Use of UV mutational signatures to distinguish between cutaneous and pulmonary primary squamous cell carcinoma.

Background: Squamous cell carcinoma (SCC) of presumed lung origin (PLO) is now the second most frequent histologic subtype of non-small cell carcinoma after adenocarcinoma. The use of clinic-genomic correlation provided by comprehensive genomic profiling (CGP) can revise clinicopathologic diagnoses of presumed primary lung SCC (PLO-SCC) to diagnoses of metastatic SCC of cutaneous origin (C-SCC).

Design: A total of 10 146 samples of clinically advanced PLO-SCC (84% known Stage IV) passed QC metrics and were designated as PLO-SCCs by review of test requisition forms, clinical notes, and pathology reports. One thousand seven hundred sixty-one cases of known primary C-SCC were also included in this study. All samples underwent hybrid capture-based CGP (Foundation Medicine, Inc.) using a targeted gene panel to evaluate all classes of genomic alterations (GA), determine MSI, TMB, and genomic ancestry status. The mutational signature (MS) of each case was called by the decomposition method using reference signatures in the COSMIC database. PD-L1 tumor cell expression was determined by IHC (22C3; Dako). All results were compared using the Fisher exact method with the false discovery rate corrected with a Benjamini-Hochberg adjustment.

Results: A total of 253 of 10 146 (2.5%) PLO-SCC cases featured a UV+ MS; 812 of 1761 C-SCC (46.1%) that also featured a UV radiation exposure MS (UV+) were also included in this study. PLO-SCC UV+ cases used for sequencing included tissue samples from the lung (162), lymph node (34), soft tissue (33), liver (8), head and neck (7), brain (5), and skin thought to be metastatic sites from primary lung SCC (4). The PLO-SCC UV+ patients were 78.7% male and had a median age of 72 years, which was younger and more frequently male gender than both the C-SCC UV+ and C-SCC UV- patients (p < 0.0001). Both the PLO-SCC UV+ and C-SCC UV+ featured greater GA per tumor than the PLO-SCC UV- cases (p < 0.0001). In the PLO-SCC UV- cases, tobacco exposure and APOBEC were the most frequent MSs. For the biomarkers associated with immune checkpoint inhibitor efficacy, when compared with the PLO-SCC UV- cases, the PLO-SCC UV+ cases featured more cases with TMB ≥10 mutations/Mb (88.5% vs. 36.5%; p < 0.0001) and ≥20 mutations/Mb (66.8% vs. 6.8%; p < 0.0001) and a trend for less frequent positive PD-L1 (≥50% TPS) IHC staining (30.2% vs. 39.6%; p = 0.062). Compared to PLO-SCC UV- cases, PLO-SCC UV+ and C-SCC UV+ cases were more likely to harbor clinically-actionable GA in PTCH1 and NOTCH1/2 (p < 0.0001) and less likely to harbor clinically-actionable GA in KRAS, PIK3CA, and PTEN (p < 0.0001). The frequency of PTCH1 GA in PLO-SCC UV+ (32% vs. 0.9% in PLO-SCC UV-) suggested that PLO-SCC UV+ may include a mixture of C-SCC and cutaneous basal cell carcinomas (C-BCC) with squamous differentiation.

Conclusions: When cases of PLO-SCC undergo CGP, a small 2.5% subset of cases that featured a UV MS emerge that indicates that these tumors may actually represent metastatic cutaneous SCC or BCC with squamous differentiation. Given the significant treatment and clinical impact associated with the resolution of the true diagnosis of these cases, the use of genomic sequencing in PLO-SCC may be clinically beneficial.