Disclaimer: I am not a chemist or a medical professional. The information provided in the following should be perceived accordingly. Any details or insights shared are based on general knowledge and understanding.
I recently learned that coffee that is brewed without using a paper filter such as brewing with a french press or espresso, contain higher levels of cafestol and kahweol. These are diterpenoid molecules, which have been found to increase cholesterol. In coffee brewed with a paper filter such as drip coffee, these compounds are only found in negligible amounts, because they generally do not pass through the paper filter.
As it happens, I switched from using paper filters to using a mesh
filter of stainless steel some time ago. I did so mainly because I had
concerns about the plastic funnel which I was using at the time, since
heating plastic is generally inadvisable, because it
increases the release of endocrine disrupters. Presumably, the plastic
used for products that are supposed to be heated is resistant to it,
but I have my doubts. I also had some concerns about chemicals in
paper filters, which could end up in the coffee when pouring boiling
water through the filter. I did not know for sure if these were
actually concerns with any basis in reality, but I generally opt for
non plastic kitchen appliances and food containers if the option
exists. There also might not be any actual reason to have concerns
about paper filters, but why use them,
when there is an alternative, which is definitely free of any chemicals
used in the production of paper?
Well, as it turns out, because it might prevent other known harmful
compounds from entering the brew. To complicate things further, as
I’m reading a bit about cafestol and kahweol, I learn that they have
also been found to have properties that lower the risk of certain
cancers. So what is the right way to brew my coffee from a health
perspective? Are my concerns about paper filters unfounded or are
there risks to using them, which outweigh the negatives of unfiltered
coffee?
I will examine three questions:
The following is quite dense with details, so I have added a couple of summaries for ease of parsing or to get the main takeaways quickly.
Paper filters come in a variety of shapes and sizes and can be bleached or non bleached. The paper is produced in a similar manner to other types of paper using a technique called pulping. There are different methods, but they are similar in that the raw material, usually wood, is separated into its useful cellulose fibres, which is beaten into pulp. Chemicals are then added and a mesh is used to screen the solution before pressing and drying it to get the paper.
There is a whole host of chemicals used dependent on the details of
the actual method and the desired properties of the paper. For filter
paper, one of the most important properties is how well it holds up
when it gets wet. This is called wet-strength and is achieved by using
strength resins such as urea-formaldehyde and melamine-formaldehyde.
Using formaldehyde causes some very toxic fumes during production, but
other options are available, which does not have this problem such as
polyamide-epichlorohydrin (PAE) and polyamidoamine-epichlorohydrin
(PAAE). These are, despite the similar names, not the same compounds,
but they often appear to be treated interchangeably in research
literature on their health risks. This, I assume, is because the risk
is associated with the formation of 1,3-dichloro-2-propanol (1,3-DCP)
and 3-monochloropropanediol (3-MCPD) as a side reaction from the use
of the epichlorohydrin and not the polyamide
or the polyamidoamine. 3-MCPD has shown clear evidence of
carcinogenic activity in rat models1 and is classified as a
possible carcinogen by The International Agency for Research on
Cancer
(IARC).
While there is no regulation that explicitly bans 1,3-DCP in food-contact
materials, they must be manufactured so that 1,3-DCP does not migrate
into food and is absent or non-detectable. Certain jurisdictions,
such as Germany, explicitly require non-detectable levels.2
Other studies on 3-MCPD has also shown adverse effects
on kidney health as well as neurotoxicity. It has even been considered
a potential male contraceptive, but was found to cause damage to the
testis of rats3.
There are also different chemicals used for sizing to control other properties relevant to filter paper like absorbability and porosity, which controls how fast water will travel through the filter. For this, sizing agents such as alkylketene dimers (AKD) and alkenyl succinic anhydride (ASA) are used. I have not been able to find much information on these in terms of health effects. Neither is listed in the IARC database, but Succinic anhydride is listed as group 3, so not classifiable as carcinogenic to humans. If this is also the case for ASA, I don’t know. I also don’t know if the limited research on health risk associated with these is because there is no cause for concern or if no one has ever looked into them. In any case, I can’t find any studies assessing them in relation to filter paper or at what levels they can be found in paper in general.
