摘要:Quantitative structure-activity relationships(QSAR)models of the antitumor activities(Aj:AC,ALand AH)and quantum chemical parameters Qt for Chinese hamster ovary(CHO),Mice leukemia cell(L1210),Human leukemic cell(HL60)are set up to study the antitumor mechanisms at the microscale of 15 s-triazole fused heterocycle derivatives of fluoroquinolone(FQTHs).The quantum chemical parameters(Qt,such as EHOMO,ELUMO,μ,QCd,QNe,QOf,QF and QS)of the above mentioned compounds are calculated using the MOPAC-AM1 method.The QSAR models of the antitumor activities for FQTHs are established using leaps-and-bounds regression.The correlation coefficients(R2)and leave-one-out(LOO) cross validation R2cv of the optimal three-parameter QSAR models are 0.910 and 0.827 for AC model,0.874 and 0.815 for AL model,0.941 and 0.894 for AH model,respectively.The QSAR models have both favorable robustness and good prediction capability by R2,R2adj,F,R2cv,VIF,AIC,FIT tests.The QSAR models mentioned above show that QCd,QNe,QOf and μ affect the antitumor activity directly.%为了探讨氟喹诺酮均三唑稠杂环类衍生物(FQTHs)体外抗肿瘤作用的微观机理,建立其对中国仓鼠卵巢细胞(CHO)、鼠白血病细胞(L1210)和人白血病细胞(HL60)等的体外抗肿瘤活性(Aj:AC、AL、AH)与量化参数Qt的定量结构-活性相关性(QSAR)模型.采用半经验计算法MOPAC-AM1计算15种化合物的量化参数(Qt,如EHOMO、ELUMO、μ、QCd、QNe、QOf、QF、QS等).运用最佳变量子集回归方法建立FQTHs体外抗肿瘤活性的QSAR模型.AC的最佳三元线性回归模型的判定系数R2和逐一剔除法交叉验证系数R2cv分别为0.910和0.827,AL对应的R2和R2cv分别为0.874和0.815,AH对应的R2和R2cv分别为0.941和0.894.通过R2、R2adj、F、R2cv、VIF、AIC、FIT等检验证明上述模型具有稳健性和预测能力.模型显示QCd、QNe、QOf和μ直接影响15种化合物的生物活性.