Single-biopsy transcriptomic heterogeneity scores and multiregional diversity in clear-cell renal cell carcinoma
Year: 2026
Type: preprint
Abstract: Background. Single-specimen transcriptomic heterogeneity scores and diversity among tumour regions describe different properties. We assessed their agreement in clear-cell renal cell carcinoma (ccRCC) and examined how reference construction and simulated aggregation influence score interpretation. Methods. In four multiregion cohorts, single-region DEPTH2 was compared with expression divergence among the remaining regions. Reference fitting and gene selection excluded the whole input patient. Additional experiments examined reference swaps and paired gene-wise maximum (MAX) versus count-sum (SUM) mixtures from eight Li donors. Monte Carlo sensitivity analyses used repeated seeds and seven independently recruited Oxford patients with a different cellular-state definition. Results. Agreement was small and imprecise in the three RNA-sequencing cohorts: Spearman ρ̂ = 0.109 (95% CI −0.204 to 0.391; 32 tumours), 0.122 (−0.192 to 0.394; 26) and 0.130 (−0.095 to 0.347; 38). The nine-tumour array cohort gave 0.522 (0.257 to 0.672), although its interval method undercovered in simulation. DEPTH2 did not demonstrate improvement over a cell-cycle comparator in TRACERx. Cross-pipeline reference swaps classified every tested input as high relative to the reference median. In mixtures, the primary DEPTH reference interaction was −0.427 (descriptive donor-bootstrap interval −0.665 to −0.186) in Li and 0.102 (−0.110 to 0.328) in Oxford. The Oxford cohort did not reproduce the Li result. Conclusions. Single-region DEPTH2 showed limited, imprecisely estimated agreement with held-out regional diversity. Reference construction changed score classification and aggregation comparisons. Small patient counts, differing targets and interval limitations constrain generalisation.
Author: Mamun, S M Abdulla Al
Institution: University of Manchester
Language: English
Subfield: Cancer Research
Domain: Life Sciences
Open Access status: green
OpenAlex ID: W7236760491
