Impact of polyphenol oxidase on the bioavailability of flavan-3-ols in fruit smoothies
Ottaviani JI, Ensunsa JL, Fong RY, Kimball J, Medici V, Kuhnle GGC, Crozier A, Schroeter H, Kwik-Uribe C · Food & Function 14(18):8217–8228 · 2023 · DOI: 10.1039/d3fo01599h
Citation impact (DOI lookup, 2026-05-20): 17 citations · FWCI = 3.25 · citation_percentile = 99th
TL;DR
A controlled, single-blinded, crossover study in healthy men found that blending flavan-3-ols into a high-PPO banana smoothie reduces peak plasma SREM concentration by 84 ± 9% compared with capsule delivery (Cmax: 96 ± 47 vs 680 ± 78 nmol/L, mean ± SEM; p = 4 × 10⁻⁵) and reduces AUC by 81 ± 11% (272 ± 159 vs 2259 ± 279 nmol/L; p = 4 × 10⁻⁵). Phase 1 was non-randomized; Phase 2 was randomized. A low-PPO mixed-berry smoothie preserved bioavailability at capsule-equivalent levels (Cmax 659 ± 104 nmol/L; p = 0.818). A follow-up phase (n = 11) showed that even preventing pre-blending contact — allowing only gastric co-ingestion via alternating sips — still reduced AUC by 37 ± 6% (p = 0.004) though Cmax was not significantly different, implicating post-ingestion stomach-phase PPO activity as a continuing degradation mechanism. The study is the strongest controlled human evidence that food matrix (specifically PPO-containing fruits co-ingested with flavan-3-ol sources) is a major determinant of polyphenol bioavailability. Registered as NCT03526094.
Background and rationale
Flavan-3-ols (catechins, epicatechins, and their oligomers — the procyanidins) are dietary bioactives found in cocoa, tea, berries, apples, and grapes, associated with cardiovascular, metabolic, and cognitive health benefits mediated partly through anti-inflammatory effects and eNOS/nitric oxide pathway modulation. Smoothies are a widely promoted vehicle for increasing fruit and vegetable intake, but many common smoothie ingredients (particularly bananas) contain high levels of polyphenol oxidase (PPO), an enzyme that oxidises polyphenol substrates — including catechins and epicatechins — to ortho-quinones, which polymerise and are poorly absorbed. Whether this enzymatic degradation is relevant at normal blending conditions, and whether physical separation until ingestion can bypass it, was the central question this study addressed.
Study design
| Feature | Detail |
|---|---|
| Design | Controlled, single-blinded, crossover. Phase 1 non-randomized (order of test materials fixed to control for natural fruit composition variation); Phase 2 randomized (computer-generated allocation) |
| Phase 1 (n=8) | Three conditions in crossover: (a) high-PPO banana-based smoothie (8926 ± 389 IU mL⁻¹ PPO activity; 417 mL serving) + standardized cocoa extract flavan-3-ols, (b) low-PPO mixed-berry smoothie (96 ± 5 IU mL⁻¹ PPO activity; 590 mL serving) + flavan-3-ols, (c) flavan-3-ols in capsule + 1% fat milk (control) |
| Phase 2 (n=11) | Flavan-3-ols co-ingested simultaneously with high-PPO banana drink by alternating sips (FD + BD), no pre-ingestion contact between the two; flavan-3-ol drink alone (FD) served as control |
| Population | Healthy male adults, age 25–60 years (mean 30 ± 8 yr Phase 1; 30 ± 7 yr Phase 2); BMI <30 kg m⁻²; BP <140/90 mmHg; no history of CVD, stroke, renal, hepatic, or thyroid disease, GI disorders |
| Primary readout | Peak plasma concentration (Cmax) of structurally related (−)-epicatechin metabolites (SREMs) over 6 h post-intake |
