Abstract
This study aims to explore the nomogram risk prediction model for chemotherapy prognosis of patients with non-small cell lung cancer (NSCLC) based on clinical features and pathological parameters, and its clinical application value. Patients with NSCLC were randomly divided into a training set (n = 168) and a validation set (n = 72) according to the ratio of 7:3. In the training set, multivariate logistic regression was used to analyze the risk factors of chemotherapy prognosis and build a nomogram prediction model. Receiver operating characteristic curve and calibration curve were drawn to evaluate the prediction efficiency of the nomogram model, which was verified in the validation set. Among the 168 patients in the training set, 68 (40.48%) had a poor prognosis, while 30 (41.67%) of the 72 patients in the validation set had a poor prognosis. There was no significant difference in the incidence, clinical features, and pathological parameters between the training set and the validation set (P > .05). The results of univariate analysis showed that there were significant differences in age, hypertension, smoking history, cough, hemoptysis, dyspnea, tumor diameter, histological subtypes, contact area between tumor and blood vessels, and short diameter of lymph nodes between patients with poor prognosis and patients with good prognosis in the training set (P < .05). Multivariate logistic regression analysis showed that age, cough, hemoptysis, dyspnea, tumor diameter, histological subtypes, and contact area between tumor and blood vessels were independent risk factors for poor prognosis (P < .05). Further construct the nomogram prediction model; the nomogram model has good calibration and fitting degree between prediction and reality in the training set and validation set (C-index is 0.923 and 0.947). The results of the Hosmer–Lemeshow test are χ2 = 10.692, P = .219 and χ2 = 6.350, P = .608, respectively; the area under the curve of the nomogram model in predicting poor prognosis is 0.918 (95% confidence interval: 0.870–0.966) and 0.946 (95% confidence interval: 0.887–1.000). The model shows good calibration and fitting in both the training set and validation set, and can accurately predict the prognosis of patients with NSCLC after chemotherapy, which can provide valuable reference for clinicians to formulate individualized treatment plans and evaluate the prognosis of patients.
Keywords: chemotherapy prognosis, clinical features, nomogram model, non-small cell lung cancer, pathological parameters
1. Introduction
As a malignant tumor with high morbidity and mortality worldwide, lung cancer has brought a heavy burden to human health.[1,2] Among them, non-small cell lung cancer (NSCLC) accounts for the vast majority of lung cancer, about 80% to 85%.[3–5] Chemotherapy plays an important role in the comprehensive treatment of NSCLC, which is of key significance for prolonging the survival time and improving the quality of life.[6] However, there are significant differences in the response and prognosis of different patients to chemotherapy, which makes it urgent to accurately predict the prognosis of patients with NSCLC after chemotherapy in clinical practice.[7] Clinical features and pathological parameters, which can be directly obtained or determined by routine examination, are closely related to the chemotherapy prognosis of patients with NSCLC.[8,9] For example, the clinical features such as the patient’s age, sex, and smoking history, as well as the pathological parameters such as the histological type, degree of differentiation, and lymph node metastasis of the tumor, may all affect the curative effect of chemotherapy and the survival outcome of the patient. By comprehensively analyzing these clinical features and pathological parameters, we can build a model that can accurately predict the prognosis of patients with NSCLC, which will help clinicians to formulate personalized treatment plans, more accurately evaluate the patient’s disease development, and take intervention measures in advance. The nomogram model, as a visual prediction tool, integrates multiple prediction factors and turns them into an intuitive graphic display, thus providing a quantitative basis for clinical decision-making.[10] In the prognosis prediction of various tumors, the nomogram model has shown its unique advantages, which can comprehensively consider the interaction of multiple factors, avoid the limitations of single-factor prediction, and improve the accuracy and reliability of prediction. Therefore, the purpose of this study is to construct and verify the nomogram prediction model of chemotherapy prognosis of patients with NSCLC based on clinical characteristics and pathological parameters, so as to provide strong support and guidance for clinical treatment of NSCLC.
2. Data and methods
2.1. Research objects
Two hundred forty patients with NSCLC in our hospital from 2022 to 2024 were selected as the research objects. To develop and validate the prognostic model, patients were randomly divided into a training set (n = 168, 70%) for model construction and a validation set (n = 72, 30%) for internal validation. This split ratio is commonly used in similar studies[1,2] to ensure a sufficient sample size for model development while retaining an adequate portion for performance assessment. General data were collected simultaneously. The study was approved by the Ethics Committee of The Second Affiliated Hospital of Air Force Medical University (No. 2022023312), and informed consent was obtained from all patients.
