Tsegay et al. 2025 — Toxicological qualities and detoxification of fruit by-products: review

Open-access narrative review (CC BY 4.0, De Gruyter) synthesizing toxicological data on by-products (peels, pomace, seeds, kernels, rinds) of the world’s highest-production fruits as of 2022, motivated by the valorization push that increasingly routes these by-products into food ingredients, nutraceuticals, cosmetics, and animal feed. Scope is global; the review pulls from refs [1–142] without restricting to any single jurisdiction, growing region, or analytical method. Hazards examined are heavy metals (As, Cd, Co, Cr, Ni, Pb, Hg), mycotoxins (aflatoxins, OTA, fumonisins, ZEN, T-2, alternaria toxins, tryptoquialanines), anti-nutritional factors (oxalates, phytates, tannins, alkaloids, saponins, HCN, glycosides), toxicant organic compounds (biogenic amines, AMG, phthalates), and pesticide/fungicide residues. Final section reviews thermal, chemical, and non-thermal detoxification methods (high-pressure processing, pulsed electric field, cold plasma, supercritical CO₂, ultrasound, ozone treatment, lyophilization, fermentation).

Key numbers

Hazard quotient ranges cited from upstream literature (ref [2]) for fruit by-products generically (the review does not attribute these to any specific commodity or sample set): As 2.71–11.38, Cd 0.60–3.32, Cr 0.81–3.18, Co 0.03–0.09, Cu 0.09–0.26, Ni 0.08–0.34, Pb 0.83–2.23 (mg/kg basis stated in the review; HQ values above 1.0 indicate concern in cited contexts).

Table 1 (dominant heavy metals in fruit by-products, all mg/kg unless noted):

• Apple peel: Pb <1, Cd <1, Cr 0.57–3.8, Ni <1 (refs [37,40,41]).
• Apricot kernel: Cd 0.1–6, Co 2.7–35.7 (ref [42]).
• Avocado seed: Cr 0.57–2.29, Co 0.00, Pb 0.00 (refs [40,43]).
• Banana peel: As <0.0001, Cd 0.0013–0.18, Cr 1.42–4.04, Co 0.4–47.2, Ni 0.592, Pb 0.0038–0.64 (refs [36,44,45]).
• Blueberry pomace: Cd 0.011, Cr 0.242, Co 0.08, Ni 0.73 (ref [46]).
• Grape peel/skin/pomace: As 0.004, Cd <0.5, Cr 0.18–2.41, Ni <0.5, Pb 0.021–1.11 (refs [34,35,37]).
• Lemon peel by-products: Cd 0.00047–0.25, Cr 1.04, Co 0.038, Ni 0.973–1.24, Pb 0.0188–0.22 (refs [34,36,39]).
• Lime by-products: As ND, Cd 0.003, Co 0.073, Ni 1.678, Pb 0.128 (ref [39]).
• Mango (peel + seed kernel composite): Cd 0.33, Co 28.0 (ref [45]).
• Orange peel: Cd <0.5, Cr 1.04–4.14, Co 0.015, Ni 0.05–2.36, Pb 0.01–1.75 (refs [34,36,37,39,47]).
• Papaya peel + seed: As 0.0287–0.03, Cd 0.0027–0.00685, Cr 0.278–7.36, Co 0.4–219, Ni 0.246–13.2, Pb 0.03–0.044 (refs [45,48,49]).
• Peach: Cr 0.17–1.38, Pb <0.10 (ref [50]).
• Pineapple peel: As <0.0001, Cd 0.0074, Cr 8.77, Co 70.3, Pb 0.0027 (refs [44,45]).
• Plum peel: As 1.2, Cd ND, Ni 2.8, Pb ND (ref [51]). Plum kernel: As <0.1, Cd 0.13, Co 0.2, Ni 1.7, Pb 0.13 (ref [52]).
• Pomelo peel: Cd 1.36 × 10⁻³, Pb 0.0296 (ref [34]).
• Raspberry pomace: Cd 0.0084, Cr 0.116, Co 0.073, Ni 0.762, Pb 0.047 (ref [46]).
• Strawberry pomace: Cd 0.0078, Cr <0.01, Ni 0.00336, Pb 0.0011 (refs [53,54]).
• Mandarin: Cd 0.00062, Pb 0.020 (ref [38]).
• Watermelon peel + seed: Cd 0.008–0.1, Cr 4.65, Pb 0.06–0.09 (refs [40,55]).

Single-paper outlier (peel): Pineapple skin reported at 0.64 mg/100 g Pb (= 6.4 mg/kg) dry weight per refs [13,14] — above the WHO 0.02 mg/100 g (= 0.2 mg/kg) maximum limit for edible parts of crops cited in the review. Six other by-products in the same comparison (orange peel, watermelon rind, banana peel, apple pomace, strawberry pomace, grape pomace) showed Pb below 0.1 µg/L (a unit the review states without converting to mg/kg; not reconciled in the source text).

