Recently, a chromatographic quantitative method using relative molar sensitivity (RMS), the so-called RMS method, was developed for the determination of target analytes in food additives, drugs, and supplements. By determining the RMS value of the analyte to a different reference compound, this method avoids the need for an analytical standard of the analyte itself to plot calibration curves for quantification. In this collaborative study performed by 10 laboratories, we demonstrated the robustness and reliability of the RMS method in the quantitative analysis of chlorogenic acid (5-O-caffeoylquinic acid, 5CQA) in apple juice. Reagent-grade 5CQA derived from natural sources is commercially available, but its purity, stability, and hygroscopic properties still need to be clarified. Therefore, there is always a risk of bias in the quantitative results, even when the calibration curve is prepared using reagent-grade 5CQA as the analytical standard. The RMS method overcomes this problem by using caffeic acid (CA) with high purity and stability as a reference compound. Prior to the collaborative study, a laboratory documented the standard operating procedure for method validation, which was then implemented in all laboratories to determine the RMS value of 5CQA to CA and quantify 5CQA in five apple juice samples. A comparison between the results obtained using a calibration curve and the RMS method validates the RMS method using CA as a reference compound for the quantitative analysis of 5CQA without an analytical standard.

Chromatography methods such as HPLC and gas chromatography (GC) are widely employed in the quantitative analysis of bioactive substances, including multicomponent products, in food, food additives, and pharmaceuticals. To ensure the reliability of the quantitative result, it is crucial to use a reference compound identical to the analyte with known purity. However, for many target analytes, obtaining reference compounds with certified purity is difficult. Even if the analytical standard or isolated analyte can be obtained, the calibration curves cannot provide reliable quantitative values if the purity of the standard is unknown and the concentration of the standard solution is not corrected with its purity. Therefore, the development of a quantitative method that can circumvent this problem is highly desirable to avoid bias in the quantitative results arising from ambiguity in the purity of the reference compounds.

Recently, quantification methods based on relative molar sensitivity (RMS) have become used in the food and pharmaceutical industries.^1–7) RMS is a coefficient defined as the detector sensitivity ratio of the analyte to the reference substance (rs) per unit amount of substance under arbitrary chromatographic conditions.⁸⁾ Standard solutions (mixture or individual) of the analyte and the reference compound are analyzed by combining ¹H quantitative NMR (¹H-qNMR) and chromatography (e.g., HPLC), and the RMS (R_r/R_n) value is calculated using the ratio of the amount of substance (R_n) obtained from ¹H-qNMR and the response ratio (R_r) obtained from HPLC. Because the ratio of the amount of reference substance to the analyte (R_n) can be expressed as n_analyte/n_rs and the response ratio (R_r) as y_analyte/y_rs (where n is the amount of substance and y is the response), the following Eq. (1) is valid:

where F_RMS is the RMS value of analyte to rs. Per Eq. (1), a means the slope of the calibration curve constructed by plotting the detector response (y) and the amount of compound (n) as the y- and x-axes, respectively. Therefore, the RMS value is calculated as the ratio of the slopes of analyte and reference compound. Once the RMS of the analyte to the reference compound is known, the analyte can be quantified using the reference compound for determining the RMS instead of an analytical standard with known purity. This quantification method using RMS, which is called the RMS method or the single-reference method with RMS, enables the accurate determination of not only unstable analytes but also constituents that should be monitored or regulated for maintaining human health in pharmaceuticals, food additives, and foods using only a conventional HPLC system. In The Japanese Pharmacopoeia Eighteenth Edition (JP18), the quantification method of perillaldehyde in the specification of the crude drug “Soyo (Perilla Herb)” was revised to the RMS method.^1,3) The original quantification method in the previous edition (JP17) used the analytical standard of perillaldehyde. However, since this substance is unstable, the RMS method uses diphenylsulfone, which is a stable substance available in reagent-grade with certified purity. Similarly, the RMS method was adopted to determine seven specifications of food additives, that is, horseradish extract,⁹⁾ mustard extract,⁹⁾ rumput roman extract, cochineal extract,²⁾ Jamaica quassia extract,¹⁰⁾ Luohanguo extract,¹¹⁾ and rosemary extract,¹²⁾ and the principle and procedure used were described in the general HPLC and GC method sections in Japan’s Specifications and Standards for Food Additives, Tenth Edition (JSFA-X).¹³⁾ In addition to these regulations, the RMS method has been applied to the quantification of bioactive or marker components in food, food additives, and supplements.^2,6,14,15)

