Steviol glycosides are the sweet compounds extracted from the leaves of Stevia rebaudiana Bertoni, a plant native to South America that has become one of the most important sources of high-intensity, zero-calorie sweeteners in the global food industry. Indigenous peoples of Paraguay and Brazil used the leaves of the plant, commonly known as sweetleaf, to sweeten beverages long before modern chemistry identified the molecules responsible for the taste. Because these compounds are several hundred times sweeter than sucrose yet contribute essentially no calories, they have been adopted widely in beverages, dairy products, and snack foods aimed at consumers seeking to reduce sugar intake. That demand has intensified as public health authorities around the world press manufacturers to lower sugar content in response to rising rates of obesity and type 2 diabetes, and as several countries have introduced sugar taxes that make non-nutritive sweeteners economically attractive. Regulatory agencies, including the European Commission and the Joint FAO/WHO Expert Committee on Food Additives, have established specifications and permitted uses for steviol glycosides, which in turn creates a need for analytical methods that can reliably measure them in the complex matrices of real foods rather than in purified standards. Without such methods, regulators cannot verify that products contain what their labels declare, nor can they estimate how much of the sweeteners consumers actually ingest.

The analytical challenge is considerable. Steviol glycosides comprise a family of structurally related molecules, including stevioside, rebaudioside A, rebaudioside B, rebaudioside C, rebaudioside D, rebaudioside E, rebaudioside F, and several others, that differ only in the number and arrangement of sugar units attached to the steviol backbone. Rebaudioside A, for example, carries additional glucose units relative to stevioside, a difference that subtly alters both its sweetness profile and its chromatographic behavior. These subtle structural differences make chromatographic separation difficult, particularly in food matrices that contain sugars, acids, proteins, fats, and other additives that can interfere with detection. A carbonated soft drink presents a very different analytical problem from a sweetened yogurt or a cereal bar, and a method that performs well in one matrix may fail in another. Previous efforts have employed capillary electrophoresis, high-performance thin-layer chromatography, and liquid chromatography coupled with tandem mass spectrometry, but each approach carries trade-offs in cost, throughput, accessibility, and the number of compounds that can be resolved in a single run. Mass spectrometry offers exceptional sensitivity and specificity, but the instrumentation is expensive, requires specialized expertise, and is not available in every food control laboratory, particularly in smaller regional facilities.

The new method relies on a conventional HPLC system equipped with a photodiode array detector, or PDA, a configuration that is far more common in food control laboratories than mass spectrometry instrumentation. Separation was achieved on a standard C18 reversed-phase column measuring 250 mm in length with a 4.6 mm internal diameter and 5 µm particles. The complete chromatographic run takes 40 minutes, a duration the authors judged acceptable for resolving all 13 target glycosides in a single analysis. Using this setup, the team obtained regression coefficients ranging from 0.9995 to 0.9999 across the calibration range, indicating excellent linearity of detector response over the concentrations relevant to food analysis. Linearity of this quality matters because quantification in routine surveillance depends on calibration curves that remain reliable across the wide span of concentrations found in commercial products, from lightly sweetened dairy drinks to confectionery items in which the sweeteners are used at much higher levels.

Sensitivity figures reported in the study place the method comfortably within the range needed for regulatory monitoring. Limits of detection ranged from 0.09 to 0.64 mg/kg depending on the individual glycoside, while limits of quantitation ranged from 0.27 to 1.93 mg/kg. These values mean that even trace amounts of the sweeteners can be detected and reliably quantified in finished products, allowing laboratories to verify compliance with maximum permitted levels and to characterize the actual exposure of consumers to steviol glycosides through their diets. Sensitivity at these levels is also relevant to unintended carryover, since tiny amounts of a sweetener can appear in products that do not declare it, either through shared production lines or through the use of flavoring preparations that contain traces of the compounds.

For full method validation, the researchers selected four representative compounds: rebaudioside A, stevioside, rebaudioside F, and rebaudioside C. These four were chosen because they are among the most commonly encountered steviol glycosides in commercial food applications, with rebaudioside A and stevioside historically dominating the market. The validation was performed across three food matrix categories designed to represent the diversity of processed foods: beverages, fermented milk, and snacks. Each category presents distinct analytical difficulties, from the acidity and coloring of beverages to the protein and fat content of fermented dairy and the heterogeneous composition of snack products. Fermented milk, in particular, contains lactic acid, live cultures, and dairy proteins that can co-elute with target analytes or degrade chromatographic peaks, making it one of the more demanding matrices for glycoside analysis.