Glyoxal is also a possible contaminant from these resins. There are other options such as polyethyleneimine and chrome-stearate-complexes, but these are no longer legal to use in the EU.
Lastly, bleached filters may also contain additional compounds from chemicals used in the bleaching process, but I’m not going to consider them here, since I see no reason to use bleached filters in the first place.
The central concern here is whether these chemicals persist in the final paper product, potentially transferring into coffee, and if their presence poses harm at the detected quantities. Specifically, formaldehyde, which is known for its carcinogenic properties, has been researched extensively as it is used in many other areas. For instance, urea-formaldehyde, utilized in agriculture as a controlled nitrogen source and also used in household appliance manufacturing for coating and other purposes, has drawn attention in literature focusing on health risks associated with its release into the air or migration into food. At higher concentrations, it can provoke respiratory irritation and pose carcinogenic risks.
The levels allowed is regulated by the FDA in the US and by the EU in Europe. Here, I’m focusing on EU regulation, since that is what is most relevant to me. In the EU formaldehyde in consumer goods is governed by Regulation (EU) 2023/1454 as part of the REACH regulation. This regulation does not specifically mention paper products and such regulations are typically closer regulated on a national level. I was able to find both German and Danish regulations for formaldehyde in paper products with food contact. In Danish regulation, the allowed amount is 0.5mg/dm^2, but this appears to be for any paper product, which is produced with the intent to have food contact. I would say that there is a significant difference between storing food in a paper wrapping and pouring hot water through the paper, but the distinction doesn’t seem to be made in neither Danish nor EU regulation for that matter. For Germany, the BfR has a regulation specifically for filter paper used for hot water. It mentions detectable levels of formaldehyde from extract from finished product must be no more than 1 mg/dm^2 2.
For 3-MCPD Commission Regulation (EU) 2020/1322 states allowed
detectable levels in food, but I haven’t been able to find any
regulation pertaining to levels in paper products specifically. The
daily maximum intake has been established at 0.002 mg/kg bodyweight by
the European Food Safety Authority4. This limit has been set based
on studies in rat models that found the lowest dose of 1.1 mg/kg
bodyweight indicates adverse effects and then applying an uncertainty
factor of 500 to derive the daily maximum intake at 0.002 mg/kg
bodyweight in 20015.
This has since been doubled to 0.004 mg/kg bodyweight 6.
The BfR (Federal Institute for Risk Assessment of Germany) has set a
limit of 0.012 mg/L for 3-MCPD and 0.002 mg/L for 1,3-DCP that must
not be exceeded in water extracts of paper intended for food contact
which has become EU standard7.
I could not find any epidemiological studies on 3-MCPD.
There appears to be very little research on formaldehyde migration from filter paper specifically, but from what I can gather, these compounds used in the production are only found in trace amounts in the final paper product.
I was however able to find a report from an Austrian university from 2011 which tests migration of these compounds from specifically coffee filters8. They tested a number of different filters, some of them commonly available in Denmark as well. They found that the levels of formaldehyde and glyoxal were in no way concerning. In fact, the levels were smaller than the German limit of 1 mg/dm^2 by a factor of 200-300 for all the tested filters. A number of other tests were also done, such as visually inspecting the fibres with microscopes for contaminants, which showed none, and chromatograms which didn’t reveal any vaporable substances. This indicates the paper wasn’t recycled, which will often have additional contaminants.
I also found a study conducted in Turkey on filter paper samples intended for hot beverages which found migration of formaldehyde well below the limits of the Turkish regulations, although these regulations are presumably less tight than EU regulations.
