Stir your blend into coffee with heavy cream and MCT oil, or into kefir with MCT oil — ideally inside a one-meal-a-day window.
The cream or the kefir is what carries the compounds across. A little is enough; more is not better. The MCT is there for fuel, not for absorption, and the two jobs are separate.
Each label’s own directions come first. Some blends are taken with a meal rather than fasted — green tea catechins (EGCG) among them — and where a label says so, the label wins.
Three things in the cup. Three different jobs.
They are not interchangeable, and the reason is chain length. Everything on this page follows from that one fact.
or full-fat kefir
Long-chain fat. It is the only one of the three that releases the gut hormone that empties the gallbladder. Bile arrives, micelles form, the compound rides out in a chylomicron.
Medium-chain fat, C8 and C10. It skips the bile machinery entirely and goes up the portal vein to the liver, which turns it into ketones quickly. That shortcut is exactly why it is good fuel.
The condition they arrive into. A fasted gallbladder is a full one, and the hormone response is fast once a long-chain fat shows up — measured at roughly six-fold within fifteen minutes.
Why fat at all
Many of the compounds in these blends dissolve in fat rather than water. For those, absorption is not a matter of swallowing them — it is a transport problem, and the body solves it with a four-step chain that only a long-chain fat sets in motion.
The chain, step by step
- Fat reaches the duodenum and CCK is released. Measured in humans: fasting plasma CCK 1.0 ± 0.2 pM rising to 6.0 ± 1.6 pM within fifteen minutes of a meal, with the gallbladder 51% emptied by thirty minutes on ultrasound.
- Bile salts build mixed micelles that hold a fat-soluble compound in suspension in the watery gut lumen.
- Micellar solubilization is the rate-limiting step. In the trial that removed fat entirely, carotenoid appearance in human chylomicrons was negligible — the authors’ word was “virtually no carotenoid absorption.”
- Long-chain fat drives chylomicron assembly, and the compound leaves through the lymph rather than the portal vein — which also spares it a first pass through the liver.
Add fat back and the effect is large. Against a fat-free comparator, β-carotene absorption from carrots rose 6.6-fold (95% CI 4.05–10.9) when avocado was eaten with them, and α-carotene rose 8.9-fold with avocado oil in a separate randomized crossover.
A little is required. More is not better.
This is the part most advice gets wrong, and it is the part that matters at the kitchen counter. Fat is a switch, not a dial. Once enough has arrived to call up bile, adding more buys nothing — and four separate randomized or controlled human trials say so.
| Trial | What it varied | What happened |
|---|---|---|
| Roodenburg 2000 A | 3 g vs 36 g fat spread, ~56–60 adults | No difference in α-carotene, β-carotene or vitamin E. Twelve times the fat bought nothing. |
| Ribaya-Mercado 2007 A | 7 / 15 / 29 g fat per day, nine weeks, 116 children | Serum β-carotene rose about 5× and the body’s vitamin A pool roughly doubled — equally in all three groups. |
| Borel 1997 A | 0 / 15 / 30 / 40 g added triglyceride | Preformed vitamin A absorption was not related to the amount of fat at any level. |
| Dawson-Hughes 2013 A | No meal vs high-fat vs low-fat, 62 adults, vitamin D | The low-fat arm absorbed most (241.1 vs 207.4 vs 200.9 nmol/L). |
Even inside the trials that support taking fat, the same ceiling shows up: in the avocado crossover, neither doubling the avocado nor swapping the fruit for its oil changed absorption significantly. A referenced review puts the practical threshold at roughly three to five grams of fat in a meal — stated there as a general figure rather than measured in a single trial, and noted as likely differing between compounds.
A splash of cream in a cup of coffee clears that bar. That is the whole requirement.
The same trial that found no fat effect on the carotenes found a clear one for lutein esters — a 207% rise on the high-fat spread against 88% on the low-fat one. Esterified compounds have to be cut apart before they are absorbed, and that step does appear to want more fat. So the rule is not universal even within one study. It is per-compound, which is how it is written here.
