The FFC was tablet-type products and made from “raw materials” and excipients. In this study, two raw material samples, lots A and B, were analyzed and compared. Lot A was an ingredient of the FFC with health hazard reports, while lot B was not. Sample solutions were prepared by extracting each lot with 75% (vol%) methanol, followed by ultra-high-performance liquid chromatography-high resolution mass spectrometry (UHPLC-HRMS, Thermo), LC–MS (Shimadzu), supercritical fluid chromatography-HRMS (SFC-HRMS, Shimadzu), and GC–MS (Shimadzu) analyses.

Differential analysis of the UHPLC-HRMS data detected compounds 1 and 2 as the components in lot A. Compound 1 was eluted at 20.2 min using UHPLC-HRMS, and its molecular formula was C23H34O7 (m/z: 421.2217 ([M-H]−) and 423.2374 ([M + H]+)). Compound 2 was eluted at 28.1 min, and its molecular formula was C28H42O8 (m/z: 505.2805 ([M-H]−), 507.2981 ([M + H]+), and 524.3225 ([M + NH4]+) (Fig. S1).

Obtained as far back as possible (three years), 29 raw material samples including lots A and B were investigated to determine whether compounds 1 and 2 were present. The analyses revealed that both compounds 1 and 2 were detected in seven raw material samples that were manufactured during a specific period (approximately six months). Compounds 1 and 2 were not detected in samples produced outside of that specific period. Moreover, two FFC samples with health hazard reports and one FFC sample with no report were analyzed, revealing that compounds 1 and 2 were detected in only products with health hazard reports. Although limited to the samples available for analyses in this study, both compounds 1 and 2 were detected in all the samples that also contained puberulic acid.

Compounds 1 and 2 were isolated from the samples, and UHPLC-HRMS, GC–MS, and NMR analyses were conducted to identify their structures. The results of 15N-NMR, 19F-NMR, and 31P-NMR analyses indicated that nitrogen, fluorine, and phosphorus atoms did not exist in compounds 1 and 2. The isotope pattern results of the UHPLC-HRMS analysis indicated that sulfur atoms were also not present in compounds 1 and 2. Furthermore, the MS/MS spectra of UHPLC-HRMS and EI-MS spectra of GC–MS revealed that compounds 1 and 2 have a similar substructure to that of lovastatin acid (3) and lovastatin (4) (Fig. S2). As references for identification, lovastatin acid (3) and lovastatin (4) (Fig. 1) were analyzed using NMR spectroscopy, as well as compounds 1 and 2.

Chemical structures of compounds 1 and 2, lovastatin acid (3), and lovastatin (4)

Compounds 1 and 2 were dissolved in dimethyl sulfoxide (DMSO)-d6 and analyzed using NMR (1H-NMR, 13C-NMR, distortionless enhancement by polarization transfer (DEPT), 1H-1H COSY, HMQC, HMBC, and 1H-13C heteronuclear 2-bond correlation (H2BC)). 1H- and 13C-NMR data of compounds 1 and 2 in DMSO-d6 are listed in Table 1, and spectrum data are shown in the supplementary materials (Fig. S3–S26). The 1H-1H COSY and HMBC correlations of compounds 1 and 2 are shown in Fig. 2. In the 1H- and 13C-NMR data of compounds 1 and 2, specific signals appeared that were not detected in lovastatin acid (3) and lovastatin (4). These signals were derived from the two acetyl groups detected at δC/δH: 21.0/1.98 and 21.0/1.94 ppm for compound 1 and δC/δH: 20.9/1.96 and 21.0/1.92 ppm for compound 2. In addition, the structure of compound 2 was probably a derivative of compound 1, whose side chain that was originally attached to C1 was removed. Consequently, the structures of the novel compounds 1 and 2 were presumed, as shown in Fig. 1.

Selected 2D NMR correlations for compounds 1 and 2

To confirm our results and determine the steric configuration, we are currently synthesizing compounds 1 and 2 from lovastatin (4) for NMR and UHPLC-HRMS analyses to determine whether they are consistent with those of compounds 1 and 2 isolated from the samples, respectively. The route of contamination of these compounds are also currently under investigation.