It is easier to find research on 3-MCPD in filter paper, probably because PAE and PAAE are used more commonly to gain wet-strength than urea-formaldehyde and melamine-formaldehyde.
In one study, they investigated coffee and tea filters among other paper products collected in Germany and Canada. They established that consumers are exposed to 3-MCPD through filter paper, but the amounts are relatively small compared to many other sources such as processed oils. However, they also mention that frequent use of coffee filters containing 3-MCPD at approximately 0.0055 mg per filter (these were bleached filters, which have higher levels) could result in non-trivial amounts when combined with overall dietary exposure9. The unbleached filters in this study obtained from Germany had an average of 0.047 mg/kg 3-MCPD and the filter with the highest amount had 0.084 mg/kg.
Another study tested a range of paper products intended for food contact and looked at coffee filters specifically, found they were among the products with lowest levels. In fact, a large percentage of the samples had no detectable levels of 3-MCPD nor 1,3-DCP7.
Both formaldehyde and 3-MCPD can be found in varying degrees in many types of food. Meat, fruits and vegetables contain naturally formed formaldehyde from various chemicals breaking down. The levels vary a lot. Watermelon, for instance, has 9 mg/kg while Cauliflower has 26.9 mg/kg formaldehyde 10. Coffee beans themselves contain some amount of formaldehyde, one study estimating 3.4-4.5 mg/L for brewed coffee.11
3-MCPD can also be found in various types of food and things like vegetable oils are a common source, but especially soy sauce has been cause for concern, because tests have shown that many samples had levels of 3-MCPD higher than what is considered safe by the EU.6 As with formaldehyde, 3-MCPD also occurs in coffee. One study assessed the levels in a variety of coffee beans. In the four samples of roasted beans, they found an average of 0.29 mg/kg 3-MCPD. They also found that the darkness of the roasted coffee beans was directly linked to the formed 3-MCPD and the longest-roasted beans had the highest levels12. This is not surprising since 3-MCPD often occurs in processed foods especially were heat is involved in the processing.
Considering that the WHO estimates the daily intake of 1.5-14 mg of formaldehyde from food alone for average adults and that the European Food Safety Authority states that daily consumption of formaldehyde from diet should not exceed 100 mg13, the levels that appear to migrate from paper filter should be negligible in terms of risk when it comes to formaldehyde.
The mean estimates of daily intake of 3-MCPD in published studies are 0.0002 mg/kg bodyweight for adults and 0.0013 mg/kg bodyweight for children. The estimated 95th percentile for these groups are 0.0026 mg/kg and 0.0038 mg/kg, which is still below the recommended maximum daily intake.6 Given that the levels found in coffee filters is approximately six times lower compared to the levels occurring in the coffee beans used to brew the coffee, there should not be much cause for concern regarding 3-MCPD either.
The available migration studies and regulatory limits suggest that chemicals used in the production of paper coffee filters, primarily formaldehyde-related resins and epichlorohydrins, are present in the finished product, but only in trace levels. Migration into coffee during brewing appears to be minimal and well below established threshold, as well as substantially smaller than background dietary exposure from other food sources.
Having established that paper filter chemicals pose minimal risk, let’s examine what they’re actually filtering out.
Coffee has lipids in it which are made up of mostly triglycerides (~75%), but also the diterpenes cafestol and kahweol (~20%) and some other fatty acids in smaller quantities. The levels at which these occur vary not only in how the coffee is brewed, but also between different species of coffee, where the coffee is grown and with the roasting of the beans.14 In fact, one study on 32 samples of Brazilian Arabica coffees produced in two different regions found that kahweol ranged between 1.75 to 10.68 g/kg and cafestol from 1.76 to 9.66 g/kg15. Both cafestol and kahweol are generally found in smaller quantities in Arabica than Robusta, although that is not necessarily the case due to variation.
