Molecular Signatures in Advanced Cutaneous Squamous Cell Carcinoma

Evidence

This retrospective study examined molecular correlates of PD-1 inhibitor response in patients with locally advanced or metastatic, histopathologically confirmed cutaneous squamous cell carcinoma treated at German skin cancer centers between May 2019 and December 2022. Eligibility required receipt of at least one PD-1 inhibitor cycle. Investigators collected clinical information and pretreatment tumor tissue, enrolling 44 cutaneous squamous cell carcinoma tissue samples from 33 patients. Tumor response was evaluated with Response Evaluation Criteria in Solid Tumors version 1.1, allowing the molecular comparisons to be anchored to recorded clinical response categories rather than to laboratory measurements alone. [pmid:42965700]

The laboratory workflow combined whole-transcriptome and protein-level assessment. RNA profiling used HTG EdgeSeq technology; sequenced RNA counts were normalized by the median-ratio method, and differential-expression analyses used multiple-testing correction through the Benjamini–Hochberg procedure within DESeq2. Candidate targets emerging from the transcriptomic work were further assessed with immunofluorescence and immunohistochemistry. The study also used CIBERSORTx-based deconvolution to estimate the composition of the immune infiltrate. This multimodal approach examined both immune-related transcriptional signals and tumor-intrinsic markers in pretreatment material. [pmid:42965700]

The reported clinical grouping included overall response in 17 patients (51.5%) and no overall response in 16 patients (48.5%). Early progression occurred in 17 patients (51.5%), whereas 16 patients (48.5%) did not have early progression. These reported categories supplied two related but distinct comparison structures: favorable clinical outcome or disease control versus poorer response, and early progression versus late or no progression. The source describes the molecular analyses as associations with these clinical patterns, not as tests of a biomarker-directed intervention. [pmid:42965700]

Tumors from patients achieving disease control had an immune-active transcriptional profile characterized by higher interferon-associated signaling, enhanced antigen-presentation pathways, and cytotoxic immune activity. Interferon-gamma and cytolytic-activity signatures were also reported as higher among patients with favorable clinical outcomes. In contrast, CIBERSORTx-based deconvolution found marked interpatient heterogeneity, with T cells, macrophages, and natural killer cells among major immune-infiltrate components, but no major differences in overall immune-cell composition between patients with early and late or no progression. Histopathological assessment showed a tendency toward higher immune-cell infiltration among patients achieving disease control. [pmid:42965700]

Among tumor-intrinsic signals, elevated SLC7A5 and SOX2 gene expression was associated with poor response and early progression. At the protein level, higher SLC7A5 expression was associated with early progression. SOX2 protein expression was heterogeneous, including particularly high expression in a subset of progressive tumors. Thus, the protein-level observations did not portray SOX2 as a uniform feature across all progressive tumors, while the SLC7A5 pattern was reported in relation to early progression. [pmid:42965700]

Analysis — Interpretation of observed associations

The observed pattern is consistent with the study’s retrospective design and selection process: pretreatment tumors from patients who later had disease control carried more immune-active transcriptomic features, while elevated SLC7A5 and SOX2 expression tracked with unfavorable response or early-progression categories. The use of pretreatment tissue gives the associations temporal separation from recorded treatment outcomes, and protein-level assessment added a separate measurement approach for selected transcriptomic candidates. However, the cohort was assembled from treated patients with available clinical data and tumor material rather than through prospective biomarker stratification or random allocation. Accordingly, the comparisons describe covariation between measured tumor features and outcomes in this selected clinical cohort. They do not establish that interferon-associated signaling, antigen-presentation pathways, cytolytic activity, SLC7A5, or SOX2 caused response, resistance, or progression. The lack of major overall immune-composition differences across progression groups also places the reported transcriptional associations alongside heterogeneous cellular infiltrates rather than a simple group-wide difference in estimated immune-cell proportions. [pmid:42965700]

Limitations

The retrospective design cannot establish causation. Clinical and tissue data were collected retrospectively from patients treated with PD-1 inhibitors, so the observed associations may reflect selection processes and characteristics of the sampled population as well as relationships between molecular measurements and outcome categories. The comparator groups were defined by recorded response and progression patterns, while the measured exposures were pretreatment gene-expression signatures, selected protein-expression assessments, estimated immune-cell composition, and histopathological immune infiltration. This structure identifies correlations within the cohort rather than causal effects of individual molecular features. [pmid:42965700]

The transcriptomic analysis was exploratory and used pretreatment tumor samples from a small cohort of 17 patients. The source notes that whole-transcriptome profiling has high dimensionality relative to this sample size, making individual differential-expression findings susceptible to sampling variability. It also reports no independent validation of the observed molecular associations and states that the candidate biomarkers require validation in larger independent cohorts. Interpatient heterogeneity in immune-cell composition and the heterogeneous SOX2 protein pattern further constrain any single-marker interpretation. These limitations mean the report identifies candidate signatures associated with outcomes in this cohort, not validated clinical biomarkers. [pmid:42965700]