4-Methoxyphenylboronic acid (4-MPBA) was purchased from Thermo Fisher Scientific (Cat. No. 30948, ≥ 96%). 4-Methoxyphenol (Phenol, Cat. No. M18655, 99.8% by GC), Anisole (Des-Br, Cat. No. 296295, 99.7% by GC), 4-Bromoanisole (Bromo, Cat. No. A11824, 98.8% by GC), and 4, 4’-Dimethoxybiphenyl (BisPh, Cat. No. 148539, 99.5% by GC), were purchased from Sigma-Aldrich. The potential impurities of compound 4-MPBA are illustrated in Figure 2, where Bromo is the leftover starting material, Phenol is the oxidation impurity, and Des-Br and BisPh are typical processing impurities.

1,3,5-Trimethoxybenzene (74599-1G, Lot BCCG8790, Potency 0.9999, qNMR reference standard), Dimethyl sulfoxide-d6 (DMSO-d6, Cat. No. 320760075, 100.0 atom %D, for NMR solvent), and Deuterium oxide (D₂O, 99.9 atom % D) were purchased from Sigma-Aldrich.

Mobile phase components were HPLC grade and purchased from Fisher Chemical or Sigma-Aldrich: Acetonitrile (ACN, HPLC grade, Fisher Scientific A998-4), Methanol (MeOH, HPLC grade, Fisher Scientific A452SK-4), Trifluoroacetic acid (TFA, HPLC grade, Sigma Aldrich 302031), O-Phosphoric Acid (PA, 85% HPLC Grade, Fisher A260), Formic Acid (FA, LC-MS grade, Thermo Scientific 28905), Ammonium Formate (AF, Optima LCMS grade, Fisher Chemical A115-50), Ammonium Acetate (AAc, ≥99.99% trace metals basis, Sigma Aldrich 431311-50G), Acetic acid (AA, glacial (HPLC) grade, Fisher Chemical A35-500), and Ammonium hydroxide (NH4OH, ≥99.99% trace metals basis, Sigma Aldrich 338818-1L).

Figure 2. Structure of 4-MPBA and its potential impurities.

The analyses are performed on a Agilent Infinity II 1260/1290 system with a UV detector (wavelength 223 nm), using a reverse phase (RP) HPLC column (Ascentis® Express 90 Å ES-Cyano, 2.7 µm, 150 mm × 4.6 mm) at 35˚C, with 0.025% TFA in both mobile phases A (water) and B (methanol), applying a simple gradient (0 – 1 min, 14%B; 1 – 22 min, 14% - 80%B; 22 – 25 min, 80%B; 25 - 25.1 min, 80% - 14%B; 25.1 - 30min, 14%B), with an injection volume of 10 µL, and at a flow rate of 0.8 mL/min. Refer to Table S1 in the supplementary material for detailed HPLC conditions. A working standard solution (WSS) was a mixture of 0.1 mg/mL of each of two standards, 4-MPBA and Bromo. The purity sample was prepared at 1.0 mg/mL in diluent (50% aqueous acetonitrile). The assay sample was prepared by 10-fold dilution from the purity sample. The resolution solution was prepared by spiking 2% w/w of each potential impurity into 1.0 mg/mL of 4-MPBA. Limit of quantitation (LOQ) solution was prepared by serial dilution (200-fold) from the WSS. To ensure consistent results, all HPLC samples were allowed to equilibrate in the aqueous diluent for approximately 30 seconds prior to injection.

The assay (% w/w) of sample can be calculated from Equation (1). Similarly individual impurity (% w/w) of sample can be calculated from Equation (2).

