CN103077328B - The application of the drug efficacy prediction model of fluoroquinolones - Google Patents

Abstract

An application for the drug efficacy prediction model of fluoroquinolones, comprising: use Accelrys? Discovery? Studio software calculates the molecular descriptor of new fluoquinolone; According to molecular descriptor select the special medicine of 4 bacteriums dynamic-pharmacodynamics index forecast model; Select the quantitative structure activity relationship model that 1 bacterium is general; Select 1 drug effect-structure prediction model; Need the new fluoquinolone of prediction whether within the estimation range 3*p/n of lever value with lever value test and judge; If new fluoquinolone is within the estimation range 3*p/n of lever value, substitutes into model with molecular descriptor and predict; Initial for difference body temperature, dosage are substituted in drug effect-structure prediction model and predicts the trend of body temperature with the change for the treatment of number of days.The present invention combines with structure, can be used in the research and development such as drug screening or the structure optimization initial stage, can provide reference for clinical dosing regimen, and the drug effect information of a large amount of flouroquinolone drugs developed is applied.

Description

The application of the drug efficacy prediction model of fluoroquinolones

Technical field

The present invention relates to a kind of drug efficacy prediction model.Particularly relate to a kind of application of drug efficacy prediction model of fluoroquinolones.

Background technology

Fluoroquinolones is the third-largest class antibacterials, has the advantages that activity is comparatively strong, has a broad antifungal spectrum, bioavilability are high.In order to shorten R&D cycle, cost-saving, the medicament research and development pattern based on model runs through the R&D process of current fluoroquinolones.

The fluoroquinolones method conventional when researching and developing the screening at initial stage or structure optimization is quantitative structure activity relationship model.Quantitative Structure-Activity Relationship Study be correlationship between the structural information of compound and activity, and by the activity of structural information predictive compound.Kumar etc. establish fluoquinolone containing C-7 piperazinyl for the quantitative structure activity relationship model of staphylococcus aureus MTCC1430, hay bacillus MTCC2423, Escherichia coli MTCC739.Abdel-Aziz etc. with the fluoquinolone established containing arenesulfonyl base for the quantitative structure activity relationship model of staphylococcus aureus ATCC29213, hay bacillus ATCC10400, Escherichia coli ATCC25922, pseudomonas aeruginosa ATCC27853.The index major part of this type of research activity is log (1/MIC), only considered the antibacterial action of medicine, does not consider medicine pharmacokinetics in vivo, does not more consider drug effect in its body.

In clinical testing process, pharmacophore model is through being commonly used to find optimal dosage, drug effect between comparison therapy scheme, the threshold value etc. selecting dosage regimen, find fAUC/MIC.The people such as Stergiopoulou investigate the interaction between dexycholate amphotericin B, Ciprofloxacin, human neutrophil with sigmoidEmax model.The people such as Andraud are the Escherichia coli bactericidal effect in time of Marbofloxacin alignment degree resistance with time-kill curve prediction.These pharmacophore models contain much information, reference can be provided for clinical dosing regimen, but do not combine with structure, which structure influence clinical efficacy cannot be described, therefore cannot be used in the research and development such as drug screening or the structure optimization initial stage, the drug effect information of a large amount of flouroquinolone drugs developed cannot be applied.

Summary of the invention

Technical matters to be solved by this invention is, provides a kind of fAUC that can predict for staphylococcus aureus ATCC29213, enterococcus faecalis ATCC29212, Escherichia coli ATCC25922, pseudomonas aeruginosa ATCC27853 ₂₄the fAUC of/MIC and average level ₂₄/ MIC and the application of the drug efficacy prediction model of the fluoroquinolones of temperature decline trend after taking fluoroquinolones.

