Pharmacokinetic/pharmacodynamic Models Of Schedule-dependent Interaction Of Erlotinib And Gemcitabine

Erlotinib （ER） is the first targeted therapeutic agent to treat advanced-stageNSCLC in clinic. It is an oral active, small-molecule tyrosine kinase inhibitor （TKI）that reversibly binds to the intracellular tyrosine kinase domain of epidermal growthfactor receptor （EGFR）. It can block the autophosphorylation of EGFR followed withinhibiting a series of down-signaling pathways, and sequentially inhibit cellproliferation and angiogenesis. ER causes cells accumulated in G1 phase andconsequently induces tumor cells apoptosis. Gemcitabine （GM） is one of the first-linechemotherapeutic agents for the treatment of NSCLC as well as pancreatic cancer.With a similar structure to cytosine arabinoside, it can inhibit ribonucleotidereductaseand incorporate into the DNA, leading to chain termination ultimately. GM arrestscells mainly in S phase and induces apoptosis in a broad spectrum of solid tumors,including NSCLC.ER and GM can both be used in the treatment of NSCLC, offering considerablepotential to combine the two drugs and achieve better anti-tumor effects. However,some large, randomized phase III trials suggested that ER given concurrently withGM showed no survival benefit compared with GM monotherapy in treatingadvanced NSCLC. To find possible mechanisms of the failed combination andoptimize the schedule for the interaction of ER and GM, we investigated thepharmacological effect of different ER and GM combination schedules in vitro and invivo, and developed a pharmacokinetic/pharmacodynamic （PK/PD） model tocharacterize and quantify the anti-cancer effect of combination therapies using time-dependent data.Firstly, two NSCLC cell lines, H1299 and A549, with variable GM sensitivitywere exposed to six different ER combined with GM schedules. Cell growthinhibition, protein modulation in relative signaling pathway, and cell cycledistribution were analyzed to evaluate the pharmacological effects of differentschedules, and combination index was calculated to assess the interaction betweentwo drugs. In GM-sensitive cell line, H1299, strong synergism was observed whenER preceded GM, but other sequences showed antagonism. However, in GM-resistantcell line, A549, moderate synergism was observed when GM preceded ER orconcurrent treatment. Removing one drug before cells exposed to the other, whichindicated as Interval schedule, can reduce the antagonism in both cell lines tested anddemonstrated considerably better anti-tumor effect in H1299 compared with otherschedules. The changes of protein expression and cell cycle can support theseperformances.Based on the in vitro results, we further conducted a preclinical in vivo study tocompare the tumor inhibition and toxicity between ER monotherapy, GMmonotherapy, ER and GM used simultaneously, ER and GM used with interval inH1299 cells xenograft tumor model in female BALB/c nude mice. Compared with ERor GM monotherapy, Simultaneous group showed additive anti-tumor interaction inthe two drugs, but the tumor inhibition of Interval group was significantly better thanSimultaneous group after drugs administered for 12 days, which confirmed that theInterval schedule reduces the antagonism between the two drugs in GM-sensitiveH1299 cells and has potential to be tested in clinical trials. The tumor size-timeprofiles of different schedules were applied in our PK/PD model as thepharmacodynamic data. The body weight loss was not observed significantly in eachexperimental group.To obtain the pharmacokinetic data of different schedules, we developed a simple,rapid and sensitive LC-MS/MS method to quantify ER and its active metabolite,OSI-420, simultaneously in BALB/c nude mice plasma. ER, OSI-420, and propranolol, as internal standard, were extracted from nude mice plasma samplesusing liquid-liquid extraction. Separation was achieved on a reverse phase C18columnwith a mobile phase of acetonitrile-water （35:65, v/v） containing 5mM ammoniumformate （pH=3.0）. All compounds were monitored by mass spectrometry withelectrospray positive ionization. The lower quantitative limit was 0.5ng/ml for bothER and OSI-420; accuracy was estimated by relative error range from 0.07% to 8.00%for ER and -2.83% to 6.67% for OSI-420; precision was validated by relative standarddeviation values between 2.28% to 15.12% for ER, and 1.96% to 11.50% for OSI-420,respectively. This method was applied to a pharmacokinetic study of BALC/c nudemice following 12.5mg/kg ER oral administration. A 2-compartment model was usedto fit the pharmacokinetics of ER and 1-compartment model for the pharmacokineticsof OSI-420. The ratio of active metabolite to parent drug in mice was about 10%,which is greater than previously reported value, 5%, in human.In the pharmacokinetic study introduced previously, we obtained thepharmacokinetic parameters of ER. CL_ER/F, V_c,ER/F, Q_ER/F, V_p,ER/F, ka were19000ml/kg/day, 885ml/kg, 763ml/kg/day, 168ml/kg, 22.6day^-1, respectively. Thepharmacokinetic parameters of GM were extracted from literature. CL_GM, V_c,GM, Q_GM,V_p,GMwere 84451ml/kg/day, 887ml/kg, 53131ml/kg/day, 970ml/kg, respectively. Withthese PK parameters, we simulated the PK data according to differentpharmacodynamic schedules. A smooth nonlinear growth function was used tocharacterize the tumor growth curve of vehicle control group, and all the cells were ina proliferating compartment hypothetically. When either ER or GM was administered,proliferating cells become non-proliferating with a rate depending on the drugconcentration in plasma and the potency factor （k₂） of the drug. Atransit compartmentmodel was applied to represent the delayed tumor shrinkage relative to the timecourse of systemic exposure, the transit rate constant is k₁. To describe thepharmacological effect of combination therapy, an interaction termφwas introducedinto the semi-mechanistic anticancer PK/PD model, to reflect the nature and intensityof drug interaction. The k₂and k₁for GM was 0.924 ml·μg^-1·day^-1 and 1.96 day^-1, indicating that GM can stop tumor cells from proliferating to a great extent, but theefficacy has little hysteresis. The k2and k1for ER was 0.0716 ml·μg-1·day-1and 0.544day-1, suggesting that the anti-tumor effect of ER delayed for a relative longer time,and its natural potency to kill H1299 was less.φof Simultaneous treatment andInterval schedule were 0.961 and 1.82 respectively, which implied that interaction ofthe two drugs was additive in Simultaneous treatment but synergistic in the Intervalschedule. With the model parameters, we can simulate the tumor size-time profiles ofvarious dosage, longer drug duration or different drug scheme. The model can beapplied to standard xenograft experiments and may assist in the selection of bothoptimal drug combinations and administration schedules. The unique feature of thepresented approach is the ability to characterize the nature of combined druginteraction as well as to quantify the intensity of such interactions by assessing thetime course of combined drug effect.Both the experimental and modeling results suggested that in Interval schedule,the two drugs can achieve synergism in anti-tumor effect, with no significantlyobserved toxicity, which is superior to traditional Simultaneous treatment. TheInterval schedule has some potential to be tested in clinical trials.