Where: Asam is the peak area of 4-MPBA in the assay sample. Astd is mean peak area (n = 6) of the Bromo in the first 6 injections of the WSS. Cstd is nominal concentration of the Bromo in the WSS (e.g., 0.1 mg/mL). Cassay is nominal concentration of the assay sample solution (e.g., 0.1 mg/mL). P is the potency of the Bromo reference standard (e.g., 0.988). RRF is the Relative response factor of 4-MPBA relative to the Bromo standard, calculated by dividing the slope of the 4-MPBA linear curve (0.05 - 0.15 mg/mL; 50% - 150% of 0.1 mg/mL) by the slope of the Bromo linear curve over the same range. Asam(i) is the peak area of individual impurity in the purity sample. Cpurity is nominal concentration of the purity sample solution (e.g., 1.0 mg/mL). RRF(i) is the relative response factor of each impurity relative to the Bromo standard, calculated by dividing the slope of the linear curve for each impurity (0.0005 - 0.02 mg/mL; 0.05% - 2.0% of 1.0 mg/mL) by the slope of the Bromo linear curve (0.05 - 0.15 mg/mL; 50% - 150% of 0.1 mg/mL).

It was challenging to separate the phenol impurity peak from the 4-MPBA peak. Initially, a Waters XBridge Shield RP18 column (4.6 × 150 mm, 3.5 µm)—selected for its embedded polar group (carbamate) to assist with phenolic compounds—and various acid/base modifiers (TFA, PA, FA, NH4OH) were used to analyze 1.0 mg/mL 4-MPBA resolution solutions, both with and without a 2% w/w phenol spike. Although baseline separation between phenol and 4-MPBA was not fully achieved, TFA and phosphoric acid (0.05% v/v) in both mobile phases A (water) and B (ACN) performed better than formic acid; see Figure 3.

Figure 3.Different acid (0.05%) in mobile phases A and B.

Figure 4.Chromatograms of two resolution solutions.

Baseline separation between the Phenol impurity and 4-MPBA was achieved using 0.05% v/v TFA in methanol as mobile phase B (MPB), as shown in Figure 4(a). However, resolution was compromised upon the addition of acetonitrile to MPB (0.05% TFA in a 1:1 methanol/acetonitrile mixture; Figure 4(b)). Furthermore, the 4-MPBA peak exhibited significant broadening and a decrease in retention when 0.05% ammonium hydroxide was utilized in both MPA and MPB (Figure 4(c)).

Figure 5.Chromatograms with different columns (PDA wavelength 225 nm, peaks from left to right are Phenol, 4-MPBA, Des-Br, Bromo).

Table 1. Peak tailing with different HPLC columns.

Inspired by Waters application notes [37][38], ten columns were screened by analyzing a mixture containing 0.2 mg/mL each of phenol, 4-MPBA, Des-Br, and Bromo, while maintaining consistent mobile phase components (0.05% TFA in water as MPA and 0.05% TFA in MeOH as MPB). All analytes were not only well retained but also well resolved (as shown in the chromatograms in Figure 5). Based on the peak tailing results listed in Table 1, the Supelco Ascentis Express 90 Å ES-CN provided the best peak shape for 4-MPBA. The cyano groups on the ES-CN column offer unique interactions that resolve tailing issues with polar compounds, such as boronic acids, more effectively than standard RP phases.

Figure 6. UV Spectra of Phenol, 4-MPBA, Des-Br, Bromo, and BisPh.

Figure 7. Chromatograms of 0.1 mg/mL mixture of Phenol, 4-MPBA, Des-Br, Bromo, and BisPh at different wavelengths.

Based on the UV spectra of Phenol, 4-MPBA, Des-Br, Bromo, and BisPh (Figure 6), several wavelengths (220 nm, 223 nm, 225 nm, and 228 nm) were evaluated. The wavelength 223 nm (Figure 7(b)) was selected, because all the resulting peak heights are comparable to that of 4-MPBA, see Figure 7 for the chromatograms at each tested wavelength.

Since BisPh is insoluble in methanol but soluble in acetonitrile, 50%-100% acetonitrile in water (with or without 0.05% TFA) were used as diluents. Figure 8presents the chromatograms of the resolution solution (1.0 mg/mL of 4-MPBA spiked with 2% of Phenol, Des-Br, Bromo, and BisPh). While 100% Acetonitrile resulted in poor peak shapes (Figure 8(a)), both 50% acetonitrile in water (Figure 8(b)) and 0.05% TFA in 50% acetonitrile (Figure 8(c)) yielded excellent peak shapes.

