Tracy Rose

Assistant Professor

University of North Carolina
Country flagUnited States

Research Interests

ApplyKite TwinLearns as you apply

Make your next move.

Prepare your application

Review drafts before you send them.

Explore related searches

Articles (21)

Radiotherapy (RT) for prostate cancer has been associated with an increased risk for the development of bladder cancer. We aimed to integrate clinical and genomic data to better understand the development of RT-associated bladder cancer. A retrospective analysis was performed to identify control patients (CTRL; n = 41) and patients with RT-associated bladder cancer ( n = 41). RT- and CTRL-specific features were then identified through integration and analysis of the genomic sequencing data and clinical variables. RT-associated bladder tumors were significantly enriched for alterations in KDM6A and ATM , whereas CTRL tumors were enriched for CDKN2A mutation. Globally, there were an increased number of variants within RT tumors, albeit at a lower variant allele frequency. Mutational signature analysis revealed three predominate motif patterns, with similarity to SBS2/13 (APOBEC3A), SBS5 (ERCC2/smoking), and SBS6/15 (MMR). Poor prognostic factors in the RT cohort include a short tumor latency, smoking status, the presence of the smoking and X-ray therapy mutational signatures, and CDKN2A copy number loss. Based on the clinical and genomic findings, we suggest at least two potential pathways leading to RT-associated bladder cancer: The first occurs in the setting of field cancerization related to smoking or preexisting genetic alterations and leads to the development of more aggressive bladder tumors, and the second involves RT initiating the oncogenic process in otherwise healthy urothelium, leading to a longer latency and less aggressive disease. Significance: Clinicogenomic analysis of radiation-associated bladder cancer uncovered mutational signatures that, in addition to a short tumor latency, smoking, and CDKN2A loss, are associated with a poor outcome. These clinical and genomic features provide a potential method to identify patients with prostate cancer who are at an increased risk for the development of aggressive bladder cancer following prostate RT.

Year:

2025

Mutational characterization of the CTRL and RT cohorts. BAM files corresponding to the FoundationOne sequencing assay were back converted to FASTQ format and aligned to GRCh38. A, The number of alterations per sample was calculated and plotted by variant type (insertions, SNVs, silent, intronic, missense, nonsense, and deletions). Pairwise comparisons between CTRL (gold), RT–short latency (dark blue), and RT–long latency (light blue) tumors were made using the Wilcoxon rank-sum test, with P values listed above their corresponding comparison. B, Histogram of VAF (alternate read count/total read depth) for CTRL (gold), RT–short latency (dark blue), and RT–long latency (light blue) tumors, variants are bin by a frequency of 0.01 (1%). Dashed line represents a VAF of 0.05 (5%). Inset boxplots represent alterations per sample either above or below the 5% VAF threshold and grouped by latency. Pairwise comparisons between CTRL (gold), RT–short latency (dark blue), and RT–long latency (light blue) tumors were made using the Wilcoxon rank-sum test, with P values listed above their corresponding comparison. Boxplots are represented by the IQR and midline at the median. Error bars equal the Q1/Q3 ± 1.5 × IQR. Oncoplots of recurrent mutations occurring above a frequency of 10% in the ( C ) CTRL and ( D ) RT cohorts. Only mutations with a VAF of ≥5% were included. E, Mutation enrichment was performed using the maftools package, and genes with an FDR >0.05 are indicated. Bars to the left of the dashed line represent CTRL vs. RT, with bars to the right of the dashed line indicating significant gene based on a 1 vs. other comparison. F, Comparison of KDM6A mutations between non-RT (TCGA and CTRL) and RT (RT and DFCI) cohorts. WT, wild type.

Year:

2025

Comparison of the copy number landscape between CTRL and RT tumors. A, Copy number status, as determined by the FoundationOne CDx assay, was plotted by the number of tumors with a given alteration (gains = positive value, losses = negative values). Groups are shown as CTRL (gold), RT–short latency (dark blue), and RT–long latency (light blue). Significant enrichment was calculated using the Fisher exact test, and P B ) CTRL, ( C ) short Latency, and ( D ) long latency, and OS differences between chr9p21.3 WT and loss (as determined by CDKN2A/B copy number status) were visualized using KM curves. Log rank P value was calculated for the pairwise comparison. Patients were then separated by CKDN2A/B copy number status to compare the difference in OS between the latency group within ( E ) deleted or ( F ) copy number neutral tumors. G, RT samples in the DFCI cohort were grouped by CDKN2A/B loss (GISTIC2 value

Year:

2024

Construction and clinical characterization of the RT and CTRL cohorts. A, Flowchart representing identification and filtering of patients included or excluded in each cohort. Barplots of distribution of ( B ) T, ( C ) N, and ( D ) M stage tumors for both the CTRL and RT cohorts. Bar height represents the proportion of cohort with the absolute number inset in the respective box. Fisher exact test was performed to calculate significance between CTRL and RT sample distribution. E, Kaplan–Meier (KM) curves representing OS for patients with CTRL (gold), RT–short latency (dark blue), and RT–long latency (light blue) tumors. F, KM curves representing OS for RT patients grouped by smoking status at the time of the bladder cancer diagnosis. For KM curves, dashed lines indicate median survival. Significance was calculated using the Cox proportional hazard model with CTRL and never smokers as the reference, respectively. G, Forest plot visualizing a multivariate Cox proportional model for OS incorporating smoking and latency. Reference groups are indicated within the figure. H, Swimmer plot showing time from RT to end of follow-up. Solid black lines represent the age of patients at the time of prostate RT. The x-axis origin denotes the time of RT, with the teal bar length proportional to the latency and the pink bar indicating length of follow-up to death or last follow-up. Vital status is noted by either a circle (alive at last follow-up) or a triangle (deceased). UTUC; upper tract urothelial carcinoma.

Year:

2024

Mutational characterization of the CTRL and RT cohorts. BAM files corresponding to the FoundationOne sequencing assay were back converted to FASTQ format and aligned to GRCh38. A, The number of alterations per sample was calculated and plotted by variant type (insertions, SNVs, silent, intronic, missense, nonsense, and deletions). Pairwise comparisons between CTRL (gold), RT–short latency (dark blue), and RT–long latency (light blue) tumors were made using the Wilcoxon rank-sum test, with P values listed above their corresponding comparison. B, Histogram of VAF (alternate read count/total read depth) for CTRL (gold), RT–short latency (dark blue), and RT–long latency (light blue) tumors, variants are bin by a frequency of 0.01 (1%). Dashed line represents a VAF of 0.05 (5%). Inset boxplots represent alterations per sample either above or below the 5% VAF threshold and grouped by latency. Pairwise comparisons between CTRL (gold), RT–short latency (dark blue), and RT–long latency (light blue) tumors were made using the Wilcoxon rank-sum test, with P values listed above their corresponding comparison. Boxplots are represented by the IQR and midline at the median. Error bars equal the Q1/Q3 ± 1.5 × IQR. Oncoplots of recurrent mutations occurring above a frequency of 10% in the ( C ) CTRL and ( D ) RT cohorts. Only mutations with a VAF of ≥5% were included. E, Mutation enrichment was performed using the maftools package, and genes with an FDR >0.05 are indicated. Bars to the left of the dashed line represent CTRL vs. RT, with bars to the right of the dashed line indicating significant gene based on a 1 vs. other comparison. F, Comparison of KDM6A mutations between non-RT (TCGA and CTRL) and RT (RT and DFCI) cohorts. WT, wild type.

Year:

2024

Radiotherapy (RT) for prostate cancer has been associated with an increased risk for the development of bladder cancer. We aimed to integrate clinical and genomic data to better understand the development of RT-associated bladder cancer. A retrospective analysis was performed to identify control patients (CTRL; n = 41) and patients with RT-associated bladder cancer ( n = 41). RT- and CTRL-specific features were then identified through integration and analysis of the genomic sequencing data and clinical variables. RT-associated bladder tumors were significantly enriched for alterations in KDM6A and ATM , whereas CTRL tumors were enriched for CDKN2A mutation. Globally, there were an increased number of variants within RT tumors, albeit at a lower variant allele frequency. Mutational signature analysis revealed three predominate motif patterns, with similarity to SBS2/13 (APOBEC3A), SBS5 (ERCC2/smoking), and SBS6/15 (MMR). Poor prognostic factors in the RT cohort include a short tumor latency, smoking status, the presence of the smoking and X-ray therapy mutational signatures, and CDKN2A copy number loss. Based on the clinical and genomic findings, we suggest at least two potential pathways leading to RT-associated bladder cancer: The first occurs in the setting of field cancerization related to smoking or preexisting genetic alterations and leads to the development of more aggressive bladder tumors, and the second involves RT initiating the oncogenic process in otherwise healthy urothelium, leading to a longer latency and less aggressive disease. Significance: Clinicogenomic analysis of radiation-associated bladder cancer uncovered mutational signatures that, in addition to a short tumor latency, smoking, and CDKN2A loss, are associated with a poor outcome. These clinical and genomic features provide a potential method to identify patients with prostate cancer who are at an increased risk for the development of aggressive bladder cancer following prostate RT.