The technical solution adopted in the present invention is: a kind of application of drug efficacy prediction model of fluoroquinolones, comprises the steps:

1) molecular descriptor of new fluoquinolone is calculated with AccelrysDiscoveryStudio software;

2) according to molecular descriptor select the special medicine of 4 bacteriums dynamic-pharmacodynamics index forecast model; Select the quantitative structure activity relationship model that 1 bacterium is general; Select 1 drug effect-structure prediction model;

3) need the new fluoquinolone of prediction whether within the estimation range 3*p/n of lever value with lever value test and judge;

4) if new fluoquinolone is within the estimation range 3*p/n of lever value, then use step 1) in molecular descriptor substitute into step 2) in the special medicine of 4 bacteriums move-pharmacodynamics index forecast model and the general quantitative structure activity relationship model prediction staphylococcus aureus ATCC29213 of bacterium, enterococcus faecalis ATCC29212, Escherichia coli ATCC25922, the fAUC24/MIC of pseudomonas aeruginosa ATCC27853 and the fAUC24/MIC of average level;

5) by initial for difference body temperature, dosage substitute into step 2) in drug effect-structure prediction model in predict body temperature with treatment number of days change trend.

Step 1) described in molecular descriptor comprise CHI_3_CH, CHI_V_0, IAC_mean, Zagreb, Num_Rings6, LogD, E_ADJ_mag, CHI_3_P and Num_H_Donors.

Step 2) described in the special medicine of 4 bacteriums dynamic-pharmacodynamics index forecast model is:

Staphylococcus aureus ATCC29213:Log ₁₀fAUC ₂₄/ MIC=-11.476+2.317*CHI_3_CH+0.396*CHI_V_0+3.928*IAC_Mean;

Enterococcus faecalis ATCC29212:Log ₁₀fAUC ₂₄/ MIC=-4.061+0.0327*Zagreb+1.492*<2.590-Num_R ings6>;

Escherichia coli ATCC25922:fAUC ₂₄/ MIC=-2564.696+1367.216*CHI_3_CH+1416.158*IAC_Mean+14.623 * Num_Rings6 ²+ 24.301*LogD;

Pseudomonas aeruginosa ATCC27853:Log ₁₀fAUC ₂₄/ MIC=0.213-1.489*CHI_3_P+0.0198*E_ADJ_mag+2.729*IAC_Mean-0.411*Num_H_Donors.

Step 2) described in the general quantitative structure activity relationship model of bacterium be Log ₁₀fAUC ₂₄/ MIC=-0.356*Num_H_Donors – 0.312*Num_Rings6+0.00347*E_ADJ_mag+1.331*CHI_3_CH.

Step 2) described in drug effect-structure prediction model be TEMP=36.739+ (A0-36.739) * exp (-(0.0261*10 ^{(-0.356*Num_H_Donors – 0.312*Num_Rings6+0.00347*E_ADJ_mag+1.331*CHI_3_CH)}* Dose*DAY).

Step 3) described in lever value be with ARTE-QSAR software or calculate with hat matrix.

The application of the drug efficacy prediction model of fluoroquinolones of the present invention, combine with structure, which structure influence clinical efficacy can be described, therefore, it is possible to be used in the research and development such as drug screening or the structure optimization initial stage, reference can be provided for clinical dosing regimen, the drug effect information of a large amount of flouroquinolone drugs developed is applied.

The predictive ability evaluation of model 1-4 of the present invention is better, and R2, adjustment R2 are more than 0.9, and the checking of inside and outside portion is all greater than 0.5, and wherein internal verification have employed the test set of scalping method, external certificate one by one uses 25% of sample when being Modling model.Result is as table 2.

Table 2 model prediction merit rating (staphylococcus aureus ATCC29213)

The edge R of model 5 ² _wbe 0.66, retrain R ² _w0.88.R ² _wcalculated by Wald inspection institute.