Banana peel ANFs (M. sinensis, fresh, ref [15]): alkaloid 0.66 g/100 g, oxalate 37.0 g/100 g, phytate 2.78 g/100 g, saponin 6.57 g/100 g. Same paper reported boiled banana fruit extract values higher: alkaloid 1.76 g/100 g, oxalate 40.2 g/100 g, saponin 8.12 g/100 g.

Regulatory limits cited in review (mg/kg unless noted):

• WHO maximum for Pb in edible parts of crops: 0.02 mg/100 g (= 0.2 mg/kg) (ref [12]).
• WHO maximum for As and Pb (monthly intake): 0.21 mg/100 g (ref [12]).
• WHO maximum for Cd: 0.25 mg/100 g (ref [12]).
• WHO maximum for Al: <0.2 mg/100 g (ref [12]).
• EU Regulation 1275/2013 / Ordinance 10/2009 (animal feed): 1 mg/kg Cd, 10 mg/kg Pb (ref [38]).
• European Commission MRL for citrus by-products: As 0.02, Cd 0.02, Hg n/a, Co n/a, Ni n/a, Pb 0.1 mg/kg (ref [56]).
• WHO/FAO RDI thresholds for orange by-product analytes: Cr 0.05–0.2, Ni 1.4, Pb 0.214 mg/day (ref [87]).
• FAO/WHO ADI: Pb 0.21–0.25 mg/day, Cd 0.06–0.07 mg/day (ref [55]).

Mycotoxin highlights (mg/kg, from Table 2 of the review):

• Apple peel/skin/seed/pomace: AOH-3-S 1.4–10.8 × 10⁻³, AME-3-S 1.7–10 × 10⁻³ (refs [57–61]).
• Apricot seeds: AFB1 and AFB2 = 0.0017–22.451 (refs [62–64]).
• Pineapple shell: Fusarium = 250 (refs [70–72]). Pineapple pomace: Aflatoxin B₂ and B₁ = 0.008 × 10⁻³, Aflatoxin G₁ = 0.013–0.033 × 10⁻³, OTA = 0.051 × 10⁻³ (refs [60,72,79]).
• Tangerine peel: tryptoquialanine A = 248.1, tryptoquialanine C = 375.80 (refs [34,60,69,72–75]).
• Watermelon peel: AOH = 0.003–0.017 (refs [60,81,82]).
• Pineapple shell mycotoxins: aflatoxin B₂, aflatoxin G₁, and OTA all measured below detection limit 2.9 × 10⁻³ mg/kg (ref [77]).
• Stępień et al. [78]: pineapple skin infected with mycotoxin mainly Fusarium 250 mg/kg (cited US FDA mycotoxin exposure guideline for total fusarium in cereal foods is 2–4 mg/kg — paper notes the cited value as hazardous without pretreatment).

ANF outliers (mg/kg):

• Orange peel oxalates: 997.8 (Table 2).
• Watermelon meal: phytate 9,900, tannin 32 × 10⁵, oxalate 2,130 (ref [81]).
• Mango kernel flour ANFs: tannins 1027.4, phytic acid 1149.8, oxalates 213.4, HCN 0.00 (ref [63]).
• Pineapple pomace: HCN 715, alkaloids 161.9.
• Apricot seed AMG (amygdalin): 52,000 mg/kg (ref [62]) — review notes 3.0 mg/kg body weight oral dose did not induce changes in rabbit spermatozoa.

Pesticide outliers (mg/kg):

• Mango peel: chlorpyrifos and methiocarb residues within Codex MRL (refs [65,66]).
• Orange pomace: imidacloprid 162.16, carbendazim 372.1, abamectin 0.261, cypermethrin 495.6, prochloraz 8.11.7 (refs [34,60,69,72–75]).
• Pomelo: total pesticides 2.143 mg/kg, procymidone 0.7 (ref [34]).

Fruit waste percentages (Figure 1b, % of fresh fruit weight discarded as by-product): pineapple peel + crown + core 60%, orange pomace 50%, lemon pomace 50%, mango peel + seed 45%, banana peel 35%, watermelon rind 33%, kiwi seeds 30%, melon peel + seed 30%, papaya peel + seed 29.5%, peach peel + seed 24.8%, grape pomace 20%, cashew-apple bagasse 20%.

Global production 2022, million tonnes (Figure 1a, FAOSTAT): bananas 135.1, watermelons 100.4, apples 95.8, oranges 76.4, mangoes/guavas/mangosteens 59.2, tangerines/mandarins/clementines 44.2, pineapples 29.4, melons 28.6, peaches/nectarines 26.4, pears 24.6, plums/sloes 12.4, papayas 13.8.

