ELEdison Tak-Bun Liu

Edison Tak-Bun Liu

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article

Previous Organizations / Institutes

The organizations / institutes from the scholar's other articles in the past ten years


Edison Tak-Bun Liu has a total of 477 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Open Access

    Data from HRDetect in Tubo-ovarian Carcinoma: Stratification and Therapeutic Implications

    by Kalyan Banda, Helen R. Davies, Yogesh Kumar, Yasin Memari, Andrea Degasperi, Marc Radke, Isabel Rodriguez, Ted A. Gooley, Ronit Katz, Francesca Menghi, et al.

    Purpose:To evaluate HRDetect, a mutational signature–based algorithm, for prognostic utility in tubo-ovarian carcinoma and to determine if whole-genome sequencing (WGS) with rearrangement signature (RS) analysis can refine genomic classification beyond the traditional homologous recombination–deficient versus homologous recombination–proficient framework. Experimental Design:WGS was performed on matched tumor–normal pairs from 185 patients with advanced-stage tubo-ovarian carcinoma from the University of Washington (UW) cohort. HRDetect scores were calculated using substitution signatures, RS, microhomology-mediated deletions, and global loss of heterozygosity. An independent validation cohort (ARIEL2, n = 77) of platinum-sensitive tubo-ovarian carcinoma treated with rucaparib was analyzed to correlate HRDetect with PARP inhibitor (PARPi) response. RS analysis and unsupervised hierarchical clustering were employed to delineate genomic subgroups. Results:In the UW cohort, 51.4% of cases were classified as HRDetect-high and had significantly prolonged median overall survival versus others [6.2 vs. 4.1 years; hazard ratio (HR), 0.6; 95% confidence interval (CI), 0.41–0.87; P = 0.007]. Furthermore, 48.4% of HRDetect-high tumors harbored BRCA1/2 mutations or BRCA1 promoter methylation, whereas 22.1% lacked BRCA1/2 alterations. A substantial fraction (23.2%) fell into an HRDetect-intermediate category, highlighting greater genomic heterogeneity than currently appreciated. RS profiling uncovered 11 genomic clusters, with specific RS profiles (RS1, RS14, and RS18) correlating with poor survival. In ARIEL2, HRDetect-high tumors showed better PARPi responses, with improved progression-free survival (11.1 vs. 7.1 months; HR, 0.44; 95% CI, 0.26–0.74; P = 0.03) and response rates (54% vs. 22.5%). Conclusions:HRDetect predicts survival and sensitivity to PARPi in tubo-ovarian carcinoma. Combined with RS-based clustering, it reveals unappreciated genomic heterogeneity and supports a nuanced stratification framework to improve precision oncology in tubo-ovarian carcinoma.

  • Open Access

    Supplementary Figure S1 from HRDetect in Tubo-ovarian Carcinoma: Stratification and Therapeutic Implications

    by Kalyan Banda, Helen R. Davies, Yogesh Kumar, Yasin Memari, Andrea Degasperi, Marc Radke, Isabel Rodriguez, Ted A. Gooley, Ronit Katz, Francesca Menghi, et al.

    Supplementary Figure 1: CONSORT diagram of the UW ovarian cancer cohort

  • Open Access

    Supplementary Figure S2 from HRDetect in Tubo-ovarian Carcinoma: Stratification and Therapeutic Implications

    by Kalyan Banda, Helen R. Davies, Yogesh Kumar, Yasin Memari, Andrea Degasperi, Marc Radke, Isabel Rodriguez, Ted A. Gooley, Ronit Katz, Francesca Menghi, et al.

    Supplementary Figure 2: CONSORT diagram of the ARIEL 2 PART 1 clinical trial cohort.

  • Open Access

    Supplementary Figure S3 from HRDetect in Tubo-ovarian Carcinoma: Stratification and Therapeutic Implications

    by Kalyan Banda, Helen R. Davies, Yogesh Kumar, Yasin Memari, Andrea Degasperi, Marc Radke, Isabel Rodriguez, Ted A. Gooley, Ronit Katz, Francesca Menghi, et al.

    Supplementary Figure 3: Additional genomic characteristics in HRDetect groups.

  • Open Access

    Supplementary Figure S4 from HRDetect in Tubo-ovarian Carcinoma: Stratification and Therapeutic Implications

    by Kalyan Banda, Helen R. Davies, Yogesh Kumar, Yasin Memari, Andrea Degasperi, Marc Radke, Isabel Rodriguez, Ted A. Gooley, Ronit Katz, Francesca Menghi, et al.

    Supplementary Figure 4: Comparative genomic characteristics of HRDetect-intermediate versus high groups

  • Open Access

    Supplementary Figure S5 from HRDetect in Tubo-ovarian Carcinoma: Stratification and Therapeutic Implications

    by Kalyan Banda, Helen R. Davies, Yogesh Kumar, Yasin Memari, Andrea Degasperi, Marc Radke, Isabel Rodriguez, Ted A. Gooley, Ronit Katz, Francesca Menghi, et al.

    Supplementary Figure 5: Comparative genomic characteristics of HRDetect-intermediate and low groups.

  • Open Access

    Supplementary Figure S6 from HRDetect in Tubo-ovarian Carcinoma: Stratification and Therapeutic Implications

    by Kalyan Banda, Helen R. Davies, Yogesh Kumar, Yasin Memari, Andrea Degasperi, Marc Radke, Isabel Rodriguez, Ted A. Gooley, Ronit Katz, Francesca Menghi, et al.

    Supplementary Figure 6: Copy number signatures in UW cohort.

