Prognostic factors for poor outcomes and the predictive role of circulating tumor DNA in immunotherapy for advanced non-small cell lung cancer.

1. Introduction
Non-small cell lung cancer (NSCLC) accounts for approximately 85% of all lung cancer cases and remains a leading cause of cancer-related mortality worldwide. Advanced NSCLC, particularly stages IIIB and IV, is characterized by poor prognosis and limited treatment options, emphasizing the critical need for effective therapeutic strategies and reliable prognostic biomarkers. The advent of immune checkpoint inhibitors (ICIs), such as programmed death-1 (PD-1) and programmed death-ligand 1 (PD-L1) inhibitors, has revolutionized the treatment landscape for advanced NSCLC, offering significant survival benefits for selected patients.[1–3] However, the clinical outcomes of immunotherapy vary considerably, with a subset of patients experiencing disease progression or limited therapeutic response. Identifying prognostic factors associated with poor outcomes and predictive markers of treatment efficacy is therefore crucial for optimizing patient management and improving clinical outcomes.[4,5]
Circulating tumor DNA (ctDNA), a fragment of cell-free DNA released into the bloodstream by tumor cells, is increasingly recognized as a valuable biomarker in cancer diagnosis, monitoring, and prognosis. ctDNA levels reflect tumor burden and genetic alterations, providing a noninvasive method for evaluating disease dynamics. In the context of immunotherapy, ctDNA offers potential as a predictive tool for assessing treatment response.[6,7] While initial studies have demonstrated the utility of ctDNA in various cancers, its role in guiding immune therapy in advanced NSCLC remains inadequately defined. Traditional prognostic factors for NSCLC, such as PD-L1 expression, tumor stage, and performance status, provide insights into patient outcomes but have limitations in predicting immunotherapy efficacy. PD-L1 expression, although widely used, does not fully account for the variability in treatment responses. Combining ctDNA analysis with established clinical factors could offer a more comprehensive approach to patient stratification and prognosis.[8,9]
This study investigates the prognostic factors associated with poor outcomes in advanced NSCLC and evaluates the predictive value of ctDNA for immune therapy. By analyzing ctDNA levels before and during treatment, we aim to assess their relationship with therapeutic response and PFS. This research seeks to clarify the role of ctDNA in predicting immune therapy outcomes, providing a basis for its integration into clinical practice.

2. Method

2.1. Study design
This retrospective study was conducted at our hospital to evaluate prognostic factors associated with poor outcomes and the predictive value of ctDNA in immune therapy for advanced NSCLC. Patients treated between January 2021 and December 2023 were included based on the following criteria: NSCLC at stage IIIB or IV with negative driver gene status and PD-L1 expression > 50%, age ≥ 18 years, and treatment with Tislelizumab at a standard dose of 200 mg every 21 days. Exclusion criteria included concurrent infections, other malignancies, or autoimmune diseases. A total of 118 eligible patients were enrolled, and therapeutic efficacy was evaluated after 2 treatment cycles using RECIST 1.1 criteria, with outcomes classified as complete remission (CR), PR, stable disease (SD), or PD. Patients achieving CR or PR were categorized as the observation group (n = 67), while those with SD or PD comprised the control group (n = 51). The study adhered to STROBE (strengthening the reporting of observational studies in epidemiology) guidelines.[10] Informed consent was obtained from all subjects and/or their legal guardian(s). The study was approved by the ethics committee of our hospital and conducted in accordance with relevant guidelines and the Declaration of Helsinki. All methods adhered to ethical principles for medical research involving human subjects. Participant privacy was protected by removing personal identifiers before data analysis.

2.2. Efficacy assessment
All patients underwent baseline evaluations before treatment and again after 2 cycles of therapy. Assessments included contrast-enhanced computed tomography (CT) of the chest, abdomen, and pelvis; contrast-enhanced magnetic resonance imaging (MRI) of the brain; whole-body bone scans; and ultrasound examination of superficial lymph nodes. Treatment efficacy was evaluated using the RECIST 1.1, which categorized responses as CR, PR, SD, or PD. The objective response rate was calculated as follows: objective response rate = (CR + PR)/ (CR + PR + SD + PD) × 100%.