2.2. Inclusion and exclusion criteria
2.2.1. Inclusion criteria
Patients diagnosed with NSCLC, receiving chemotherapy treatment, and having complete clinical and pathological data were included.
2.2.2. Exclusion criteria
Patients with other serious diseases, incomplete follow-up information, or who were not diagnosed with NSCLC for the 1st time were excluded.
2.3. Imaging examination methods
2.3.1. Chest computed tomography scan
2.3.1.1. Detection of tumor size and location
The patient was supine and scanned after breathing calmly. The scanning range was from the apex of the lung to the costal diaphragm angle, and the slice thickness was generally set at 5 to 10 mm. For suspicious parts, thin-layer scanning (slice thickness of 1–3 mm) was performed to determine the tumor boundary and measure the size more accurately. On the axial image, we measured the long diameter and short diameter of the largest slice of the tumor, calculated the tumor volume by multiplying them or directly recorded the maximum diameter as the tumor size index. At the same time, according to the anatomical structure of the lung lobes and segments, the tumor location was determined, and it was classified as central type (near hilum, main bronchus, etc) or a peripheral type (located at the periphery of the lung).
2.3.1.2. Lymph node assessment
On the chest computed tomography (CT) image, lymph nodes in the mediastinum, hilum, and other areas were carefully observed. The diameter of the short axis of the lymph nodes was measured. When the diameter of the short axis was >1 cm, it was marked as suspicious lymph node metastasis. For suspicious lymph nodes, further enhanced scanning was performed, and a contrast agent (such as iohexol; the dose was generally 1.5–2 mL/kg according to body weight) was injected intravenously, and the injection speed was about 2 to 3 mL/second, and scanning was performed after a delay of 30 to 60 seconds. We observed the enhancement characteristics of lymph nodes, including enhancement degree (compared with surrounding blood vessels) and enhancement uniformity. For example, if the lymph nodes were annularly enhanced with low density in the center, it is suggested that there may be necrosis, and metastasis is more likely.
2.3.1.3. Vascular invasion judgment
During CT angiography examination of the chest, the contrast agent was injected and scanned at specific phases (such as pulmonary artery phase and aortic phase) to clearly show the relationship between the tumor and surrounding blood vessels. Through post-processing techniques such as multi-planar reconstruction and volume rendering, the contact area between tumor and blood vessel, morphological changes of blood vessel wall (such as smoothness, local thickening), stenosis or occlusion of lumen, and filling defect in lumen were observed directly. If the tumor surrounds the blood vessel more than 180° and the blood vessel wall is irregularly thickened, it is highly suggestive of blood vessel invasion. A preliminary judgment of histological subtype tendency was made based on tumor morphological characteristics, such as adenocarcinoma often presenting as ground-glass density nodules, and squamous cell carcinoma often accompanied by central necrosis.
2.3.2. Positron emission tomography-computed tomography (PET-CT)
2.3.2.1. Detection and staging of tumor metabolic activity
Patients need to fast for 4 to 6 hours before examination, and control the blood sugar level in the normal range (fasting blood sugar is generally lower than 7.0 mmol/L). Intravenous injection of radioactive tracer 18F-fluorodeoxyglucose at a dose of about 3.7 to 7.4 MBq/kg according to body weight. After the injection, the patient rested quietly for 60 to 90 minutes, and then the whole body was scanned by PET-CT. When scanning, the anatomical images were obtained by CT scanning, and then the metabolic images were obtained by positron emission tomography scanning. Finally, the 2 images were fused by image fusion technology. On the fused image, the radioactive uptake of the lung tumor and other tissues and organs in the whole body was observed, and the metabolic activity of the tumor was quantified by the standard uptake value (SUV). Generally, an SUV >2.5 indicates that the tumor is highly metabolized and may be a malignant lesion. Through whole-body scanning, distant metastases are found, such as bone metastases showing radioactive concentration in the corresponding parts, and brain metastases can cause abnormal metabolism in specific areas of the brain, thus determining the tumor stage.
