Home > Research & Publications > atom connectivity dynamics analysis

Analysis of the relationship between lung cancer drug response level and atom connectivity dynamics based on trimmed Delaunay triangulation

Bin Zou, Debby D. Wang, Lichun Ma, Lijiang Chen, and Hong Yan


In this work, we investigated the relationship between the number of EGFR residues connected with gefitinib and the response level for each EGFR mutation type. Three-dimensional trimmed Delaunay triangulation was applied to construct connections between EGFR residues and gefitinib atoms. Through molecular Dynamics (MD) simulations, we discovered that when the number of EGFR residues connected with gefitinib increases, the response level of the corresponding EGFR mutation tends to descend, which means patients with the corresponding EGFR mutation are more likely to response well to gefitinib. The findings here can lead us a better understanding of dynamic features of EGFR mutant-gefitinib complexes, and establish a useful link between these features and patients’ responses to gefitinib. Our studies provide a useful reference for personalized NSCLC treatment plan.


TABLE I. Gefitinib response levels and mutant-gefitinib connectivity measure for 31 EGFR mutants, summarized from 137 patients with EGFR-mutated NSCLC. Specifically, 3 (II) represents that 3 patients have the response level of II to gefitinib. The connectivity measure corresponds to the specific number of mutant residues that are connected with gefitinib.

No. Mutation name Response level Connectivity Measure Comments
delE709_T710insD IV 16.354 1 (IV)
delE746_A750 II 21.019 1 (I), 26 (II), 2 (III), 1 (IV)
delE746_A750insAP II 18.519 1 (II)
delE746_S752insV II 18.801 1 (II)
delE746_T751insA III 16.743 1 (III)
delE746_T751insI II 19.679 1 (II)
delE746_T751insV I 21.79 1 (I)
delE746_T751insVA I 20.584 1 (I)
delL747_A750insP III 18.123 1 (II), 1 (IV)
delL747_A755insSKG II 20.829 1 (II)
delL747_K754insANKG IV 15.103 1 (IV)
delL747_P753insS II 17.729 7 (II), 1 (III)
delL747_T751 II 20.826 1 (II)
delT751_I759insN II 19.554 1 (II)
dulH773 IV 17.871 1 (IV)
dulN771_H773 IV 16.034 1 (IV)
dulS768_D770 III 18.089 1 (II), 1 (III), 1 (IV)
E709A_G719A IV 17.393 1 (IV)
E709K_L858R II 19.465 1 (II)
G719A_L858R II 19.373 1 (II)
G719A_L861Q II 17.887 1 (II)
G719C_S768I III 16.071 1 (III)
G724S_L861Q IV 19.503 1 (IV)
K757R IV 14.379 1 (IV)
L858R II 22.186 2 (I), 50 (II), 12 (III), 4 (IV)
L861Q II 18.542 2 (II)
L861R IV 18.225 1 (IV)
R776H_L858R II 22.35 1 (II)
R831H III 19.493 1 (III)
S768I_V774M III 18.075 1 (III)
L858R_T790M IV 14.464

Figure 1. Comparison of connectivity for different EGFR mutation pairs: (a) L858R and L858R_T790M, (b) delE746_A750 and L858R_T790M. We can see that L858R and delE746_A750 have more residues connected with gefitinib than L858R_T790M does, among the 1000 frames. It is widely acknowledged that these EGFR mutations respond well to gefitinib while L858R_T790M causes strong gefitinib-resistance in most cases, which are consistent with our results.

Figure 2. Correlation analysis of EGFR mutations and the corresponding average connectivity measure. The horizontal axis is the mutation index with their drug response level ascending from I to IV. Within each response group, we sorted their average number of residues connected with gefitinib in a descending order. We can see that the relation is approximately linear. The average value (shown as stars) of each group further verifies our conclusion. If we define response level IV representing a drug-resistance case while response levels I to III a non-resistance case, we can set a two-section boundary that divides EGFR mutations into drug-resistance group and drug-control group, and we can get a very low error rate of 4/31.


Figure 3. The methods. We collected 30 EGFR TK mutation types from 137 EGFR-mutation induced NSCLC patients (Lee, 2013). For all these patients, gefitinib was used in their treatments and their drug response levels were recorded. Moreover, L858R_T790M, a well-acknowledged gefitinib resistance mutation, was also added into our dataset and we set the corresponding response level as IV. Crystal structures of the 31 EGFR mutants modeled in (Wang, 2013) were adopted in our studies. Then we performed Molecular Dynamics (MD) Simulations and collected a trajectory of 1000 position frames for each EGFR mutant. 3D trimmed Delaunay triangulation was applied to calculate the connectivity between EGFR residues and gefitinib for each trajectory frame. At last, we correlated the connectivity between EGFR residues and gefitinib to lung cancer drug response level, if more EGFR residues are connected to the drug molecule, the binding affinity between the protein and the drug will likely be higher.