The chemistry of heterocyclic compounds comprises at least half of all researches in the field of organic chemistry and forms the basis of many pharmaceutical industries, veterinary products and agrochemicals [8] . In the last decade, as a result of a wide range of applications of heterocycles in the pharmaceutical and medicinal chemistry, the synthesis of these compounds has become a big target in synthetic organic chemistry [9] .

In recent years, with major advances in the synthesis of heterocyclic structures, the literature contains more than one class of biologically active compounds [10] . Among them, the nitrogenous heterocyclic 4-(3H)-qui- nazolinones and substituted quinazolines represent a very important class of drugs with several biological properties, such as anticancer [11] , diuretic [12] , anti-inflammatory [13] , anti-convulsant [14] and anti-hypertensive [15] activities.

The interest in the medicinal chemistry of quinazolinones derivatives was stimulated in the early 1950s with the elucidation of the structure of Febrifugina [16] , an alkaloid, which was effective against malaria. The methaqualone [17] was first synthesized in 1951 and is the best known quinazolinone derivative, famous for its hypnotic-sedative effects [18] . From these data, there has been a growing scientific interest in the fields of isolation, synthesis and pharmacological properties of compounds related to quinazolinones [18] .

Like quinazolinones, the quinoline pharmacophoric group is widely recognized in organic synthesis and can be found in a wide variety of compounds, such as 4-anilinoquinazoline derivatives with known biological properties. These compounds are reported in the literature as potent and selective inhibitors of tyrosine kinase pertaining to the epidermal growth factor (EGF) family of receptors [19] . In addition, knowledge of these enzymes inhibition process appears to be the path for the therapy of many diseases, such as cancer, “psoriasis” as diabetes, cardiovascular diseases, among others [20] . From this evidence, there have been more detailed studies on the biological function of a number of derivatives of this structural class [21] .

Numerous studies of structure-activity relationships (SAR) involving many series of quinazoline derivatives have led to advances in power, specificity and the pharmacokinetics properties of these inhibitors [22] . For instance, three drugs, Gefitnib (Iressa) [23] , Erlotinib (Tarceva) [24] and Lapatinib (Tykerb) [25] have been approved by the FDA and have been marketed for the treatment of lung cancer cells. In addition, several reversible and irreversible inhibitors of epidermal growth factor receptor inhibitors (EGFR) tyrosine kinase are currently being investigated [12] [26] . These small molecules mimic region of the ATP adenine and therefore are potent competitive inhibitors of ATP [27] .

Many of the tyrosine kinase enzymes which are early components of the growth signal transduction pathway in mammalian cells are encoded by proto-oncogenes, and their transformation or overexpression has been shown to occur in a large percentage of clinical cancers. These tyrosine kinase enzymes, especially the receptors for growth factors, such as EGF and platelet-derived growth factor (PDGF), have thus become important targets for drug design [11] . Previous evidence has shown the importance of correlating the pharmacokinetics parameters with the drug’s effectiveness. This study has the objective of investigating whether experimentally available quinazolines as EGFR inhibitors have good in silico pharmacokinetics parameters. A particular importance of this study is to reinforce that not always the most potent inhibitor is the one that presents the best phar-macoki- netics parameters and, therefore, is the most promising drug candidate.

Table 1 shows 69 quinazoline molecules studied to assess the in silico pharmacokinetics profiles. According to the calculations performed to obtain the parameters of the Lipinski’s rule of five, molecules 56, 57, 58, 68 and 69 violated the rule about Log P (>5). However, according to calculations, other molecules are prone to exhibit good oral bioavailability.

A drug should have good pharmacokinetics parameters, being absorbed acting as a potent inhibitor. The absorption includes the transference of the drug into the bloodstream [28] . In the past, many scientific studies in drug discovery were based on the synthesis of inhibitors and inhibition using in vitro tests to choose the best molecules and continuing the drug development. Such a methodology does not take into account the phar-ma- cokinetics properties of molecules and, therefore, many potent inhibitors were discovered, but not approved as drugs [28] . The modern methodology in drug development allows the rational design of new drug candidates by screening not only potent inhibitors, but also molecules with improved pharmacokinetics properties [28] . This work is intended to apply such an approach to develop potential drug candidates with lower risk to fail.