2.3.2.2. Comprehensive evaluation of tumor status
PET-CT examination was performed before and after treatment, and the metabolic activity of the tumor site was compared. After chemotherapy, if the SUV value of the tumor site is significantly lower than that before treatment (such as a reduction of more than 30%), it indicates that chemotherapy is effective; if there is no obvious change or increase in SUV value, it indicates that the tumor is resistant to chemotherapy drugs and the treatment plan needs to be adjusted.
2.4. Pathological examination method
2.4.1. Tissue biopsy
For peripheral lung tumors, a percutaneous lung biopsy can be performed under CT guidance. The patient is placed in an appropriate position (e.g., prone or supine), and after local anesthesia, a biopsy needle (such as a spring-loaded needle or coaxial biopsy needle) is inserted through the skin into the tumor tissue to obtain 2 to 3 tissue specimens, typically 1 to 2 cm in length. For central tumors, a bronchoscopic biopsy may be conducted. The bronchoscope is inserted through the nasal or oral cavity into the trachea and bronchi until reaching the tumor site, where biopsy forceps are used to collect tissue specimens. The obtained specimens are immediately fixed in formalin, followed by paraffin embedding, sectioning (approximately 4–5 μm thick), and hematoxylin and eosin staining. A pathologist then examines the morphology, structure, and arrangement of tumor cells under a microscope to determine the histological subtype (adenocarcinoma/squamous cell carcinoma/large cell carcinoma) and histological type. Adenocarcinoma: tumor cells exhibit glandular structures or papillary growth, with abundant cytoplasm containing mucin vacuoles. Squamous cell carcinoma: intercellular bridges and keratin pearl formation are observed, with eosinophilic cytoplasm. Large cell carcinoma: tumor cells are large in size, with prominent nucleoli and abundant cytoplasm, lacking definitive glandular or squamous differentiation. The degree of cellular differentiation is assessed based on nuclear atypia and cellular maturity: well-differentiated tumors: tumor cells appear more uniform, with few mitotic figures and high resemblance to normal tissue cells; and poorly differentiated tumors: tumor cells exhibit large, irregular nuclei, a high nuclear-to-cytoplasmic ratio, and frequent mitotic figures.
2.4.2. Lymph node biopsy
For patients with suspected lymph node metastasis by CT and other imaging examinations, lymph node biopsy can determine whether there are tumor cells in the lymph nodes.
2.5. Evaluation method of chemotherapy prognosis
The Lung Cancer Symptom Scale was used for evaluation 3 months after the end of chemotherapy. If the total score exceeded 25, the prognosis was judged to be poor.
2.6. Statistical methods
Using SPSS 26.0 (IBM Corporation, Armonk), the data were counted by chi-square test, and the t value was substituted into the measurement data, with statistical significance (P < .05).
3. Results
3.1. Comparison of chemotherapy prognosis, clinical features and pathological parameters between the training set and validation set
Sixty-eight cases (40.48%) of the 168 patients in the training set had a poor prognosis, and 30 cases (41.67%) of the 72 patients in the validation set had a poor prognosis. To ensure the robustness of the random split, we compared the baseline characteristics between the training and validation sets. As shown in Table 1, there were no statistically significant differences in the incidence of poor prognosis, clinical features, or pathological parameters between the 2 sets (all P > .05), indicating a balanced distribution and the absence of significant selection bias.
Table 1.
Comparison of clinical features and pathological parameters between the training set and verification set.
