Detection of exogenous sugars in pineapple juice using compound-specific stable hydrogen isotope analysis

Stable isotope analysis has long been used to detect economically motivated adulteration in fruit juices, leveraging differences in stable carbon (δ¹³C) and hydrogen (²H/¹H) isotope ratios among plant sources. Conventional approaches include EA-IRMS for bulk δ¹³C to identify C4 sugar addition (e.g., cane, corn) due to distinct photosynthetic pathway signatures, and SNIF-NMR for site-specific ²H/¹H measurements of ethanol fermented from juice sugars. These methods are codified in Codex standard 234 and recognized by AOAC. While δ¹³C analysis readily detects C4 sugar addition to C3-derived products, it cannot reveal addition of C3 beet sugar when sugar profiles are matched and other ingredients mask dilution. Site-specific deuterium analyses can detect C3 beet syrups in C3 juices because hydrogen isotope ratios vary strongly with geography and plant physiology. However, pineapple juice presents a unique challenge: it is produced by a CAM plant (Ananas comosus). Pineapple sugars’ global δ¹³C values and SNIF-NMR ²H/¹H values overlap with those from cane and maize (C4 plants), diminishing the ability of standard methods to detect cane sugar addition. A ¹³C SNIF-NMR method targeting ethanol methyl and methylene positions can detect C4 sugars at around 15% of total sugars but is time-consuming and costly (multi-day fermentation and high-field NMR). This study addresses the gap by applying a compound-specific stable hydrogen isotope approach via GC-IRMS of trifluoroacetate (TFA) derivatives, enabling direct measurement of carbon-bound non-exchangeable (CBNE) hydrogen in sugars. The research question is whether δ²H of sucrose CBNE hydrogen measured by GC-CrAg/HTC-IRMS can discriminate authentic pineapple juices from those adulterated with exogenous beet (C3) or cane (C4) sucrose, more rapidly and sensitively than existing methods.

- Stable isotope techniques in juice authentication are established, including δ¹³C and δ¹⁸O analyses and site-specific ²H/¹H by SNIF-NMR (refs 1–6). Codex and AOAC methods formalize these approaches. - δ¹³C distinguishes C4 sugars (corn, cane; typically −10 to −12‰) from C3 products (−23 to −28‰). Internal isotopic correlations (e.g., pulp vs sugars, LC-IRMS of individual carbohydrates) enhance detection limits (refs 11–12). - Limitations: δ¹³C cannot detect C3 beet sugar addition to C3 juices. ²H/¹H (CBNE) reflects geographic and physiological influences; beet sugars (temperate C3) often show lower deuterium content than tropical fruit sugars. - Pineapple (CAM) complicates detection because its global δ¹³C and SNIF-NMR ²H/¹H overlap with C4 sugars. Prior ¹³C SNIF-NMR on ethanol methyl/methylene positions improved C4 detection in pineapple but with ~15% detection limit and significant time/cost (ref 17). - Recent development (ref 18): derivatization of carbohydrates with N-methyl-bis(trifluoroacetamide) enables GC-IRMS measurement of CBNE δ²H of individual sugars with ≤3‰ repeatability, suggesting potential for rapid pineapple juice authentication within ~48 hours.