Why the cream, and not the MCT
Both are fat. Only one of them is the right tool, and the reason is measurable in humans twice over — once at the trigger, once at the outcome.
Conversion of β-carotene to vitamin A was unchanged between the two meals, so nothing was gained somewhere else in that trial. The compound simply did not arrive.
A 2023 randomized crossover adds the same verdict from a different angle: olive oil beat coconut oil by 55.2% on total carotenoid absorption, with α-carotene 110.8% higher and lycopene 45.8% higher. And a 2026 review states the mechanism plainly — medium-chain fatty acids “do not require bile salt-mediated micelle formation for absorption” and are “transported predominantly via the portal venous system rather than the lymphatic circulation.”
Coconut oil is not MCT oil. It is mostly C12 lauric acid, which sits above the twelve-carbon threshold, not below it. So that 55.2% gap is the conservative number: a true C8/C10 oil would be expected to do worse than coconut, not better. We are labeling that as an inference from the chain-length data, not as something the trial measured.
The case for MCT, examined
There is a real argument on the other side, and it deserves to be met rather than ignored. It runs: MCT dissolves lipophilic compounds well, forms good emulsions, is cleaved faster than long-chain fat, and in at least one cell model delivers more of a compound into the cell. Each of those is true of something. None of them is true of a spoon of MCT oil in a cup of coffee. Here is each one, named, with what its own paper reports.
| The claim | Its source | What the paper actually shows |
|---|---|---|
| MCT beats corn oil at getting β-carotene into cells | Zheng 2022 10.1021/acs.jafc.2c02660 | True in a dish: 56.30 ± 20.13% vs 14.01 ± 1.04%. But in that same paper, far less β-carotene ever reached the micellar phase with MCT (22.22% vs 69.47%), the MCT figure carries a standard deviation a third of its own mean, and a Caco-2 monolayer has no gallbladder, no CCK, no lymph. It cannot measure the step where MCT fails. |
| A high-MCT curcumin emulsion raised bioavailability 10.6-fold | Teixé-Roig 2023 10.1016/j.foodres.2023.113595 | The control arm was a curcumin suspension — there is no long-chain-oil arm in the study. It shows an emulsion beating unformulated powder, which nobody disputes. Its own introduction concedes that “recent in vivo studies reported no differences on lipophilic compounds bioavailability when comparing them.” Rats. |
| MCT is the right lipid for a carrier | Feng 2020 10.1039/D0FO00473A | Tested MCT at 0, 10, 20, 40, 60 and 100%. Bioaccessibility reached “a maximum at 20% (w/w) and decreasing thereafter” — pure MCT finished last of six, behind even the zero-MCT solid lipid. In vitro. |
| MCT emulsions improve transport across the gut wall | Luo 2022 10.1016/j.foodchem.2021.130980 | Compared MCT against corn oil and olive oil in the same model. Verdict: “olive oil-based systems exhibiting the highest efficacy.” A different lab from Zheng, so an independent result — and it also favors the long-chain oil. In vitro. |
| Curcumin is highly soluble in MCT — 1.32 mg/mL | Tan 2021 10.3390/ma14112783 | The figure is real: “the solubility of CUR in MCT was 1.32 ± 0.03 mg/mL.” It was reached by heating to as much as 120 °C. That is a manufacturing condition, not a description of a drink. |
A manufactured emulsion is not a bottle of oil. Nearly all of the MCT evidence comes from lipid systems built with an emulsifier and high-pressure homogenisation, compared against a dry powder. Those are products, and they work. Pouring free MCT oil into a cup reproduces none of that.
And on the question that is actually on the table — free oil, swallowed, measured in a person — there is one human head-to-head, it used the same sixteen men for both arms, and MCT lost. A cell result does not overturn a human result on a different endpoint.