| SREM panel | Structurally related (−)-epicatechin metabolites (SREMs): main circulating forms confirmed as (−)-epicatechin-3′-sulfate, (−)-epicatechin-3′-glucuronide, 3′-methoxy-(−)-epicatechin-5-sulfate, and 3′-methoxy-(−)-epicatechin-7-sulfate; quantified by UPLC-MS/MS (Waters Acquity UPLC with Waters Quattro Micromass spectrometer; negative-ion MRM mode) |
| Flavan-3-ol dose | Phase 1: 541 mg total flavan-3-ols added to capsule condition (75 mg (−)-epicatechin); 638 mg added to banana smoothie (88 mg (−)-epicatechin); 484 mg added to mixed berry smoothie (68 mg (−)-epicatechin). Phase 2: 88 mg (303 µmol) (−)-epicatechin per drink from standardized cocoa extract (CocoaproTM, Mars Inc.) |
| Washout period | Minimum 6 days between study days (habitual diet during washout; low flavan-3-ol diet protocol on day before and on study day) |
| Preregistration | Registered at ClinicalTrials.gov as NCT03526094 |
Phase 2 rationale: If blending-induced PPO activity (during the ~minutes of blending and storage before consumption) were the sole mechanism, preventing pre-contact should fully restore absorption. That it did not implicates active PPO persisting in the gastric environment — banana PPO remains partially active at gastric pH during the early post-prandial window.
Results
Phase 1 — Cmax and AUC comparison 1
All values are mean ± SEM (not SD). Statistical analysis: linear mixed model with treatment as fixed effect and subject and sequence as random effects; p-values vs capsule control.
| Condition | n | Cmax (nmol/L) | p vs capsule | AUC₀₋₆ h (nmol/L) | p vs capsule |
|---|---|---|---|---|---|
| Capsule (control) | 8 | 680 ± 78 | — | 2259 ± 279 | — |
| Low-PPO mixed-berry smoothie | 6 | 659 ± 104 | 0.818 (ns) | 1941 ± 294 | 0.362 (ns) |
| High-PPO banana smoothie | 6 | 96 ± 47 | 4 × 10⁻⁵ | 272 ± 159 | 4 × 10⁻⁵ |
- The banana smoothie Cmax was 84 ± 9% lower than capsule; AUC was 81 ± 11% lower than capsule (as reported in text — these are means ± SEM of the within-subject percentage differences).
- The mixed-berry smoothie achieved capsule-equivalent bioavailability for both Cmax and AUC (both p > 0.3).
- Secondary readout: gut microbiota-derived γ-VLM (5-(3′,4′-dihydroxyphenyl)-γ-valerolactone metabolites) plasma concentrations at 6 h: capsule 342 ± 97 nmol/L; banana smoothie 194 ± 96 nmol/L; mixed-berry smoothie 600 ± 169 nmol/L. No significant differences between conditions for γ-VLMs.
Phase 2 — separated co-ingestion 1
Even when physical contact between banana drink and flavan-3-ols was prevented until swallowing (alternating-sip protocol), plasma flavan-3-ol levels were still substantially reduced. Analysis by paired t-test (n = 11):
| Condition | Cmax (nmol/L) | p vs FD alone | AUC₀₋₆ h (nmol/L) | p vs FD alone |
|---|---|---|---|---|
| FD alone (flavan-3-ol drink, no banana) | 622 ± 115 | — | 1702 ± 300 | — |
| FD + BD (simultaneous alternating-sip co-ingestion) | 429 ± 80 | n.s. | 988 ± 150 | 0.004 |
- Cmax was not significantly different (p = n.s.), but AUC was reduced by 37 ± 6% (p = 0.004 vs FD alone).
- 24-h urine SREMs and γ-VLMs were also significantly reduced after FD + BD vs FD (both p < 0.05, paired t-test).
- The partial-but-significant AUC reduction despite no pre-ingestion contact implicates post-ingestion gastric PPO activity as a contributing degradation mechanism.
