Serum tumor markers and outcomes in lung cancer patients with brain metastases: a retrospective longitudinal cohort study.

Background: Serum tumor markers (STMs) are recommended for cancer diagnosis and surveillance. However, their role in lung cancer with brain metastases (BM) is not yet clear. We aim to analyze the roles of baseline levels of STMs or ongoing STM surveillance on survival.

Methods: This retrospective longitudinal cohort study included 1,169 lung cancer patients with BM. The STM data during disease course were collected. Distinct trajectory groups were identified using the latent class growth mixed model (LCGMM). The roles of STMs on survival were further analyzed using Kaplan-Meier analysis and Cox proportional hazard models.

Results: Serum levels of cytokeratin-19 fragment (CYFRA21-1) (P<0.001), carcinoembryonic antigen (CEA) (P=0.005) and neuron-specific enolase (NSE) (P<0.001) at baseline exhibited significant correlation with overall survival (OS) of patients with BM, serving as independent prognostic factors. Further analysis indicated that baseline CYFRA21-1, CEA, NSE as well as status of key driver genes were independent prognostic factors in non-small cell lung cancer (NSCLC) patients with BM, while for small cell lung cancer (SCLC) patients with BM, baseline NSE and receiving chemotherapy show independent correlations with survival. Furthermore, we delineated the dynamic trajectories of STMs based on changes in disease course. More specifically, compared to those showing a baseline-high trend in CEA levels, the survival of patients with either persistently-rising or consistently normal levels seemed to be more promising. For CYFRA21-1, both early-rising and later-rising trends were observed, indicating a prognosis inferior to that of individuals with normal-level trajectory. Likewise, for NSE, patients with persistently-rising or persistently-descending trends showed no significantly survival difference. However, in comparison with the status of driver genes, receiving radiotherapy and targeted therapy, the dynamic changes in STM levels lacked independent prognostic significance. Further analysis indicated that among BM patients lacking key driver genes, NSE trajectory (P<0.05), CYFRA21-1 trajectory (P<0.05) and receiving chemotherapy (P<0.001) were independent prognostic factors.

Conclusions: Baseline levels of serum CYFRA21-1, CEA and NSE, as well as status of key driver genes are recommended for evaluating BM patients' outcome. Dynamic changes of STMs during disease course were not significantly associated with the final outcome of BM patients.