Paper Filter vs. Metal Mesh: The Lipid Cafestol Impact on Cholesterol

There is one coffee decision with a clear, reproducible, measurable effect on a number your doctor tracks. It is not the bean, the origin, the roast or the price. It is whether the brew passes through paper.

The compounds

Coffee beans contain two lipid-soluble diterpenes: cafestol and kahweol. They sit in the coffee oil. They do not dissolve in water in any meaningful quantity, which means the question is entirely mechanical — does your brewing method let the oil through, or trap it?

Paper traps it. Almost everything else does not.

What it does to LDL

Controlled trials and meta-analysis agree: unfiltered coffee raises serum LDL cholesterol, and filtered coffee does not. Cai and colleagues' meta-analysis of randomised controlled trials put the effect in the region of 5–10 mg/dL at several cups a day, with the size depending on dose and on the individual.

The mechanism is usually described backwards, including by sources that ought to know better. Cafestol is an agonist at the farnesoid X receptor (FXR) and the pregnane X receptor — it switches them on. Activating FXR suppresses cholesterol 7α-hydroxylase (CYP7A1), the rate-limiting enzyme that converts cholesterol into bile acids. Less conversion means more cholesterol retained in circulation (Ricketts et al., 2007).

So: cafestol does not "block" or "down-regulate" FXR. It activates it, and the downstream suppression of bile acid synthesis is what moves LDL.

How much does this actually matter?

It depends entirely on where you are starting.

If your LDL is well controlled and you drink two cups of French press a day, this is a small effect among much larger ones — diet as a whole, exercise, genetics, body composition. It is not worth anxiety.

If your LDL is a live clinical concern, or you are on a statin, or you have a family history of early cardiovascular disease, then a habit of four or five cups of unfiltered coffee daily is a modifiable input with a known direction and a free fix. Switching to paper costs nothing and changes nothing about how much coffee you drink.

The Norwegian cohort work is where much of this originates, and for a good historical reason: boiled unfiltered coffee was the national default, and the shift toward filtered brewing through the 1970s and 80s produced a natural experiment at population scale.

What you lose by filtering

Honesty requires the other side. Paper does not only trap diterpenes.

• Body. Coffee oil carries mouthfeel. A paper-filtered cup is cleaner and lighter; a French press is heavier and rounder. Some people strongly prefer the latter and are not wrong to.
• Some aromatics bind to oil and go with it.
• Antioxidant polyphenols largely pass through. Chlorogenic acids are water-soluble, so filtering does not remove them.

If you want a compromise, a metal mesh with a paper insert, or a cloth filter, sits between the two. Cloth retains more oil than paper and less than metal.

The methods this is really about

• Turkish coffee is unfiltered by definition — the grounds settle in the cup. Traditional, delicious, and the highest-diterpene preparation there is.
• French press is the most common unfiltered method in Western kitchens and the one most people are unknowingly drinking daily.
• Espresso sits in the middle. A single shot is a small volume, so the per-serving dose is moderate even though the concentration is not low.
• Pour-over — a Panama Geisha through a V60 or a Kenya AA through a paper cone — puts diterpenes at the bottom of the range while showing origin character at its clearest.

Frequently asked

Do paper filters remove caffeine? No. Caffeine is water-soluble and passes straight through. Filtering changes lipids, not stimulants.

What about pod machines? It depends on the pod. Many use a fine metal or plastic mesh rather than paper, which puts them closer to espresso than to filter coffee.

Is decaf different? The decaffeination process does not target diterpenes. An unfiltered decaf carries them much as caffeinated would.

Should I stop drinking French press? Only if your LDL is a concern you are actively managing. Otherwise this is one small input among many, and enjoying your coffee is not nothing.

How much oil each method actually passes

The mechanism is a sieve, so the numbers follow the aperture.

Paper filters retain the coffee oil almost completely, along with the diterpenes dissolved in it. A metal mesh has openings orders of magnitude larger; it holds back grounds and passes oil freely. Cloth sits between the two, retaining more than metal and less than paper.

Measured cafestol content follows that ordering closely. Unfiltered preparations — boiled, press pot, Turkish — have been reported at several milligrams per cup, while paper-filtered coffee is a small fraction of a milligram. Urgert and Katan's review remains the clearest summary of the range and of the variation between studies.

Espresso deserves a separate note because it is routinely misplaced. Its concentration of diterpenes is not low: the crema is emulsified oil, and no paper is involved. But a serving is 30 ml. Per shot, the dose lands in the middle of the table rather than at the top, and someone drinking two espressos is in a very different position from someone drinking four mugs of press coffee.

Kahweol, and the rest of the diterpene story

Cafestol has a constant companion. Kahweol occurs alongside it in arabica in roughly comparable quantity and behaves similarly in the brew. Robusta contains very little kahweol, which is why the related compound 16-O-methylcafestol is used analytically as a marker for robusta in coffee sold as pure arabica.

Two further points, both of which cut against a simple story.

Some controlled trials of high unfiltered intake have reported modest rises in liver transaminases alongside the lipid changes. The effect is small and reverses on stopping, but it is a reminder that diterpene exposure is not exclusively a cholesterol question.

And in cell and animal models, cafestol and kahweol show induction of detoxification enzymes and some anti-carcinogenic activity. This is real laboratory work and it is regularly quoted to argue that unfiltered coffee is fine. It should not be. In-vitro activity at concentrations no drinker reaches does not offset a reproducible human lipid effect measured in randomised trials. The evidence tiers are not comparable.

If you like the body, keep it

Filtering is not the only way to reduce diterpenes, and enjoying your coffee counts for something.

• Paper insert in a metal device. A paper filter dropped into a mesh cone, or a paper disc under a press plunger, retains most of the oil while keeping the brewing style.
• Cloth. A cloth-filtered brew keeps more body than paper and passes less oil than metal — the genuine middle position, at the cost of laundering.
• Split the day. Press pot in the morning, paper the rest of the time, cuts exposure by most of it without giving up the cup you actually want.
• Volume beats method, past a point. Two cups of press coffee is a smaller diterpene load than five cups of anything unfiltered. If you are only willing to change one thing, changing how much is as effective as changing how.

Coarser grinding, longer steeping and settling the grounds do not help. The oil is dissolved, not suspended, and none of those separate it.

The Scandinavian natural experiment

Norway is where most of this was worked out, and the reason is historical accident. Boiled, unfiltered coffee was the national default, consumed in large quantities. Through the 1970s and 80s the country switched substantially to paper-filtered brewing, and serum cholesterol at population level fell over the same period alongside other dietary changes.

That shift is what made the question visible, and later Norwegian cohort work has continued to report differences in cardiovascular outcomes between filtered and unfiltered drinkers, with unfiltered faring worse. Cohort evidence is observational and cannot separate brewing method from everything correlated with it, so it supports the randomised trial data rather than standing alone. But the two point the same way, which is more than can be said for most coffee-and-health claims.

Sources

• Cai L et al. The effect of coffee consumption on serum lipids: a meta-analysis of randomized controlled trials. Eur J Clin Nutr 2012. 10.1038/ejcn.2012.68
• Ricketts ML et al. The cholesterol-raising factor from coffee beans, cafestol, is an agonist ligand for the farnesoid and pregnane X receptors. Mol Endocrinol 2007. 10.1210/me.2007-0133
• Urgert R, Katan MB. The cholesterol-raising factor from coffee beans. Annu Rev Nutr 1997. PubMed 9240927