| Clinical features | Training set (n = 168) | Verification set (n = 72) | Statistical values | P value |
|---|---|---|---|---|
| Age (yr) | 57.18 ± 9.03 | 56.24 ± 8.56 | 0.757 | .450 |
| Gender | ||||
| Man | 94 | 42 | 0.116 | .733 |
| Woman | 74 | 30 | ||
| BMI (kg/m2) | 22.89 ± 2.08 | 22.46 ± 2.21 | 1.487 | .138 |
| Diabetes | ||||
| Have | 42 | 25 | 2.367 | .124 |
| Without | 126 | 47 | ||
| Hypertension | ||||
| have | 46 | 24 | 0.864 | .353 |
| Without | 122 | 48 | ||
| Smoking history | ||||
| Have | 71 | 36 | 1.221 | .269 |
| Without | 97 | 36 | ||
| Drinking history | ||||
| Have | 82 | 33 | 0.179 | .672 |
| Without | 86 | 39 | ||
| Cough (times) | 9.52 ± 5.71 | 8.65 ± 5.04 | 1.081 | .281 |
| Hemoptysis (times) | 4.91 ± 3.03 | 4.73 ± 3.54 | 0.374 | .709 |
| Dyspnea (times) | 3.94 ± 2.84 | 3.78 ± 2.46 | 0.410 | .682 |
| Tumor diameter (cm) | 3.85 ± 2.84 | 3.68 ± 2.67 | 0.498 | .620 |
| Tumor location | ||||
| Central type | 64 | 31 | 0.519 | .471 |
| Circumferential type | 104 | 41 | ||
| Degree of cell differentiation | ||||
| Highly differentiated | 31 | 12 | 0.165 | .921 |
| Moderate differentiation | 66 | 30 | ||
| Poorly differentiated | 71 | 30 | ||
| Histological subtypes | ||||
| Adenocarcinoma | 95 | 48 | 0.111 | .945 |
| Squamous cell carcinoma | 57 | 41 | ||
| Large cell carcinoma | 16 | 11 | ||
| Contact area between tumor and blood vessel (cm2) | 3.25 ± 0.78 | 3.22 ± 0.46 | 0.334 | .739 |
| Short diameter of lymph nodes (cm) | 1.36 ± 0.27 | 1.38 ± 0.23 | 0.461 | .645 |
| IL-6 (ng/L) | 10.66 ± 2.54 | 10.35 ± 2.46 | 0.879 | .380 |
| CRP (mg/L) | 8.55 ± 3.16 | 8.84 ± 3.08 | 0.670 | .503 |
BMI = body mass index, CRP = C-reactive protein, IL-6 = interleukin-6.
3.2. Comparison of clinical features and pathological parameters between the training set with good prognosis and the group with poor prognosis
The results of univariate analysis showed that there were significant differences between patients with poor prognosis and patients with good prognosis in age, hypertension, smoking history, cough, hemoptysis, dyspnea, tumor diameter, contact area between tumor and blood vessels and short diameter of lymph nodes (P < .05, Table 2).
Table 2.
Comparison of clinical features and pathological parameters between the poor prognosis and good prognosis groups.
| Clinical features | Bad group (n = 68) | Good group (n = 100) | Statistical values | P value |
|---|---|---|---|---|
| Age (yr) | 61.24 ± 8.64 | 59.43 ± 8.26 | 1.368 | .173 |
| Gender | ||||
| Man | 38 | 56 | 0.001 | .988 |
| Woman | 30 | 44 | ||
| BMI (kg/m2) | 23.24 ± 2.03 | 22.67 ± 2.11 | 1.779 | .077 |
| Diabetes | ||||
| Have | 12 | 30 | 3.294 | .070 |
| Without | 56 | 70 | ||
| Hypertension | ||||
| Have | 25 | 21 | 5.059 | .024 |
| Without | 43 | 79 | ||
| Smoking history | ||||
| Have | 35 | 36 | 3.970 | .046 |
| Without | 33 | 64 | ||
| Drinking history | ||||
| Have | 37 | 45 | 1.439 | .231 |
| Without | 31 | 55 | ||
| Cough (times) | 11.24 ± 5.64 | 8.35 ± 5.49 | 3.304 | .001 |
| Hemoptysis (times) | 6.24 ± 2.06 | 4.01 ± 3.27 | 5.405 | .001 |
| Dyspnea (times) | 5.62 ± 3.57 | 2.81 ± 1.34 | 6.210 | .001 |
| Tumor diameter (cm) | 4.29 ± 1.53 | 3.56 ± 1.26 | 3.368 | .001 |
| Tumor location | ||||
| Central type | 29 | 35 | 1.004 | .316 |
| Circumferential type | 39 | 65 | ||
| Degree of cell differentiation | ||||
| Highly differentiated | 11 | 20 | 0.625 | .722 |
| Moderate differentiation | 26 | 40 | ||
| Poorly differentiated | 31 | 40 | ||
| Histological subtypes | ||||
| Adenocarcinoma | 47 | 48 | 7.398 | .024 |
| Squamous cell carcinoma | 16 | 41 | ||
| Large cell carcinoma | 5 | 11 | ||
| Contact area between tumor and blood vessel (cm2) | 2.84 ± 0.36 | 3.54 ± 0.85 | 7.272 | .001 |
| Short diameter of lymph nodes (cm) | 1.28 ± 0.12 | 1.41 ± 0.32 | 3.956 | .001 |
| IL-6 (ng/L) | 10.87 ± 2.03 | 10.52 ± 2.83 | 0.934 | .352 |
| CRP (mg/L) | 9.03 ± 3.85 | 8.23 ± 2.55 | 1.505 | .135 |
BMI = body mass index, CRP = C-reactive protein, IL-6 = interleukin-6.