Accompanying drawing explanation

Fig. 1 (a) is the residual error histogram of staphylococcus aureus ATCC29213 forecast model;

Fig. 1 (b) is the residual error P-P figure of staphylococcus aureus ATCC29213 forecast model;

Fig. 1 (c) is the residual-normal expected value figure of staphylococcus aureus ATCC29213 forecast model;

Fig. 2 (a) is the residual error histogram of enterococcus faecalis ATCC29212 forecast model;

Fig. 2 (b) is the residual error P-P figure of enterococcus faecalis ATCC29212 forecast model;

Fig. 2 (c) is the residual-normal expected value figure of enterococcus faecalis ATCC29212 forecast model;

Fig. 3 (a) is the residual error histogram of Escherichia coli ATCC25922 forecast model;

Fig. 3 (b) is the residual error P-P figure of Escherichia coli ATCC25922 forecast model;

Fig. 3 (c) is the residual-normal expected value figure of Escherichia coli ATCC25922 forecast model;

Fig. 4 (a) is the residual error histogram of pseudomonas aeruginosa ATCC27853 forecast model;

Fig. 4 (b) is the residual error P-P figure of pseudomonas aeruginosa ATCC27853 forecast model;

Fig. 4 (c) is the residual-normal expected value figure of pseudomonas aeruginosa ATCC27853 forecast model;

Fig. 5 is the residual error histogram of the general quantitative structure activity relationship model of bacterium;

Fig. 6 is drug effect-structure prediction model criteria residual error-prognostic chart.

Embodiment

Below in conjunction with embodiment and accompanying drawing, the application to the drug efficacy prediction model of fluoroquinolones of the present invention is described in detail.

The application of the drug efficacy prediction model of fluoroquinolones of the present invention, comprises the steps:

1) molecular descriptor of new fluoquinolone is calculated with AccelrysDiscoveryStudio software;

Described molecular descriptor comprises CHI_3_CH, CHI_V_0, IAC_mean, Zagreb, Num_Rings6, LogD, E_ADJ_mag, CHI_3_P and Num_H_Donors.

With AccelrysDiscoveryStudio software or CHI_3_CH, CHI_V_0, IAC_mean, Zagreb, Num_Rings6, LogD, E_ADJ_mag, CHI_3_P, Num_H_Donors equimolecular descriptor with the new fluoquinolone of following formulae discovery.Wherein CHI_3_CH is calculated the P (chain_3) of each three-membered ring by formula (2-5), and wherein chain_3 represents 3 former molecular rings, δ _vifor the quantivalency of atom, formula (2-6) is used to calculate all three-membered ring P (chain_3) sums afterwards.CHI_V_0 is calculated by formula (2-7), formula (2-8), first uses formula (2-8) to calculate the δ of each atom ^v, wherein Z ^vbe valence electron, Z is atomic number, and h is connected numbers of hydrogen atoms, then uses formula (2-7) by the δ of all atoms ^vadd and.The expression formula of IAC is formula (2-9), wherein A ^hwhat represent is all atom numbers (comprising hydrogen atom), A _gbelong to the atom number of same atomic number.Zagreb is calculated by formula (2-10), δ _ifor summit quantivalency.Num_Rings6 represents the number of hexatomic ring.It is lipid (octanol/water) under 7.4 that LogD refers to pH value.The computing formula of E_ADJ_mag is as being formula (2-11), wherein A _efor the number that key is connected with key.The computing formula of CHI_3_P is as formula (2-12) and (2-13), in its Middle molecule, three keys are linked to be a paths and (can not form ring, can not three keys from a summit) be called one group of subgroup namely in formula, δ _virefer to the quantivalency on each summit in each group, owing to being that three keys are connected, therefore each group has 4 summits, and χ is the value of CHI_3_P.

$P(chain\_3)=\underset{i=1}{\overset{3}{\&Pi;}}1/\sqrt{{\mathrm{\&delta;}}_{{\mathrm{\&nu;}}_i}}---(2-5)$

χ _{CHI_3_CH}＝∑P(chain_3)(2-6)

δ ^v＝(Z ^v-h)/(Z-Z ^v-1)(2-8)

${\stackrel{\&OverBar;}{I}}_{AC}=-{\mathrm{\&Sigma;}}_g\frac{A_g}{A^h}\&CenterDot;{\log }_2\frac{A_g}{A^h}---(2-9)$