2.3. Measurement of peripheral blood ctDNA
Peripheral blood ctDNA levels were measured through plasma separation and DNA extraction prior to treatment and after 2 cycles of therapy. The process was performed as follows: 5 mL of fasting venous blood was collected from each patient and centrifuged at 2000 rpm for 10 minutes to separate the upper plasma layer; plasma DNA was extracted using the Qiagen plasma DNA extraction kit (Qiagen, Germany); the collected plasma was further centrifuged at 3500 rpm for 15 minutes, and the precipitate was retained; lysis buffer was added to the precipitate, followed by incubation for 60 minutes, after which reagent 2 was added and the sample incubated at 50°C for 3 hours; reagents 3 to 5 were sequentially added, followed by centrifugation to collect the precipitate; the DNA precipitate was resuspended in 70% ethanol, centrifuged, and the supernatant discarded, after which the remaining DNA pellet was dissolved using the kit’s dissolution buffer; and ctDNA was isolated using the Qiagen (Hilden, North Rhine-Westphalia , Germany) column-based adsorption kit according to the manufacturer’s instructions. Finally, ctDNA fragments were quantified using fluorescence-based quantitative polymerase chain reaction (qPCR) to monitor plasma ctDNA levels.

2.4. Follow-up and outcome assessment
All patients were followed up monthly through telephone consultations, outpatient visits, or inpatient reviews after initiating treatment. The follow-up period ended on December 31, 2024, with a median follow-up duration of 8 months. During follow-up, PFS was recorded, defined as the time from the initiation of treatment to either disease progression or the date of the last follow-up. This systematic approach ensured accurate and consistent monitoring of patient outcomes.

2.5. Statistical analysis
All statistical analyses were conducted using SPSS software (Version 27.0; IBM Corporation, Armonk). For quantitative variables following a normal distribution, inter-group comparisons were performed using independent sample t-tests, with results expressed as mean ± standard deviation. Categorical variables were summarized as frequencies and percentages, and their associations were examined using Chi-square (χ²) tests. If the Chi-square test assumptions were not met, Fisher exact test was applied. The predictive value of peripheral blood ctDNA levels for achieving an objective response to immunotherapy was evaluated using receiver operating characteristic (ROC) curve analysis. Prognostic factors influencing posttreatment outcomes were identified through univariate and multivariate regression analyses. The correlation between ctDNA levels and PFS was assessed using Spearman rank correlation coefficient. Kaplan–Meier survival curves were utilized for survival analysis, and statistical significance was set at P < .05.

3. Results

3.1. Comparison of clinical characteristics between observation and control groups
Patients who achieved CR or PR after 2 treatment cycles were categorized as the observation group (n = 67), while those with SD or PD comprised the control group (n = 51). A comparative analysis of baseline clinical characteristics was conducted between the 2 groups to identify any significant differences that might influence treatment outcomes. The distribution of sex was similar between the groups, with male patients constituting 64.18% of the observation group and 64.71% of the control group (P = .953). The mean age was 66.02 ± 3.80 years in the observation group and 64.83 ± 3.96 years in the control group, showing no significant difference (P = .101). Smoking history was comparable as well, with 67.16% of patients in the observation group and 72.55% in the control group being smokers (P = .529). Pathological types were predominantly adenocarcinoma in both groups, accounting for 74.63% in the observation group and 72.55% in the control group (P = .799). The staging of the disease showed no significant difference; stage IV cases comprised 76.12% of the observation group and 74.51% of the control group (P = .841). The maximum tumor diameter was also similar between the groups, with the majority of patients having tumors measuring 3 to 5 cm (64.18% in the observation group vs 56.86% in the control group, P = .786). However, a significant difference was observed in the treatment modality. Patients in the observation group were more likely to have received a combination of immunotherapy and chemotherapy (80.60%) compared to the control group (60.78%), while immunotherapy alone was more common in the control group (39.22% vs 19.40% in the observation group), with this difference being statistically significant (P = .018) (Table 1).