Not every compound needs fat
This is where most supplement advice overreaches, and where a page like this earns its keep. “Take it with fat” is true for a large and important class of compounds. It is not a universal law, and for two compounds the measured effect runs the other way.
| Compound | Fat’s measured effect | Grade |
|---|---|---|
| Carotenoids — α- and β-carotene, lutein, lycopene, zeaxanthin | Large increase. Near-zero absorption without fat. | A |
| Vitamin E (α-tocopherol) | Increase, and blocking fat digestion cut absorption 43–60%. | A |
| Vitamin K1 (phylloquinone) | Butter with spinach raised bioavailability roughly three-fold (13.3% vs 4.1% of a reference preparation). | B+ |
| Quercetin | Cmax +45%, 24-hour exposure +32% on a high-fat breakfast vs fat-free. | A |
| Vitamin D | Mixed. One trial +32%; the largest (n=88) found no difference fasted vs fed. | A |
| Vitamin A (preformed) | No effect, confirmed twice by unrelated designs: adding fat did not help, and blocking fat did not hurt. | A |
| Sulforaphane | Null across a five-fold difference in meal fat (P = 0.459). | A |
| Anthocyanins | Null. Cream delayed the peak but 24-hour excretion was unchanged. | B+ |
| Resveratrol | Worse. A high-fat breakfast cut exposure and peak by 45% and 46%. | B+ |
| Green tea catechins (EGCG) | Worse. Fasted exposure was 2.7× higher than with a light breakfast. | B+ |
Green tea catechins are the clearest case on this page of two good rules pointing opposite ways. The absorption data says fasted. Our label says with a meal, never on an empty stomach — and that instruction is set by safety guidance, not by absorption.
We keep the label rule. Getting more of a compound into the blood is not the goal when the reason for the instruction is something else entirely. Where a label and this page disagree, follow the label.
Taking them inside a fast
A fast does not undo any of the above, and it does not need to be broken to get absorption. What matters is not how long the fast has run — it is whether a long-chain fat arrives in the same sitting as the compound.
Two measured facts make that work. First, across a 10- to 20-hour fast the total lipid concentration of gallbladder bile was unchanged — the bile is waiting, and its solubilizing capacity has not degraded. Second, the fasted state is a low-CCK state with a full gallbladder, and the response is fast once a stimulus arrives: CCK rises roughly six-fold within fifteen minutes, with the gallbladder 51% emptied by thirty.
So a splash of cream in black coffee, taken at hour eighteen, finds a loaded gallbladder and empties it. That is the entire mechanism, and it is why the protocol works.
And the MCT has a measured job inside that same window. Ketone production rises on a clock as a fast runs: in a randomized crossover sampling blood every two hours, β-hydroxybutyrate reached 0.5 mmol/L at a mean of 21.1 hours without exercise and 17.5 hours with it, and a separate randomized controlled trial in 106 people measured 0.369 mM at roughly eighteen hours. MCT is a ketogenic fat — C8 and C10 are converted in the liver — which is why it earns its place in the cup even though it is not the thing carrying the compounds across.
Declared gap: no human trial was located that measured absorption of a lipophilic compound specifically after a 16–23 hour fast. The reasoning above chains four measured human results; it is labeled an inference, not a finding.
The receipts
Every figure on this page, with the study it came from. A is a randomized or controlled human trial measuring the outcome directly. B+ is weaker human data, or animal work replicated by a different lab. — is mechanism, review, cell or animal work, carried as mechanism only and never as a human result.
