The global cancer statistics for 2022 reveal that lung cancer (1), following other cancers, has become the second most common cancer worldwide and is also the malignancy with the highest mortality rate. Lung cancer accounts for 21% of all cancer-related deaths (2), making it the leading cause of cancer mortality globally. Approximately 340 people die from lung cancer each day, with a 5-year relative survival rate of 25% (3, 4). In China, lung cancer ranks highest in both incidence and mortality among all malignant tumors (5). The incidence and mortality rates of lung cancer continue to rise worldwide. Lung cancer is categorized into small cell lung cancer (SCLC) and non-small cell lung cancer (NSCLC) on the basis of histological characteristics, with NSCLC accounting for 80% to 85% of cases (6, 7).

Precision treatment technologies for lung cancer are advancing rapidly, with immune checkpoint inhibitors (ICIs) emerging as a focal point of research and significantly improving survival outcomes for some lung cancer patients. Particularly for patients with advanced NSCLC who lack driver gene mutations, ICIs used alone or in combination with other therapies have become a first-line standard treatment (8–10). Cancer immunotherapy combats cancer by enhancing the body’s immune system. In recent years, immunotherapy has rapidly developed and is now used alongside surgery, radiation therapy, chemotherapy, targeted therapy, and endocrine therapy to fight cancer. Immunotherapy employs two main strategies: the first is to stimulate the immune system through vaccine therapy to recognize and attack cancer cells (active immunotherapy), and the second is to use checkpoint inhibitors to increase the inhibitory state of the immune system, thereby restoring its ability to eliminate cancer cells (passive immunotherapy) (11). Monoclonal antibodies targeting immune checkpoints, including programmed cell death-1 (PD-1) and its ligand PD-L1, have shown response rates of up to 40% in monotherapy (12).

Given the complex fluid dynamics of tumors and the immunogenic variations within the tumor microenvironment (TME), relying solely on a single biomarker to accurately predict the efficacy of ICIs remains an unstable approach (13, 14). Numerous studies have established a prognostic evaluation framework for NSCLC patients treated with ICIs, integrating multidimensional information from clinical practice, pathological histology, genomics, and radiomics to provide a more comprehensive understanding of treatment prospects (15, 16). Therefore, this study aimed to develop a comprehensive overall survival risk prediction model for patients with NSCLC receiving immunotherapy that incorporates demographic information, clinical characteristics, and laboratory indicators.

This study aimed to develop an OS risk prediction model for NSCLC patients undergoing immunotherapy incorporating demographic information, clinical characteristics, and laboratory indicators. A nomogram was constructed based on age, KPS, BMI, diabetes status, targeted therapy, Hb, WBC, LDH, CRP, PLR, and LMR. The model demonstrated good predictive ability and was developed as an online NSCLC prediction tool.

Previous studies on predicting OS in NSCLC patients are numerous. Mezquita et al. found that the pre-treatment Lung Immune Prognostic Index (LIPI) was closely associated with poor efficacy of ICIs in patients with NSCLC. LIPI consists of two key indicators: the derived neutrophil-to-lymphocyte ratio (dNLR) and LDH levels, where a dNLR greater than 3 and LDH exceeding the upper limit of normal (ULN) are independent predictors of poor prognosis (17, 18). In patients with advanced NSCLC (aNSCLC), continuing immunotherapy after disease progression was associated with prolonged OS and PFS (19). A prognostic risk model for NSCLC based on seven smoking-related genes demonstrated stable predictive ability, with FCGBP having the highest mutation frequency (20). Additionally, studies have shown that in Asian patients with advanced NSCLC receiving immune checkpoint inhibitors, obesity (BMI) is associated with better OS (18). Cheng Lu et al. reported that CellDiv features (computer-extracted tumor cell diversity features) strongly predict 5-year OS in early-stage NSCLC patients and are related to apoptotic signalling and cell differentiation pathways (21). In 2020, the Mazzaschi team prospectively collected baseline peripheral blood from 109 NSCLC patients receiving continuous ICI treatment to construct the immune effect score (IeffS) based on a tumor−host interaction model. IeffS revealed that elevated PD-L1, reduced CD8+PD-1+ cells, and the absence of NK cells were risk factors for reduced ICI efficacy (P<0.01). In combination with LIPI, it significantly impacted PFS (HR=4.61), OS (HR=4.03), and ICI response rates (P<0.001) (22). A phase III trial in stage I-III NSCLC patients reported a median PFS of 20.8 months for the standard chemotherapy group compared with 31.6 months for the nivolumab group, with 24% of patients achieving pathological complete response (23). However, some previous studies have been limited in their choice of predictive factors, focusing on single dimensions (24–26). In contrast, this study enhances predictive performance and comprehensiveness by incorporating a comprehensive set of predictive factors from three domains.

