One partner raises how much of the other is absorbed
Shown by blood or chylomicron concentrations over time (AUC). Examples below: dietary fat with carotenoids; ascorbic acid with non-heme iron.
Combination science Peer-reviewed studies NIH ODS and NCCIH Illustrative models
Most measured combination effects change absorption or metabolism, and testing synergy in people needs a factorial trial
Published findings for botanical, vitamin and mineral pairs, sorted by mechanism and evidence level, and why “synergy” depends on the reference model and the trial design.
10 of 12catalogued findings concern absorption or metabolism; none tests synergy at a target
18% → 82%inhibition of iron absorption as added phytate rose from 2 to 250 mg phosphorus
29%power to detect an interaction as large as the main effect in a trial with 80% power for that effect
4×the participants needed to test that interaction with 80% power
Status. EUTECNIQ is pre-operational and has tested no combination. The catalog reports other researchers' published results. The surfaces, isobologram and sample-size calculator are educational models with stated, illustrative parameters; they are not data about any ingredient.
Academy fundamentals: Course 11, Multi-objective ingredient design
Mechanisms
Combination effects fall into five mechanisms, and only one of them is synergy at a biological target
Five mechanisms, five kinds of evidence; only the pharmacodynamic class is synergyPublished mechanisms
Text version of the five mechanisms
One partner changes the enzymes or transporters that clear the other, or a drug
Shown by probe-drug kinetics in people, or by enzyme and transporter assays. It carries drug-interaction risk: piperine inhibits CYP3A4 and P-glycoprotein in vitro; St John's wort induces CYP3A4 in people.
Both act on the same pathway or on complementary ones
Shown only by effect measurements against a stated reference model (Loewe or Bliss) and, in people, by a factorial trial. This is the only class that the word synergy describes precisely.
One partner keeps the other soluble, dispersed or chemically intact
Shown by digestive-recovery models and absorption studies. Ascorbic acid with green tea catechins is an example in rats and cell models.
One partner binds, competes with or speeds the removal of the other
Polyphenol-rich beverages and calcium reduce non-heme iron absorption; enzyme induction can weaken medicines. These are as important for a formula as any gain.
Each step in evidence needs its own study
- Pharmacokinetic enhancement≠Pharmacodynamic synergy
- Bioavailability gain≠Efficacy or safety
- In vitro synergy≠Clinical synergy
Explanation, details and sources
Each mechanism is shown by a different kind of measurement: an absorption study cannot show a combined effect on a target, and a cell assay cannot show what reaches the blood.
Evidence catalog
Ten of twelve findings concern absorption or metabolism, and the one botanical-pair efficacy trial lacked the arm needed to test synergy
Twelve findings by mechanism; ten concern absorption or metabolism, and none tests synergy at a targetPrimary studies
Seven quantified findings on one scale: what the partner did to absorption or exposure, with versus without itPrimary studies
Human studyAnimal studyBars: reported ranges
Data table
| Finding | Ratio (with ÷ without) | As reported | Study |
|---|---|---|---|
| Piperine 20 mg + curcumin 2 g: serum curcumin (8 volunteers) | ≈20 (plotted) | "the increase in bioavailability was 2000%" | Shoba et al. 1998 |
| Green tea extract + sucrose + ascorbic acid, rats: EGCG AUC₀–₆ h | 2.98 (derived) | 181.8 versus 61.0 pmol·h/L | Peters et al. 2010 |
| St John's wort 300 mg three times daily, 14 days: alprazolam AUC | ≈0.5 (plotted) | "a 2-fold decrease" | Markowitz et al. 2003 |
| Chili with an iron-fortified meal: iron absorption | 0.62 (derived) | 6.0% versus 9.7% | Tuntipopipat et al. 2006 |
| Calcium 300–600 mg with wheat rolls: iron absorption | 0.40–0.50 (derived) | reduced by 50–60% | Hallberg et al. 1991 |
| Phytate 250 mg phosphorus with wheat rolls: iron absorption | 0.18 (derived) | inhibited by 82% | Hallberg et al. 1989 |
| Black tea with a bread meal: iron absorption | 0.06–0.21 (derived) | reduced by 79–94% | Hurrell et al. 1999 |
Not plotted because no value is reported in the abstract: turmeric 0.5 g with the same iron-fortified meal (no inhibition, P = 0.91; Tuntipopipat et al. 2006); ascorbic acid against phytate (significantly counteracted the inhibition; Hallberg et al. 1989); dietary fat with salad carotenoids (Brown et al. 2004).