Epidemiological studies have shown that fruit consumption effectively maintains biological homeostasis and prevents cerebrovascular and cardiac diseases.^16–18) The disease-preventive function of fruits is thought to operate by means of the antioxidant effect of polyphenols. Apple is a worldwide cultured fruit with high phytochemical polyphenol content. At the end of the 19th century, Japan started culturing apples and improving their productivity, taste, and quality to improve breeding varieties. As a result, the distribution ratio of apple varieties in Japan is currently different from that in other countries. Since most Japanese varieties have a high brix/acid ratio, rich juice, and pulp, this kind of apple has attracted worldwide attention. In addition, considering that the high polyphenol content in the juice provides added value to apple products for export, Japanese producers are interested in the development of internationally standardizable quantitative methods for such polyphenols. One of these polyphenols is chlorogenic acid (5-O-caffeoylquinic acid, 5CQA (1)), which has been reported to reduce body fat by acting on mitochondria, promoting lipid uptake and fat burning.¹⁹⁾ Although the 5CQA content in apples can be easily quantified using HPLC, reagent-grade 5CQA derived from natural sources, which is used as an analytical standard to plot the calibration curves for quantification, is unstable, which results in bias in the quantitative results. To overcome this issue, we designed an RMS method using caffeic acid (CA (2)), which has high purity and stability, as the reference substance (Fig. 1).

Fig. 1. The Structures of 5-O-Caffeoylquinic Acid (5CQA) and Caffeic Acid

In the present study, we organized a collaborative study between 10 laboratories in Japan to evaluate the RMS method using HPLC to quantify 5CQA in apple juice and to compare this method with the common quantitative HPLC analysis using absolute calibration curves.

Laboratory A designed this collaborative study, distributed the apple juice samples, reagents, and analytical columns, and documented the standard operating procedure (SOP) for method validation. All collaborative laboratories implemented the method following the SOP for determining the RMS value and 5CQA content.

Ten investigators from 10 laboratories (A–J) performed this collaborative study. Supplementary Table S1 summarizes the HPLC systems and parameters used to determine the RMS value of 5CQA to CA, which was used as the RMS reference compound, and to quantify the 5CQA content in apple juice samples in each laboratory.

Five apple juice samples (samples 1–5, Table 1) were purchased in the Japanese market. 5CQA hemihydrate and CA (3,4-dihydroxycinnamic acid) were purchased from FUJIFILM Wako Pure Chemical Corporation (Osaka, Japan) and used to determine the RMS value of 5CQA to CA and the 5CQA content in the apple juice samples. Before conducting this collaborative study, the purities of 5CQA and CA were evaluated using ¹H-qNMR with (0.9449 ± 0.0086) kg/kg (k = 2) and (0.9794 ± 0.0088) kg/kg (k = 2), respectively.²⁰⁾ Methanol-d₄ was used for ¹H-qNMR, HPLC-grade distilled water and acetonitrile were used as the HPLC mobile phase, and other reagents were of special grade or higher.

All laboratories used a semimicrobalance with a maximum weight of over 40 g and an HPLC system equipped with a photodiode array (PDA) or a variable wavelength detector (VWD).

All laboratories used the HPLC conditions summarized in Table 2.

Approximately 50 mg of CA was precisely weighed and dissolved and diluted to 25 mL with methanol. Then, 1.0 mL of this solution was mixed with 9.0 mL of water, allowed to reach room temperature, and diluted exactly to 20 mL with a 1 : 9 methanol–water mixture. This solution was used as a 100 μg/mL standard stock solution of RMS reference (CA). In the same manner, approximately 50 mg of 5CQA was precisely weighed to prepare the corresponding standard stock solution of 5CQA analyte. The concentrations of all the standard solutions were corrected by the purities of CA and 5CQA before performing the experiments. Standard stock solutions of CA and 5CQA were prepared in triplicate to identify any variation between preparations. Each standard stock solution of CA and 5CQA was further diluted to approximately 1, 5, 10, 20, 40, and 50 μg/mL with 1 : 9 methanol–water, and the resulting solutions were used as the standard solutions of CA and 5CQA to prepare the respective calibration curves.

Using the proportional relationship between the peak area (y) and the molar concentration (μmol/mL; x), the calibration curves of CA and 5CQA through the origin (y = ax) were prepared by measuring in triplicate the standard solutions of CA and 5CQA using HPLC under the conditions shown in Table 2. The results were used to plot three calibration curves, and the RMS value (F_RMS) of 5CQA to CA was calculated using the slopes of the calibration curves using Eq. (2) as follows:

where a_5CQA and a_CA are the slopes of the calibration curves of 5CQA and CA, respectively, and F_RMS is the RMS value of 5CQA to CA.

