Enhanced Detection of BRCA Copy Number Alterations Within a Commercial HRD Assay: Implications for Precision Oncology in Ovarian Cancer

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

Reviewer Comments on Manuscript:

Validation of a Novel CNA Detection Algorithm for BRCA and HRR Genes in Ovarian Cancer: A Large Clinical Cohort Study (IJMS-4263292)

 General Assessment

This manuscript reports the validation of a copy number alteration (CNA) detection algorithm integrated into a commercial homologous recombination deficiency (HRD) assay (SOPHiA DDM™ Dx HRD Solution) using a large cohort of ovarian cancer samples (n=760). The authors demonstrate that the algorithm reliably identifies exon-level large genomic rearrangements (LGRs) in BRCA1/2 and other HRR genes, with high concordance to orthogonal methods such as MLPA. The study also describes the landscape of CNAs in HRR genes and their correlation with genomic instability (GI) status. While the technical performance data are solid and the cohort size is commendable, the innovation and novelty of the work are limited and do not meet the high bar expected for a journal with a broad molecular sciences readership.

1. Incremental novelty in a crowded field of HRD assays

The core claim of novelty rests on the integration of a “novel CNA detection algorithm” into an HRD assay. However, multiple commercial HRD assays already incorporate CNA or LGR detection, including the widely used Myriad myChoice® CDx (which detects BRCA LGRs and provides a genomic instability score) and FoundationOne® CDx (which reports copy number alterations across many genes). The authors do not provide a direct comparison of their algorithm’s performance against these established platforms, nor do they demonstrate a unique advantage (e.g., superior sensitivity for small exon-level events, lower input DNA requirement, or better performance in highly fragmented FFPE samples). The manuscript therefore reads as a validation study of a vendor‑specific algorithm rather than a conceptually novel contribution to the field.

2. Lack of algorithmic transparency and reproducibility

The CNA detection algorithm is proprietary to SOPHiA GENETICS, and the manuscript provides only a high-level description (coverage normalization, HMM segmentation, three-step calling). No source code, mathematical details, or parameter settings are disclosed. This lack of transparency severely limits the reproducibility of the work and prevents the scientific community from building upon or independently validating the method. For a study that positions novelty on the algorithm itself, the absence of open methodology is a major shortcoming.

3. Limited methodological comparison with state-of-the-art CNA tools

The authors compare their algorithm to MLPA and an amplicon‑based NGS assay (Devyser BRCA) in a subset of 73 samples, achieving 96–99% concordance. However, they do not benchmark their algorithm against other open‑source or widely used computational CNA detectors from NGS data (e.g., CNVkit, EXCAVATOR2, DECoN, or CODEX). Such comparisons would be necessary to demonstrate that the proprietary algorithm offers superior sensitivity, specificity, or resolution for exon‑level events, especially in challenging FFPE samples. Without this, the claim of “novel algorithm” is not substantiated.

4. Overstated clinical implications without prospective validation

The authors propose that a “tumor‑first testing workflow” using this algorithm can efficiently screen for BRCA LGRs, followed by reflex germline testing. While this is a reasonable strategy, it has already been advocated in several guidelines and previous publications (e.g., Grafodatskaya et al., J Med Genet, 2022). The manuscript does not provide prospective data or cost‑effectiveness analysis to support the superiority of this specific algorithm over existing approaches. The discussion of therapeutic implications (e.g., PARPi sensitivity based on exon‑level events) is speculative, as no clinical outcome data linking specific CNA calls to treatment response are presented.

 Recommendation

Major revision required – but given the limited conceptual novelty, the manuscript may be better suited for a specialized diagnostic or technical journal (e.g., Journal of Molecular Diagnostics, Clinical Chemistry) rather than a broad-scope journal like IJMS.

Without these changes, the manuscript does not offer sufficient innovative contribution for publication in International Journal of Molecular Sciences.

Author Response

Please see the attachment

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

This manuscript evaluates the performance of a copy number alteration (CNA) detection algorithm integrated into a commercial HRD assay for identifying BRCA1/2 large genomic rearrangements (LGRs) in ovarian cancer using FFPE samples. The manuscript presents a sizable dataset and demonstrate its analytical performance. However, I feel that the manuscript is somewhat marketing-focused. I have several questions listed below.

1. Tumor purity and heterogeneity are major challenges in CNA detection from FFPE samples. However, there is no mention of tumor purity estimation for the samples used in this study. It would be helpful if the authors could clarify whether tumor purity affects CNA detection using this HRD assay and its integrated CNA detection algorithm.
2. In Table 1, the percentages reported in this table are difficult to interpret because different denominators appear to be used. Specifically, “Excluded events” and “Evaluable events” seem to be calculated using the total number of assessed events (1,592) as the denominator, whereas “Normal” and “CNA-positive” appear to be calculated using the number of evaluable events (1,564). To improve clarity, the authors should either explicitly state the denominators used for each percentage or present this information as a stepwise filtering workflow (e.g., total assessed events → excluded events → evaluable events → CNA classification).
3. On Figure 3, please add panel labels (A, B, C, and D) directly to the figure to improve clarity and ensure consistency with the legend.
4. I am confused about the number of patients reported in the manuscript. The authors state that 760 samples were analyzed (page 6, line 83), but later indicate that 759 samples were evaluable (page 4, line 108). However, the manuscript does not explain why one sample was excluded or filtered.
5. The authors did not describe any parameters used in the SOPHiA DDM™ Dx HRD solution, and the description of the algorithm remains limited. Providing additional methodological details would improve the scientific value of the manuscript.
6. The manuscript claims to evaluate the algorithm's performance, however, it does not include comparisons with any open-source pipelines. Therefore, it is difficult to draw a clear conclusion about whether the algorithm performs well.

Author Response

Please see the attachment

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

Recommendation of  accept in present form

Reviewer 2 Report

Comments and Suggestions for Authors

The authors have resolved all my questions.