It has been known for some time, that coffee consumption is associated with a reduced risk of certain cancers, and cafestol and kahweol are believed to be at play in this. A study performed on animal models found that cafestol and kahweol produce biological effects compatible with anticarcinogenic properties and they hypothesize that similar effects may also be present in humans. 16
There is also evidence of anti-inflammatory properties since studies have found that cafestol and kahweol can significantly inhibit inflammatory mediators, which was also confirmed in vivo in rat models17.
In terms of neuroprotective effects, there is limited evidence that cafestol and kahweol could be beneficial, although the evidence in this area is less well established and is largely based on preclinical models.18
Almost all evidence of these effects is derived from in vitro and animal studies. If controlled randomized studies in humans exist, I have not been able to find them. Because of this, I do not value these effects highly as the evidence is limited.
Quite a few studies have found an association between an increase in serum cholesterol and coffee consumption and the cause has been isolated to cafestol and kahweol. The hypothesized mechanism for why the increase happens is that the bile acid receptors farnesoid X and pregnane X, which are located in the intestines and liver, are activated by cafestol. This leads to signaling to the liver that reduces conversion of cholesterol into bile acids, resulting in reduced clearance of cholesterol and an increase in serum LDL cholesterol.19 The mechanism is similar, but weaker and less well established for kahweol.
In the 1980s, when the association was first made, some studies showed a clear correlation between coffee consumption and others didn’t, leading to the theory that brewing method had an impact20. A meta-analysis from 2001 found that the consumption of unfiltered coffee increased serum levels of total and LDL cholesterol21. They note that the average effect size of the included studies when comparing filtered with unfiltered coffee consumption was relatively large with an increase in LDL cholesterol of 17.8 mg/dl. The American Heart Association considers LDL levels lower than 100 mg/dl to be optimal, so I would agree that an average increase of 17.8 mg/dl is a lot, as it could easily put someone with otherwise low cholesterol outside of optimal simply by drinking unfiltered coffee.
Nevertheless, considering that there are both positive and negative effects of these diterpenes, does the benefits outweigh the higher cholesterol? Probably not. While previous studies 22 23 have found that coffee consumption is inversely associated with all-cause mortality, they did not differentiate between filtered and unfiltered coffee, but a large study from 2020 did and found that unfiltered coffee consumption was associated with higher mortality24. They also found that filtered coffee consumptions was associated with lower mortality compared to no coffee consumption and hypothesize that the reason could be beneficial antioxidants, which coffee has aplenty. This is of course an observational study with all its limitations, but a very large one, and since unfiltered coffee drinkers have higher mortality, it does not indicate that the positive health benefits outweigh the negative effects. They mention that they did not account for use of sugar, milk or creamer, which is strange since it would have been a simple question to include. In this study, unfiltered meant boiled coffee, which some studies refer to as Scandinavian-style coffee, where the grounds are boiled directly in the pot. This is not very common in Denmark anymore, but still used to some degree in Norway and Sweden, although to a much lesser extent than in the past. Without any certainty, I have a suspicion that this is associated with lower social economic status, which could be a major confounder.
The diterpene levels in the brewed coffee varies not just between using a paper filter and not using one. A study from 1995 investigated the levels of cafestol and kahweol in different brewing methods. They looked at drip coffee, Scandinavian boiled coffee, Turkish coffee, French press coffee and Italian espresso as well as instant coffee. They conclude from their samples, that Turkish and French press coffee have similar effect on serum cholesterol levels as Scandinavian coffee. For espresso type coffee the effect is smaller and for drip coffee and instant coffee, the effect is negligible. They also found that the strength of the brew had a large impact on the cholesterol raising effect25. A similar study found largely the same results, although Turkish coffee had the highest levels of cafestol and kahweol 26. These findings have been replicated a number of times, although there is some variation in the levels for each brewing method most likely due to differences in beans species, roast and brewing strength.