3.2. Changes in peripheral blood ctDNA levels before and after treatment
The levels of ctDNA in peripheral blood demonstrated significant changes between the observation group (CR or PR, n = 67) and the control group (SD or PD, n = 51) before and after treatment. Pretreatment ctDNA levels were comparable between the 2 groups (4.61 ± 1.18 ng/μL vs 4.47 ± 1.12 ng/μL, P = .515), indicating no baseline differences. posttreatment ctDNA levels showed a substantial reduction in the observation group (2.72 ± 1.11 ng/μL), whereas they remained relatively elevated in the control group (4.18 ± 1.54 ng/μL). The difference in posttreatment ctDNA levels between the 2 groups was statistically significant (P < .001). Within-group analyses revealed that the observation group experienced a pronounced decrease in ctDNA levels after treatment (t = 9.549, P < .001), whereas the control group did not show a significant change (t = 1.088, P = .279) (Table 2).

3.3. Predictive value of peripheral blood ctDNA for objective response to immunotherapy
The predictive utility of ctDNA levels for achieving an objective response (CR or PR) to immunotherapy in advanced NSCLC was evaluated. ROC curve analysis demonstrated that the AUC for ctDNA as a predictor of objective response was 0.821, indicating high predictive accuracy. The Youden index was calculated to be 0.620, with an optimal ctDNA cutoff value of 3.59 ng/μL. Using this threshold, the sensitivity and specificity for predicting an objective response were 80.9% and 79.8%, respectively.

3.4. Correlation between peripheral blood ctDNA levels and PFS in advanced NSCLC patients
The relationship between ctDNA levels and PFS in advanced NSCLC patients was analyzed using Spearman correlation. Results revealed a significant negative correlation between ctDNA levels and PFS (r = −0.784, P < .001), indicating that higher ctDNA levels were associated with shorter PFS.

3.5. Cox regression analysis of factors influencing PFS in advanced NSCLC patients
To identify factors associated with PFS in advanced NSCLC patients, univariate Cox regression analysis was performed on clinical and pathological characteristics, including circulating tumor DNA (ctDNA) levels. The analysis revealed that pathological type, maximum tumor diameter, disease stage, treatment modality, and ctDNA levels were significant predictors of PFS (P < .05, Table 3). Subsequently, variables identified as statistically significant in the univariate analysis were included in a multivariate Cox regression model. This analysis demonstrated that treatment modality and ctDNA levels were independent predictors of PFS in advanced NSCLC patients (P < .05, Table 4). Specifically, combination therapy involving immunotherapy and chemotherapy, as well as lower ctDNA levels, were associated with improved PFS.

3.6. Kaplan–Meier survival analysis of PFS based on ctDNA levels
Based on the posttreatment ctDNA cutoff value of 3.59 ng/μL, patients were stratified into a high ctDNA group (ctDNA ≥ 3.59 ng/μL, n = 53) and a low ctDNA group (ctDNA < 3.59 ng/μL, n = 65). Kaplan–Meier survival analysis demonstrated a significant difference in median PFS between the 2 groups. The low ctDNA group had a median PFS of 8.6 months, which was significantly longer than the 5.5 months observed in the high ctDNA group (χ² = 15.869, P < .001).