3.3. Analysis of risk factors of chemotherapy prognosis in patients with NSCLC
The occurrence of poor prognosis was taken as the dependent variable (good = 0, bad = 1), and the factors with P < .05 in univariate analysis were taken as the covariate. Further multivariate logistic regression analysis showed that age, cough, hemoptysis, dyspnea, tumor diameter, and contact area between tumor and blood vessels were independent risk factors for poor prognosis (P < .05, Table 3).
Table 3.
Logistic regression analysis of risk factors of chemotherapy prognosis in patients with non-small cell lung cancer.
| Factor | B | SE | Wald | P | OR | 95% CI |
|---|---|---|---|---|---|---|
| Haemoptysis | 0.400 | 0.094 | 18.002 | .001 | 1.492 | 1.240–1.795 |
| Dyspnea | 0.584 | 0.148 | 15.525 | .001 | 1.794 | 1.341–2.399 |
| Tumor diameter | 0.645 | 0.180 | 12.896 | .001 | 1.905 | 1.340–2.709 |
| Histological subtypes | −1.913 | 0.400 | 22.902 | .001 | 0.148 | 0.067–0.323 |
| Contact area between tumor and blood vessel | −0.991 | 0.389 | 6.481 | .011 | 0.371 | 0.173–0.796 |
CI = confidence interval, OR = odds ratio, SE = standard error.
3.4. Construction of a nomogram prediction model for chemotherapy prognosis of patients with NSCLC
Based on the independent risk factors identified by multivariate logistic regression analysis, a nomogram prediction model for chemotherapy prognosis of patients with NSCLC was constructed. Each independent risk factor in the model was assigned, and the total score for predicting chemotherapy prognosis of patients with NSCLC was calculated, which was reflected by the probability of predicting poor prognosis. The higher the total score, the higher the accuracy of predicting poor prognosis of patients with NSCLC (Fig. 1).
Figure 1.
Nomogram of chemotherapy prognosis prediction model for patients with non-small cell lung cancer. X1–X6 are hemoptysis, dyspnea, tumor diameter, histological subtypes and contact area between tumor and blood vessels.
3.5. Evaluation and validation of the prognosis prediction model for patients with NSCLC after chemotherapy
In the training set and validation set, the C-index of the nomogram model was 0.921 and 0.960, respectively. The average absolute error of the coincidence degree between the predicted value and the real value is 0.116 and 0.086, respectively. The results of the Hosmer–Lemeshow test are χ2 = 4.427, P = .817 and χ2 = 12.229, P = .141, respectively. The receiver operating characteristic curve shows that in the training set and validation set, the area under the curve (AUC) predicted by nomogram model with poor prognosis is 0.919 (95% confidence interval: 0.873–0.965) and 0.956 (95% confidence interval: 0.908–1.000), the sensitivity and specificity were 0.898, 0.797, 0.947 and 0.867, respectively (Figs. 2 and 3).
Figure 2.
Calibration curve of chemotherapy prognosis prediction model for patients with non-small cell lung cancer. (A) The training set, and (B) the validation set.
Figure 3.
Receiver operating characteristic (ROC) curves. (A) The training set; (B) the validation set. AUC = area under the curve, CI = confidence interval, FPR = false positive rate.
3.6. Decision curve analysis of nomogram prediction model for chemotherapy prognosis of patients with NSCLC
The decision curve shows that when the threshold probability is between 0.05 and 0.95, the nomogram model constructed in this study has more clinical benefits in predicting the prognosis of chemotherapy (Fig. 4).
Figure 4.
Decision curves. (A) The training set; (B) the validation set.
4. Discussion
In the field of treatment of NSCLC, the uncertainty of chemotherapy effect has been puzzling clinicians and patients. How to accurately predict the prognosis of patients before chemotherapy has become a key link to improving treatment strategies and improving the quality of life of patients. In this study, we successfully constructed and verified the nomogram prediction model for chemotherapy prognosis of patients with NSCLC based on clinical features and pathological parameters. The model integrates key factors such as age, cough, hemoptysis, dyspnea, tumor diameter, histological subtypes and contact area between tumor and blood vessel, and shows good prediction efficiency in the training set and validation set. The C-index is 0.923 and 0.947, and the AUC value is 0.918 and 0.946, respectively. The calibration curve shows that the predicted value is in high agreement with the real value, and the average absolute error is small. The construction of this model provides a more accurate and intuitive tool for predicting the prognosis of patients with NSCLC after chemotherapy.