Before preparing the sample solutions, 2.125 mL of 70% perchloric acid or 2.75 mL of 60% perchloric acid was diluted to 250 mL with water. This solution (90 mL) was mixed with 10 mL of methanol to prepare a 1 : 9 methanol–0.1 mol/L perchloric acid solution mixture as the dilution solvent. The apple juice was shaken well, and 5 mL was precisely taken immediately after opening. Approximately 15 mL of 1 : 9 methanol–0.1 mol/L perchloric acid solution mixture was added, and the mixture was sonicated in an ultrasonic bath for 5 min. After the volume was precisely filled up to 25 mL with the same dilution solvent, the resulting solution was filtered through a 0.45 μm membrane filter and used as the sample solution for quantification of 5CQA. If filtration was difficult, the solution was centrifuged and the supernatant was filtered to prepare the sample solution. Two sample solutions were prepared for each sample and measured in triplicate using HPLC.

Approximately 100 μg/mL solution was prepared in the same way as for the RMS standard stock solution (CA), and 5 mL of this solution was precisely diluted to 10 mL with 1 : 9 methanol–water to obtain the RMS standard solution containing approximately 50 μg/mL of CA. The CA concentration in this standard solution was corrected with the purity of CA prior to the experiments.

The apple juice sample solutions and the RMS standard solution containing CA as an RMS reference were analyzed under the same HPLC conditions used for determination of the RMS value. The peak areas of 5CQA in the sample solutions and the peak area of CA in the RMS standard solution (CA) were measured, and the 5CQA contents (μg/mL) were calculated using Eq. (3):

where y_5CQA and y_CA are the peak areas of 5CQA and CA, respectively, m_5CQA is the molecular mass of 5CQA (354.31), C_CA is the CA concentration in the RMS standard solution containing CA (μg/mL), C_5CQA is the 5CQA concentration in the apple juice sample (μg/mL), and N is the dilution factor of apple juice (5).

To validate the RMS method, the 5CQA content (μg/mL) was also quantified using the calibration curve method. The calibration curves were plotted using the standard solutions of 5CQA with different concentrations prepared for determining the RMS value.

Since the RMS value is the ratio of the amount of substance ratio (the amount ratio) of an analyte to a reference compound to the respective sensitivity ratio determined using the chromatographic detector, the RMS can be calculated from the ratio of the slope of the analyte and that of a reference compound when a proportional relationship between the amount of substance and the sensitivity is observed during the chromatographic analysis. The RMS reference should be a chemically stable, inexpensive, and high purity substance with approximately the same maximum absorption as that of the analyte. 5CQA contains a caffeoyl moiety and a UV absorption maximum at approximately 325 nm. Considering that CA is commercially available with high purity and stability, we selected CA as a candidate for RMS reference.

Before determining the RMS value of 5CQA to CA, the HPLC conditions were optimized to obtain sufficient separation of 5CQA from other components in apple juice and allow an isocratic than a gradient mobile phase to prevent the variation in the solvent composition during the mobile phase from affecting the spectral shapes and absorption maxima at approximately 325 nm of 5CQA and CA. A screening of HPLC conditions revealed that an isocratic mobile phase could separate the peaks of 5CQA, CA, and impurities in apple juice (Fig. 2). The collaborative study was performed using the optimized HPLC conditions shown in Table 2.

Fig. 2. Typical HPLC Chromatograms Obtained Using Photodiode Array Detection at 325 nm under Optimized HPLC Conditions

Peak 1, chlorogenic acid (5CQA); peak 2, caffeic acid (CA). (A) Standard solution of 5CQA (50.9 μg/mL, 0.14 μmol/mL); (B) Standard solution of CA (50.0 μg/mL, 0.27 μmol/mL); (C) Sample solution of apple juice sample 1; (D) Apple juice sample 2; (E) Apple juice sample 3; (F) Apple juice sample 4; (G) Apple juice sample 5.