4. Discussion
NSCLC remains a leading cause of cancer-related mortality worldwide, despite recent advances in systemic therapies. ICIs, targeting pathways such as PD-1 and PD-L1, have revolutionized treatment, offering durable responses in a subset of patients. However, the heterogeneous nature of NSCLC leads to variable outcomes, with many patients experiencing poor response or progression. This variability underscores the urgent need to identify reliable prognostic factors and predictive biomarkers to optimize patient selection and tailor therapeutic approaches. ctDNA, a component of cell-free DNA released into the bloodstream by tumor cells, has emerged as a promising noninvasive biomarker.[11,12] This study explored the prognostic factors influencing poor outcomes and evaluated the predictive value of ctDNA in immune therapy for advanced NSCLC. The findings reveal that ctDNA serves as a significant biomarker for predicting treatment response and PFS, providing critical insights into patient stratification and personalized management.
The strong predictive value of ctDNA levels for objective response to immune therapy underscores its clinical utility. ROC curve analysis demonstrated an AUC of 0.821, with optimal sensitivity (80.9%) and specificity (79.8%) achieved at a cutoff value of 3.59 ng/μL. This suggests that ctDNA not only reflects tumor burden but also captures the underlying tumor biology that determines immune sensitivity. Tumors with lower ctDNA levels may exhibit less genetic heterogeneity and immune evasion, leading to better responses to ICIs. The observed reduction in ctDNA levels posttreatment in patients achieving complete or partial remission (PR) further supports its role as a dynamic biomarker. The decline in ctDNA levels likely represents effective tumor control and reduced cellular turnover, consistent with the mechanisms of action of ICIs, which reinvigorate T cell-mediated anti-tumor responses. Conversely, persistently elevated ctDNA levels in patients with stable or progressive disease (PD) may indicate ongoing tumor proliferation or resistance mechanisms, including alterations in immune checkpoint pathways or mutations in genes like JAK1/2.[13,14]
The negative correlation between ctDNA levels and PFS (r = −0.784) highlights its prognostic relevance. Elevated ctDNA levels were associated with shorter PFS, suggesting that ctDNA could serve as a surrogate marker for aggressive disease phenotypes. Tumors shedding higher ctDNA levels might have enhanced metastatic potential or greater tumor volume, both of which are linked to poor outcomes.[15] Additionally, ctDNA may capture mutations associated with immune resistance, such as loss of antigen presentation or activation of alternative immune suppression pathways. Kaplan–Meier survival analysis confirmed these findings, revealing a significantly longer median PFS of 8.6 months in the low ctDNA group compared to 5.5 months in the high ctDNA group. This significant difference underscores the potential of ctDNA to distinguish patients likely to benefit from ICIs versus those who may require alternative therapeutic strategies.[16,17]
Multivariate Cox regression analysis identified ctDNA levels and treatment modality as independent prognostic factors for PFS. Combination therapy involving ICIs and chemotherapy was associated with improved outcomes compared to immunotherapy alone. This highlights the synergistic effects of combining systemic therapies to overcome resistance and enhance anti-tumor responses. Chemotherapy may prime the tumor microenvironment by increasing antigen presentation or reducing immunosuppressive cells, thereby enhancing the efficacy of ICIs. The identification of ctDNA as an independent predictor provides a noninvasive tool for risk stratification. Routine monitoring of ctDNA levels could enable early detection of treatment failure or disease progression, prompting timely intervention. Moreover, ctDNA profiling offers opportunities to identify actionable mutations, enabling precision medicine approaches to address resistance mechanisms.[18,19]
Several mechanisms may explain the observed dynamics of ctDNA in immune therapy. First, ctDNA levels reflect the overall tumor burden and apoptotic activity, both of which decrease with effective therapy. Second, ctDNA captures tumor heterogeneity, providing a comprehensive overview of the genetic landscape, including mutations driving immune escape. Lastly, ctDNA may reflect the tumor microenvironment’s immune status, such as infiltration by cytotoxic T cells or activation of immune checkpoints.[20,21] Despite the promising results, this study has limitations. The retrospective design and single-institution setting may limit the generalizability of the findings. Additionally, the analysis focused on ctDNA levels without examining specific genetic alterations or resistance mechanisms. Future prospective studies with larger cohorts are needed to validate these findings and explore the integration of ctDNA monitoring with other biomarkers, such as tumor mutational burden or PD-L1 expression.

5. Conclusion
Our study demonstrates that ctDNA levels are significant independent prognostic factors in advanced NSCLC patients receiving immunotherapy. Lower posttreatment ctDNA levels are associated with better treatment responses and longer progression-free survival (PFS). Incorporating ctDNA monitoring into clinical practice may enhance patient stratification and optimize therapeutic strategies, leading to improved clinical outcomes.

Acknowledgments
We sincerely thank everyone who participated in this research.

Author contributions
Conceptualization: Wei-Qing Zhang, Jin-Long Huang, Zhen-Dong Xu.
Data curation: Wei-Qing Zhang, Jin-Long Huang, Dong-Liang Lin, Zhen-Dong Xu.
Formal analysis: Wei-Qing Zhang, Jin-Long Huang, Dong-Liang Lin, Zhen-Dong Xu.
Funding acquisition: Zhen-Dong Xu.
Investigation: Dong-Liang Lin.
Methodology: Dong-Liang Lin.
Supervision: Wei-Qing Zhang, Jin-Long Huang, Zhen-Dong Xu.
Validation: Wei-Qing Zhang, Jin-Long Huang, Zhen-Dong Xu.
Visualization: Wei-Qing Zhang, Jin-Long Huang, Zhen-Dong Xu.
Writing – original draft: Wei-Qing Zhang, Zhen-Dong Xu.
Writing – review & editing: Wei-Qing Zhang, Zhen-Dong Xu.