Study design and samples: - Twenty authentic commercial production samples of pineapple products (single-strength juice n=3, puree n=2, concentrate n=15) were collected by SGF International representatives from production lines in Brazil (n=4), China (1), Costa Rica (1), Indonesia (2), Kenya (1), Philippines (1), South Africa (1), Thailand (8), Vietnam (1). A retail single-strength pineapple juice drink (declared added sugars and citric acid) was also analyzed for demonstration. Samples stored at −18 °C. Concentrates/purees diluted to 12 °Brix before processing. - Retail beet and cane sucrose samples were obtained; their C3/C4 identities were confirmed by bulk δ¹³C. Reagents and references: - Stable isotope CRMs: USGS70, USGS71 (for δ²H normalization to VSMOW-SLAP), USGS40, USGS41 (for δ¹³C normalization to VPDB-LSVEC), IAEA-CH-3 (QC for δ¹³C). n-Hexadecane used as in-house δ²H QC (assigned −89.09‰ VSMOW-SLAP). Chemicals: calcium hydroxide, pyridine, N-methyl-bis(trifluoroacetamide) (MBTFA), standard sugars. Sugar isolation and derivatization: - Pulp removed by centrifugation (10,397×g, 10 min). Calcium hydroxide added to supernatant to pH 8.5, heated at 80 °C ~15 min to precipitate calcium citrate and other acid salts. Precipitate filtered. Supernatant deep-frozen in liquid N2 and lyophilized overnight to yield sugar fraction. - Derivatization: 25 mg sugars reacted with MBTFA (1 mL) in pyridine (1 mL), substituting exchangeable hydroxyl hydrogens with trifluoroacetate (TFA) groups to render sugars GC-amenable and preserving CBNE hydrogens for δ²H analysis. GC-CrAg/HTC-IRMS for δ²H of CBNE hydrogen: - Sugars (sucrose, glucose, fructose) as TFA derivatives separated by GC and passed into a capillary furnace with chromium particles and silver wool at 1200 °C (GC5-BioVisION, Elementar UK). Furnace retains C, O, F; H2 released and measured for isotopologue ratios (¹H¹H, ²H¹H) by IRMS. - δ²H values first measured vs monitoring gas in IonOS, then normalized to VSMOW-SLAP using USGS70/71; n-hexadecane checked normalization per batch. - Precision: Prior repeatability for sucrose-TFA CBNE δ²H ±2.0‰ (1σ, n=18 over 3 days). In this study, triplicate analyses per sample; sample SD across 20 authentic juices 0.2–2.2‰ (mean 1.1‰). EA-IRMS for bulk δ¹³C of sugars: - ~0.2 mg of lyophilized sugars combusted in Elementar Pyrocube coupled to Bio-Vision IRMS. Reactor with WO₃ at 1020 °C; reduction over Cu at 850 °C. - δ¹³C first measured vs monitoring gas, then normalized to VPDB using USGS40/41; IAEA-CH-3 used as QC. Triplicate precision 0.02–0.40‰ (mean 0.12‰). Data analysis: - Summary statistics of sucrose CBNE δ²H and bulk sugar δ¹³C by country and overall. Comparative datasets for retail cane and beet sucrose. - Normal and t-distribution-based interval estimation: Using sample mean (x̄) and SD (σx) from 20 authentic juices (Df=19), critical t-values 2.095 (95% CI) and 2.865 (99% CI) used in t = (x − x̄)/σx to derive expected 95% and 99% population intervals for sucrose CBNE δ²H. - Bivariate assessment: 95% prediction interval ellipses for δ²H (sucrose-TFA CBNE) vs bulk δ¹³C (sugars) computed in Excel to visualize class separation of pineapple vs cane/beet sucrose. - Mixing experiment: Fresh pineapple juice (sucrose CBNE δ²H +14.17±0.69‰, n=4) adulterated with cane sucrose (−101.12±2.8‰, n=4) at 0, 10, 20, 100% w/w of sucrose; measured vs calculated δ²H compared for linearity (reported R, slope, intercept).

- Measurement precision: For 20 authentic pineapple juices, sucrose-TFA CBNE δ²H triplicate SD: 0.2–2.2‰ (mean 1.1‰). Bulk sugar δ¹³C triplicate SD: 0.02–0.40‰ (mean 0.12‰). - Bulk δ¹³C (pineapple sugars): Mean −13.29‰ (range −11.29 to −14.87‰), consistent with AIJN guidance (−13.5 to −11.0‰), with some variation due to CAM dynamics. - Bulk δ¹³C (controls): Beet sucrose mean −26.80‰; cane sucrose −11.66‰ (range −12.26 to −10.80‰), overlapping pineapple range for cane and thus not diagnostic for cane adulteration. - Sucrose CBNE δ²H (pineapple): Overall mean +0.34‰ with SD 18.7‰; country-specific ranges broad (Thailand: −39.6 to +31.2‰; n=8). Indicates substantial natural variation but generally enriched relative to local meteoric water by ~32‰ on average (up to ~75‰). - Sucrose CBNE δ²H (controls): Cane sucrose mean −84.2‰ (SD 17.6‰; range −101.0 to −55.0‰). Beet sucrose mean −157.9‰ (SD 12.0‰; range −169.6 to −145.6‰). - Discrimination capability (univariate δ²H): Using authentic pineapple mean and SD (n=20), expected pineapple sucrose CBNE δ²H ranges: • 95% CI: −38.9 to +39.5‰. • 99% CI: −53.3 to +54.0‰. Values below −53.3‰ (or above +54.0‰) are highly unlikely (P≤0.01) to be authentic. Cane sucrose values (−101 to −55‰) largely lie below the 99% lower bound, with slight potential overlap near −53.3‰. - Mixing linearity: Simulated adulteration of fresh pineapple juice with cane sucrose produced δ²H values following a linear mixing model (R=0.9997, slope 1.0040, intercept +0.6‰), validating quantitative interpretation. - Case example: A retail “50% pineapple juice” drink (label declared added sugars/citric acid) had sucrose CBNE δ²H −86.3±1.5‰ (n=3), well below the −53.3‰ 99% cutoff, classed as highly unlikely to be authentic. - Throughput advantage: Method delivers results within ~48 hours and offers compound-specific information (sucrose focus), improving over time-intensive SNIF-NMR ethanol approaches.