  • HRDetect in Tubo-ovarian Carcinoma: Stratification and Therapeutic Implications

    by Kalyan Banda, Helen R. Davies, Yogesh Kumar, Yasin Memari, Andrea Degasperi, Marc Radke, Isabel Rodriguez, Ted A. Gooley, Ronit Katz, Francesca Menghi, et al.

    Purpose:To evaluate HRDetect, a mutational-signature-based algorithm, for prognostic utility in tubo-ovarian carcinoma (OC), and to determine if whole-genome sequencing (WGS) with rearrangement signature (RS) analysis can refine genomic classification beyond the traditional homologous recombination deficient (HRD) versus proficient (HRP) framework. Experimental Design: WGS was performed on matched tumor-normal pairs from 185 patients with advanced-stage OC from the University of Washington (UW) cohort. HRDetect scores were calculated using substitution signatures, rearrangement signatures, microhomology-mediated deletions, and global loss of heterozygosity. An independent validation cohort (ARIEL2, n = 77) of platinum-sensitive OC treated with rucaparib, was analyzed to correlate HRDetect with PARP inhibitor (PARPi) response. RS analysis and unsupervised hierarchical clustering were employed to delineate genomic subgroups. Results: In the UW cohort, 51.4% of cases were classified as HRDetect-high and had significantly prolonged median overall survival versus others (6.2 vs 4.1 years; HR=0.6; 95% CI=0.41-0.87; p=0.007). 48.4% of HRDetect-high tumors harbored BRCA1/2 mutations or BRCA1 promoter methylation, while 22.1% lacked BRCA1/2 alterations. A substantial fraction (23.2%) fell into an HRDetect-intermediate category, highlighting greater genomic heterogeneity than currently appreciated. RS profiling uncovered eleven genomic clusters, with specific RS profiles (RS1, RS14, RS18) correlating with poor survival. In ARIEL2, HRDetect-high tumors showed better PARPi responses, with improved progression-free survival (11.1 vs 7.1 months; HR=0.44; 95% CI=0.26-0.74; p=0.03) and response rates (54% vs 22.5%). Conclusions: HRDetect predicts survival and sensitivity to PARPi in OC. Combined with RS-based clustering, it reveals unappreciated genomic heterogeneity, and supports a nuanced stratification framework to improve precision oncology in OC.


  • Top Cited Papers
    Open Access

    Differential assembly of mouse and human tumor microenvironments

    by Tristan Courau, Rebecca G. Jaszczak, Bushra Samad, Emily Flynn, Nayvin W. Chew, Gabriella C. Reeder, Jessica Tsui, Semhar Teklu, Lomax F. Pass, Austin W. Edwards, et al.

    Mouse models are frequently used to develop treatments for human cancer. However, the degree to which their tumor microenvironments (TMEs) are synonymously assembled is particularly poorly characterized. Through systematic immunoprofiling of 15 commonly used mouse models, we found that most murine TMEs recapitulate the composition of poorly infiltrated human tumors, extensively biased toward high macrophage densities. We discovered substantial species-specific biases of chemokine expression networks known to drive TMEs assembly, together with discoordinated frequencies of T and myeloid cell subtypes. Even with variable alignment, conserved cell-type-specific gene expression programs emerged across species and cohorts. Dissecting the coordinated T cell–myeloid gene expression programs revealed a conserved axis between interferon-responsive myeloid states and ongoing T cell cytotoxicity that transcends tissue of origin and predicts clinical outcome. Collectively, this work provides a practical atlas outlining both the hazards and opportunities of using mice to model human cancer.


  • Open Access

    Prevalence of BRCA1 Promoter Methylation in Cohorts of Individuals With and Without Cancer Assessed by Circulating Cell-Free DNA

    by Edison T. Liu, Margaret Antonio, Tracy Nance, Francesca Menghi

    PURPOSE: BRCA1 promoter methylation ( BRCA1 meth) affects 20%-27% of triple-negative breast cancer (TNBC) and ovarian cancer (OvCa) and has therapeutic predictive value in both cancer types. BRCA1 meth is established during early embryonic development as a constitutional event, and its detection in the peripheral blood cells of unaffected women associates with increased risk for breast cancer and OvCa. Thus, facile and sensitive assessment of BRCA1 meth is an important component for both risk assessment and the therapeutic management of these cancer types. METHODS: We used the GRAIL circulating cell-free DNA (cfDNA)–based targeted methylation platform to quantify BRCA1 meth in cfDNA (cf BRCA1 meth) from plasma samples of individuals with and without cancer. RESULTS: Detection of cf BRCA1 meth was accurate and sensitive, with an empirical LoD 95 of 0.0081. cf BRCA1 meth was found in 4.1% (113/2,790) of individuals without cancer, and it was significantly more prevalent in females compared with males (4.8% v 2.9%, P = .016), suggesting that the dynamics of BRCA1 promoter methylation and its consequences for cancer development may differ by sex. In a pan-cancer cohort comprising 2,849 patients and representing 15 tissue types, cf BRCA1 meth was enriched only in TNBC (15.2%, P = .001) and OvCa (13.9%, P = .014). Importantly, in these cf BRCA1 meth-positive cases, the fraction of cf BRCA1 meth was correlated with a methylation-based estimate of tumor burden. CONCLUSION: Our findings demonstrate that BRCA1 meth can be robustly detected using cfDNA assessment both as a constitutional event in noncancer samples and as a metric of BRCA1 meth tumor burden in individuals with BRCA1 meth cancers. Therefore, the GRAIL assay provides a single quantitative platform that can be used to explore the role of BRCA1 meth in cancer risk and therapeutic response.