Integrated mutational footprints in prostate cancer: where the 85% explanatory claim stops

From single-base signatures to integrated mutational footprints

Earlier mutational signature work established that APOBEC enzymes leave a dominant footprint in many cancers, initially understood as restricted to TpC sites [2]. Langenbucher et al. showed that DNA secondary structure is an orthogonal influence on APOBEC3A substrate optimality, with VpC sites in optimal hairpins outperforming TpC sites and resolving the genomic Twin Paradox [2]. In prostate cancer specifically, Lotan et al. demonstrated that HRD scores are low in primary disease and higher in germline BRCA2-altered cases than in ATM- or CHEK2-altered cases, establishing that not all homologous recombination gene mutations produce equivalent genomic scarring [3]. The anchor paper builds on this foundation by integrating de novo extracted single-base substitution, insertion–deletion and copy-number signatures with six novel complex structural variant signatures, yielding eight IMFs [1]. This is a methodological expansion rather than a replacement: it absorbs prior signature concepts into a higher-dimensional framework that also captures structural variant processes previously invisible to single-base analyses [1].

What the 85% explanation actually covers

The headline claim is that eight IMFs collectively explain the mutational processes in 85% of primary prostate cancer genomes from 959 donors in the Pan Prostate Cancer Group [1]. This figure derives from a cohort of 961 primary prostate cancers with diverse histopathological stages and clinical outcomes, plus metastatic samples, and it includes clock-like mutational processes in non-CIN tumours as part of the explanatory framework [1]. The 85% is therefore a cohort-level explanatory statistic: it describes how much of the mutational landscape can be assigned to defined processes across a population, not the probability that any individual patient's genome is fully explained. The remaining 15% is not trivial—it represents genomes where the current IMF catalogue does not capture the dominant mutational process, and it may include rare aetiologies, technical artefacts, or processes that require even higher-dimensional integration [1]. The study also reports that IMFs were strongly influenced by regional biases in the genome, most prevalently androgen receptor-mediated mutagenesis and replication stress [1]. This regional bias finding is important because it means IMF activity is not uniformly distributed across the genome, and bulk estimates may mask localised mutational hotspots relevant to driver gene discovery [1].

HRD, non-canonical impairment and African ancestry enrichment

Four IMFs, present in 37% of primary tumours, were significantly associated with shorter time to metastasis, including reactive oxygen-species-driven mutagenesis and both canonical and non-canonical homologous recombination deficiency [1]. The non-canonical HRD footprint, IMF5, was enriched in patients of African ancestry (P = 5.36 × 10⁻⁵, n = 32, Fisher's test) and was associated with CDK12 mutations, which phenocopy impaired HR in the absence of BRCA1/2 mutations [1]. This finding aligns with prior evidence that HRD scores are not uniformly elevated across all homologous recombination gene alterations: Lotan et al. showed that germline BRCA2-altered prostate cancers had median HRD scores of 27 versus 16.5 for ATM-altered and 9 for CHEK2-altered cases [3]. The anchor paper's IMF5 enrichment in African ancestry patients is consistent with a previous study finding more tandem duplications in this group [1]. However, the African ancestry subgroup analysis rests on only 32 patients, and the study was not designed to develop a risk stratification biomarker [1]. The clinical implication—that non-canonical HRD might expand PARP inhibitor eligibility beyond BRCA1/2—remains an open question because the study did not test PARP inhibitor response directly [1].

Metastasis prediction and ARPI sensitivity: emulated, not proven

The anchor paper reports that IMF6 activity was predictive of reduced risk of treatment failure following androgen receptor pathway inhibitors (ARPIs), using an approach to emulate phase III randomized controlled trials [1]. IMF6 is linked to SPOP mutations and replication stress, corroborating recent findings of improved ARPI response in SPOP-mutated mCRPC and potentially extending prediction beyond isolated SPOP-mutation-based approaches [1]. However, only seven patients in the HMF cohort had SPOP alterations, and the hazard ratio for IMF6-predicted sensitive patients was 0.21 (0.05–0.79) with P = 0.021 in a small subgroup [1]. This is a hypothesis-generating signal, not a validated predictive biomarker. The broader ARPI resistance landscape involves AR amplification, AR mutations, AR splice variants, and non-AR-dependent pathway activation [4][7][8]. Shiota et al. reviewed that specific AR mutations (L702H, W742L/C, H875Y, F877L, T878A/S) are frequently identified after treatment resistance and change the biology of the receptor [4]. The anchor paper's IMF6 signal operates at the level of mutational processes rather than specific AR mutations, and the two approaches have not been directly compared [1][4]. Until prospective validation in independent cohorts with adequate SPOP-mutant representation, IMF6 should not be used to select or deselect patients for ARPI therapy [1].

Boundaries: germline variants, AR mutations and prospective validation

The anchor paper's conclusions are based on 959 donors from the Pan Prostate Cancer Group, and the authors explicitly state that the study was not designed to develop a biomarker for risk stratification [1]. The 85% explanatory rate and the treatment sensitivity predictions require prospective cohort validation and cannot be extrapolated to all populations [1]. Germline rare variant evidence provides a complementary and non-overlapping layer: Lin et al. identified germline nonsynonymous rare variants in DDR genes with significantly higher burden in metastatic versus non-metastatic prostate cancer (p = 4.57 × 10⁻⁶), and functionally validated BRCA2-I1962T as sensitising to olaparib [5]. This germline contribution is not captured by IMFs, which are derived from tumour mutational footprints [1][5]. Similarly, AR mutation testing and circulating tumour DNA approaches address resistance mechanisms that operate at the level of the receptor or pathway rather than the mutational process [4][6]. Wan et al. showed that mutational signatures can be identified in plasma cell-free DNA at low coverage, suggesting a future path for non-invasive IMF assessment, but this remains proof-of-concept [6]. The clinical decision points that IMFs cannot yet replace include germline genetic counselling, AR mutation testing for treatment resistance, and prospective risk stratification for localised disease [1][4][5].

About These Sources

This research page is built on 8 studies (7 peer-reviewed, 1 preprint) — published from 2021 to 2026, 2 from 2024 or later, 3 in Q1 journals, collectively cited 516 times — selected as the most relevant from 13 studies that passed quality screening, drawn from 173 papers retrieved from a database of over 500 million.

Sources used in this answer

Related research questions