Most measured partners lowered absorption; the two that raised exposure acted through metabolism or chemical protection, and none of these measurements tests action at a shared target.
Evidence key:Traditional useIn vitroAnimalHuman PK/absorptionHuman RCTNot assessedDocumented safety concern
All sources were opened in PubMed, NIH ODS or NCCIH on October 22, 2026; full references are on the Sources page. The NCCIH and ODS pages are summaries for consumers and health professionals, not new studies.
Piperine is a neutral amide: a benzodioxole ring and a trans,trans-diene joined to piperidineReference structure, computed conformer
Explanation, details and sources
Each row comes from a primary study, a meta-analysis or an NIH fact sheet that we opened and checked. The tag gives the strongest evidence type for the stated finding, not for any health benefit.
| Combination | Mechanism | What was measured | Result | Evidence | Source |
|---|---|---|---|---|---|
| Ascorbic acid + non-heme iron (with phytate) | Absorption enhancement | Radio-iron absorption from wheat rolls with added phytate, adults | Phytate inhibited absorption by 18% at 2 mg, 64% at 25 mg and 82% at 250 mg phytate phosphorus; added ascorbic acid significantly counteracted the inhibition | Human PK/absorption | Hallberg et al. Am J Clin Nutr 1989;49:140–144. NIH ODS: vitamin C "enhances the bioavailability of nonheme iron" |
| Ascorbic acid across a whole diet | Absorption enhancement, weaker in context | Iron absorption from a complete, freely chosen diet over three 5-day periods, 12 adults, 51–247 mg/day vitamin C | No significant difference in mean absorption between periods; the authors conclude the effect is "far less pronounced" than in single-meal tests | Human PK/absorption | Cook & Reddy. Am J Clin Nutr 2001;73:93–98 |
| Polyphenol-rich beverages + non-heme iron | Antagonism (polyphenol binding) | Erythrocyte incorporation of radio-iron from a bread meal taken with black tea, herb teas, cocoa or coffee, adults | Black tea reduced absorption by 79–94%, peppermint tea 84%, cocoa 71%, chamomile 47%; 100–400 mg total polyphenols per serving reduced absorption by 60–90% | Human PK/absorption | Hurrell, Reddy & Cook. Br J Nutr 1999;81:289–295 |
| Turmeric or chili + iron-fortified meal | Antagonism for chili; none detected for turmeric | Stable-isotope iron absorption, randomized crossover, 10 women per study | Chili (25 mg polyphenols) reduced absorption by 38% (6.0% vs 9.7%); turmeric (0.5 g, 50 mg polyphenols) did not inhibit it (P = 0.91) | Human PK/absorption | Tuntipopipat et al. J Nutr 2006;136:2970–2974 |
| Calcium + iron | Antagonism (mucosal transfer) | Iron absorption from wheat rolls with 40–600 mg added calcium, 126 adults | Dose-related inhibition up to 300 mg; 300–600 mg calcium reduced absorption by 50–60%, and heme iron absorption also fell. NIH ODS notes the effect "has not been definitively established" | Human PK/absorption | Hallberg et al. Am J Clin Nutr 1991;53:112–119; NIH ODS iron fact sheet (Sep 4, 2025) |
| Dietary fat + carotenoids (salad vegetables) | Solubilization for absorption | Chylomicron carotenoids for 12 h after salads with 0, 6 or 28 g canola oil, crossover, 7 adults | Negligible appearance of α-carotene, β-carotene and lycopene with fat-free dressing; more with reduced-fat (P < 0.04) and more again with full-fat dressing (P < 0.02) | Human PK/absorption | Brown et al. Am J Clin Nutr 2004;80:396–403 |
| Piperine + curcumin | Metabolism inhibition (glucuronidation) | Serum curcumin after 2 g curcumin with or without 20 mg piperine, healthy volunteers; rats in parallel | Bioavailability increased by 2000% in humans and 154% in rats. The human arm had 8 participants, 6-hour sampling and curcumin alone mostly below detection (UK COT 2024); a 30-person crossover found no significant difference between a piperine product and a standard extract at matched curcuminoid dose (Fança-Berthon et al. 2021). NCCIH: combining curcumin with piperine is one way to raise bioavailability, and "highly bioavailable formulations of curcumin … may harm your liver" | Human PK/absorption Documented safety concern | Shoba et al. Planta Med 1998;64:353–356; COT/2024/07; Fança-Berthon et al. J Nutr 2021;151:1802–1816; NCCIH turmeric (Apr 2025). Details: Course 8 |