This work was based on 69 quinazoline compounds synthesized by Bridges [11] , whose ability to inhibit EGFR was described by IC₅₀. In that study, the most active compound was named 56 and then considered as the most promising EGFR inhibitor. Further studies for the development of EGFR inhibitors were then inspired on molecule 56. The present work demonstrates that compound 56, as well as 57, 58, 68 and 69 molecules, violated a key parameter of the Lipinski’s rule of five, the Log P, thus rising the chance of having problems with oral bioavailability [5] , i.e. the dose of the drug fraction that is found in the general circulation [29] . Preliminary computational studies can support the selection of compounds with prospective good bioavailability perfor- mance from a pool of molecules [28] . Molecules violating the rule for Log P can be checked in Table 1; con- sequently, these molecules may have poor absorption when administered orally.

Figure 1 demonstrates that molecule 56, which has the best experimental bioactivity, is not expected to show acceptable pharmacokinetic profile, such as low oral bioavailability, according to the low drug-likeness score obtained from the calculations using Molsoft [7] .

Figure 2 shows a series of molecules with Log P and pIC₅₀ values within an optimal range, i.e. molecules that

^*Log P > 5 is a violation in the Lipinski´s rule of five. ^#Molecule with the highest biological activity, represented by the lower IC₅₀

may have a high correlation between biological activity and lipophilicity. This is appropriate for the construction of a parabolic bilinear model.

The dataset in this study presents molecules with pIC₅₀ higher than 9.0 (IC₅₀ lower than 1 nM and, therefore, highly active [30] ), which are selected and colored in red Figure 2(a). Among these compounds, some can be identified as potentially having favorable pharmacokinetics properties by calculating the corresponding Log P. Kubinyi [4] have demonstrated an ideal range in Log P for selecting molecules, which was used to select some quinazolines in this study; molecules with calculated Log P values between 2.0 and 3.0 were selected, as shown in red Figure 2(b). These authors have also demonstrated that a bilinear model describes a correlation between bioactivity and lipophilicity of a series of similar molecules. This model indicates that there is optimal lipophilicity, which can reflect pharmacokinetic and pharmacodynamic requirements ideals, which increase or decrease can lead to progressive reduction of the biological activity. Based on this study, a similar model was built to correlate the biological activity data and Log P for the series of quinazoline EGFR inhibitors

Figure 3 shows the variation of Log P as a function of the biological activity (pIC₅₀). According to the bilinear parabolic correlation, six molecules have been identified as having the best profile by considering both biological activity and lipophilicity. Consequently, molecules 11, 19, 20, 43, 44 and 55 can be considered the best candidates obtained in our studies. This result excludes the molecule 56.

According to the proposed model, six molecules of Table 1, which are colored in red in the Figure 3 (11, 19, 20, 43, 44, 55) showed the best profile when analyzing the ideal values for both Log P and bioactivity. This result indicates that such molecules have high bioactivity and are probably favorable pharmacodynamic and pharmacokinetically. Thus, the previously proposed compound 56 does not match ideal requirements to be the best drug candidate from that pool of quinazoline EGFR inhibitors. In addition, there is no scientific evidence that the compound 56 has good in vivo pharmacokinetics parameters. Probably, the nonpolar O-ethyl and propyl groups in molecules 56 and 57 contribute to increase the inhibition of the receptor. However, these groups are responsible for increasing the Log P value, thus possibly causing reduction in bioavailability. Thus, incorporation of polar moieties, e.g. as terminal groups at the O-ethyl and propyl chains, could be attempt to improve the bioavailability without loosing the interaction with the receptor.

Figure 4 shows the pharmacokinetics results for the compounds proposed as EGFR inhibitors.

The result of this study confirms the necessity of previous computational modeling to select the most promising drug candidates. A potent inhibitor, which has poor absorption and, hence, low oral bioavailability, may not be easily accepted by the patients. So, it can be discarded by the pharmaceutical industry even at later stages of drug development.