Precision testing demonstrated that the method performs consistently both within a single day and across different days. Intraday precision values fell between 1.93 and 3.91 percent relative standard deviation, while interday precision ranged from 2.13 to 5.92 percent. Both figures are well within the acceptance criteria typically applied in analytical validation guidelines, such as those issued by the International Council for Harmonisation, the Association of Official Analytical Chemists, and the European Commission’s SANTE guidance document, all of which the authors cite as methodological references. Alignment with these internationally recognized frameworks is significant because it allows laboratories outside Korea to evaluate the method against familiar performance benchmarks, facilitating potential adoption beyond the country where it was developed.

Matrix effects, a critical consideration in food analysis because components of the sample can suppress or enhance the apparent signal, were quantified for each food category. The effects were modest in beverages, ranging from 5.43 to 6.88 percent, somewhat larger in snacks at 9.63 to 14.32 percent, and largest in fermented milk, where they reached 13.29 to 15.57 percent. These results suggest that while the method is robust across all three categories, laboratories analyzing dairy products in particular should be attentive to matrix-related bias, potentially through the use of matrix-matched calibration or standard addition procedures. The relatively low matrix effects overall reflect the effectiveness of the sample preparation approach, which the authors developed in light of prior work on pretreatment methods for steviol glycosides in diverse food samples, including earlier studies on fermented milk by some of the same research groups. That continuity of research effort is evident in the way the new protocol consolidates lessons learned from earlier, narrower applications into a single broadly validated procedure.

To establish that the method is not merely reproducible within a single laboratory, the team conducted an interlaboratory validation, in which the procedure was performed by additional laboratories to confirm that results could be reproduced elsewhere. This step is essential for any method intended to serve as a reference procedure for national food safety monitoring, since enforcement actions and exposure assessments depend on measurements that different laboratories can obtain consistently. Analytical methods that look strong in the hands of their developers sometimes falter when transferred to other facilities, where differences in equipment, column lots, and operator technique can erode performance. The successful interlaboratory outcome supports the method’s candidacy for adoption in official food control contexts in Korea and potentially beyond.

The practical value of the method was demonstrated by applying it to 130 commercially available processed foods purchased in Korea. This survey allowed the researchers to quantify the actual content of steviol glycosides in products on the market, generating data that can inform dietary exposure assessments conducted by the Ministry of Food and Drug Safety. Such monitoring data are increasingly important as reformulation trends drive greater use of non-nutritive sweeteners, and as regulators seek to verify that product labeling and additive usage comply with national standards. Exposure assessments typically combine analytical concentration data with national food consumption surveys, so the quality of the concentration measurements directly determines the reliability of the resulting risk estimates. The reference list of the paper indicates that Korean agencies had previously conducted safety evaluations of food additives, and this new method provides the analytical backbone for continued surveillance.

The study builds on a substantial body of prior analytical work. Earlier researchers developed fast isocratic HPLC methods for analyzing steviol glycosides in stevia leaves, LC-MS/MS approaches for stevia leaf extracts and commercial soju, UHPLC-MS/MS methods for foods and beverages, and high-performance thin-layer chromatography benchmarks for sugar-free products. A 2020 single-laboratory validation published in the Journal of Agricultural and Food Chemistry similarly targeted 13 steviol glycosides in foods, dietary supplements, and ingredients. The Korean team’s contribution lies in combining comprehensive separation of 13 glycosides with a widely accessible PDA detector, rigorous single- and interlaboratory validation across multiple food matrices, and direct application to a large set of market products, thereby bridging a gap between methods developed for pure ingredients and the needs of routine food surveillance.

Several limitations should be noted. The validation focused on four of the 13 separated glycosides, so quantitative performance for the remaining compounds, while presumably covered by the calibration data, was not subjected to the same depth of matrix-specific validation. The matrix categories, though representative, do not exhaust the range of foods in which steviol glycosides may appear, and the 40-minute run time, while acceptable, is longer than some rapid or mass-spectrometric alternatives. The authors also note that data will be made available on request.