征。

方法

本研究引入自然语言提取和大语言模型等技术，通过人机协同模式构建SCC结构化数据库，回顾性收集2015年至2023年的晚期SCC患者，以纳武利尤单抗在我国获批肺癌适应证为节点（2018年6月15日），将患者划分为ICIs上市前和ICIs上市后两组，分析治疗方案演变和生存状况。基于Captra-Lung多中心数据库，纳入2017年12月至2024年4月PD-L1阴性且一线应用帕博利珠单抗联合化疗（Pembro组）的晚期SCC患者，以同期含铂双药化疗的患者为对照（Chemo组），分析生存情况。选取对免疫治疗差异应答的SCC患者分成两组（定义无进展生存期[progression-free survival，PFS]≤6月为疗效不佳组，PFS≥18月为疗效良好组），利用肺组织病理切片进行数字空间组学检测和分析，探索疗效差异患者的肿瘤微环境特征。

结果

本研究成功利用人工智能技术辅助建立了晚期SCC结构化数据库，共纳入患者601例，ICIs上市后患者占69.7%（419/601），其中69.7%（292/419）一线应用化疗联合ICIs治疗。中位随访39.0月，ICIs上市后人群的PFS显著延长（8.1月 ，[95%置信区间[confidence interval, CI]：7.3-10.1] vs 6.2月[95%CI:5.2-7.4]，P<0.0001）；总生存期（overall survival，OS）明显改善（25.8月，95%CI：22.3-31.4] vs 13.7月，95%CI：10.7-17.8；P<0.001）。ICIs上市后20.3%（85/419）的患者一线未使用ICIs，其中位PFS仅5.0月。多个多因素COX比例风险回归模型显示一线仅化疗者疾病进展（HR=2.3-2.6；P<0.001）和死亡风险（HR=2.1-3.5；P<0.001）均显著增加。ECOG-PS>1也是死亡（HR=1.8-2.5；P<0.003）的独立危险因素。ICIs上市前后，3级及以上的不良事件发生率相近（41.2% vs 42.2%；P=0.82）。

在Captra-Lung数据库中，5个中心的87例SCC患者（Pembro组：PD-L1晚期SCC 40例，46.0%）和159例非鳞非小细胞肺癌（non-squamous non-small cell lung cancer, non-SCC，Pembro组中PD-L1阴性的non-SCC共74例，占46.5%）纳入分析。中位随访28.3月，在SCC组中，与化疗相比，Pembro组的PFS显著延长（13.8月 [95%CI：3.2-24.1] vs 4.8月，95%CI：3.4-6.2; P<0.001），OS明显改善（NR vs 14.2月，95%CI：6.3-22.1; P=0.005）；而non-SCC患者的PFS在治疗组间无明显差异（Pembro：9.3月 [95%CI：7.6-11.0] vs Chemo：8.0月 [95%CI：6.0-10.0]; P=0.61）；OS亦无显著差别（Pembro：18.1月[95%CI：13.4-22.8] vs Chemo：21.3月 [95%CI：18.3-21.3]; P=0.99）。

选取8例PD-L1阴性的SCC患者肺组织病理标本（疗效不佳和疗效良好组各4例）进行数字空间组学分析。转录组数据显示在疗效不佳组中，浆细胞在肿瘤区室中浸润比例显著降低（0.75% vs 5.57%, P=0.03）；而在淋巴细胞区室中则相反（11.53% vs 3.69%，P=0.008）。蛋白质组富集分析表明：在疗效不佳组中，PI3K/AKT、白细胞介素-4/13信号通路在肿瘤区室内显著富集；而淋巴细胞区室内则存在调节性T细胞通路的激活。

结论

真实世界数据表明：ICIs上市后，晚期SCC的治疗格局发生了重大变革，化疗联合免疫已经成为主流一线方案，且患者的PFS和OS均显著延长。PD-L1阴性晚期SCC一线使用帕博丽珠单抗联合化疗可显著延长患者生存。数字空间组学分析显示PD-L1阴性SCC化疗联合免疫治疗效果不佳的患者肿瘤微环境存在明显特征。

Background
Squamous cell lung cancer (SCC) is a common pathological subtype of lung cancer, often diagnosed at an advanced stage with poor outcomes under traditional chemotherapy. While immune checkpoint inhibitors (ICIs) have demonstrated improved prognosis in clinical trials, real-world data reflecting the impact of evolving treatment paradigms remain scarce. Additionally, the benefits of immunotherapy for PD-L1-negative patients require further validation through real-world evidence. Heterogeneity in treatment responses necessitates exploration of tumor microenvironment (TME) characteristics.