To calculate the RMS value of 5CQA to CA, all collaborative laboratories plotted the calibration curves of 5CQA and CA using both standard solutions with different concentrations, which were prepared using reagent-grade compounds with purities determined via ¹H-qNMR analysis. As shown in Table 3, each laboratory plotted three calibration curves of 5CQA and CA using PDA or VWD detectors to obtain linear relationships between the peak areas and the concentrations (μmol/mL) and confirmed the linearity with coefficient of determination (R²) values over 0.999. Laboratory D gave R² values of 0.988–0.995 and 0.999 for the 5CQA and CA slopes, respectively; considering there was no deviation from the SOP method in laboratory D, all slope data measured in the collaborative laboratories were used for calculating the RMS value (F_RMS) of 5CQA to CA. The F_RMS values calculated from the data of five PDA detectors of five laboratories ranged between 1.037 and 1.148 with small relative standard deviation (RSD) values of 0.2–1.0%, with the average value being 1.100 with an RSD of 2.9%. Meanwhile, the F_RMS values calculated from the data of eight VWD detectors of eight laboratories were 1.077–1.142 with small RSD of 0.1–0.8%, and the average value was 1.099 with an RSD of 1.5%. This suggested that using the VWD detector provided more stable F_RMS values, but there was no significant difference of performance between PDA and VWD detectors. Therefore, we concluded that the average F_RMS calculated from all the data obtained using PDA and VWD, which was 1.099 with an RSD of 2.0%, could be used for the quantification of 5CQA in apple juice samples.

The browning in apples has been reported to be caused by the oxidation of 5CQA to the intermediate chlorogenic acid quinone in the presence of polyphenol oxidase, which further polymerizes with amino acids and other compounds.²¹⁾ To avoid lowering of the 5CQA content in apple juice samples during preparation and analyses of their test solutions, 1 : 9 methanol–0.1 mol/L perchloric acid solution mixture was used as the dilution solvent to inactivate the enzyme. Before this collaborative study, laboratory A confirmed that the 5CQA content in the apple juice sample solutions prepared using this dilution solvent did not change considerably during the HPLC measurement, being the variation within ±1%. In addition, three sample solutions prepared by spiking 50 and 500 μg of 5CQA into 5 mL of apple juice sample 4 were subjected to a recovery test, yielding recovery rates of 100.6% (RSD 5.9%) and 100.3% (RSD 1.7%), respectively. These results suggest that the preparation of the sample solutions was well established to demonstrate the applicability of the RMS method.

Table 4 shows the results of the 5CQA content in apple juice samples 1–5 determined in laboratories A–J using the calibration curve method and the RMS method. Laboratory H was excluded because of an error in the injection volume of the sample solution. The 5CQA content determined using the calibration curves differed between the five samples. Values between 27 and 153 μg/mL were obtained, and no significant differences were observed between the values obtained with the PDA and VWD detectors for each sample. Meanwhile, the 5CQA contents in samples 1–5 determined using the RMS method with the average F_RMS value of 1.099 were almost identical to those obtained using the calibration curve method for each sample. Since the RSDs of the values determined using the RMS method were also as small as those determined via the calibration curve method, the RMS method using CA as the RMS reference compound can be used as an alternative method for determination of 5CQA in apple juice samples instead of the calibration curve method.

We developed and validated a quantitative method for determination of 5CQA, which is one of the bioactive components in apple juice, based on the RMS value with CA as the RMS reference compound. The RSD of the RMS values determined using all the data obtained with PDA and VWD detectors in 10 laboratories was 2.0%, suggesting that the obtained RMS value of 1.099 can be used for quantitative analysis in different laboratories. A comparison between the RMS and calibration curve methods demonstrated that both methods provided almost the same quantitative values for 5CQA in five apple juice samples. Therefore, the RMS method, which overcomes the lack of reference compounds with certified or well-determined purity in the quantitative analysis of bioactive components in agricultural products and low-processed foods, is comparable to the calibration curve method for the quantitative analysis of 5CQA in apple juice.

The results obtained from this collaborative study indicated that the RMS method can serve as an alternative for quantifying the calibration curve method. As without the reference material which identical compound to analyte for the RMS method, the proposed method is beneficial for various products for quality control, of which most constituents were unavailable commercially. The RMS method can be popularly utilized in the future, especially in regulatory science for natural products.

This collaborative study was implemented as part of the 2022FY research project to improve the export environment through international standardization of the Japanese Agricultural Standards and other standards with support from the Ministry of Agriculture, Forestry and Fisheries. This work was supported in part by Health and Labour Sciences Research Grants (Nos. 20KA1008 and 23KA1012) from the Ministry of Health, Labour and Welfare.

The authors declare no conflict of interest.

This article contains supplementary materials.