The nomogram model has been widely adopted tumor prognosis risk assessment, demonstrating excellent predictive accuracy (27). In 2021, Zeng et al. conducted a retrospective analysis of immunotherapy plus chemotherapy data from 130 stage IIIA-IVB NSCLC patients and developed a PFS prediction nomogram based on bone metastasis, the dNLR, smoking status, and PD-L1 expression. The low-risk group had a significantly longer median PFS (P<0.001). The model demonstrated good predictive performance, with C-indexes of 0.725 and 0.688 in the training and validation sets, respectively (28). In 2020, the Hopkins team developed and validated a prognostic tool based on clinical trials to identify advanced NSCLC patients who would benefit from atezolizumab treatment. This tool includes factors such as PD-L1 expression levels, dNLR values, CRP concentrations, LDH activity, albumin (ALB) levels, patient performance status, the period since self-diagnosed metastasis, and the number of metastatic sites. The low-risk group benefited the most, and this study was the first to identify CRP as an OS predictor, with CRP reduction associated with prolonged OS (P<0.001, c=0.66) (29). Other studies have constructed nomogram models based on patients’ general characteristics, histological features, pathological and immunohistochemical results, inflammation, and nutritional indicators to effectively predict OS and PFS in NSCLC patients after surgery (30). Mengmeng Song et al. identified the ALI, LCR, and NLR as the top three inflammation/nutrition indicators for predicting the prognosis of lung cancer patients (31). A study on local tumor response and survival outcomes following combined treatment of stereotactic radiosurgery and immunotherapy in NSCLC with brain metastases revealed that a KPS score of < 80 (p = 0.001) and a lung-specific molecular marker graded prognostic assessment (lung-mol GPA) score of < 1.5 (p = 0.02) were predictive of poorer survival outcomes (32). The variables included in this study are broadly consistent with previous studies, and Hb and WBC count have been newly identified as independent prognostic factors for NSCLC patients. This finding provides a new perspective and an essential basis for evaluating disease prognosis.

The biological complexity and heterogeneity of NSCLC underscore the necessity of personalized treatment. Molecular targeted therapies based on driver gene mutations, such as EGFR, ALK, and ROS1, have achieved significant advances in the treatment of advanced NSCLC. However, various resistance mechanisms, such as secondary mutations and bypass activation, limit the long-term efficacy of targeted therapies. The limitations of single-agent therapies suggest that combination treatments may improve outcomes. ICIs, in combination with chemotherapy, targeted therapy, or radiotherapy, have shown promising results in clinical studies (33). Traditionally, radiotherapy has been thought to kill tumor cells mainly by damaging DNA. However, radiotherapy can also control metastatic lesions outside the irradiated field through the “abscopal effect.” This mechanism involves radiotherapy-induced tumor cell death, releasing tumor-associated antigens, which activate a systemic immune response to control tumor proliferation (34). For patients with unresectable stage III NSCLC and wild-type driver genes, combining radiotherapy and immunotherapy in initial treatment is crucial for significantly improving efficacy (35). The PACIFIC trial was a landmark study that demonstrated durvalumab, an ICI, significantly prolonged PFS and OS in patients after concurrent chemoradiotherapy. The real-world effectiveness and safety of durvalumab are consistent with the results of the PACIFIC trial, further supporting its use as the standard treatment for patients with unresectable stage III NSCLC (36).

The limitations of this study primarily lie in the lack of a multicenter research design and external validation. Although the sample size was relatively large, the findings need to be confirmed in larger-scale prospective cohorts. This limitation may introduce selection bias, thereby restricting the generalizability and representativeness of the results. Future research should aim to establish a multicenter collaborative framework to enhance the comprehensiveness and reliability of the data. Moreover, personalized prediction has become possible with advancements in genomics, translational medicine, and precision medicine technologies. The study found that AI models can accurately predict tumor response and PFS of ICIs in advanced NSCLC by assessing PD-L1 tumor proportion score (TPS). In the future, it may be feasible to offer more efficient and convenient personalized risk assessment strategies by integrating advanced gene testing techniques and the assistance of artificial intelligence (37, 38).

In this study, we developed an OS risk prediction model for patients undergoing immunotherapy for NSCLC that incorporates demographic information, clinical characteristics, and laboratory indicators. The model demonstrated good predictive ability and was developed as an online NSCLC prediction tool. This tool facilitates convenient and efficient individualized risk assessment, providing real-time prognostic prediction services for healthcare providers and patients.