Fifteen fruit sticker samples were collected from apples and pears purchased in Southeast Queensland, Australia, during July and September 2022. Three stickers from each type of fruit were collected. Typically, each fruit had one sticker labelled on it. The stickers were carefully removed using clean tweezers. For one variety of apples, stickers were double-labeled. Both the top and bottom labels were collected and analyzed separately.

Naturally dried stickers (three combined per sample) were fortified with 100 ng of ¹³C₁₂-1,5,5,6,6,10-hexachlorodecane, and sonicated for 30 min using 4 mL of hexane:acetone (v:v, 1:1). The extracts were concentrated under N₂ to approximately 3 mL, vortexed twice with H₂SO₄, and supernatants after centrifugation (4,000 rpm for 15 min) were concentrated to approximately 1 mL. A self-packed column based on acidified silica gel (44%) was used for further purification. Detailed information on the sample pre-treatment is provided in Supplementary Texts 1 and 2. After the concentration, samples were then directly injected, without using an analytical column, into a quadrupole time-of-flight mass spectrometer (QToF-HRMS, Triple TOF 5600 Sciex, Concord, Ontario, Canada), scanning m/z 250-1050 [Supplementary Text 2].

The concentrations of PCAs were calculated using the algorithm based on response factors for individual PCA groups (all isomers)^[24]. This method is based on the modelled response factors of PCA congeners using QToF-HRMS, and the calculated concentrations of each PCA congener group. For each PCA congener group, the internal-standard-corrected peak area was converted to concentration using the group-specific response factor. The individual response factor method outperforms the traditional deconvolution algorithm^[25], because it does not require similarity to any standard patterns, offering good accuracy and precision.

Procedure blank samples (n = 5) were prepared by extracting glass fiber filter using the analytical method described above. The levels detected in the blanks were, on average, 1.1 ± 0.3, 4.5 ± 1.4, and 4.6 ± 1.4 ng/sample for PCAs-C_10-13, PCAs-C_14-17, and PCAs-C_>17, respectively. Blank correction was applied to all samples. Limits of detection (LODs) were defined as the average concentrations detected in the procedure blanks plus three times of the standard deviation. The calculated LODs were 2.0, 8.7, and 8.8 ng/sample. The recovery of internal standard (¹³C₁₂-1,5,5,6,6,10-hexachlorodecane), calculated by comparing the internal standard detected in the real sample are those with a non-extracted standard sample, ranged from 72%-104% in the samples. Quality control (QC) spiked samples were prepared by spiking known amounts of PCAs standards into blanks (n = 3) and analyzed together with sticker samples. The differences between calculated results and assigned results of the QC spiked samples were < 20%. Duplicate samples were analyzed for five samples, with the differences < 15% for all the analytes. Average results of the duplicates were therefore used in this study.

The PCA concentrations in this study were reported as ng/sticker. However, for comparison with other products, dry weight-based results (ng/g d.w.) were also estimated by dividing the ng/sticker by the mass of each sticker. Statistical analyses were carried out using SPSS 26.0 (Chicago, IL, USA). Data were log₁₀ transformed prior to statistical analyses. Pearson correlation coefficients were used for the correlation between the concentrations of PCAs estimated from two different methods. Criteria for significance were set at P < 0.05.

Estimated daily intakes (EDIs, ng/day) of PCAs via fruit sticker consumption was calculated based on Equation (1)^[26]

(1) $$ \mathrm{EDIs=DC_i\times C\times R} $$

where DC_i is the daily consumption of fruit (units/day), C is the concentration of PCA in fruit sticker (ng/sticker), and R is the transfer rate of PCAs from fruit stickers to the fruit.

Hazard quotients (HQ) were calculated using Equation (2)

(2) $$ \mathrm{HQ=EDI/(TDI\times BW)} $$

where BW is the body weight for studied population, and TDI is the tolerable daily intake of PCAs (100 μg/kg body weight/day for all PCAs was used in this study^[27]).

PCAs were detected in all 15 fruit sticker samples. A total of 202 PCA congener groups were investigated in this study, including 20 PCA-C_<10, 42 PCA-C_10-13, 62 PCAs-C_14-17, and 78 PCAs-C_18-21 congener groups. Among these, 11 (55%) PCA-C_<10, 36 (86%) PCA-C_10-13, 40 (65%) PCA-C_14-17, and 33 (42%) PCA-C_18-21 congeners were detected in at least one sticker sample. Detailed detection for each congener is presented in the Supplementary Table 1. Detection frequencies declined with the increase in the Cl number of the compounds. For example, Cl_1-3 congeners were detected in 100% of these samples, decreasing to 50% for Cl₁₀ congeners. This indicated the predominance of lower chlorinated congeners in the stickers.