In terms of clinical features, age was identified as an independent risk factor. With the increase in age, patients’ physical function declines, their tolerance to chemotherapy may decrease, and there are more potential complications, thus affecting the effectiveness and prognosis of chemotherapy. For example, the organ function of elderly patients declines, the hematopoietic function of bone marrow is weakened, and adverse reactions such as bone marrow suppression are more likely to occur during chemotherapy, which further affects the follow-up treatment process and quality of life, which is consistent with the conclusion that age affects the prognosis of tumor patients in previous studies.[11,12] Symptoms such as cough, hemoptysis and dyspnea are also related to poor prognosis, which may reflect the progress of the tumor, the invasion of the respiratory tract and the change of general state. Severe cough and hemoptysis often indicate that the tumor locally invades blood vessels or bronchi, while dyspnea may mean that lung function is seriously damaged, which indicates that patients are facing greater challenges in the process of chemotherapy, and the prognosis is poor.[13–15] Among pathological parameters, histological subtypes of NSCLC (adenocarcinoma, squamous cell carcinoma, large cell carcinoma) were identified as independent risk factors in multivariate analysis. This finding aligns with clinical observations that different subtypes exhibit distinct responses to chemotherapy and prognosis. For example, adenocarcinoma patients in our cohort showed a higher prevalence of poor prognosis (47/95, 49.5%) compared to squamous cell carcinoma (16/57, 28.1%), which may be attributed to adenocarcinoma’s higher tendency for vascular invasion and distant metastasis. In contrast, squamous cell carcinoma was associated with stronger correlations with tumor-vessel contact area, potentially reflecting its reliance on angiogenesis for progression. These subtype-specific differences highlight the importance of integrating histological typing into prognosis models, as generic models may overlook subtype-specific risk profiles. Tumor diameter and the contact area between tumor and blood vessels have an important influence on prognosis. The larger tumor diameter indicates that the tumor load is larger, and there may be more tumor cell heterogeneity, which makes it more difficult for chemotherapy drugs to completely remove tumor cells.[16] The large contact area between tumor and blood vessels suggests that tumor cells are more likely to enter the blood circulation, and the risk of distant metastasis increases, which also makes the risk of poor prognosis of patients significantly increase.[17] This is consistent with the theory of tumor size, vascular invasion and prognosis in oncology, and further emphasizes the key position of these pathological parameters in evaluating the prognosis of patients with NSCLC after chemotherapy. Moreover, the predictive strength of our nomogram is underpinned by the integration of quantitatively assessed imaging parameters. Critically, variables such as tumor diameter and contact area between tumor and blood vessels were not merely clinical estimates but were objectively derived from standardized preoperative imaging (CT and CT angiography). The inclusion of these metrics provides a more direct anatomical and pathophysiological basis for prognosis prediction, reflecting tangible tumor characteristics like burden and local invasiveness. This approach enhances the model’s credibility by moving beyond basic clinical descriptors and incorporating crucial imaging-based evidence that is routinely available in clinical practice. Compared with other studies, the nomogram model constructed in this study has its own characteristics and advantages. Some previous studies may only focus on the influence of a single factor or a few factors on the prognosis of chemotherapy for NSCLC, but this model integrates multiple clinical features and pathological parameters, which reflects the overall state of patients more comprehensively and avoids the one-sidedness of single-factor prediction. For example, some studies only focus on the genotyping of tumors or only consider the clinical stages of patients to predict prognosis, while ignoring important factors such as symptoms and the microscopic relationship between tumors and blood vessels. In addition, this study fully verified the model internally by strictly dividing the training set and validation set, which ensured the reliability and generalization ability of the model and made it more valuable in clinical applications.