| Piperine + drugs cleared by CYP3A4 or P-glycoprotein | Enzyme and transporter inhibition (interaction risk) | Digoxin and cyclosporine transport in Caco-2 cells; verapamil metabolism in human liver microsomes | P-glycoprotein transport inhibited (IC50 15.5 and 74.1 µM); CYP3A4 metabolite formation inhibited (Ki 36–77 µM). The authors conclude dietary piperine could affect plasma levels of such drugs | In vitro Documented safety concern | Bhardwaj et al. J Pharmacol Exp Ther 2002;302:645–650 |
| Green tea catechins + ascorbic acid + sucrose | Physicochemical protection | Plasma catechins in rats after 50 mg green tea extract in four formulations; in vitro digestion with Caco-2 uptake | EGCG AUC0–6h 181.8 vs 61.0 pmol·h/L, and EGC 3237.0 vs 1304.1, with sucrose and ascorbic acid versus plain extract; catechin retention higher with ascorbic acid | Animal In vitro | Peters, Green, Janle & Ferruzzi. Food Res Int 2010;43:95–102 |
| St John's wort + CYP3A4 substrates | Enzyme induction (adverse interaction) | Alprazolam kinetics before and after 14 days of St John's wort 300 mg three times daily, 12 healthy adults | Alprazolam AUC fell 2-fold and half-life fell from 12.4 to 6.0 h. Mechanism: hyperforin binds the pregnane X receptor (Ki 27 nM) that regulates CYP3A4. NCCIH lists weakened oral contraceptives, cyclosporine, HIV drugs and warfarin | Human PK/absorption Documented safety concern | Markowitz et al. JAMA 2003;290:1500–1504; Moore et al. PNAS 2000;97:7500–7502; NCCIH St John's wort (May 2025) |
| Green tea catechins + caffeine | Not separated (mixture tested) | Random-effects meta-analysis of 11 articles on weight loss and maintenance | Catechins or an EGCG–caffeine mixture: −1.31 kg; the authors call the effect "small positive", with habitual caffeine intake and ethnicity as possible moderators | Hursel, Viechtbauer & Westerterp-Plantenga. Int J Obes 2009;33:956–961 | |
| Curcumin extract + boswellic acid (botanical pair) | Claimed pharmacodynamic synergy, not tested | 12-week double-blind trial, 201 patients with knee osteoarthritis, three arms: curcumin complex, curcumin plus boswellic acid, placebo | The combination beat placebo on physical-performance tests and WOMAC pain; curcumin alone beat placebo only on performance tests. With no boswellic-acid-alone arm, synergy cannot be separated from boswellia's own effect | Human RCT | Haroyan et al. BMC Complement Altern Med 2018;18:7 (NCT02390349) |
Dose dependency
Inhibition of iron absorption rose with the amount of phytate, and every polyphenol beverage tested reduced absorption
Iron absorption from a test meal, reduced by phytate dose and by polyphenol-rich beveragesHuman studies
Data table
| Added to the meal | Reduction in iron absorption | Study |
|---|---|---|
| Phytate, 2 mg phytate phosphorus | 18% | Hallberg et al. 1989 |
| Phytate, 25 mg | 64% | Hallberg et al. 1989 |
| Phytate, 250 mg | 82% | Hallberg et al. 1989 |
| Black tea | 79–94% | Hurrell et al. 1999 |
| Peppermint tea | 84% | Hurrell et al. 1999 |
| Pennyroyal tea | 73% | Hurrell et al. 1999 |
| Cocoa | 71% | Hurrell et al. 1999 |
| Vervain tea | 59% | Hurrell et al. 1999 |
| Lime-flower tea | 52% | Hurrell et al. 1999 |
| Chamomile tea | 47% | Hurrell et al. 1999 |
| Beverages with 20–50 mg total polyphenols per serving | 50–70% | Hurrell et al. 1999 |
| Beverages with 100–400 mg total polyphenols per serving | 60–90% | Hurrell et al. 1999 |
Hallberg et al. tested seven phytate dose levels between 2 and 250 mg; the abstract reports these three. All three reductions had P < 0.001.
A combination's effect depends on the amount of each partner, so a formula's interaction screen states doses, not just ingredient names.