The concentrations of ∑PCAs-C_10-21 in the fruit stickers ranged from 36 to 150 ng/sticker [Figure 1 and Supplementary Table 2]. In addition, the results estimated based on their individual response factors were compared to those quantified by deconvolution, which decomposes the peak intensity of each PCA into a linear combination of patterns from several technical PCAs products. Although significant correlations were found between two sets of data for PCA-C_10-13 (r = 0.722, P < 0.001) and PCAs-C_14-17 (r = 0.631, P = 0.001), results from individual response were consistently lower for PCA-C_10-13 (average, -105%), but higher for PCAs-C_14-17 (43%) and PCAs-C_18-21 (64%). In fact, the goodness of fit (R²) of the multiple regression model for deconvolution in this study was consistently below 0.5, indicating PCA patterns differed from the known standards. Notably, all PCA standard mixtures are dominated by Cl_4-8 congeners, contrasting with the predominance of Cl_1-3 in the fruit stickers observed in this study. Even when Cl_1-2 congeners were removed, the R² values were still below 0.5 for most samples, with the only exception of sticker #11 for PCA-C_10-13 (R² = 0.516). Those results indicated that the individual response factor method may be more reliable and accurate when the profile did not match the patterns in the standards, suggesting broader applications to a variety of samples with complex sources and matrix effects.

The PCA concentrations measured in fruit stickers (0.72-3.0 μg/g) fall within the same order of magnitude as those reported for several established food-contact materials [Supplementary Table 3]. For comparison, PCA-C_10-13 levels in paper packaging (2.8 μg/g) and aluminum foil (2.1 μg/g)^[17], as well as in PP packaging (4.0 μg/g)^[15], were similar to the concentrations detected in fruit stickers in our study. In contrast, fruit stickers showed substantially higher levels of PCAs than PE (0.15 μg/g) and polyethylene terephthalate (PET, 0.23 μg/g) food packaging^[15]. There was no data available on the concentrations of LCCPs in food-contact materials. In addition, the PCA concentrations in the fruit stickers (0.72-3.0 μg/g) were 2-3 orders of magnitude lower than those detected in glue samples (up to 10%) collected in China^[4,23], and other household products (0.2%-50%)^[15,28,29], possibly because of the regulations on food-contact materials. However, these results were higher than the PCA concentrations in fruit, including those collected in Belgium (PCA-C_10-13, < LOQ-1.3 ng/g ww, PCAs-C_14-17, < LOQ-5.2 ng/g ww)^[30], and China (PCA-C_10-13, 59-2,000 ng/g dw)^[31], indicating that the application of fruit sticker on the fruits may result in the potential transfer of PCAs from the stickers to fruits.

PCAs-C_18-21 were the predominant PCAs group in the fruit stickers, contributing 40% (Sticker #11) - 70% (Sticker #4) to ∑PCAs-C_10-21 concentrations [Figure 2]. This is distinct from most known matrices. Due to their physicochemical properties, PCAs-C_18-21 have lower vapor pressures^[32], and higher viscosities^[33]. Therefore, the application of PCAs-C_18-21 in the products, including fruit stickers, could enhance their stickiness. More specifically, C₂₁ congeners had the highest average contribution (24%) to ∑PCAs-C_18-21, followed by C₂₀ (14%), C₁₉ (12%). It’s important to note that recent studies identified C_21-30 congeners^[34,35], therefore, we further investigated the detection of PCAs-C_21-30 congeners. Overall, 186 congener groups were examined, of which 78 were detected. Among these, congeners with Cl number of 1, 2, and 3 were detected in 74%, 92% and 72% of the samples, respectively, whereas detection frequencies for all other congener groups were all below 40%. Due to the lack of the quantification method (i.e., response factors for QToF), we were unable to estimate their concentrations. Given the high contributions of C_19-21 congeners, the concentrations of PCAs-C_18-21, and therefore ∑PCAs-C_10-21, could be underestimated due to the potential presence of C_>21 congeners.

In terms of Cl numbers, Cl₁ and Cl₂ congeners contributed 46%, and 37% to ∑PCAs-C_10-21, respectively. The Cl content of PCA-C_10-13, PCAs-C_14-17, and PCAs-C_18-21 measured in the fruit stickers were 33% ± 5%, 25% ± 3%, and 16% ± 2%, respectively, which were lower than all the known PCA technical mixtures (36%-63%). The distinct profiles detected in the fruit stickers suggested the use of a specific but unknown mixture of PCAs used in their manufacture (i.e., hard paraffins and microcrystalline waxes^[36]), though other factors such as potential degradation or contamination during processing could also contribute to these profiles^[37]. Alternatively, the high contribution of Cl₁ and Cl₂ congeners may also be caused by the breakdown of C–Cl bond during the manufacturing, i.e. thermal degradation, photodegradation, or mechanical stress, though further investigation is needed to confirm the mechanism^[38].

Given the high abundance of long-chain (C_>18) and low chlorine (Cl_1-3) PCAs in the fruit stickers, their potential toxicity needed to be carefully evaluated. To assess this, the 96h LC50 values for Pimephales promelas were estimated from Toxicity Estimation Software Tool (TEST) from US EPA^[39]. Despite the variation among isomers, the overall toxicity increased with both the carbon chain length and chlorine number. For example, the average LC50 of C₂₁H₄₃Cl was similar to that of C₂₀H₃₉Cl₃ and C₁₉H₃₅Cl₅. Thus, the long-chain, low-chlorinated PCAs predominant in fruit stickers were not necessarily less toxic than the more chlorinated PCAs.