However, this study also has some limitations. First of all, the sample size of the study is relatively limited, and although it has certain effectiveness in statistical analysis, it may not fully cover all the clinical features and pathological types of patients with NSCLC, which may have a certain impact on the universality of the model. In the future, multi-center and large sample research is needed to further verify and improve the model. Second, due to objective constraints, this study did not perform external validation with original data from other centers, nor did it conduct cellular experiments or clinical intervention trials. As a result, the specific biological pathways by which risk factors such as age, cough, and hemoptysis influence chemotherapy prognosis cannot be elucidated at the mechanistic level. For example, we were unable to verify the sensitivity of different pathological tumor cells to chemotherapy drugs through cellular drug sensitivity assays, nor could we validate the model’s predictive efficacy across different medical settings using multi-center clinical data. This limits the mechanistic interpretation and external applicability of the model, and relevant conclusions need to be further confirmed through basic experiments and multi-center clinical data in subsequent studies. Additionally, this study only included some clinical features and pathological parameters, and did not consider some emerging biomarkers, such as tumor mutation load.[18,19] With the development of precision medicine, these biomarkers play an increasingly prominent role in predicting the prognosis of patients with NSCLC, and their inclusion in the model may further improve the accuracy of prediction.[20,21] Regarding innovation, this study offers practical value by constructing a nomogram based on readily available clinical and pathological parameters, enabling quick clinical application without additional molecular testing. The model’s integrated multi-dimensional approach (AUC: 0.918–0.946) outperforms single-factor models, providing a more comprehensive prognostic tool. However, the study’s innovation is limited by the absence of cellular/molecular experiments, which prevents exploration of the biological basis for risk factors (e.g., how tumor-vessel contact area influences chemoresistance). Future directions to enhance innovation include: incorporating multi-center data and molecular markers (e.g., circulating tumor DNA, tumor mutational burden) to develop “clinical-pathological-molecular” hybrid models; using cell lines (e.g., A549) for drug sensitivity assays to validate risk factor mechanisms; and applying machine learning algorithms to optimize model performance and analytical depth. Finally, this study is retrospective. Although the bias is minimized in the process of data collection and analysis, the potential problems, such as recall bias and selection bias, cannot be completely ruled out. A prospective study design will help to evaluate the predictive efficacy and clinical practicability of the model more accurately.
To sum up, the nomogram prediction model for chemotherapy prognosis of patients with NSCLC constructed in this study has important clinical significance and certain innovation, but it still needs to be further optimized and verified in the research with larger samples, multiple centers and more indicators, to be better applied to clinical practice, provide more reliable basis for individualized treatment and prognosis evaluation of patients with NSCLC, and promote the continuous improvement of clinical diagnosis and treatment level of NSCLC.
Author contributions
Conceptualization: Sanhu Yang, Xiangbing Xin.
Methodology: Yongshi Liu, Yan Li.
Data curation: Xiang Ji, Yuhui Yun.
Supervision: Guang Yang, Xiangbing Xin.
Writing – original draft: Sanhu Yang.
Writing – review & editing: Guang Yang, Xiangbing Xin.
Abbreviations:
- AUC
- area under the curve
- CT
- computed tomography
- NSCLC
- non-small cell lung cancer
- PET-CT
- positron emission tomography-computed tomography
- SUV
- standard uptake value
The authors have no funding and conflicts of interest to disclose.
The study was approved by the Ethics Committee of The Second Affiliated Hospital of Air Force Medical University (No. 2022023312), and informed consent was obtained from all patients. This study was conducted in accordance with the Declaration of Helsinki.
The datasets generated during and/or analyzed during the current study are not publicly available, but are available from the corresponding author on reasonable request.
How to cite this article: Yang S, Liu Y, Li Y, Ji X, Yun Y, Yang G, Xin X. Construction and validation of a nomogram for predicting chemotherapy prognosis in non-small cell lung cancer patients based on clinical features and pathological parameters: An observational study. Medicine 2026;105:8(e47382).
Contributor Information
Sanhu Yang, Email: paiwe5993@163.com.
Yongshi Liu, Email: yon19719@163.com.
Yan Li, Email: zhu57133@163.com.
Xiang Ji, Email: bile151@163.com.
Yuhui Yun, Email: toul775@163.com.
Guang Yang, Email: paiwe5993@163.com.