While instant coffee has almost no cafestol or kahweol, it has higher levels of 3-MCPD. This is probably due to the harder roasting of beans used for instant coffee, which is done to preserve more taste.
In contrast to paper-filter chemicals, cafestol and kahweol are present in meaningful quantities in unfiltered coffee and have a well-established effect on serum LDL cholesterol. While they may have beneficial biological properties suggested by research, the evidence for this in human subjects is fairly weak. The levels of cafestol and kahweol vary substantially between brewing methods in general.
Considering the small trace amounts of chemicals used in the production, that can be found in the final paper filter product, there is probably little cause for concern. The research on cafestol and kahweol seem to strongly indicate that cholesterol is affected and risk of CVD increased. For me personally, cholesterol and coronary heart disease are probably of much greater concern than cancer, based on family history. I also already consume a diet that is somewhat high in saturated fat, so brewing my coffee with a paper filter is probably advisable. There is some debate on whether you should rinse paper filters prior to brewing the coffee, but this usually seems to be in relation to the taste of the coffee, but it would presumably have the added benefit of rinsing out any trace chemicals left over from the production process.
Unfortunately I vastly prefer the taste of coffee brewed with my steel mesh filter, as it produces coffee with a richer and fuller taste, but for now I will go back to using paper filters and maybe every now and then brew a coffee with the mesh filter. Since espresso has lower levels of diterpenes, I am considering buying a moka pot to maybe use on occasion instead of the mesh filter.
Cho, W. S., Han, B. S., Nam, K. T., Park, K., Choi, M., Kim, S. H., … & Jang, D. D. (2008). Carcinogenicity study of 3-monochloropropane-1, 2-diol in Sprague–Dawley rats. Food and Chemical Toxicology, 46(9), 3172-3177. ↩︎
Bundesinstitut für Risikobewertung (2009). “XXXVI. Papier, Kartons und Pappen für den Lebensmittelkontakt. Stand vom 01.02.2023 ↩︎
Mou, Y., Sun, L., Geng, Y., Xie, Y., Chen, F., Xiao, J., … & Ma, L. (2023). Chloropropanols and their esters in foods: Exposure, formation and mitigation strategies. Food Chemistry Advances, 3, 100446. ↩︎
EFSA Panel on Contaminants in the Food Chain (CONTAM), Knutsen, H. K., Alexander, J., Barregård, L., Bignami, M., Brüschweiler, B., … & Hogstrand, C. (2018). Update of the risk assessment on 3‐monochloropropane diol and its fatty acid esters. EFSA Journal, 16(1), e05083. ↩︎
European Commission, 2001, Opinion of the Scienti®c Committee on Food on 3-Monochloro-propane-1,2-diol (3-MCPD) Updating the SCF Opinion of 1994 (adopted on 30 May 2001), SCF/CS/ CNTM/OTH/17 Final. ↩︎
World Health Organization. (2011). Evaluation of certain contaminants in food. World Health Organization technical report series, (959), 1. ↩︎
Korte, R., Schulz, S., & Brauer, B. (2021). Chloropropanols (3-MCPD, 1, 3-DCP) from food contact materials: GC-MS method improvement, market survey and investigations on the effect of hot water extraction. Food Additives & Contaminants: Part A, 38(6), 904-913. ↩︎
Verfahrenstechnik, U., & Fellner, M. DIPLOMARBEIT MIGRATION FROM COFFEE FILTER PAPERS. ↩︎
Becalski, A., Zhao, T., Breton, F., & Kuhlmann, J. (2016). 2-and 3-Monochloropropanediols in paper products and their transfer to foods. Food Additives & Contaminants: Part A, 33(9), 1499-1508. ↩︎
