Construction and validation of a scoring system to predict resistance to chemotherapeutic drugs using gene expression profiles in canine lymphoma.

ABSTRACT

The present study aimed to verify the changes in the expression levels of 13 candidate genes associated with chemotherapy resistance and to construct a scoring system to predict resistance to these drugs. The expression levels of the 13 candidate genes were compared between 20 dogs with lymphoma that were sensitive to drugs used in CHOP-based protocol and 16 dogs with lymphoma that were resistant to these drugs. The expression levels of six genes; ASNS, CCR3, CALCA, FCER1A, LOC448801, and EDNRB were significantly different between the two groups. A scoring system to predict resistance to cyclophosphamide, doxorubicin and vincristine, which are used in CHOP-based protocol, was constructed based on expression levels of the six genes in these 36 dogs using logistic regression models. After internal validation, sensitivity and specificity of the scoring system were 0.759 and 0.853, respectively. External validation was conducted in another cohort of 33 dogs with lymphoma, and sensitivity and specificity of the scoring system were 0.800 and 0.696, respectively. In conclusion, this study identified six genes associated with resistance to drugs used in CHOP-based protocol in canine lymphoma and proposed a novel scoring system to predict resistance to these drugs. This system might be beneficial in selecting the most appropriate chemotherapy protocol for individual dogs with lymphoma.