The study demonstrates that δ²H analysis of CBNE hydrogen in sucrose via GC-CrAg/HTC-IRMS provides strong discriminatory power for detecting exogenous sucrose in pineapple juice. While bulk δ¹³C cannot distinguish cane sucrose from pineapple sugars due to CAM-driven δ¹³C overlap, CBNE δ²H separates CAM-derived pineapple sucrose (centered near 0‰; broad distribution −39 to +39‰ at 95% CI) from cane (C4; ~−84‰) and beet (C3; ~−158‰) sucrose. The pronounced hydrogen isotope differences arise from hydrological gradients and plant physiology (evapotranspiration, morphology), with pineapple grown in tropical regions generally more ²H-enriched than temperate beet and distinct from cane. Although expected latitudinal correlations in δ²H are documented for other fruits, the pineapple dataset (largely tropical latitudes, dominated by Thailand) did not show a strong latitude–δ²H trend, potentially due to narrow latitude span and irrigation confounders. Population modeling using t-distribution-derived bounds indicates values below −53.3‰ (99% CI) are highly unlikely for authentic pineapple. Cane sucrose values mostly fall below this threshold, enabling detection of adulteration, supported by the linear mixing experiment that links added cane sucrose proportion to δ²H shifts. A slight potential overlap exists near the 99% lower boundary, and the cane population range may be broader than the small control set, but overall separation remains robust. Bivariate analysis (δ²H vs δ¹³C) with 95% prediction ellipses further assists in classifying samples and assessing the likelihood of C3 (beet) or C4 (cane) additions. Practically, the method offers improved speed (≤48 h), sensitivity, and compound specificity relative to SNIF-NMR approaches that require fermentation and high-field NMR, positioning GC-IRMS CBNE δ²H as a valuable addition to juice authenticity testing. Incorporating internal or intermolecular isotopic references (e.g., citric acid converted to aconitic acid for CBNE δ²H) may further lower detection limits for added cane sugar by leveraging internal isotopic correlations, analogous to protein–sugar δ¹³C pairing in honey authentication.

This feasibility study introduces and validates a GC-IRMS-based compound-specific hydrogen isotope method that measures δ²H of carbon-bound non-exchangeable hydrogen in sucrose (as TFA derivative) to detect exogenous sugars in pineapple juice. The approach reliably differentiates authentic pineapple juices from those adulterated with cane (C4) or beet (C3) sucrose, overcoming limitations of bulk δ¹³C and reducing analysis time compared with SNIF-NMR. Quantitative interpretation is supported by linear mixing behavior and statistically derived authenticity intervals (95% and 99% CIs). The method provides results within about 48 hours and achieves per-sample precision near 1‰ for δ²H. Future work should: (i) expand authentic pineapple datasets across geographies/seasons to better characterize natural δ²H variability and refine decision limits; (ii) evaluate additional sugars and potential internal isotopic references (e.g., CBNE δ²H of citric acid after conversion to aconitic acid) to enhance sensitivity; (iii) further explore bivariate/multivariate models (δ²H with δ¹³C and other isotopes) to improve robustness; and (iv) assess detection limits across a broader range of cane syrups and processing conditions.

- Sample size and representativeness: Authentic dataset limited to n=20 and biased toward Thailand, potentially not fully representative of global pineapple δ²H variability; assumptions of normality may not strictly hold. - Overlap near decision boundary: Slight potential overlap between the lower tail of pineapple δ²H distribution (99% CI at −53.3‰) and the upper tail of cane sucrose distribution (as low as −55‰ in this study; broader literature suggests possible wider range), which may complicate borderline cases. - Geographic and irrigation effects: Hydrogen isotopes in plant sugars are influenced by local meteoric water and evapotranspiration; irrigation with remote waters can confound geographic correlations, adding uncertainty in provenance-based expectations. - Single-compound focus: Primary discrimination uses sucrose CBNE δ²H; variations in sugar composition or processing could affect signal strength or interpretation; extension to other compounds and internal references is proposed but not yet validated here. - Feasibility scope: Study demonstrates concept and initial performance; comprehensive validation (e.g., method detection limits for specific adulteration levels across matrices) requires further work.