Research ArticleOriginal Article
Open Access

Temporal radiomics for non-invasive preoperative prediction of pathologic complete response to neoadjuvant chemoimmunotherapy in non-small cell lung cancer

Sunyi Zheng, Shuo Wang, Ziwei Feng, Jing Liang, Jiaxin Liu, Xiaomeng Yang, Zhanshuo Zhang, Yuechen Cui, Jiping Xie, Shuxuan Fan, Jing Wang, Guoqing Liao, Haiyu Zhou, Zhaoxiang Ye, Jianyu Xiao, Lei Shi, Xiaonan Cui and Dongsheng Yue
Cancer Biology & Medicine January 2026, 20250327; DOI: https://doi.org/10.20892/j.issn.2095-3941.2025.0327

Article Figures & Data

Figures

  • Figure 1
    Figure 1

    Flowchart of the study design. Tumors from non-small cell lung cancer (NSCLC) patients who underwent neoadjuvant chemoimmunotherapy were segmented on pre- and post-treatment CT scans. Radiomics features were extracted separately from the pre- and post-treatment images and delta-radiomics features were calculated as the relative change between them. These features were used to build three models including pre-treatment, post-treatment, and delta-radiomics models. The optimal features from each model were integrated to develop a combined model. All models were evaluated on internal and external test sets using the area under the curve and decision curve analysis to predict pathologic complete response (pCR). SEN and SPE represent sensitivity and specificity, respectively.

  • Figure 2
    Figure 2
    Figure 2

    Comparison of model performance of four models in predicting pathologic complete response for non-small cell lung cancer patients to neoadjuvant chemoimmunotherapy. The combined model integrates the optimal radiomics features from pre-treatment, post-treatment and delta radiomics models.

  • Figure 3
    Figure 3

    Boxplot comparison of predicted probabilities for pathologic complete response (pCR) across radiomics models.

  • Figure 4
    Figure 4

    Decision curves of different radiomics models in prediction of pathologic complete response.

  • Figure 5
    Figure 5

    Feature importance plots based on Shapley Additive Explanations (SHAP) for the pre-treatment, post-treatment, delta, and combined radiomics models (A–D). Features are ranked in descending order according to the mean absolute Shapley values, reflecting the relative contribution to the overall model prediction. Features at the top have a greater influence on the model output.

  • Figure 6
    Figure 6

    Summary plots based on Shapley Additive Explanations (SHAP) showing the global influence of each radiomic feature on the prediction of pathologic complete response across the pre-treatment, post-treatment, delta, and combined radiomics models (A–D). Features are ranked from top to bottom according to the overall impact. Positive SHAP values are associated with a higher predicted probability of pCR, while negative values suggest a lower predicted probability.

  • Figure 7
    Figure 7

    Waterfall plots based on Shapley Additive Explanations (SHAP) visualizing individual-level feature contributions in the combined radiomics model. Red bars indicate features that increased the predicted probability of a pCR, while blue bars indicate those that decreased the predicted probability of a pCR. The final model output reflects the cumulative influence of all features. In the upper plot, a large change in wavelet-HLL_ngtdm_Strength may suggest tumor shrinkage, contributing to the prediction of pCR. In the lower plot, original_shape_Flatness_posttreatment exhibited a relatively large negative influence, indicating that the lesion with a more spherical post-treatment shape was associated with a lower probability of achieving a pCR.

Tables

  • Table 1

    Patient characteristics in the training and test sets

    VariablesTraining set (n = 106)Internal test set (n = 47)External test set (n = 110)
    Non-pCR (n = 56)pCR (n = 50)P valueNon-pCR (n = 25)pCR (n = 22)P valueNon-pCR (n = 71)pCR (n = 39)P value
    Age (years)58.73 ± 6.9162.06 ± 6.100.01062.20 ± 6.5162.36 ± 7.730.93862.2 ± 9.065.4 ± 7.20.092
    Gender0.3200.2520.053
     Male44 (78.6%)43 (86.0%)20 (80.0%)21 (95.5%)60 (84.5%)38 (97.4%)
     Female12 (21.4%)7 (14.0%)5 (20.0)1 (4.5%)11 (15.5%)1 (2.6%)
    Smoking history0.1560.4260.669
     No14 (25.0%)7 (14.0%)4 (16.0%)1 (4.5%)28 (39.4%)13 (33.3%)
     Yes42 (75.0%)43 (86.0%)21 (84.0%)21 (95.5%)43 (60.6%)26 (66.7%)
    Pathology< 0.0010.2780.794
     Non-SCC25 (44.6%)4 (8.0%)8 (32.0%)4 (18.2%)25 (35.2%)12 (30.8%)
     SCC31 (55.4%)46 (92.0%)17 (68.0%)18 (81.8%)46 (64.8%)27 (69.2%)
    ICIs1.0001.0001.000
     Anti-PD-(L)152 (92.9%)46 (92.0%)22 (88.0%)19 (86.4%)71 (100.0%)39 (100.0%)
     Non-PD-(L)14 (7.1%)4 (8.0%)3 (12.0%)3 (13.6%)0 (0.0%)0 (0.0%)
    Clinical stage0.7840.4470.683
     II25 (44.6%)21 (42.0%)13 (52.0%)9 (40.9%)26 (36.6%)12 (30.8%)
     III31 (55.4%)29 (58.0%)12 (48.0%)13 (59.1%)45 (63.4%)27 (69.2%)

    SCC, ICIs, PD-1, and PD-L1 refer to squamous cell carcinoma, immune checkpoint inhibitors, programmed cell death protein 1, and programmed death-ligand 1, respectively.

    P < 0.05 was considered statistically significant.

    Back to top

    In this issue

    Print
    Download PDF
    Email Article
    Citation Tools

    Jump to section

    Keywords