| Study | What it measured | Identifier | |
|---|---|---|---|
| The transport chain | |||
| Liddle 1985 | CCK 1.0 → 6.0 pM in 15 min; gallbladder 51% emptied by 30 min. n=22. | A | 10.1172/JCI111809 · PMID 2580857 |
| McLaughlin 1999 | CCK release gated between C11 and C12. n=15. | A | 10.1016/S0016-5085(99)70227-1 |
| Hofmann & Borgström 1964 | The micellar phase of fat digestion in man. Named, title read only — full text is a scanned PDF, not reachable. | — | 10.1172/JCI104909 · PMID 14162533 |
| Pavlović 2018 | Bile salts solubilize lipid digests in mixed micelles. Review. | — | 10.3389/fphar.2018.01283 |
| Reboul 2019 | Carotenoid uptake routes: passive diffusion and/or transporter. Review. | — | 10.3390/nu11040838 |
| Porter 2007 | Long-chain and unsaturated lipids recruit intestinal lymphatic transport. Review. | — | 10.1038/nrd2197 |
| Heidt 2026 | MCFAs do not require bile-salt micelles; portal rather than lymphatic transport. Review. | — | 10.3390/nu18071027 |
| Fat is required | |||
| Brown 2004 | Fat-free dressing: “virtually no carotenoid absorption.” 0/6/28 g. n=7. † | A | 10.1093/ajcn/80.2.396 · PMID 15277161 |
| Kopec 2014 | β-carotene 6.6× (CI 4.05–10.9) from carrots with avocado. n=12. | A | 10.3945/jn.113.187674 |
| Unlu 2005 | α-carotene 8.9× with avocado oil; β-carotene 17.4×. n=11. | A | 10.1093/jn/135.3.431 · PMID 15735074 |
| Jeanes 2004 | Vitamin E: amount of fat and food matrix both matter. Isotope tracer, true zero-fat arm. n=8. | A | 10.1079/BJN20041249 |
| Melia 1996 | Blocking fat digestion cut vitamin E absorption 43–60% — and did not reduce vitamin A. n=12. | A | 10.1002/j.1552-4604.1996.tb04230.x · PMID 8844448 |
| Gijsbers 1996 | Vitamin K1 from spinach: 13.3% with butter vs 4.1% without. n=5. | B+ | 10.1079/BJN19960027 · PMID 8813897 |
| Guo 2013 | Quercetin Cmax +45%, AUC0–24h +32%, high-fat vs fat-free. n=9. | A | 10.1002/mnfr.201200619 |
| Dawson-Hughes 2015 | Vitamin D peak +32% (95% CI 11–52%) with fat present. | A | 10.1016/j.jand.2014.09.014 · PMID 25441954 |
| More fat is not better — the disagreeing trials | |||
| Roodenburg 2000 | 3 g vs 36 g: no effect on α-carotene, β-carotene or vitamin E. Lutein esters +207% vs +88%. | A | 10.1093/ajcn/71.5.1187 · PMID 10799382 |
| Ribaya-Mercado 2007 | 7/15/29 g/day over nine weeks: rises equal across all three. n=116. ‡ | A | PMID 17413103 — no DOI located |
| Borel 1997 | 0/15/30/40 g: vitamin A not related to meal triglyceride. n=8. | A | 10.1038/sj.ejcn.1600466 |
| Dawson-Hughes 2013 | Low-fat arm absorbed most (241.1 vs 207.4 vs 200.9 nmol/L). n=62. | A | 10.1002/jbmr.1896 · PMID 23427007 |
| Cavalier 2016 | Vitamin D: no difference fasted vs fed. The largest trial here, n=88. | A | 10.3390/nu8050309 |
| MCT versus long-chain fat | |||
| Borel 1998 | Chylomicron β-carotene AUC 68.1 (MCT) vs 301.4 (LCT) nmol/(L·h), P<0.05. No chylomicron TG rise on MCT. n=16. | A | 10.1093/jn/128.8.1361 · PMID 9687557 |
| Yao 2023 | Olive oil +55.2% total carotenoid vs coconut oil. n=16. | A | 10.1016/j.ajcnut.2023.03.011 |
| Zheng 2022 | Caco-2 cell uptake favors MCT; micellization favors corn oil 69.5% vs 22.2%. Cells. | — | 10.1021/acs.jafc.2c02660 · PMID 35861437 |
| Teixé-Roig 2023 | MCT emulsion 10.6× vs a curcumin suspension. No long-chain arm. Rats. | — | 10.1016/j.foodres.2023.113595 · PMID 37986458 |
| Feng 2020 | MCT 0–100% in a lipid carrier: maximum at 20%, pure MCT last of six. In vitro. | — | 10.1039/D0FO00473A · PMID 32458895 |
| Luo 2022 | MCT vs corn vs olive: olive highest efficacy. In vitro, independent of Zheng. | — | 10.1016/j.foodchem.2021.130980 · PMID 34628238 |