• 1)  Ministry of Health, Labour and Welfare, Japan. “The Japanese Pharmacopoeia Eighteenth Edition (2021)”: <https://www.mhlw.go.jp/content/11120000/000788459.pdf>, cited 23 February, 2025.
• 2)  Nishizaki Y., Sato-Masumoto N., Nakanishi A., Hashizume Y., Tandia M., Yamazaki T., Kuroe M., Numata M., Ihara T., Sugimoto N., Sato K., Food Hyg. Saf. Sci., 59, 1–10 (2018).
• 3)  Masumoto N., Nishizaki Y., Maruyama T., Igarashi Y., Nakajima K., Yamazaki T., Kuroe M., Numata M., Ihara T., Sugimoto N., Sato K., J. Nat. Med., 73, 566–576 (2019).
• 4)  Sakai Y., Masumoto N., Nishizaki Y., Ohtsuki T., Matsufuji H., Sugimoto N., Sato K., Jpn. J. Food Chem. Safety, 27, 123–134 (2020).
• 5)  Uchiyama N., Hosoe J., Ishizuki K., Arai R., Sugimoto N., Suzuki A., Asano R., Igarashi Y., Miura T., Muto Y., Suematsu T., Komatsu T., Higano T., Furukawa T., Shimada N., Goda Y., Yakugaku Zasshi, 143, 951–962 (2023).
• 6)  Iwasaki D., Kanazawa M., Kawamoto F., Araho D., Murakami T., Nishizaki Y., Masumoto N., Sugimoto N., Food Chem., 427, 136647 (2023).
• 7)  Mizumoto T., Nakano F., Nishizaki Y., Masumoto N., Sugimoto N., Food Hyg. Saf. Sci., 60, 134–143 (2019).
• 8)  Nishizaki Y., Masumoto N., Sugimoto N., “Comprehensive natural products III,” 7.04—Application of ¹H-quantitative NMR from the viewpoint of regulatory science., ed. by Liu H. W., Begley T. P., Elsevier, Amsterdam, 2020, pp. 50–60.
• 9)  Masumoto N., Nishizaki Y., Nakajima K., Sugimoto N., Sato K., Food Hyg. Saf. Sci., 62, 73–78 (2021).
• 10)  Nishizaki Y., Tada A., Ishizuki K., Ito Y., Onoda A., Sugimoto N., Akiyama H., Food Hyg. Saf. Sci., 56, 185–193 (2015).
• 11)  Masumoto N., Ishizuki K., Nishizaki Y., Ohtsuki T., Kuroe M., Yamazaki T., Numata M., Matsufuji H., Sugimoto N., Sato K., Chem. Pharm. Bull., 69, 18–25 (2021).
• 12)  Nishizaki Y., Masumoto N., Nakajima K., Ishizuki K., Yamazaki T., Kuroe M., Numata M., Ihara T., Tada A., Sugimoto N., Sato K., Food Addit. Contam. Part A Chem. Anal. Control Expo. Risk Assess., 36, 203–211 (2019).
• 13)  Consumer Affairs Agency, Government of Japan, “Japan’s Specifications and Standards for Food Additives tenth Edition (2024).”: <https://www.caa.go.jp/policies/policy/standards_evaluation/food_additives/official_documents_002/assets/001208056.pdf>, cited 23 February, 2025.
• 14)  Koyama K., Sasako H., Higashi Y., Ichikawa H., Nagoya A., Hirao T., Food Hyg. Saf. Sci., 63, 202–209 (2022).
• 15)  Takahashi M., Nishizaki Y., Sugimoto N., Sato K., Inoue K., J. Chromatogr. A, 1555, 45–52 (2018).
• 16)  Liu S., Manson J. E., Lee I. M., Cole S. R., Hennekens C. H., Willett W. C., Buring J. E., Am. J. Clin. Nutr., 72, 922–928 (2000).
• 17)  He K., Hu F. B., Colditz G. A., Manson J. E., Willett W. C., Liu S., Int. J. Obes. (Lond.), 28, 1569–1574 (2004).
• 18)  Dauchet L., Amouyel P., Hercberg S., Dallongeville J., J. Nutr., 136, 2588–2593 (2006).
• 19)  Murase T., Yokoi Y., Misawa K., Ominami H., Suzuki Y., Shibuya Y., Hase T., Br. J. Nutr., 107, 1757–1765 (2012).
• 20)  ISO24583:2022, International Organization for Standardization. “Quantitative nuclear magnetic resonance spectroscopy – Purity determination of organic compounds used for foods and food products – General requirements for 1H NMR internal standard method”: <https://www.iso.org/standard/78997.html>, cited 23 February, 2025.
• 21)  Murata M., Sugiura M., Sonokawa Y., Shimamura T., Homma S., Biosci. Biotechnol. Biochem., 66, 2525–2530 (2002).