Objective

This study aimed to establish a real-world structured database for advanced SCC, systematically analyze the evolution of treatment patterns and survival outcomes before and after ICI approval, and identify key prognostic risk factors. Focusing on PD-L1-negative advanced SCC patients, we evaluated survival benefits of pembrolizumab-chemotherapy combination therapy. Furthermore, digital spatial profiling was employed to characterize TME features in patients with differential treatment responses.

Methods

Using natural language processing and large language model-assisted human-AI collaboration, we constructed a structured SCC database, retrospectively enrolling advanced SCC patients (2015–2023). Patients were stratified into pre- and post-ICI eras based on nivolumab's approval in China (June 15, 2018) to analyze treatment evolution and survival. From the Captra-Lung multicenter database, PD-L1-negative advanced SCC patients receiving first-line pembrolizumab-chemotherapy (Pembro group, n=40) or platinum-doublet chemotherapy (Chemo group, n=47) between December 2017 and April 2024 were compared. For TME analysis, SCC patients were dichotomized by immunotherapy response (poor: progression-free survival [PFS] ≤6 months; favorable: PFS ≥18 months), with lung tissue sections subjected to digital spatial omics profiling.

Results

A structured database of 601 advanced SCC patients was established. Post-ICI approval, 69.7% (419/601) received first-line treatment, with 69.7% (292/419) receiving chemotherapy-ICI combinations. Median follow-up was 39.0 months. Post-ICI patients exhibited significantly prolonged PFS (8.1 months, 95% CI: 7.3–10.1 vs. 6.2 months, 95% CI: 5.2–7.4; P< 0.0001) and OS (25.8 months, 95% CI: 22.3–31.4 vs. 13.7 months, 95% CI: 10.7–17.8; P< 0.001). Patients not receiving first-line ICIs (20.3%, 85/419) had a median PFS of 5.0 months. Multivariate Cox analysis confirmed increased risks of progression (HR = 2.3–2.6; P< 0.001) and death (HR = 2.1–3.5; P< 0.001) in chemotherapy-only groups. ECOG-PS >1 independently predicted higher mortality (HR = 1.8–2.5; P< 0.003). Grade ≥3 adverse events were comparable between eras (41.2% vs. 42.2%; P= 0.82).

In the Captra-Lung cohort, Pembro group showed significantly longer PFS (13.8 months, 95% CI: 3.2–24.1 vs. 4.8 months, 95% CI: 3.4–6.2; P< 0.001) and OS (not reached vs. 14.2 months, 95% CI: 6.3–22.1; P= 0.005). No survival differences were observed in non-SCC.

Spatial profiling revealed reduced plasma cell infiltration in tumor compartments (0.75% vs. 5.57%; P= 0.03) and increased infiltration in lymphocyte compartments (11.53% vs. 3.69%; P= 0.008) in the poor-response group. Proteomic enrichment analysis highlighted activation of PI3K/AKT and interleukin-4/13 (IL-4/IL-13) signaling pathways in tumor regions, alongside regulatory T-cell pathway activation in lymphocyte regions of poor responders.

Conclusion

Real-world data demonstrate a paradigm shift in advanced SCC treatment post-ICI approval, with chemotherapy-ICI combinations becoming the mainstream first-line regimen, significantly improving PFS and OS. First-line pembrolizumab-chemotherapy enhances survival in PD-L1-negative advanced SCC. Spatial omics identified distinct TME features in poor responders, providing mechanistic insights into differential treatment outcomes.