Heavy Metal Index
Ask the IndexAboutIndex Pro

Impact of extraction methods on Monte Carlo based dietary health risk assessment of potentially harmful elements in edible plants

Cite this pageVersion history

Overview

Stolecka and Gruszecka-Kosowska evaluated how the choice of extraction method systematically alters carcinogenic and non-carcinogenic risk estimates for eight potentially harmful elements (arsenic, cadmium, chromium, nickel, lead, antimony, tin, thallium) in commonly consumed edible plants (vegetables, fruits, cereals). Seven extraction methods were applied to the same samples: total content (TC), BCR sequential extraction, US EPA method 1340, ISO method, UBM (Unified Bioaccessibility Method), SBET1, SBET2, PBET1, PBET2. Monte Carlo simulations used Polish dietary consumption data. The TC method consistently produced the highest carcinogenic and non-carcinogenic risk estimates, as it assumes complete gastrointestinal absorption. Gastric-phase extractions yielded higher HQ values than intestinal-phase methods. Sensitivity analysis identified PHE concentration as the main contributor to risk variability. The study demonstrates that bioavailability-based extraction methods yield more realistic and materially lower risk estimates than TC methods.

Key numbers

  • Sample set: 25 plant-based commercial botanical products from southern Poland, collected from fresh markets; CRMs analysed alongside: BCR-679 (white cabbage), WEPAL-IPE-907 (spinach), WEPAL-IPE-951 (aubergine), WEPAL-IPE-182 (wheat).
  • Analytical method: ICP-MS (ELAN 6100, Perkin Elmer) according to US EPA Method 6020A and ISO 17294-2:2003 protocols.
  • LOQs (mg/L in extract): arsenic < 0.001, cadmium < 0.0003, chromium < 0.005, nickel < 0.001, lead < 0.0001, antimony < 0.0002, tin < 0.001, thallium < 0.0001.
  • CRM recoveries fell within 80-120% for most analysed PHEs. RSD for replicate measurements less than or equal to 5%.
  • Reference doses (RfD) used for HQ calculations (mg/kg body weight/day): arsenic 3.0 x 10^-4, cadmium 1.0 x 10^-4, chromium 9.0 x 10^-4, nickel 2.0 x 10^-2, lead 1.5 x 10^-3, antimony 4.0 x 10^-4, tin 6.0 x 10^-1, thallium 1.0 x 10^-5.
  • Cancer slope factors (SFo, (mg/kg body weight/day)^-1): arsenic 1.5; chromium 0.16 (for Cr-VI).
  • Mean PHE concentrations in vegetables (TC method, mg/kg dry weight): arsenic below LOQ; cadmium 0.165 +/- 0.071; chromium 0.822 +/- 0.292; nickel 1.32 +/- 0.870; lead below LOQ; antimony 0.029 +/- 0.025; tin 0.298 +/- 0.105; thallium 0.032 +/- 0.052 (Table 3).
  • Mean PHE concentrations in fruits (TC method, mg/kg dry weight): arsenic below LOQ; cadmium 0.045 +/- 0.034; chromium 1.69 +/- 1.83; nickel 1.13 +/- 1.32; lead below LOQ; antimony 0.016 +/- 0.009; tin 0.285 +/- 0.296; thallium 0.005 (Table 3).
  • Mean PHE concentrations in cereals (TC method, mg/kg dry weight): arsenic below LOQ; cadmium 0.054 +/- 0.036; chromium 2.58 +/- not stated; nickel 0.970 +/- 0.435; lead 0.632 +/- 0.128; antimony 0.010; tin below LOQ; thallium 0.007 +/- 0.005 (Table 3).
  • Exposure parameters (Monte Carlo probability distributions): body weight lognormal (67.55-72.85 kg); exposure duration uniform (0-24 years); ingestion rates vegetables 200.5-292.5 g/day, fruits 135.17-261.83 g/day, cereals 142.57-299.43 g/day; averaging time 365 x exposure duration.
  • Key finding on method comparison: TC-based HQ consistently exceeded bioavailability-based HQ estimates; gastric-phase methods produced higher HQs than intestinal-phase methods for most PHEs. SBET gastric produced the highest non-carcinogenic risk values after TC.
  • Sensitivity analysis: PHE concentration was the dominant input variable; ingestion rate had a smaller but notable effect. Body weight had negligible influence.
  • MOE for lead: calculated using EFSA lead (Pb) Reference Dose (1.2 micrograms/kg body weight/day, BMDL for cardiovascular effects in adults). DIR (Daily Intake Rate of Pb from edible plant category) = (C x IR) / BW. MOE values < 1 indicate concern; values > 1 indicate low and acceptable risk.

Methods (brief)

Edible plant samples were prepared as for consumption (washed, peeled if needed, cut to small pieces), dried at 70 degrees Celsius under radiant lamps, ground to coarse powder, and stored in sealed bags. Seven extraction procedures were applied uniformly to all plant samples and PHEs. TC: microwave digestion 130 degrees Celsius, 2 h. BCR: 0.11 M acetic acid, room temperature, 16 h shaking. US EPA: 0.4 M glycine pH 1.5, 37 degrees Celsius, 1 h shaking. SBET1: 0.4 M glycine pH 1.5, 37 degrees Celsius, 1 h; SBET2: same with 0.5 M NaHCO3 pH 7. PBET1 (gastric): pepsin, malate, citric acid, lactic acid, pH 2, 37 degrees Celsius, 1 h shaking. PBET2 (intestinal): 0.1 M NaHCO3, pancreatin, bile salts, pH 6-8, 37 degrees Celsius, 1 h. UBM (gastric): saliva, gastric digestive fluids, pH 1.15-1.25, 37 degrees Celsius, 1 h; UBM (intestinal): saliva, duodenal, bile digestive fluids, pH 5.8-6.8. ISO: ISO gastric digestive fluids, pH 1.15-1.25, 37 degrees Celsius, 1 h; ISO intestinal: ISO gastric, duodenal, bile digestive fluids pH 5.8-6.8, 37 degrees Celsius, 4 h. ICP-MS used for all elemental determinations after extraction. Monte Carlo simulations with 10,000 iterations using @RISK software (Lumivero). Analysis restricted to adult population because child consumption data were limited or incomplete.

Implications

  • Certification (HMTc): This paper is primarily methodological, not an occurrence study. Its principal contribution is the demonstration that choice of extraction method can alter non-carcinogenic HQ estimates by factors of several-fold. For HMTc standards based on literature occurrence data, it is important to record whether reported concentrations are TC or bioavailability-adjusted, as mixing these in a pooled distribution without correction will bias the percentile calculation.
  • Courses: Excellent case study for teaching the distinction between total content and bioaccessible content, and why the same reported concentration can produce materially different risk estimates depending on the extraction method used by the original study.
  • App: The PHE concentrations in this source are not intended as representative occurrence data for general population exposure estimation. They are from a specific 25-sample set in southern Poland used to compare methods, not to characterise market-level distributions.
  • Microbiome: No microbiome data.

Update history

No substantive edit history is available in this build. The full commit record is available in git.

Referenced by