Figure 8.Chromatograms of resolution solution (2% of Phenol, Des-Br, Bromo, and BisPh in 1.0 mg/mL 4-MPBA with different diluents).

The robustness of the method was assessed by varying several parameters, including flow rate, injection volume, initial %B, column temperature, and %TFA in mobile phases (see Table 2). Under all tested conditions, the peaks remained well-resolved with comparable peak shape. Chromatograms of the resolution solution and the 0.1 mg/mL mixture solution of five compounds are provided in the supplementary material (see Figures S6-S15).

Table 2. Robustness.

Table 3. USP resolution of peaks in resolution solution.

Note:^a: An unknown/unspecified impurity.

Figure 9.Chromatograms of resolution solution (a) and WSS (b).

No interference peaks were observed at the retention time of Phenol, 4-MPBA, Des-Br, Bromo, and BisPh in the diluent injections. The USP resolution of each peak in resolution solution (containing a 2% spike of Phenol, 4-MPBA, Des-Br, Bromo, and BisPh in 1.0 mg/mL 4-MPBA) was greater than 2.0 (see Table 3). A typical chromatogram of resolution solution is shown in Figure 9(a), and a typical chromatogram of WSS is shown in Figure 9(b). No carryover was detected in the blank injections.

The LOD was determined to be 0.00025 mg/mL for each of the 5 components, representing 0.025% of nominal concentration of the purity sample (1.0 mg/mL); LOQ was 0.0005 mg/mL for each component, or 0.05% of nominal purity sample concentration. The signal-to-noise ratio (S/N) of each component across three LOD injections was more than 3 (see Table S-2 in the supplementary material). For the six LOQ injections, the S/N of each component was greater than 10 (see Table S-3 in the supplementary material), and the %RSD of each component was less than 6.0% (Table S-3). Figure 10 presents representative chromatograms of the diluent, LOD, and LOQ.

Figure 10.Chromatograms of diluent (a), LOD (b), and LOQ (c).

The Identification (ID) was verified by the relative retention time (RRT), calculated by dividing the retention time (RT) of 4-MPBA peak in the assay samples by the mean RT (n = 6) of 4-MPBA peak in the WSS. The RRT of 4-MPBA cross all assay sample injections was within the acceptance criteria (0.98 - 1.02, see Table 4).

Table 4. Identification.

The linearity of all the potential impurities (Phenol, Des-Br, Bromo, and BisPh) was verified over the low concentration range (0.05% (LOQ) to 2.0% of 1.0 mg/mL), with correlation coefficient (R²) better than 0.995 (see Table 5). Linearity of 4-MPBA and Bromo was also confirmed over the assay concentration range (50% to 150% of 0.1 mg/mL), with R² better than 0.998 (Table 6). The relative response factor (RRF) of each impurity was calculated by dividing its slope (low concentration range) by the slope of the Bromo (assay concentration range). The RRF of 4-MPBA is calculated by dividing its assay-range slope by the Bromo assay-range slope (see Table 5 and Table 6). The RRF of Bromo is defined as 1.00.

Table 5. Linearity of impurities (0.05% (LOQ) – 2.0% of 1.0 mg/mL).

Table 6. Linearity of 4-MPBA and Bromo (50% – 150% of 0.1 mg/mL).

The accuracy and precision for assay quantitation of 4-MPBA were verified at three levels (50%, 100%, 150%). Recovery of each replicate at all three levels was within 98.0% - 102.0%, with averages of 99.2%, 98.6%, 98.2%, respectively; the %RSD was less than 1.0% at all levels (see Table S5 in the supplementary material). Intermediate precision for the assay was verified by two analysts using two HPLC systems (Agilent Infinity II 1290 and 1260) over multiple days, yielding a % RSD of 1.1% (n = 12, see Table S6 in the supplementary material). Similarly, the accuracy and precision of the impurities (Phenol, Des-Br, Bromo, and BisPh) were verified at three levels (0.05% [LOQ], 1.0%, 2.0% of 1.0 mg/mL, see Table S7 in the supplementary material). Recovery of each replicate at these levels ranged between 93% - 114%, with a %RSD of less than 4.0%. Intermediate precision of all impurities was also confirmed, with a %RSD value less than 6.5% (n = 12, see Table S-8 in the supplementary material).