Cytotoxicity findings noted:

• Banana peel aqueous extract: no cytotoxicity to CAL 27, Caco-2, or HepG2 at 0.014–10 mg/mL across 0.5/1/2 h treatments; 10 mg/mL polysaccharide-free aqueous extract showed slight proliferation in Caco-2 (ref [105]).
• Pomelo seed oil: IC₅₀ in LO₂ and HepG2 cells 16.31–28.8 mg/mL (non-cytotoxic per the >1 mg/mL criterion stated by the review) (ref [93]).
• Avocado peel/seed (Hass and Fuerte): no cytotoxicity to RAW 264.7 cells at 0.1–100 µg/mL (ref [103]).
• Citrus deliciosa peel EO (mature fruit): IC₅₀ 110 µg/mL against HT-29 cells (ref [94]).

Methods (brief)

Narrative review, not a systematic review. The authors prioritized fruits with global production exceeding 1.2 million tonnes in 2022 per FAOSTAT (ref [22]). No PRISMA-style search strategy, inclusion/exclusion criteria, or risk-of-bias assessment is reported. No pooled estimates or meta-analytic synthesis. All numerical values are extracted from cited primary literature (refs [1–142]) and presented in Tables 1 (heavy metals), 2 (mycotoxins/toxicant organic compounds/ANFs/pesticides), and narrative subsections. Sample sizes, analytical methods, LODs, and matrix-prep conditions are not consistently reported for each entry — many cells in Table 1 give a single concentration or a range without n or method context.

Editorial timing: received 14 October 2024, accepted 24 March 2025. Open-access publication under CC BY 4.0.

Limitations explicit to this review’s evidence quality:

1. Tertiary synthesis. All concentrations are second-hand from cited primary studies. Disagreements between primary sources are noted in Table 1 (e.g., banana peel Cd 0.0013–0.18 mg/kg spans two orders of magnitude across refs [36,44,45]) but not arbitrated.
2. Units are inconsistent across the review body and tables. mg/kg, µg/L, mg/100 g, g/100 g, and dry-weight versus wet-weight bases are used in adjacent sentences without harmonization. The pineapple Pb 0.64 mg/100 g vs other by-products “below 0.1 µg/L” contrast is the clearest example.
3. Speciation absent. Heavy metal entries are reported as total element (As, Hg, Cr) without iAs/tAs, MeHg/tHg, or Cr-VI/Cr-III breakdown. Toxicologically critical for As, Hg, and Cr.
4. Hazard quotient ranges (As 2.71–11.38, etc.) are cited from a single upstream review (ref [2]) without identifying the underlying contamination concentrations, exposure scenarios, body-weight assumptions, or consumption rates used to compute them.
5. Pineapple peel Pb outlier (0.64 mg/100 g dry weight) is presented without sample n, geographic origin, growing-region soil background, or replication context. A single dry-weight measurement reported as a hard maximum should not be over-weighted in any downstream synthesis.
6. No regulatory-event documentation. The review cites WHO and EC limits as comparators but does not document any enforcement actions, recalls, or jurisdictional MRL exceedances tied to by-product valorization.
7. Suspected source-printing errors preserved verbatim in Key numbers (the review is the authoritative record; we do not silently correct):
   • Watermelon meal tannin “32 × 10⁵ mg/kg” (Table 2, ref [81]) implies 3,200 g/kg, which exceeds physical possibility (>100% of dry mass). Most likely intended 32 × 10² (= 3,200 mg/kg) or similar; do not propagate as a quantitative input without re-checking ref [81].
   • Orange pomace “Prochloraz = 8.11.7 mg/kg” (Table 2, refs [34,60,69,72–75]) is a malformed numeric (extra decimal). Do not propagate as a quantitative input without re-checking the cited refs.

Implications

For certification: Pineapple peel, mango peel/kernel, orange peel, and watermelon rind concentrate metals (notably Cd, Cr, Co, Pb) and ANFs relative to whole-fruit edible flesh. Programs certifying derivative ingredients sourced from these by-products (fruit-fiber concentrates, peel-derived pectin, seed-meal flours, pomace press-cakes used as functional ingredients) cannot rely on whole-fruit contamination profiles. This review does not provide enough primary-data structure to set thresholds directly, but it establishes that by-products as a functional ingredient class warrant separate analytical treatment from the edible fraction.

For app modeling: Insufficient sample-size, LOD, and method-context disclosure to support ppb-level exposure modeling for any specific by-product ingredient. Use as a routing pointer to the primary studies cited in Tables 1 and 2 rather than as a quantitative input. The review’s chief contribution to app modeling is signaling which fruit by-products are studied at all — many tropical fruits (papaya, pomelo, cherimoya) and stone fruits beyond apricot have only one or two primary citations.