$Zagreb=\underset{i}{\&Sigma;}{\mathrm{\&delta;}}_i^2---(2-10)$

${}^E{I}_{adj}^M=A_E\&CenterDot;{\log }_2{A}_E---(2-11)$

$P\left(subgroup\right)=\underset{i=1}{\overset{4}{\&Pi;}}1/\sqrt{{\mathrm{\&delta;}}_{v_i}}---(2-12)$

χ＝∑P(subgroup)(2-13)

2) according to molecular descriptor select from the drug efficacy prediction model of fluoroquinolones the special medicine of 4 bacteriums dynamic-pharmacodynamics index forecast model; Select the quantitative structure activity relationship model that 1 bacterium is general; Select 1 drug effect-structure prediction model;

The drug efficacy prediction model of described fluoroquinolones is referring specifically to academic dissertation-Li Yan. the drug efficacy prediction model [D] of fluoroquinolones. and University Of Tianjin, 2012.

The special medicine of 4 described bacteriums is dynamic-and pharmacodynamics index forecast model is (1 ~ No. 4 model in table 1):

Staphylococcus aureus ATCC29213:Log ₁₀fAUC ₂₄/ MIC=-11.476+2.317*CHI_3_CH+0.396*CHI_V_0+3.928*IAC_Mean;

Enterococcus faecalis ATCC29212:Log ₁₀fAUC ₂₄/ MIC=-4.061+0.0327*Zagreb+1.492*<2.590-Num_R ings6>;

Escherichia coli ATCC25922:fAUC ₂₄/ MIC=-2564.696+1367.216*CHI_3_CH+1416.158*IAC_Mean+14.623 * Num_Rings6 ²+ 24.301*LogD;

Pseudomonas aeruginosa ATCC27853:Log ₁₀fAUC ₂₄/ MIC=0.213-1.489*CHI_3_P+0.0198*E_ADJ_mag+2.729*IAC_Mean-0.411*Num_H_Donors.

The general quantitative structure activity relationship model of described bacterium is (No. 5 models in table 1) Log ₁₀fAUC ₂₄/ MIC=-0.356*Num_H_Donors – 0.312*Num_Rings6+0.00347*E_ADJ_mag+1.331*CHI_3_CH.

Described drug effect-structure prediction model is (No. 6 models in table 1) TEMP=36.739+ (A0-36.739) * exp (-(0.0261*10 ^{(-0.356*Num_H_Donors – 0.312*Num_Rings6+0.00347*E_ADJ_mag+1.331*CHI_3_CH)}* Dose*DAY).

Table 1 model expression

3) need the new fluoquinolone of prediction whether within the estimation range 3*p/n of lever value with lever value test and judge;

Described lever value is with ARTE-QSAR software or calculates with hat matrix.

Lever value is calculated with ARTE-QSAR software or with hat matrix.By the molecular descriptor of new fluoquinolone together with the molecular descriptor information of fluoquinolone in table 3 in Input Software, or one reinstates hat matrix and calculates, if the H of new fluoquinolone _iivalue is greater than 3*p/n and then cannot predicts with the present invention.P is that to add 1, n be number of samples to the variable number in model, equals 13 in the present invention.Whether belong in model scope it is noted herein that model 1-5 needs to investigate new fluoquinolone respectively.

H _ii＝X _i’(X’X) ^-1x _i(2-14)

The molecular descriptor of table 312 kind of fluoroquinolones

4) if new fluoquinolone is within the estimation range 3*p/n of lever value, then use step 1) in molecular descriptor substitute into step 2) in the special medicine of 4 bacteriums move-pharmacodynamics index forecast model and the general quantitative structure activity relationship model prediction staphylococcus aureus ATCC29213 of bacterium, enterococcus faecalis ATCC29212, Escherichia coli ATCC25922, pseudomonas aeruginosa ATCC27853 fAUC ₂₄the fAUC of/MIC and average level ₂₄/ MIC.

5) by initial for difference body temperature, dosage substitute into step 1) in drug effect-structure prediction model in can predict body temperature with treatment number of days change approximate trend.

