Immunotherapy targeting PD-1/PD-L1 has become a standard treatment for advanced lung cancer, yet only 20-30% of patients experience lasting benefit. Identifying in advance who will respond is a major unsolved clinical challenge.
The current standard biomarker, PD-L1 expression measured on tumor tissue (tPD-L1), has significant practical limitations. Obtaining tissue requires invasive biopsies that carry procedural risks, and tumor biopsies only sample one location in a tumor that can vary considerably from one area to another.
PD-L1 expression is also dynamic - it can change over time in response to prior treatments such as chemotherapy or radiation, meaning a result from one biopsy may not reflect the tumor's current biology when the patient starts immunotherapy. Different laboratory assays for measuring PD-L1 also produce inconsistent results.
These limitations are especially acute in small cell lung cancer (SCLC), where the role of tPD-L1 as a predictive biomarker is poorly established and not routinely standardized. This study sought a better, non-invasive alternative using liquid biopsy.
Circulating tumor cells (CTCs) are cancer cells that shed from tumors into the bloodstream. They can be captured from a simple blood draw, offering a non-invasive window into the biology of a patient's cancer at any point in time.
Unlike tissue biopsies, liquid biopsies can be repeated frequently and may better represent the overall heterogeneity of a patient's disease by sampling cells from multiple tumor sites simultaneously. CTC counts have already been shown to correlate with prognosis in several cancer types.
This study asked a new question: can the PD-L1 status of CTCs (CTC PD-L1) predict which patients will respond to chemo-immunotherapy? Prior studies in this area produced conflicting results - some found CTC PD-L1 positivity associated with worse prognosis, others with better response - leaving the clinical picture unclear.
The researchers developed a novel assay using the LiquidBiopsy platform to simultaneously detect CTCs and measure PD-L1 expression on those cells, then prospectively tested whether this blood-based measurement could predict outcomes in both NSCLC and SCLC patients receiving combination chemo-immunotherapy.
Fifty-two patients with advanced lung cancer (stages III-IV) were enrolled at the Fourth Affiliated Hospital of Soochow University in a prospective, double-blind, real-world study. The cohort included adenocarcinoma (29%), squamous cell carcinoma (44%), and small cell lung cancer (27%), all of whom were negative for targetable driver gene mutations.
Blood samples (10 mL each) were collected at four time points: before treatment (baseline), after two treatment cycles (approximately day 42), after four cycles (approximately day 84), and at disease progression. All CTC analysis was performed at an independent laboratory to maintain the blind.
CTCs were identified using a specific protein signature: cells positive for cytokeratin (CK, a marker of epithelial cancer cells), positive for DAPI (a nuclear dye), and negative for CD45 (a marker of white blood cells). The ratio of CK to CD45 fluorescence signals above 1.7 was used to classify cells as tumor-derived, with near-perfect sensitivity and specificity.
CTC PD-L1 positivity was defined by an additional PD-L1 fluorescence signal with a PD-L1/CD45 ratio above 3.2, validated to distinguish tumor cell PD-L1 from PD-L1 expressed on immune cells. Patients with at least one PD-L1-positive CTC were classified as CTC PD-L1+.
Of the 52 enrolled patients, CTCs were detected in 71.2% at baseline. Half the patients (50%) were classified as CTC PD-L1+. Among the 38 patients who received at least two cycles of chemo-immunotherapy and were evaluable for response, results were striking.
CTC PD-L1+ patients achieved an objective response rate (ORR) of 84.2% - meaning 84% had meaningful tumor shrinkage. In contrast, CTC PD-L1- patients had an ORR of only 36.8%. This difference in response rates was large and statistically significant.
Comparing to tissue PD-L1 (tPD-L1), the ORR pattern was similar (76% for tPD-L1+ vs. 41% for tPD-L1-), but critically, the concordance between CTC PD-L1 status and tissue PD-L1 status was only 52.6% - meaning the two tests were in agreement barely more often than chance. This strongly suggests they are capturing different biological information.