References
- [1].Leiter A, Veluswamy RR, Wisnivesky JP. The global burden of lung cancer: current status and future trends. Nat Rev Clin Oncol. 2023;20:624–39. [DOI] [PubMed] [Google Scholar]
- [2].Li Y, Yan B, He S. Advances and challenges in the treatment of lung cancer. Biomed Pharmacother. 2023;169:115891. [DOI] [PubMed] [Google Scholar]
- [3].Qi C, Sun SW, Xiong XZ. From COPD to lung cancer: mechanisms linking, diagnosis, treatment, and prognosis. Int J Chron Obstruct Pulmon Dis. 2022;17:2603–21. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [4].Chen P, Liu Y, Wen Y, Zhou C. Non-small cell lung cancer in China. Cancer Commun (Lond). 2022;42:937–70. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [5].Wang K, Hu X, Chen Y, et al. Study on the application of percutaneous closed pleural brushing combined with cell block technique in the diagnosis of malignant pleural effusion. Clin Respir J. 2024;18:e13705. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [6].Khasraw M, Yalamanchili P, Santhanagopal A, et al. Clinical management of patients with non-small cell lung cancer, brain metastases, and actionable genomic alterations: a systematic literature review. Adv Ther. 2024;41:1815–42. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [7].Mendes AS, Romão R, Febra J, Azevedo SX, Fidalgo P, Araújo A. Chemotherapy: a partnership with immunotherapy in non-small cell lung cancer. Thorac Cancer. 2023;14:437–41. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [8].Felip E, Altorki N, Zhou C, et al. Overall survival with adjuvant atezolizumab after chemotherapy in resected stage II-IIIA non-small-cell lung cancer (IMpower010): a randomised, multicentre, open-label, phase III trial. Ann Oncol. 2023;34:907–19. [DOI] [PubMed] [Google Scholar]
- [9].Memon D, Schoenfeld AJ, Ye D, et al. Clinical and molecular features of acquired resistance to immunotherapy in non-small cell lung cancer. Cancer Cell. 2024;42:209–24.e9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [10].Bao X, Bian D, Yang X, et al. Multiparametric MRI for evaluation of pathological response to the neoadjuvant chemo-immunotherapy in resectable non-small-cell lung cancer. Eur Radiol. 2023;33:9182–93. [DOI] [PubMed] [Google Scholar]
- [11].Xie Y, Lei C, Ma Y, et al. Prognostic nomograms for breast cancer with lung metastasis: a SEER-based population study. BMC Womens Health. 2024;24:16. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [12].Horie T, Fukasawa K, Yamada T, et al. Erk5 in bone marrow mesenchymal stem cells regulates bone homeostasis by preventing osteogenesis in adulthood. Stem Cells. 2022;40:411–22. [DOI] [PubMed] [Google Scholar]
- [13].Bates JHT, Hamlington KL, Garrison G, Kinsey CM. Prediction of lung cancer risk based on age and smoking history. Comput Methods Programs Biomed. 2022;216:106660. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [14].Ramcharran H, Wallen J. Robotic-assisted thoracoscopic resection of anterior mediastinal cystic teratoma: a case report and literature review. J Cardiothorac Surg. 2022;17:67. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [15].Shah M, Duong B, Karnath B, Nishi S, Rasmussen P. Carcinoid tumor presenting as hemoptysis and elevated diaphragm. Cureus. 2023;15:e40586. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [16].Sharma VS, Yadav V. Effect of prehabilitation in lung cancer patients undergoing lobectomy: a review. Cureus. 2023;15:e49940. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [17].Suzuki K, Watanabe SI, Wakabayashi M, et al. A single-arm study of sublobar resection for ground-glass opacity dominant peripheral lung cancer. J Thorac Cardiovasc Surg. 2022;163:289–301.e2. [DOI] [PubMed] [Google Scholar]
- [18].Yoo J, Lee JM, Joo I, et al. Post-neoadjuvant treatment pancreatic cancer resectability and outcome prediction using CT, 18F-FDG PET/MRI and CA 19-9. Cancer Imaging. 2023;23:49. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [19].Gabbia D, De Martin S. Tumor mutational burden for predicting prognosis and therapy outcome of hepatocellular carcinoma. Int J Mol Sci. 2023;24:3441. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [20].Kim ES, Velcheti V, Mekhail T, et al. Blood-based tumor mutational burden as a biomarker for atezolizumab in non-small cell lung cancer: the phase 2 B-F1RST trial. Nat Med. 2022;28:939–45. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [21].Megyesfalvi Z, Gay CM, Popper H, et al. Clinical insights into small cell lung cancer: tumor heterogeneity, diagnosis, therapy, and future directions. CA Cancer J Clin. 2023;73:620–52. [DOI] [PubMed] [Google Scholar]