PPO activity survey (Table 4)
PPO activity was measured in 18 fruits, vegetables, and plant-derived dietary products using (−)-epicatechin as substrate. Units: KU per 100 g edible portion (mean ± SEM, n = 3). Selected values:
| Product | PPO activity (KU/100 g) |
|---|---|
| Banana | 3258 ± 71 (highest) |
| Beet greens | 1594 ± 24 |
| Apple (red delicious) | 570 ± 27 |
| Pear | 147 ± 4 |
| Beets | 94 ± 5 |
| Peach | 41 ± 2 |
| Avocado | 24 ± 5 |
| Strawberry | 18 ± 1 |
| Wheatgrass | 15 ± 1 |
| Blueberry, highbush | 12 ± 1 |
| Cucumber | 10 ± 1 |
| Parsley | 6 ± 1 |
| Mango | 6 ± 1 |
| Orange, pineapple, kale | below level of detection |
| Spirulina (powder) | below level of detection |
| Cocoa powder, unsweetened | below level of detection |
Banana PPO activity was ~204-fold higher than blueberry, confirming the extreme range across common smoothie ingredients. Beet greens and apple also carry substantial PPO activity. Berries (strawberry, blueberry), kale, and spirulina are low-PPO and safe bases for flavan-3-ol co-delivery.
Implications for flavan-3-ol delivery
Actionable guidance from this study:
- Banana as a smoothie base substantially impairs flavan-3-ol bioavailability. Blending banana with cocoa, berries, matcha, or green tea — common smoothie combinations — is expected to substantially reduce the absorbed dose of flavan-3-ols from those sources.
- Separation does not fully protect. Even consuming banana and flavan-3-ol sources simultaneously but without pre-blending contact still reduces bioavailability, suggesting gastric enzymatic degradation continues after ingestion. Eating them at separate meals is the more protective strategy.
- Berry smoothies are a low-PPO alternative. Mixed-berry smoothies preserved capsule-equivalent bioavailability, making them a suitable base when flavan-3-ol delivery is the goal.
- Heat denaturation of PPO (e.g., briefly heating banana before blending, or using frozen banana where enzymatic activity may be reduced) is a potential mitigation, though this was not directly tested in this study. PPO is sensitive to temperatures >70°C. needs-replication — heat-denaturation mitigation not tested in this study.
Relevance to the Mediterranean diet polyphenol-delivery question: The Mediterranean diet achieves much of its cardiovascular benefit through high polyphenol intake from multiple sources (olive oil polyphenols, red wine/grape-derived procyanidins, cocoa, berries, tea). The food-matrix findings here imply that co-preparation method affects realized intake even when food selection is correct — a variable not typically controlled in observational diet studies.
Extrapolation table
| Dimension | Status |
|---|---|
| Pathway conserved in humans? | yes — human controlled study; no extrapolation required |
| Phenotype conserved in humans? | yes — plasma SREM absorption measured directly in humans |
| Replicated in humans? | partial — Phase 1 and Phase 2 arms internally replicate the finding; independent replication in larger/mixed-sex cohort pending needs-replication |
Limitations
- Small n: Phase 1 n=8, Phase 2 n=11. Effect size is very large (84%) so statistical power is adequate for the primary comparison, but subgroup or dose-response analyses are underpowered.
- Healthy men only, age 25–60: No women enrolled (mean age 30 ± 8 yr Phase 1; 30 ± 7 yr Phase 2). Results may not generalise across sex or age groups given differences in gastric pH, transit time, and gut microbiome composition. The authors explicitly note this limitation (Discussion).
- Single acute dose: No chronic intake measured. Whether habitual banana + flavan-3-ol co-consumption over weeks causes equivalent bioavailability impairment or triggers adaptive gut responses (e.g., microbiome-mediated flavan-3-ol ring-fission metabolism) is unknown. long-term-unknown
- Bioavailability endpoint only: No downstream biomarker endpoints measured (FMD, flow-mediated dilation; blood pressure; ICAM-1; eNOS activity). Whether the 84% Cmax reduction translates to a proportional reduction in functional outcomes is not established by this study.