Green tea's four main catechins share one flavan-3-ol core and differ by a third B-ring hydroxyl and a gallate esterReference structures
Reference models
Whether a combination looks synergistic depends on the reference model: Loewe and Bliss expect different effects from the same two curves
Single agents (Hill, Emax = 1) · Loewe · BlissEA = unAunA + 1, u = aEC50,A
uU(E) + vV(E) = 1
EBliss = EA + EB − EAEB
uU(E) + vV(E) = 1
EBliss = EA + EB − EAEB
A alone50.0%
B alone50.0%
Highest single agent50.0%
Loewe expectation72.4%
Bliss expectation75.0%
Bliss − Loewe+2.6 pts
Assumptions
- Both agents reach the same maximum effect (Emax = 1) and follow Hill curves; doses are in units of each agent's own EC50, so the EC50 values themselves drop out
- Slopes are illustrative, not fitted to any ingredient
- These are expectations for non-interaction; no observed data are shown
Bliss minus Loewe expectation, percentage pointsIllustrative model
Show data: expectations at equal normalized doses for the current slopes
Hill slope of A 1.0, of B 2.0
| Dose of each (× own EC50) | A alone | B alone | Highest single agent | Loewe | Bliss | Bliss − Loewe |
|---|---|---|---|---|---|---|
| 0.1 | 9.1% | 1.0% | 9.1% | 12.1% | 10.0% | −2.1 pts |
| 0.3 | 23.1% | 8.3% | 23.1% | 34.0% | 29.4% | −4.6 pts |
| 1 | 50.0% | 50.0% | 50.0% | 72.4% | 75.0% | +2.6 pts |
| 3 | 75.0% | 90.0% | 90.0% | 93.5% | 97.5% | +4.0 pts |
| 10 | 90.9% | 99.0% | 99.0% | 99.2% | 99.9% | +0.7 pts |
Drag the surface or use the buttons to rotate it. The amber band marks dose pairs where the two models agree within 2 percentage points; outside it, the same observed effect could be called synergy under one model and antagonism under the other.Equations: Tang J, Wennerberg K, Aittokallio T. Front Pharmacol. 2015;6:181 (Eq. 1–2); Greco WR, Bravo G, Parsons JC. Pharmacol Rev. 1995;47:331–385. Computed by EUTECNIQ in the browser.
Method, assumptions and limits: reference-model surfaces
- What it calculates
- The effect expected if two agents do not interact, under Loewe additivity and under Bliss independence, and their difference, over a grid of doses from 0.03 to 30 times each agent's EC50.
- Equations
- Single agent: E(u) = un/(un + 1), inverse U(E) = (E/(1 − E))1/n. Loewe: the E that solves u/U(E) + v/V(E) = 1, found by bisection on logit(E). Bliss: EA + EB − EAEB. Highest single agent: max(EA, EB).
- Variables and units
- u, v: doses divided by each agent's EC50 (dimensionless). nA, nB: Hill slopes, 0.5–3. Effects in % of the common maximum.
- Assumptions
- Equal maximum effects; monotonic Hill curves; no observed combination data. When maxima differ, Loewe additivity is not defined above the lower maximum.
- Checks
- With both slopes equal to 1, Bliss − Loewe = uv/((1 + u)(1 + v)(1 + u + v)), which is 8.33 points at u = v = 1. A drug combined with itself (equal slopes) gives Loewe E = E(u + v), the "sham combination" property. Browser readouts match an independent Python reference (scipy brentq).
- Does not establish
- That any real pair is synergistic, additive or antagonistic; a dose; or an effect in people. It shows only why the reference model must be stated.
Isobologram at effect level x, normalized axess = aAx, w = bBx
Loewe: s + w = 1
CI = s + w on the Bliss isobole
Loewe: s + w = 1
CI = s + w on the Bliss isobole
Hill slopes come from the surface controls above.
Loewe CI of a Bliss-additive mixture, equal shares1.087
Bliss isobole: w at s = 0.50.577
outside the Loewe line: a Bliss-additive mixture would be called antagonistic under Loewe
Isoboles for the same pair under three reference modelsIllustrative model
Points below the Loewe line need less of both partners than dose additivity predicts. A mixture that exactly meets the Bliss expectation has a Loewe combination index (CI) below 1 when its isobole bends inward and above 1 when it bends outward.Loewe line and combination index as in Chou TC. Cancer Res. 2010;70:440–446; Bliss isobole computed from Tang et al. 2015, Eq. 1.
Method, assumptions and limits: isobologram
- What it calculates
- At effect level x, the dose pairs that give x under Bliss independence, plotted in units of each agent's single-agent dose for x (Ax, Bx), with the Loewe line and the highest-single-agent boundary.