Trézl, L., Csiba, A., Juhasz, S., Szentgyörgyi, M., Lombai, G., Hullán, L., & Juhász, A. (1997). Endogenous formaldehyde level of foods and its biological significance. Zeitschrift für Lebensmitteluntersuchung und-Forschung A, 205, 300-304. ↩︎
Hayashi T, Reece CA, Shibamoto T. Gas chromatographic determination of formaldehyde in coffee via thiazolidine derivative. J Assoc Off Anal Chem. 1986 Jan-Feb;69(1):101-5. PMID: 3949681. ↩︎
Doležal, M., Chaloupská, M., Divinová, V., Svejkovská, B., & Velišek, J. (2005). Occurrence of 3-chloropropane-1, 2-diol and its esters in coffee. European Food Research and Technology, 221, 221-225. ↩︎
European Food Safety Authority. (2014). Endogenous formaldehyde turnover in humans compared with exogenous contribution from food sources. EFSA Journal, 12(2), 3550. ↩︎
Kurzrock, T., & Speer, K. (2001). Diterpenes and diterpene esters in coffee. Food Reviews International, 17(4), 433-450. ↩︎
Zanin, R. C., Kitzberger, C. S. G., & Benassi, M. D. T. (2020). Characterization of roasted Coffea arabica species by the relationship between caffeine and diterpenes contents. Brazilian Archives of Biology and Technology, 63. ↩︎
Cavin, C., Holzhaeuser, D., Scharf, G., Constable, A., Huber, W. W., & Schilter, B. (2002). Cafestol and kahweol, two coffee specific diterpenes with anticarcinogenic activity. Food and chemical toxicology, 40(8), 1155-1163. ↩︎
Ren, Y., Wang, C., Xu, J., & Wang, S. (2019). Cafestol and kahweol: A review on their bioactivities and pharmacological properties. International journal of molecular sciences, 20(17), 4238. ↩︎
Socała, K., Szopa, A., Serefko, A., Poleszak, E., & Wlaź, P. (2020). Neuroprotective effects of coffee bioactive compounds: a review. International Journal of Molecular Sciences, 22(1), 107. ↩︎
Ricketts, M. L. (2007). Does coffee raise cholesterol?. Future Lipidology, 2(4), 373-377. ↩︎
Thelle, D. S., Heyden, S., & Fodor, J. G. (1987). Coffee and cholesterol in epidemiological and experimental studies. Atherosclerosis, 67(2-3), 97-103. ↩︎
Jee, S. H., He, J., Appel, L. J., Whelton, P. K., Suh, I. I., & Klag, M. J. (2001). Coffee consumption and serum lipids: a meta-analysis of randomized controlled clinical trials. American journal of epidemiology, 153(4), 353-362. ↩︎
Malerba, S., Turati, F., Galeone, C., Pelucchi, C., Verga, F., La Vecchia, C., & Tavani, A. (2013). A meta-analysis of prospective studies of coffee consumption and mortality for all causes, cancers and cardiovascular diseases. European journal of epidemiology, 28, 527-539. ↩︎
Crippa, A., Discacciati, A., Larsson, S. C., Wolk, A., & Orsini, N. (2014). Coffee consumption and mortality from all causes, cardiovascular disease, and cancer: a dose-response meta-analysis. American journal of epidemiology, 180(8), 763-775. ↩︎
Tverdal, A., Selmer, R., Cohen, J. M., & Thelle, D. S. (2020). Coffee consumption and mortality from cardiovascular diseases and total mortality: Does the brewing method matter?. European Journal of Preventive Cardiology, 27(18), 1986-1993. ↩︎
Urgert, R., van der Weg, G., Kosmeijer-Schuil, T. G., van de Bovenkamp, P., Hovenier, R., & Katan, M. B. (1995). Levels of the cholesterol-elevating diterpenes cafestol and kahweol in various coffee brews. Journal of agricultural and food chemistry, 43(8), 2167-2172. ↩︎
Gross, G., Jaccaud, E., & Huggett, A. C. (1997). Analysis of the content of the diterpenes cafestol and kahweol in coffee brews. Food and Chemical Toxicology, 35(6), 547-554. ↩︎