- PPO activity of specific commercial products may vary. Banana PPO activity varies with ripeness and variety; the study presumably used a standardised banana preparation, but real-world smoothie preparation will vary.
- EGCG and gallated catechins are also PPO substrates and are important flavan-3-ols in tea; whether they are equally or differentially degraded compared with epicatechin is not studied here (the study used (−)-epicatechin only as the model flavan-3-ol). The authors note PPO likely affected other flavan-3-ols present in the smoothies (procyanidins) but focused their quantification on SREMs. needs-replication — effect on gallated catechins and procyanidins not quantified in this study.
- Industry-funded: The study was funded by an unrestricted grant from Mars, Inc.; three co-authors (JO, HS, CKU) are employed by Mars, Inc. Results are biologically plausible and mechanism-consistent, but independent replication is warranted.
Significance and wiki role
This is the primary controlled human evidence for PPO-mediated impairment of dietary flavan-3-ol bioavailability (Phase 1 was a non-randomized controlled crossover; Phase 2 was randomized). The finding is directly actionable for smoothie preparation guidance. It is cited 17 times as of 2026-05-20 with a FWCI of 3.25 (99th citation percentile), indicating high relative impact within the food-polyphenol bioavailability field.
This study is relevant to:
- flavan-3-ols — compound page (primary home for the class-level bioavailability discussion and food-matrix considerations)
- egcg — EGCG is a major flavan-3-ol in tea; PPO susceptibility of gallated catechins should be noted there
- mediterranean-diet — diet page; polyphenol delivery efficiency from food preparation method
- chronic-inflammation — chronic inflammation is a downstream target of flavan-3-ol bioactives; impaired delivery reduces anti-inflammatory benefit at target hallmark level
Cross-references
| Page | Relationship |
|---|---|
| flavan-3-ols | Compound class page; this study is the primary human RCT anchor for food-matrix PPO effect |
| egcg | Key gallated flavan-3-ol in tea; PPO susceptibility context |
| mediterranean-diet | Dietary intervention page; food-preparation method affects polyphenol delivery |
| chronic-inflammation | Downstream hallmark targeted by flavan-3-ol bioactives |
Limitations and gaps
needs-replication — Phase 1 n=8 (6 per smoothie arm), Phase 2 n=11; healthy men age 25–60 only (mean ~30 yr); no women enrolled; independent replication in larger, mixed-sex, age-diverse cohort not yet published.
long-term-unknown — Single-dose acute study; chronic co-ingestion effects on flavan-3-ol absorption and downstream biomarkers (FMD, BP, inflammation) not measured.
no-mechanism — Phase 2 result implicates post-ingestion gastric PPO activity but does not directly quantify gastric pH effects on PPO activity, gastric emptying kinetics, or the relative contribution of oxidative polymerisation vs. direct conjugation reactions. The gastric-simulation experiment (pH 3, 37°C, 2 h) showed PPO retaining 68 ± 5% of pre-ingestion activity, supporting the mechanism, but this is in-vitro evidence only.
Footnotes
Footnotes
-
ottaviani-2023-ppo-flavanol-bioavailability (this page) · n=8 (Phase 1, non-randomized crossover) + n=11 (Phase 2, randomized crossover) · model: healthy men age 25–60, acute oral flavan-3-ol dose (88 mg (−)-epicatechin from standardized cocoa extract) in different vehicle conditions · Phase 1: Cmax 680 ± 78 nmol/L (capsule) vs 96 ± 47 nmol/L (high-PPO banana smoothie), 84 ± 9% Cmax reduction, p = 4 × 10⁻⁵; AUC 81 ± 11% reduction, p = 4 × 10⁻⁵; mixed-berry smoothie: Cmax 659 ± 104 nmol/L, p = 0.818 (ns) · Phase 2: AUC 37 ± 6% reduction with simultaneous co-ingestion (no pre-contact), p = 0.004 · all values mean ± SEM · NCT03526094 · DOI: 10.1039/d3fo01599h · PMID: 37615673 · Food & Function 14(18):8217–8228, 2023 ↩ ↩2