Introduction: Lung squamous cell carcinoma (SCC) is the second most frequent histologic type of non-small cell lung carcinoma (NSCLC). Next to pTNM staging, grading is also important for histomorphologic classification of tumors into relevant prognostic groups. SCC forms cohesive tumor sheets of various sizes, ranging from large pushing boundaries to small detached clusters. The presence of small clusters such as single cells, tumor buds 15 cells) could also contain prognostic information and hence be pertinent for tumor grading. Herein we propose a computer-based morphometric analysis for the global assessment of lung SCC fragmentation.

Methods: Surgically resected lung SCC tumors (discovery tissue microarray cohort n = 208, 2 cores per patient; validation whole section cohort n = 99, 2 blocks per patient) were immuno-histochemically stained with pan-cytokeratin. Color-based segmentation (tumor epithelia brown, stroma blue-grey) allowed to separate tumor from stroma and to compute the number of tumor fragments. Associations with disease-free (DFS) and overall survival (OS) were assessed by univariate Cox regressions.

Results: Tumor fragments were defined as tumor clusters >3000μm2 (circa >15 cells) entirely surrounded by stroma. Criteria for high fragmentation were: > 5 fragments summed over 2 TMA cores (range: 2-18, median = 5) and > 700 fragments summed over 2 whole sections (range: 94-3988, median = 570). Respectively, 44% and 42% of the patients had highly fragmenting tumors in the discovery (D) and validation cohort (V), correlating with blood vessel infiltration (cc = 0.181, p<0.001 (D); cc = 0.220, p = 0.029 (V)). High fragmentation was also associated with decreased OS (HR = 1.77, p = 0.001) and DFS (HR = 1.83, p = 0.027) in (D). This was confirmed in (V): OS (HR = 2.09, p = 0.004), DFS (HR = 2.60, p = 0.005).

Conclusions: We used an automated morphometric approach for unbiased scoring of tumor fragmentation on digitalized histological sections. This versatile method could potentially be applied to virtually any solid human carcinoma. Our analysis revealed that tumor fragmentation is a poor prognosticator for lung squamous cell carcinoma, e.g. due to increased likelihood of vessel infiltration.

Citation Format: Ruben Casanova, Alex Soltermann. Tumor fragmentation is a poor prognostic histologic biomarker for lung squamous cell carcinoma. [abstract]. In: Proceedings of the 107th Annual Meeting of the American Association for Cancer Research; 2016 Apr 16-20; New Orleans, LA. Philadelphia (PA): AACR; Cancer Res 2016;76(14 Suppl):Abstract nr 2264.