https://orcid.org/0000-0001-8654-6896
We thank Drs Xuewei Wu and Bin Zhang for their interest in our study and their correspondence regarding the spatial heterogeneity of programmed cell death 1 ligand (PD-L1) expression in non–small cell lung cancer. We appreciate the opportunity to address their concerns and clarify our findings.
We calculated our novel metric for intratumor heterogeneity, spatial heterogeneity index of PD-L1 (SHIP), on the basis of PD-L1 expression in tumor cells within each patch. Preliminary investigations led us to select a patch size of 25 mm2 because it effectively captured intratumor heterogeneity and correlated well with pathologists’ overall assessments. This size was also convenient for routine PD-L1 evaluation and provided consistent SHIP values between 2 pathologists. Additionally, our patch size scale aligns with the scale used for sampling sizes and intervals in multiregional genome analyses for assessing tumor heterogeneity (1,2). Studies in digital pathology, especially those using artificial intelligence, generally use high-resolution patches for cancer classification (3). However, our finding regarding intratumor heterogeneity that a 5-mm patch size yielded noteworthy results is particularly intriguing. Furthermore, multiresolutional analysis has become more common in recent research (4), and incorporating this approach could enhance our study.
In our study, we focused on the expression of a specific protein, PD-L1, to analyze intratumor heterogeneity of lung cancer tissues. However, hematoxylin and eosin staining also provides detailed information on cell morphology and tissue structure. With advancements in image analysis tools, feature extraction has become more accessible, allowing us to combine the morphological analysis of hematoxylin and eosin–stained images with SHIP (5,6). By integrating these methodologies, we aim to gain deeper insights into intratumor heterogeneity.
We acknowledge the limitation of our single-center study and the importance of controlling for treatment variations, including pneumonectomy and adjuvant therapy, which can confound our results. To address these confounders, we performed propensity score matching on our original cohort of 239 cases, using adjuvant therapy status and surgical procedures with a caliper value of 0.2 to achieve a 3:1 ratio of SHIP-low to SHIP-high groups (n = 150 and n = 50, respectively). After matching, the SHIP-low and SHIP-high groups had 49 (32.7%) cases and 12 (24.0%) cases receiving adjuvant therapy (P = .329) and 13 (8.7%) cases and 6 (12.0%) cases undergoing pneumonectomy (P = .676), with no statistical differences (χ2 tests). Notably, SHIP-high cases demonstrated statistically higher rates of tumor recurrences and disease-specific deaths compared with SHIP-low cases (5-year recurrence-free survival: 21.2% vs 44.2%, P < .001; 5-year disease-specific survival: 42.7% vs 66.1%, P < .005), underscoring the prognostic importance of SHIP even after controlling for confounders (Figure 1).
Prognostic analysis in the original cohort with propensity score matching. Kaplan–Meier analysis was performed for the SHIP-low group (n = 150) and SHIP-high group (n = 50). A) Recurrence-free survival. B) Disease-specific survival. SHIP = spatial heterogeneity index of PD-L1.
In this study, we aimed to share an approach for analyzing intratumor heterogeneity, rarely explored in the pathology field. Whereas pathologists typically focus on microscopic analysis, our intention was to highlight that even within microscopic details, adopting a more “macro” perspective can be valuable.
Drs Xuewei Wu and Bin Zhang’s insightful observations on our approach to evaluating tumor protein expression heterogeneity highlight the importance of further investigations. We will consider these points carefully as we expand the scope of our research and pursue further analysis.
Data availability
The data related to the presented in this correspondence will be made available upon reasonable request.
Author contributions
Tetsuro Taki, MD, PhD (Conceptualization; Data curation; Funding acquisition; Investigation; Methodology; Project administration; Validation; Visualization; Writing—original draft) and Genichiro Ishii, MD, PhD, (Supervision; Writing—review & editing).
Funding
This work was supported in part by the Japan Society for the Promotion of Science KAKENHI grant number 21K20821.
Conflicts of interest
The authors have no conflicts of interest to report.