The special medicine of what Fig. 1 (a) ~ Fig. 4 (c) showed is 4 bacteriums moves-the residual error histogram of pharmacodynamics index forecast model, residual error P-P schemes, residual-normal expected value figure, the medicine that 4 bacteriums are special as can be seen from Figure moves-and the residual error of pharmacodynamics index forecast model is normal distribution and do not occur Singular variance phenomenon.Fig. 5 shows the residual error histogram of the general quantitative structure activity relationship model of bacterium, can find out that residual error is normal distribution in figure.Fig. 6 shows the model criteria residual error-prognostic chart of drug effect-structure prediction model, and residual error is uniformly distributed about 0.Can find out that model meets model hypothesis by Fig. 1 ~ 6, therefore model is reliable.

Claims (1)

1. an application for the drug efficacy prediction model of fluoroquinolones, is characterized in that, comprise the steps:

1) calculate the molecular descriptor of new fluoquinolone with AccelrysDiscoveryStudio software, described molecular descriptor comprises CHI_3_CH, CHI_V_0, IAC_mean, Zagreb, Num_Rings6, LogD, E_ADJ_mag, CHI_3_P and Num_H_Donors;

2) according to molecular descriptor select the special medicine of 4 bacteriums dynamic-pharmacodynamics index forecast model; Select the quantitative structure activity relationship model that 1 bacterium is general; Select 1 drug effect-structure prediction model, wherein,

The special medicine of 4 described bacteriums is dynamic-and pharmacodynamics index forecast model is:

Staphylococcus aureus ATCC29213:Log ₁₀fAUC ₂₄/ MIC=-11.476+2.317*CHI_3_CH+0.396*CHI_V_0+3.928*IAC_Mean;

Enterococcus faecalis ATCC29212:Log ₁₀fAUC ₂₄/ MIC=-4.061+0.0327*Zagreb+1.492*<2.590-Num_R ings6>;

Escherichia coli ATCC25922:fAUC ₂₄/ MIC=-2564.696+1367.216*CHI_3_CH+1416.158*IAC_Mean+14.623 * Num_Rings6 ²+ 24.301*LogD;

Pseudomonas aeruginosa ATCC27853:Log ₁₀fAUC ₂₄/ MIC=0.213-1.489*CHI_3_P+0.0198*E_ADJ_mag+2.729*IAC_Mean-0.411*Num_H_Donors;

The general quantitative structure activity relationship model of described bacterium is Log ₁₀fAUC ₂₄/ MIC=-0.356*Num_H_Donors – 0.312*Num_Rings6+0.00347*E_ADJ_mag+1.331*CHI_3_CH;

Described drug effect-structure prediction model is TEMP=36.739+ (A0-36.739) * exp (-(0.0261*10 ^{(-0.356* num_H_Donors – 0.312*Num_Rings6+0.00347*E_ADJ_mag+1W331*CHI_3_CH)}* Dose*DAY), wherein, initial body temperature when A0 represents that treatment starts, Dose represents dosage, and unit is that 100mg, DAY represent treatment number of days, and TEMP is body temperature;

3) need the new fluoquinolone of prediction whether within the estimation range 3*p/n of lever value with lever value test and judge, wherein, p is that in model, variable number adds 1, n is number of samples, and described lever value is with ARTE-QSAR software or calculates with hat matrix;

4) if new fluoquinolone is within the estimation range 3*p/n of lever value, then use step 1) in molecular descriptor substitute into step 2) in the special medicine of 4 bacteriums move-pharmacodynamics index forecast model and the general quantitative structure activity relationship model prediction staphylococcus aureus ATCC29213 of bacterium, enterococcus faecalis ATCC29212, Escherichia coli ATCC25922, the fAUC24/MIC of pseudomonas aeruginosa ATCC27853 and the fAUC24/MIC of average level;

5) by initial for difference body temperature, dosage substitute into step 2) in drug effect-structure prediction model in predict body temperature with treatment number of days change trend.