Year:

2024

Identification of de novo mutational signatures. A, Heatmap of the cosine similarity between the de novo mutational signatures and known Catalogue of Somatic Mutations in Cancer SBS profiles. Boxplots of the signature contribution to the overall SNVs for each sample grouped by ( B ) signature and latency or ( C ) smoking status and signature for CTRL and ( D ) RT tumors. Pairwise comparisons between CTRL (gold), RT–short latency (dark blue), and RT–long latency (light blue) tumors were made using the Wilcoxon rank-sum test, with P values listed above their corresponding comparison. Boxplots are represented by the IQR and midline at the median. Error bars equal the Q1/Q3 ± 1.5 × IQR. E, Samples were grouped by the mutational signature with the greatest signature contribution. KM curves were generated based on this max signature assignment for OS for RT patients. Dashed lines indicate median survival. G, Forest plot visualizing a multivariate Cox proportional model for OS incorporating the max signature group, latency, and smoking status. WT, wild type.

Year:

2024

Comparison of the copy number landscape between CTRL and RT tumors. A, Copy number status, as determined by the FoundationOne CDx assay, was plotted by the number of tumors with a given alteration (gains = positive value, losses = negative values). Groups are shown as CTRL (gold), RT–short latency (dark blue), and RT–long latency (light blue). Significant enrichment was calculated using the Fisher exact test, and P B ) CTRL, ( C ) short Latency, and ( D ) long latency, and OS differences between chr9p21.3 WT and loss (as determined by CDKN2A/B copy number status) were visualized using KM curves. Log rank P value was calculated for the pairwise comparison. Patients were then separated by CKDN2A/B copy number status to compare the difference in OS between the latency group within ( E ) deleted or ( F ) copy number neutral tumors. G, RT samples in the DFCI cohort were grouped by CDKN2A/B loss (GISTIC2 value

Year:

2024

Construction and clinical characterization of the RT and CTRL cohorts. A, Flowchart representing identification and filtering of patients included or excluded in each cohort. Barplots of distribution of ( B ) T, ( C ) N, and ( D ) M stage tumors for both the CTRL and RT cohorts. Bar height represents the proportion of cohort with the absolute number inset in the respective box. Fisher exact test was performed to calculate significance between CTRL and RT sample distribution. E, Kaplan–Meier (KM) curves representing OS for patients with CTRL (gold), RT–short latency (dark blue), and RT–long latency (light blue) tumors. F, KM curves representing OS for RT patients grouped by smoking status at the time of the bladder cancer diagnosis. For KM curves, dashed lines indicate median survival. Significance was calculated using the Cox proportional hazard model with CTRL and never smokers as the reference, respectively. G, Forest plot visualizing a multivariate Cox proportional model for OS incorporating smoking and latency. Reference groups are indicated within the figure. H, Swimmer plot showing time from RT to end of follow-up. Solid black lines represent the age of patients at the time of prostate RT. The x-axis origin denotes the time of RT, with the teal bar length proportional to the latency and the pink bar indicating length of follow-up to death or last follow-up. Vital status is noted by either a circle (alive at last follow-up) or a triangle (deceased). UTUC; upper tract urothelial carcinoma.

Year:

2024

Prognostic value of the smoking and XRT signatures in TCGA and proposed scheme of RT-associated bladder cancer development. The Pearson correlation between the UNC signature motif frequency profile and TCGA BLCA samples was calculated, and TCGA samples were split into high and low groups at the median correlation. Samples were further split by CDKN2A status, and KM curves were generated. A multivariate Cox proportional hazard model was calculated for both the ( A ) smoking signature and ( B ) XRT signature. For all KM plots, dashed lines indicate median survival. C, Proposed model for the interplay between preexisting alteration in the urothelium and RT. WT, wild type.

Year:

2024

Year:

2024

Year:

2024

Year:

2021

Year:

2021

Year:

2020

Year:

2020

Year:

2020

Year:

2020

Year:

2020

Year:

2020

Year:

2020

Collaborators (7)

William Y. Kim

Professor

University of North Carolina

UNITED STATES
View details

Sara E Wobker

University of North Carolina at Chapel Hill

UNITED STATES
View details

Krishna Kanchi Kanchi

National Institutes of Health

UNITED STATES
View details

Jeffrey Damrauer

Assistant Professor

University of North Carolina

UNITED STATES
View details

Matthew D. Galsky

Radboud University Medical Center

NETHERLANDS
View details

Ersan Altun

Clinical Associate Professor

University of North Carolina Hospitals

UNITED STATES
View details

David Corcoran

Assistant Professor

University of North Carolina

UNITED STATES
View details