Solution stability was assessed by testing all the solutions (LOQ, Resolution, WSS, Assay sample, and Purity sample) stored at ambient temperature for 24 hours. The peak area of each component in the LOQ, Resolution, and WSS solutions was compared with the corresponding initial value. Assay (% w/w) and impurity (%w/w) were determined against a freshly prepared WSS at 24 h using Equations (1) and (2), respectively. Based on the results (see Table S9 in the supplementary material), the LOQ, resolution, WSS, Assay, and purity sample solutions are stable for 24 h at ambient temperature.

Using a diluent with approximately 50% water improves peak shape for 4-MPBA and facilitates the in situ conversion of anhydrides or oligomers into the monomer, as shown in Figure 11. A ¹H NMR study confirms that 4-MPBX (boroxine trimer, a representative anhydride/oligomer) rapidly converts to 4-MPBA upon quenching with deuterated water (Figure S3 and Figure S4 in the supplementary material). This conversion allows both the identification and assay value to be reported in the monomeric form, as all oligomers are converted in situ and subsequently utilized in the downstream aqueous reaction.

Figure 11. Inter-conversion of 4-MPBA and 4-MPBX.

The choice of organic solvent in mobile phase B is critical to achieving baseline separation of Phenol and 4-MPBA. While the aprotic solvent acetonitrile failed to resolve the phenol and 4-MPBA peaks, methanol (a polar protic solvent) achieved baseline resolution (Figures 3-4). This improvement is due to methanol’s ability to act as both a hydrogen-bond donor and acceptor, allowing it to effectively compete for hydrogen-bonding sites between the analytes and the cyano stationary phase. In contrast, acetonitrile lacks the protic hydrogen necessary to disrupt these interactions.

During in-process control (IPC) testing of the reaction producing the actual GMP PBA-type intermediate, the formation of two process-related impurities (Phenol and Des-Br types) was observed at levels below the LOQ (0.05%w/w), while the Bromo-type starting material and the BisPh-type impurity were below the LOD (0.025% w/w).

The established low-level linearity range (0.05% - 2.0% w/w of 1.0 mg/mL) enables the quantification of trace amounts of impurities, while the high-level range (50% - 150% w/w of 0.1 mg/mL) enables accurate quantification of the assay as the monomer of 4-MPBA. The established range meets QC requirements and ICH guidelines.

The RRF for 4-MPBA is critical to the RP-HPLC-UV assay method, as the actual concentration of 4-MPBA in the linearity solutions directly affects the RRF calculation. Therefore, determining an accurate potency (as monomer) of 4-MPBA is key. Using an HPLC mass balance calculation (see Equation (3)), a potency of 0.877 (as 4-MPBX) was obtained. Since one molecule of 4-MPBX yields three molecules of 4-MPBA, the potency was calculated as 0.995 (as monomer) according to Equation (4). This potency value is consistent with the assay results (99.4%) obtained via an orthogonal qNMR method (see Table 7).

where: Timp is the total HPLC impurities, and TV is % total volatiles obtained from duplicate measurements of loss of drying. This is supported by thermogravimetric analyses TGA and differential scanning calorimetry (DSC) data (see Figure S1 and Figure S2 in the supplementary material).

where: n = 3, MW1 is molecular weight (151.956) of the monomer (4-MPBA), and MWn is molecular weight (3 x 133.941) of the trimer (4-MPBX).

The method was further verified via an orthogonal quantitative NMR (qNMR) method. The mean assay as monomer obtained by qNMR was 99.4%, while the mean assay as monomer by HPLC (n = 12) was 99.2%, representing a 0.2% difference (see Table 7). A difference of less than 2.0% between an HPLC method and another orthogonal method is considered acceptable. Refer to Figure S5 and Tables S10-S14 in the supplementary material for the qNMR experimental data.

Table 7. Assay by HPLC and qNMR methods.