RE: Spatial intratumor heterogeneity of programmed cell death ligand 1 expression predicts poor prognosis in resected non–small cell lung cancer

The article authored by Yusuke Nagasaki et al. (1) introduces a novel mathematical model for assessing the spatial intratumor heterogeneity of programmed cell death ligand 1 (PD-L1) and investigates its correlation with tumor recurrence and patient prognosis. This study included 239 consecutive surgically resected specimens of non–small cell lung cancer with pathological stages ranging from IIA to IIIB. Utilizing a texture image analysis–based model, the study determined the spatial heterogeneity index of PD-L1 for each tumor. The findings demonstrated that elevated spatial intratumor heterogeneity of PD-L1 values correlated with histological subtype and vascular invasion, and patients exhibiting high spatial intratumor heterogeneity of PD-L1 values experienced significantly poorer recurrence-free rates. However, several issues warrant further discussion.

First, the authors segmented the tumor areas into 25-mm² square patches and assessed the proportions of PD-L1–positive tumor cells within each patch. However, it’s crucial to evaluate whether the 25-mm² square patches represent the optimal size for quantifying PD-L1 expression’s intratumor heterogeneity. As noted in a previous study (2), an ensemble model integrating magnifications of 5×, 10×, and 20× with 64 patches from whole slide images yielded the highest performance. A dedicated analysis comparing the impact of different patch sizes would hold meaningful clinical relevance.

Second, the spatial heterogeneity index of PD-L1 was defined as the average squared difference in the proportions of PD-L1–expressing tumor cells between a focal patch and its neighboring patches. Su et al. (3) computed the morphological intratumor heterogeneity on hematoxylin and eosin–stained tumor slides. Morphological intratumor heterogeneity was assessed for each feature as the standard deviation of the feature values across all tumor cells, stromal cells, or inflammatory cells within the sample. We are eager to explore the varying effectiveness of these 2 methodologies, that is, morphological intratumor heterogeneity and spatial heterogeneity index of PD-L1, in capturing tumor heterogeneity. We anticipate gaining deeper insights from a comparative analysis in future research endeavors.

Third, given that a statistically higher proportion of patients who underwent pneumonectomy or received adjuvant therapy in the spatial heterogeneity index of PD-L1 high group compared with the spatial heterogeneity index of PD-L1 low group (P = .001 and P = .003, respectively) (1), it may be prudent to use the propensity score matching approach to evaluate the impact of spatial heterogeneity index of PD-L1 status on prognosis. Furthermore, as the study is confined to a single-center investigation, additional independent cohorts are necessary to train and validate the robustness of the model’s performance.

In conclusion, the spatial heterogeneity index of PD-L1 proposed by the authors holds promise in predicting treatment response or resistance. This innovative approach to evaluating tumor protein expression heterogeneity paves the way for future investigations.

Data availability

No data are reported in this correspondence.

Author contributions

Xuewei Wu, PhD (Conceptualization; Writing—original draft) and Bin Zhang, PhD, MD (Conceptualization; Supervision; Writing—review & editing).

Funding

No funding was used for this study.

Conflicts of interest

The authors have no conflicts of interest to report.