(M1330-04-22) Investigation of Molecular Interactions between Active Pharmaceutical Substances and Polymer Excipients Using 1H-NMR

Purpose: For formulation screening of poorly water-soluble drugs in amorphous solid dispersions (ASDs), it is crucial to investigate molecular interactions between the active substance and polymer excipients. A measure for interactions is the peak shift in 1H-NMR spectra of drug-polymer-mixtures compared to either one of the pure substances [1]. This technique is related to the determination of binding/dissociation constants known from ligand-receptor affinity analysis in drug discovery.

Methods: For the titration experiments, solutions of pure cinnarizine (CIN; 20 mM) and pure polymer (methacrylic acid – ethyl acrylate copolymer (1:1) (MAEA), basic butylated methacrylate copolymer (BBM), hypromellose acetate succinate (HPMCAS), hypromellose (HPMC); 120µM each) were prepared using methanol-d4 (deutero GmbH). A mixture of both solutions was created with concentrations of 10 mM and 60 µM for CIN and the polymer, respectively. After recording a 1H-NMR spectrum of the three solutions (pure CIN, pure polymer, 1:1-mixture by volume), the 1:1 mixture was added to the pure polymer solution in steps of 10-200 µL, each time recording a new spectrum. The polymer concentration was kept constant, whilst the drug concentration increased steadily. To verify the method used, experiments with well-characterised drugs (e.g. lidocaine) and serum albumin were performed according to the same setup. The results obtained from NMR-shift experiments and the calculated K_D values were compared to data from literature. The NMR-measurements were performed on a Bruker Ascend™ 400 MHz magnet. For each sample, 16 scans were collected at 298 K. Samples were analysed in 5 mm NMR tubes (Boro400-5-7, deutero GmbH). Evaluation was performed using TopSpin 4.1.4 and Microsoft Excel. To linearize the data, the three mathematical models of a Scott-plot, a Scatchard-plot and a Benesi-Hildebrand-plot were utilised [2]. The equations are shown in Figure 2.

Results: Figure 1 shows the peak shifts of protons in four different positions in the cinnarizine molecule with HPMCAS. For those positions, clear shift differences were detectable that could be plotted in a linear manner when applying the respective mathematical models (Figure 2). Depending on the polymer, the shift was observable for more protons (e.g. MAEA with additional shifts on position 15 and 16) or less protons (e.g., BBM with only minor changes that did not follow any of the applied models). Qualitative information could be obtained from the recorded spectra. If no peak shift difference was observable during the titration experiments, it was assumed that no interactions happen on this specific proton position. The number of influenced proton positions gave a first hint on drug-polymer interactions that correspond to prior expectations: with the basic drug cinnarizine, more interactions were observed for acidic polymers (MAEA, HPMCAS) than for neutral (HPMC) or basic (BBM) polymers. Additionally, proton positions within or near the piperazine moiety were more likely to be affected by interactions, whereas the signals of the aromatic regions were subject to little or no changes. Quantitative calculations were made based on the models known from drug-protein binding. The three models led to comparable results for K_D- and δ_max-values (Table 1). In contrast to protein binding studies, K_D may not be defined as dissociation constant for polymers, due to the larger number and high variance in possible binding sites. A more appropriate measure is δ_max, which implies the largest possible shift difference in case of a polymer excess and complete binding of the available ligand. Comparing δ_max, interactions between CIN and MAEA were most pronounced. For all included polymers, proton position 14 underwent most interactions.

Conclusion: The experiments showed a different shift behaviour of protons in cinnarizine in dependence of the polymer type. For distinctive shifts, a linearization with three different mathematical models was possible. By this, constants of K_D and δ_max could be calculated, from which δ_max seems more appropriate for the comparison of polymer types and proton positions. In this study, expectations on stronger interactions based on basic-acidic combinations were confirmed.

References: [1] A. Pugliese et al., “New Development in Understanding Drug-Polymer Interactions in Pharmaceutical Amorphous Solid Dispersions from Solid-State Nuclear Magnetic Resonance,” Molecular pharmaceutics, vol. 19, no. 11, pp. 3685–3699, 2022, doi: 10.1021/acs.molpharmaceut.2c00479.
[2] L. Fielding, “NMR methods for the determination of protein–ligand dissociation constants,” Progress in Nuclear Magnetic Resonance Spectroscopy, vol. 51, no. 4, pp. 219–242, 2007, doi: 10.1016/j.pnmrs.2007.04.001.

Figure 1: 1H-NMR spectra for cinnarizine and HPMCAS. Concentration-dependent peak shift of cinnarizine protons in mixtures with HPMCAS in methanol-d4. Coloured arrows indicate the position of the respective protons; grey arrows show the shift direction with increasing cinnarizine concentration.

Figure 2: Quantitative analysis of 1H-NMR spectra from drug-polymer titration experiments. Evaluation of peak-shifts shown in Figure 1 by means of a Scott-plot (left), a Scatchard-plot (middle) and a Benesi-Hildebrand-plot (right); shown are the graphs for the protons on atoms number ● =14, ♦ = 7, ▲ = 3 and 5, ■ = 2 and 6. Parameters in equation are: [L] = ligand concentration, δobs = observed peak shift difference, δmax = maximum possible shift difference, KD = dissociation constant.

Table 1: Results of KD- and δmax- calculations for three different polymers on various proton positions. Not all proton positions showed detectable peak shifts with the used polymers.