| Tan 2021 | Curcumin solubility in MCT 1.32 ± 0.03 mg/mL — reached with heating to 120 °C. | — | 10.3390/ma14112783 · PMID 34073859 |
| Where fat does nothing, or the wrong thing | |||
| Rungapamestry 2007 | Sulforaphane: null across a five-fold meal-fat difference (P = 0.459). n=12. | A | 10.1017/S0007114507381403 |
| Mullen 2008 | Anthocyanins with cream: peak delayed, 0–24 h excretion unchanged. | B+ | 10.1021/jf072000p |
| la Porte 2010 | Resveratrol: high-fat breakfast cut AUC12 45% and Cmax 46%. n=8. | B+ | 10.2165/11531820-000000000-00000 |
| Naumovski 2015 | EGCG: fasted AUC 2.7× that with a light breakfast. n=4. | B+ | 10.3390/antiox4020373 |
| Fasting | |||
| Bloch 1980 | Across a 10–20 h fast, total biliary lipid concentration unchanged. n=9. | B+ | PMID 7461468 |
| Deru 2021 | Time to BHB 0.5 mmol/L: 21.07 h without exercise, 17.5 h with. Randomized crossover, n=20. | A | PMID 33731648 |
| Huang 2021 | BHB 0.369 mM at roughly 18–19 h of fasting. Randomized controlled, n=106. | A | PMID 33946428 |
† The verbatim sentence and the per-carotenoid fold figures for Brown 2004 were read in the author’s thesis version (10.31274/rtd-20201118-61); the journal paper was not reachable. Both identifiers are given so either can be checked.
‡ Ribaya-Mercado 2007 carries a PMID that could not be title-confirmed at a second source, and no DOI was located. It is printed as found rather than reconstructed.
What this page does not claim
The gaps are named here rather than left for someone else to find.
- That more fat is better. Four controlled human trials say it is not, and they are listed above with the rest.
- That every plant compound needs fat. Vitamin A, sulforaphane and anthocyanins showed no benefit; resveratrol and EGCG measured worse.
- That MCT is useless. It is an efficient ketogenic fuel, which is why it is in the cup. It is simply not the thing doing the absorbing.
- That the MCT finding extends to curcumin. The human head-to-head evidence located is carotenoid evidence. Extending it to curcumin or another lipophilic phytochemical is a mechanism argument, not a measured result, and it is written here as one.
- That fast duration changes absorption. No human trial was found measuring a lipophilic compound after a 16–23 hour fast. The reasoning on this page chains four measured results and is labeled an inference.
- Anything at all about vitamin K2 (MK-7), CoQ10, tocotrienols, alpha-lipoic acid, phytosterols or squalene and fat. No qualifying human comparison was found for any of them. Searched, found nothing, said so.
- Any figure for what fasting does to insulin. Widely circulated percentages for insulin and insulin sensitivity under intermittent fasting did not survive checking against the primary human trials, and several controlled trials matched for calories found no effect at all. Fasting is on this page only for what it does to bile and ketones, both measured.
- That exogenous ketones aid absorption. Twelve search constructions across four languages returned no human or animal study testing it. A clean null, stated as one.
Every identifier above was read at a source. Where only a DOI could be read, only a DOI is printed. Where a source could not be opened, it is named and nothing on this page rests on it.
Ask your physician before fasting or starting a supplement — above all if you are pregnant or nursing, take medication (especially for blood sugar or blood pressure), or manage a health condition. Each blend’s own label carries its directions, and the label comes first.