For courses and synthesis pages: Useful as the entry point for the fruit-by-product valorization domain. Sections 7 (“Current trends on fruit by-product toxicant reduction”) and Figure 4 catalog the detoxification methods (HPP, PEF, cold plasma, supercritical CO₂, ultrasound, lyophilization, fermentation) that are increasingly relevant to ingredient suppliers — material for an ingredient-handling course module.

Verification notes

2026-05-20 (merge-enhance pass). Pre-existing page dated 2026-05-14 carried defects from earlier ingest:

• raw_handle: manual-fetch-kimi replaced with MFK_toxicological-qualities-and-detoxification-trends- per current MFK_ handle convention.
• raw_path was truncated to ...fruit by-pro.pdf; corrected to the full PDF filename ...fruit by-products_ A review.pdf.
• raw_sha256 was missing; added (8f66d00ee3ae954d5e379415834e79667cf20b21a02edb6e95752154c8d69e36).
• access_url was missing; added DOI link.
• metals was [tAs, Pb, Cd, Cr, Ni, tHg]; Co added — the paper’s Table 1 reports Co for apricot kernels, avocado seeds, banana peel, blueberry pomace, lemon, lime, mango, orange, papaya, pineapple, plum, raspberry, watermelon, with single-paper outliers up to 219 mg/kg (papaya, ref [45,48,49]).
• ingredients previously listed invented sub-variant slugs (banana-peel, apple-pomace, orange-peel, grape-pomace, mango-peel, pineapple-peel, watermelon-rind, strawberry-pomace) that do not exist in the wiki’s ingredient taxonomy and would route to nothing. Replaced with the seventeen valid whole-fruit parent slugs that match the species discussed: banana, apple, oranges, grapes, mango, pineapple, watermelon, avocado, peach, plums, blueberries, strawberries, raspberries, grapefruit, clementines-tangerines, pear, fruit. The by-product framing is preserved via matrices and the body.
• matrices retained as [fruit-by-products, fruit-peels, fruit-pomace, fruit-seeds] — these accurately describe the review’s subject (by-product fractions, not fresh fruit) and the existing routing audit treats this set as advisory-only.
• products: [] and jurisdictions: [] retained — review article with no specific commercial product framing and global scope.
• Key numbers expanded from ~5 lines to full Table 1 commodity-by-commodity breakdown (all 19 commodities with their per-fraction metal concentrations and refs), plus Table 2 mycotoxin/ANF/pesticide outliers, regulatory-limit citations, global production 2022 in million tonnes from Figure 1a, by-product waste percentages from Figure 1b, and the four cytotoxicity findings reported in Sections 3–6.
• Methods section expanded to flag six explicit evidence-quality limitations: tertiary synthesis nature, unit inconsistency, absent speciation, hazard-quotient citation gap, single-measurement Pb outlier risk, and absence of regulatory-event documentation. The original page omitted these.
• Legacy “Wiki pages updated on ingest” section removed (replaced by the new schema’s expectation that downstream propagation happens during synthesis, not ingest).
• ingest_method and created_by fields added per current frontmatter schema.

No primary-data values were silently altered. The review’s pineapple peel Pb outlier (0.64 mg/100 g dry weight) and the contrasting “below 0.1 µg/L” framing for the other six by-products are preserved verbatim as they appear in the source on p.2 even though the unit basis is inconsistent — the review’s own framing is the authoritative record.

Audit subagent (2026-05-20) — REVISE verdict; three ⚠️ findings, all applied:

1. Pineapple shell vs pomace mycotoxin attribution — the original “Pineapple shell/pomace” collapsed row was split per Table 2 p.7 into “Pineapple shell: Fusarium = 250” (refs [70–72]) and “Pineapple pomace: Aflatoxin B₂ and B₁ = 0.008 × 10⁻³, Aflatoxin G₁ = 0.013–0.033 × 10⁻³, OTA = 0.051 × 10⁻³” (refs [60,72,79]). The aflatoxin label was also corrected from “B₁” to the source’s “B₂ and B₁”.
2. Watermelon meal tannin (32 × 10⁵ mg/kg) and orange pomace Prochloraz (8.11.7 mg/kg) added as bullet 7 in the limitations list, explicitly flagged as suspected source-printing errors preserved verbatim. The review is the authoritative record; we do not silently correct primary-source typesetting.
3. Matrices vocabulary clarification: matrices: [fruit-by-products, fruit-peels, fruit-pomace, fruit-seeds] are descriptive of the review’s actual subject (by-product fractions, not fresh fruit) and are treated as advisory-only by the routing audit. They are not enumerated in the taxonomy snapshot’s controlled-vocabulary list, but they remain in frontmatter because removing them would mis-describe the source.

Page history

The five most recent substantive edits to this page. The full version history lives in git; when DOI minting comes online (see schema docs), each entry below will also link to a version-pinned DataCite DOI.