Fig. 3: Landscape of cell type-specific subtype characteristics.

a UMAP plot of single-cell types specific to Subtypes 1 to 5, using each DEG_subtype. The color of each point represents the module score of each cell; the more relevant the module is to the cell type, the higher the score and the redder the color. UMAP information was obtained from the original study²³. b Representative histologic images of the subtypes. The tumor cell (T) and stromal (S) components are separately labeled. Note the irregularly fused tumor glands in subtype 1 tumor compared to those in subtype 2 tumors composed of small, uniform tumor cells lying within the elastic stroma similar to the normal alveolar wall. Dense stromal inflammatory cell infiltration in Subtype 5 tumors. Scale bars for b = 100 μm. c Proportions of samples with different pathologic diagnoses within each subtype. LUAD was predominant in Subtypes 1 and 2, whereas LSCC was predominant in Subtype 3. d–f Histologic patterns of LUADs in each subtype. The predominant patterns of Subtype 1 and 2 LUADs were most commonly acinar or papillary but were quite heterogeneous (d). The proportion of the lepidic pattern, considered to indicate noninvasive LUAD, was enriched mostly in Subtype 2 LUADs, suggesting that Subtype 2 is most like early LUAD (Subtype 1, n = 55; Subtype 2, n = 34, Subtype 3, n = 10; Subtype 4, n = 26; Subtype 5, n = 20) (e). Consistently, the proportion of high-grade histologic patterns (including solid, micropapillary, cribriform, and complex glandular patterns) was lowest in Subtype 2 LUADs. The high-grade histologic pattern was more extensive in Subtype 1 than in Subtype 2, but these subtypes were remarkably heterogeneous compared to Subtype 3–5 LUADs, which were mostly composed of high-grade histologic patterns (Subtype 1, n = 55; Subtype 2, n = 34, Subtype 3, n = 10; Subtype 4, n = 24; Subtype 5, n = 20) (f). For box-plots, middle line, median: box edges, 25th and 75th percentiles; whiskers, most extreme points that do not exceed ±1.5 × IQR. g–I Lymphovascular invasion (g), lymph node metastasis (h), and tumor necrosis (i) were less common in Subtype 2 tumors, which also implies that Subtype 2 tumors are in a clinically early, nonprogressed stage. j Microscopically, the stromal component was more extensive in Subtype 2, 4, and 5 tumors (Subtype 1, n = 55; Subtype 2, n = 43, Subtype 3, n = 52; Subtype 4, n = 43; Subtype 5, n = 34). For box-plots, middle line, median: box edges, 25th and 75th percentiles; whiskers, most extreme points that do not exceed ±1.5 × IQR. k–l Tumor-infiltrating lymphocytes (k) and stromal neutrophilic infiltration (l) were most extensive in Subtype 5 tumors. The p-value was calculated using the chi-square test (c, d, g–I, k, and l) the Kruskal-Wallis test (e, f, and j). m Summary of the histopathologic characteristics of the NSCLC subtypes.