Establishment And Validation Of Doxorubicin Sensitivity Prediction Model In Gastric Cancers

Chemotherapy is the standard treatment for patients with advanced gastric cancer; Doxorubicin is widely used in various chemotherapy regimens in combination with other chemicals such as 5-Fu and DDP. However, these regimens are not entirely satisfactory. According to literature, the response rate of doxorubicin in gastric cancer chemotherapy is only 30%. This poor therapy efficacy may lie in the fact that many gastric cancers are naturally resistant to many anticancer drugs, or acquire resistance during prolonged treatment. Therefore, predicting the occurrence of drug resistance becomes a major topic for successful chemotherapy of gastric cancers, unfortunately, up to date, clinical tests for predicting cancer chemotherapy response are not available. The phenotype test based on cell biological response usually requires long culture time, therefore is prone to contamination and difficult to standardize. Recently predicting chemotherapy response by microarray data has been reported, but also has a hard time to be applied in clinical practice since the facility standard requirement is usually way beyond the level which clinical laboratory could achieve. In current study, we attempted to develop a prediction model for individual response to doxorubicin chemotherapy in gastric cancer patients .【Objective】To establish a prediction model of doxorubicin response based on multiple gene markers to help the strategy decision in chemotherapy.【Methods】1. Gene expression profiles of SGC7901 and SGC-7901/ADR were analyzed using cDNA microarray technology ; 2. The EPG-257P and its doxorubicin resistant variant EPG-257RDB described by Lage and Gyorffy was obtained from the Stanford Microarray Database ; 3. The discriminatory genes were identified by compraring the microarray data of the two gastric cancer cell lines ; 4. The Prediction Analysis for Microarrays (PAM) training analysis was performed to get target genes for the following research; 5. Expression data of selected seven candidate genes were quantified by real-time RT-PCR in 20 gastric cancer specimens; 6. Doxorubicin sensitivity of the 20 gastric cancer specimens by HDRA technology ; 7. A Prediction model of doxorubicin response on gastric cancer specimens was established by multiple regression analysis between the expression data of selected seven candidate genes and doxorubicin response using NLReg software; 8. To confirm the prediction accuracy of the fixed formula, the prediction model was validated in validation set of 19 gastric cancer specimens in the same way using the same set of genes.【Results】1. By comparing gastric cancer cell line SGC7901 and its doxorubicin resistant variant SGC7901/ADR, 471 differentially expressed genes were revealed, in the other set of comparison (EPG-257P vs. EPG-257RDB), 10,144 differentially expressed genes were revealed. 2. To select discriminatory genes, we compared these two set of resistance-associated cell lines (SGC7901 vs. SGC-7901/ADR , EPG-257P vs. its doxorubicin resistant variant EPG-257RDB ) genes. Ninety sequences were found to have similar changes in both sets, all of these genes except PJA1 were upregulated in doxorubicin resistant cells. 3. The Prediction Analysis for Microarrays (PAM) training analysis was performed and 7 candidate genes which were correlated with doxorubicin response were seleceted including ADAM22, CYR61, IFITM1, FN1,SPHK1,G1P2 and GNAI1, all of these genes were upregulated in resistant cells. 4. Using expression data of selected seven candidate genes quantified by real-time RT-PCR and doxorubin sensitivity by HDRA, we did multiple regression analysis using NLReg software to compose prediction models for the in vitro activity of doxorubicin. IFITM1 and G1P2 were eliminated from the final formula since the estimated P for eachθp was much higher than those of the others. 5. Nineteen cases were subjected to real-time RT-PCR analysis to quantify the expression levels of 5 selected marker genes, and doxorubicin response was also tested using Histoculture Drug Response Assay. The results showed that the current prediction model reliably predicted the response of cancer specimens to doxorubicin (r = 0.7316); 6. The HDRA assay showed that the response rate of doxorubicin in gastric cancer was less than 30%.【Conclusions】1. A predicting model of doxorubicin sensitivity in gastric cancer is established using 5 genes ADAM22, CYR61, FN1,SPHK1 and GNAI1; the drug response prediction model was validated to be of practical usefulness to evaluate patients before chemotherapy; 2. Gastric cancer is very likely to be resistant to doxorubicin, therefore determining the existence of resistance will help to improve the outcome of chemotherapy involving doxorubicin; 3. In a whole genome screening of resistance related genes, pre-culture without drugs would help to eliminated the interference of activation of apoptosis, thus key regulators determining drug sensitivity might have bigger chance to be revealed.