To evaluate the potential transfer, a preliminary test was conducted by comparing the concentrations of PCA between a double layer sticker, where the top sticker was placed on the bottom sticker, and the bottom sticker was directly applied on the fruit. The bottom stickers (in contact with the fruit surface) and corresponding top stickers were carefully separated and analyzed individually. The top sticker was assumed to represent the initial PCA concentration prior to any migration, as it remained unexposed to the fruit surface, whereas the bottom sticker was in direct contact with the fruit and thus susceptible to transfer to the fruits. The concentrations of PCAs in the top sticker were consistently higher than those in the bottom one, with the difference of 26%, 36%, and 23% for PCA-C_10-13, PCAs-C_14-17, and PCAs-C_18-21, respectively [Supplementary Table 4]. These differences indicated that PCAs may have migrated from the bottom sticker to the fruit, while those in the top layer remained unchanged. However, it should be noted that the entire sticker (including both adhesive and non-adhesive layers) was analyzed without isolating the adhesive layer. Consequently, the observed concentration reduction in the bottom sticker could also partly result from transfer between the adhesive layer and the backing layer during the experiment. Despite the limited number of samples, this result was consistent with a previous study where PCAs migrated from six types of snack plastic food packaging materials into food simulants^[16].

The relatively high transfer rate (29% on average) is worth noting. The migration of PCAs may result from the exchange between adhesives and fruit surface through micro abrasions during handling and/or storage. However, this mechanism requires further validation. Moreover, there is no study on the migration dynamics of PCAs under different temperatures or contact time. In addition, common consumer practices, such as washing or peeling fruit before consumption were not considered in the exposure assessment. These practices may remove a certain proportion of PCAs from the fruit skin, therefore reducing human intake of PCAs from contaminated fruits. To address these gaps, future research should utilize paired fruit samples with and without stickers to directly estimate the actual transfer rate under realistic consumption conditions.

On average, Australian consumed 143 g/day fruit per day fruits^[40], consisting 0.18 apples or pears per day^[41]. Using the 0.18 apples or pears alone, the average ∑PCAs-C_10-21 concentrations in the fruit stickers of 87 ng/sticker and transfer rate of 29% resulted in the EDI was 4.5 ng/day for Australians, consisting of 0.54 ng/day PCA-C_10-13, 1.8 ng/day PCAs-C_14-17, and 1.9 ng/day PCAs-C_18-21 [Table 1]. Furthermore, when the intake of 143 g/day of fruit consisted entirely of apples or pears, this would be equivalent to approximately one apple or pear per person per day. In the worst scenario, assuming a 100% transfer rate from the fruit sticker and one apple or pear per person per day, the maximum EDI of PCAs increased to 83 ng/day via the consumption of contaminated fruits. It should be noted that the concentrations of PCAs, as well as the associated exposure assessment, were based on quantification using technical mixtures. Although the individual-response model minimizes pattern mismatch, the high proportion of congeners with lower chlorine atoms and longer carbon chains introduced uncertainty due to the lack of certified reference materials and validated quantification methods for these unusual patterns. The EDIs via contaminated fruits were 1-3 order of magnitude lower than the exposure from other pathways, including inhalation^[42], dietary^[43], dust ingestion and dermal contact^[44], indicating that the exposure resulting from fruit stickers were limited.

Risk assessment using International Programme on Chemical Safety (IPCS) thresholds revealed negligible health concerns, where EDIs remained substantially below the tolerable daily intake (100 μg/kg body weight/day for all PCAs)^[27], with the maximum HQ ranging from 1.4 × 10^-6 to 5.7 × 10^-6 for different PCAs. Current regulations, including EU No 10/2011^[20] and U.S. FDA 21 CFR^[45], do not provide specific limits for PCAs in food-contact adhesives. Despite their limited contributions to total EDIs, our results tentatively suggested a potential pathway of PCAs that may be worth considering by policymakers. Other labelling techniques, such as electrically branded fruit label, could be used as alternatives.

This is a preliminary study on the occurrences and levels of PCAs in fruit stickers, with a few samples collected from apples and pears purchased in the markets. The limited number of samples could not reflect the full picture of the contamination status in the fruit stickers. Future research is required to further investigate the levels of PCAs in fruit stickers from more regions and countries. Nevertheless, the high frequency of PCAs in the samples raises concerns regarding their pervasive presence in these stickers.

Furthermore, we employed a newly developed algorithm for the quantification of PCAs, as deconvolution was found to be inaccurate in this study. While the former method passed all QAQC checks, we were unable to assess its accuracy for sticker samples due to their distinct congener patterns.

Additionally, the potential transfer of PCAs between the stickers and fruit was only investigated for a single type of sticker, using double-labelled stickers on the same fruit. We were unable to investigate the migration kinetics at different temperatures and contacting time. A more comprehensive lab scale study would deepen our understanding of the potential risks to human health.