Combining CTC PD-L1 and tPD-L1 together improved predictive accuracy further, with sensitivity rising to 86.9% and specificity to 78%. Multivariate analysis confirmed that CTC PD-L1 was an independent predictor of response even after accounting for age, sex, cancer subtype, and other clinical factors (p=0.003).
Survival differences were dramatic. CTC PD-L1+ patients had a median progression-free survival (PFS) of 16 months compared to only 4 months in CTC PD-L1- patients - a hazard ratio of 0.28, meaning CTC PD-L1+ patients had a 72% lower risk of progression or death at any given time point (p=0.004).
For overall survival (OS), the CTC PD-L1+ group's median survival was not yet reached by the end of follow-up, while the CTC PD-L1- group had a median survival of approximately 15 months. The hazard ratio was 0.19, representing an 81% reduction in risk of death (p=0.017).
In contrast, tissue PD-L1 (tPD-L1) status showed no significant association with either PFS (p=0.74) or OS (p=0.85) in this cohort. This is a critical finding - the blood test outperformed the established standard tissue biomarker in predicting actual patient survival.
Multivariate survival analysis confirmed CTC PD-L1 as the sole independent predictor of both PFS and OS among all clinical variables tested, including cancer subtype, performance status, tumor markers, smoking history, and comorbidities.
Serial blood samples allowed the researchers to track how CTC counts changed during treatment. The overall CTC detection rate fell from 71.2% before treatment to 33.3% after two cycles, consistent with tumor cells being eliminated by effective therapy.
Among responders (partial response patients), 87.5% showed decreased or stable CTC counts at the first follow-up time point, indicating that declining CTCs correlated with clinical tumor shrinkage. Non-responding patients tended to have stable or rising CTC counts.
By the four-cycle time point, all seven responders in the extended monitoring group showed decreased or stable CTC counts, while the single non-responder showed increasing CTCs - an early signal of treatment failure potentially visible weeks before imaging would confirm progression.
Among conventional tumor markers, NSE (neuron-specific enolase) showed the best correlation with treatment response compared to CEA, CYFRA21-1, and SCC, suggesting that combining CTC monitoring with NSE measurement could further improve real-time treatment tracking.
The low concordance of only 52.6% between CTC PD-L1 and tissue PD-L1 status is a key finding. If the two tests were measuring the same thing, they would agree much more often. The discordance suggests that PD-L1 expression on circulating tumor cells reflects a different aspect of cancer biology than what is captured in a single tissue biopsy.
Tissue biopsies sample one location in a spatially heterogeneous tumor. CTCs, by circulating through the bloodstream, may better represent the overall tumor PD-L1 landscape, including metastatic sites that are not accessible for biopsy. This could explain why CTC PD-L1 better predicts systemic treatment outcomes.
CTCs may also represent the most aggressive or invasive fraction of cancer cells - those capable of surviving in the bloodstream and seeding metastases. The immunological properties of these cells, including PD-L1 expression, may be particularly relevant for predicting how the immune system and immunotherapy drugs will interact with the cancer.
The finding that CTC PD-L1 predicts outcomes in both NSCLC and SCLC is notable, as tissue PD-L1 testing is not standardized for SCLC. A liquid biopsy approach could fill this gap for SCLC patients, who currently lack reliable predictive biomarkers for immunotherapy.
Key limitations include the small sample size of 38 evaluable patients, which limits statistical power and the ability to draw definitive conclusions within subgroups like SCLC alone. The single-center design and use of a proprietary detection platform also limit generalizability.
The heterogeneity in tissue samples - including surgically resected specimens versus small biopsies from primary versus metastatic sites - may have contributed to the poor concordance with CTC PD-L1 results and could be a source of bias.
Despite these limitations, the clinical promise is significant. A blood test that can be repeated before, during, and after treatment - with no additional risk to the patient - that outperforms the established tissue biomarker in predicting survival is a meaningful advance worth pursuing in larger trials.
If validated, CTC PD-L1 testing could transform clinical practice by enabling real-time monitoring of treatment response, earlier detection of resistance, and more informed decisions about when to change therapy - all based on a simple blood draw rather than repeated invasive biopsies.