A transcriptomic biomarker predicts response to genotoxic cancer therapy
PD-0825
Abstract
A transcriptomic biomarker predicts response to genotoxic cancer therapy
Authors: Ines Guix1, Ann Lazar2, Jian-Hua Mao3, Miquel Angel Pujana4, Mary Helen Barcellos-Hoff1
1University of California San Francisco (UCSF), Radiation Oncology, San Francisco, USA; 2University of California San Francisco (UCSF), Division of Biostatistics, San Francisco, USA; 3Lawrence Berkeley National Laboratory, BioEngineering & BioMedical Sciences, Berkeley, USA; 4Catalan Institute of Oncology, Bellvitge institute for Biomedical Research (IDIBELL), Barcelona, Spain
Show Affiliations
Hide Affiliations
Purpose or Objective
Transforming growth factor β (TGFβ) enforces
effective DNA repair by promoting homologous recombination and
suppressing the use of less efficient alternative end-joining (alt-EJ). We
previously translated this mechanistic relationship into gene expression
signatures of chronic TGFβ signaling and alt-EJ DNA repair
to establish a score, βAlt, that predicts patient outcomes in response to
genotoxic cancer therapy with radiation or platinum chemotherapy (doi:
10.1126/scitranslmed.abc4465). Here we sought to verify this biology in live
human tumors and to further refine the βAlt biomarker to improve its
predictive capacity.
Material and Methods
Explants of head and neck squamous cell carcinoma (HNSC) tumors and
patient-derived xenographs were immunostained to measure TGFβ signaling,
indicated by SMAD2 phosphorylation, and unrepaired DNA damage, indicated by
persistent 53BP1 foci at 5 hours after 5 Gy irradiation. Each signature gene
was then weighted by its association with percent pSMAD2 and 53BP1 positive cells,
as well as by its centrality degree within their respective gene coexpression
network. These results were used to define a modified score, βAltw, based
on the sum of gene expression levels multiplied by their estimated weights. The
βAltw
score was
compared to the original βAlt biomarker for association with patients’ response
to genotoxic therapies and validated in independent datasets.
Results
Most
TGFβ genes were positively correlated with the frequency of pSMAD2 positive
cells (37/50) and negatively correlated with 53BP1 positive cells (36/50),
whereas most alt-EJ genes were positively correlated with residual DNA damage
marked by 53BP1 (30/36) and negatively correlated with pSMAD2 (31/36),
supporting that the signatures accurately report TGFβ
signaling competency and inefficient DNA repair, respectively. As expected, βAltw was correlated with the surviving fraction after exposure to 2 Gy of
NCI-60 pancancer cell lines (Spearman correlation coefficient = -0.36, P =
0.005). βAltw predicted overall survival in TCGA pancancer
patients whose standard of care includes radiotherapy and/or
genotoxic chemotherapy based on their cancer type and stage (P<0.001; Figure 1); its
performance was significantly superior to the original βAlt (P
< 0.001; Figure 1). The association between the βAltw
and survival was significant in three HNSC and ovarian cancer patients data
sets: HNSC GSE41613 (P = 0.015; Figure 2B), HNSC TCGA (P = 0.019; Figure 2A), and ovarian cancer GSE41613
(P = 0.036; Figure 2C,D).
Conclusion
The integration of biological response and signature expression validates
TGFβ competency as a key mediator of DNA repair. If prospectively validated, βAltw
could be a useful biomarker to assist in clinical decision-making to
personalize treatment.