- Equations
- Ax = EC50,A(x/(1 − x))1/nA. For each s, w solves EBliss(s Ax, w Bx) = x by bisection. On the ray s = w = t, the Loewe combination index of the Bliss-additive mixture is CI = 2t.
- Checks
- Slopes 1 and 1 at x = 50%: CI = 2(√2 − 1) = 0.828. Slopes 2 and 2: CI = 2(√2 − 1)1/2 = 1.287. Matches the Python reference.
- Does not establish
- A combination index from data; real CIs need replicated dose-response data and confidence intervals. For the Chou–Talalay calculator and a Bliss dose-matrix calculator, see Clinical research.
Explanation, details and sources
Synergy means "more than expected," so the expectation must be stated. Loewe additivity treats the partners as dilutions of one another; Bliss independence treats them as acting separately. With Hill slopes of 1, Bliss expects more effect than Loewe everywhere; with steeper slopes it can expect less.
Proving synergy in people
A 2 × 2 factorial trial can test an interaction, but one as large as the main effect needs four times the participants
The four arms of a 2 × 2 factorial trial and the contrasts they supportDesign schematic
Total sample size, 2 × 2 factorial, equal armsNmain = 4σ²(z1−α/2 + z1−β)²δ²
Nint = 16σ²(z1−α/2 + z1−β)²(kδ)²
Nint = 16σ²(z1−α/2 + z1−β)²(kδ)²
N for the main effect128
N for the interaction504
Inflation factor× 4.00
Interaction power at the main-effect N29.3%
With fewer than about 40 participants the normal approximation understates the sample size; use a t-based calculation.
Method, assumptions and limits: factorial-trial sample size
- What it calculates
- Total participants for a 2 × 2 factorial trial with a continuous outcome to detect (a) the main effect of one ingredient and (b) the interaction between the two, and the power that the main-effect sample size gives for the interaction.
- Equations
- Equal allocation, N/4 per arm. Main effect of A (mean of the two A arms minus mean of the two non-A arms): variance 4σ²/N. Interaction contrast (μAB − μB) − (μA − μ0): variance 16σ²/N. Required N from the normal approximation, rounded up to a multiple of 4. Interaction power = Φ(zI − z1−α/2) + Φ(−zI − z1−α/2), zI = kδ/√(16σ²/N).
- Variables and units
- σ 1–50 and δ 0.5–30, in the outcome's units; k 0.1–2; α 0.05 or 0.01; power 80% or 90%.
- Assumptions
- Known, equal variance in all arms; no dropout; one primary outcome; normal approximation. "Interaction of the same size" means the difference of differences equals δ, as in Brookes et al.
- Valid range
- Reliable for totals above about 40; smaller trials need t-distribution methods.
- Benchmark
- With k = 1 the inflation factor is exactly 4, and at the exact main-effect N (80% power, α 0.05) interaction power is 28.8%, the 29% reported by Brookes et al.; the readout uses N rounded up to a multiple of 4 (29.3% at the defaults). Browser values match an independent Python reference (scipy.stats.norm).
- Does not establish
- The effect size, SD or dropout for any product; these must come from prior data. It does not replace a statistician's design.
- References
- Brookes ST et al. J Clin Epidemiol. 2004;57:229–236. Montgomery AA, Peters TJ, Little P. BMC Med Res Methodol. 2003;3:26.
Explanation, details and sources
In a factorial trial participants receive placebo, A, B or both. Main effects compare halves of the trial; the interaction compares a difference of differences between quarters, so its variance is four times larger. Brookes et al. found that a trial with 80% power for an overall effect had only 29% power for an interaction of the same size.
Proposed practice
EUTECNIQ would claim only what each level of combination evidence supports, and would screen every formula for interaction risk first
A management proposal for combination projects; none has been performed.
Four steps, each able to support only its own kind of claimProposed practice
Text version of the four steps
- Interaction screenCheck each partner for documented effects on CYP enzymes, transporters and mineral absorption before formulation, using sources such as those in the catalog above.
- Dose-matrix studyReplicated in vitro dose matrices, analyzed against both Loewe and Bliss with the ratio, effect level and uncertainty stated.
- Human PK studyMeasure whether one partner changes the other's exposure, and reassess safety at the higher exposure.
- Factorial trialOnly where a synergy claim is wanted: placebo, A, B and A + B arms, powered for the interaction.
Related pages: Evidence and pharmacokinetics, Bioavailability design and Clinical research, which hosts the combination-index and Bliss dose-matrix calculators.
Planning a combination formula?
Start with the interaction screen and the claim you want to make. EUTECNIQ proposes to map both to the evidence each claim would need.