The following is an excerpt from Most Delicious Poison: The Story of Nature’s Toxins—From Spices to Vices by Noah Whiteman.

Disclaimer: When you purchase products through the Bookshop.org link on this page, Science Friday may earn a small commission which helps support our journalism.

Buy The Book

Most Delicious Poison: The Story of Nature's Toxins—From Spices to Vices

Buy

No suitor comes in my house
unless he has promised to me himself
and has it also inserted into the marriage contract
that I shall be permitted
to brew coffee whenever I want.
— Johann Sebastian Bach, Coffee Cantata

Filter It

Just before bed, I precisely measure out the amount of coffee Shane and I will need to make it through the next day as I ready the automatic drip machine. If I prepare too little, I will be sleepy and unable to mentally focus. If I overshoot, I will be anxious and feel sick.

At lower doses, caffeine causes euphoria and increases our alertness and cognitive performance. At higher doses, it causes nausea and increases anxiety and overall shakiness. I am addicted to caffeine and couldn’t feel better about it. You’ll see why.

Although I used to use a French press, I now only make coffee with an automatic drip machine or by pour-over. I made the switch to making only filtered coffee after reading a 2020 study of more than half a million people in Norway that found the adults who consumed unfiltered coffee had a significantly higher chance of dying over twenty years than did those who drank filtered coffee or no coffee. The increased mortality risk is probably associated with drinking unfiltered coffee, at least partly due to the presence of two coffee terpenoids that are largely removed during filtration. Both are associated with increased cardiovascular and heart disease because they raise cholesterol levels.

The offending terpenoids, cafestol and kahweol, elevate low-density lipoprotein (LDL, or “bad”) cholesterol levels in our blood. In fact, cafestol is the most potent LDL-inducing chemical known from the human diet. The mechanism, at least according to studies in laboratory mice, is the binding of these two sterols to a hormone receptor in the small intestine. This binding sends the wrong message to the liver, which then cranks up the production of LDL.

Scandinavian boiled coffee, French press (cafetière) coffee, and Turkish coffee have the highest levels of these terpenoids. Espresso and mocha have more modest levels, and instant, percolated (with a filter), and filtered coffee have the lowest, by a wide margin: unfiltered coffee contains thirty times more cafestol and kahweol than does filtered coffee.

I found several articles on coffee or wellness-related websites claiming that reusable metal mesh filters (often stainless steel or 24k plated gold) are less effective at trapping kahweol and cafestol than paper filters are. I could find no evidence supporting this claim in the peer-reviewed literature. But absence of evidence is not evidence of absence.

So, I dug into this question of the effectiveness of metal filters. Instead of an absence of evidence, I found a rigorous experimental study that resolved the question of whether metal mesh filters did the trick at removing these terpenoids. When used in automatic drip machines, paper and metal filters (the investigators used Swissgold brand of mesh filters) were equally effective in preventing most kahweol and cafestol from passing through to the brew below. However, there is a catch. In this study and a few others, pour-over coffee from a metal mesh filter produced higher levels of kahweol and cafestol than did coffee from an automatic drip machine using the same kind of filter. Although speculative, the cause of this difference may be that the water added to the automatic drip machines drips slowly and gently onto the ground coffee in the filter below, allowing a thick “filter cake” to form on the bottom, potentially acting as a kind of first-pass filter itself. During pour-overs, on the other hand, a stream of water rapidly causes the grounds to continue to swirl and remain suspended, likely allowing more of the smaller-sized particles rich in the terpenoids to escape through the mesh. Why didn’t we get to do experiments like this in home economics class?

I know what you are going to ask next: what about coffee pods? Pods have a paper filter built into them, so the cafestol and kahweol levels are comparable to automatic drip coffee prepared with paper or metal mesh filters or pour-over coffee prepared with a paper filter. What’s more, the characteristics of the bean (variety, roast), the water temperature, the grind size, and the water‑to‑coffee ratio all affect the levels of terpenoid in coffee.

I was convinced enough by what I gleaned from the scientific literature to change my habits. I stopped my twenty-year practice of using a French press to make coffee. Although I occasionally splurge on an espresso drink (which has only modest levels of the terpenoids) when I’m out and about or traveling, at home I now only make filtered coffee using the automatic drip machine with a gold mesh filter. Or I make pour-over coffee using a paper filter—and an unbleached one at that.

In the United Kingdom, people generally drink coffee that contains low to modest levels of cafestol and kahweol (filtered coffee, instant coffee, or espresso). A UK study followed 171,616 people from 2009 to 2018 to determine whether drinking coffee was associated with reduced risk of death, and if so, by how much. In general, those who drank coffee had a lower risk of death than those who did not, and this finding held for the risk of dying from cancer or cardiovascular disease.

But the devil is in the details. A slightly reduced risk of dying was found for those who drank up to 2.5 drinks per day compared to those who didn’t drink coffee. The people drinking 2.5 to 4.5 coffee drinks per day were 29 percent less likely to die during the study than were people who drank no coffee. Above 4.5 drinks per day, and the risk of dying during the study was the same as those who consumed up to 2.5 drinks per day.

Those in the middle of the spectrum (i.e., those who drink 2.5 to 4.5 cups a day) were the least likely of all the participants to die. This study comports with a much larger umbrella review of studies that found that drinking 3 to 4 cups per day is associated with a roughly 17 percent lower risk of death than the risk for nondrinkers.

Finally, there was a big potential downside to high caffeine or coffee consumption in pregnancy. Compared with low consumption of coffee, high consumption was associated, on average, with a 31 percent higher risk of low birth weight, 46 percent higher risk of pregnancy loss, and 22 percent higher risk of first trimester preterm birth, and 12 percent higher risk of second trimester preterm birth. There were no elevated risks observed in the third trimester. Chapter 10 will examine what may cause these patterns.

The protective association of coffee drinking is also observed for those who drink decaffeinated coffee. One explanation is that for most of us, coffee, whether caffeinated or decaffeinated, is the single largest source of antioxidants in our diet in the form of polyphenols. As discussed in an earlier chapter, the term antioxidant connotes chemicals that protect against oxidants, cellular stressors produced by our own bodies and the environment.

The potential protective effect of coffee in our diet may be largely due to the presence of these antioxidants in coffee beans. Among these, the chlorogenic acids are the prime suspects. Double-blind, placebo-controlled studies have found that chlorogenic acids in the diet can improve cardiovascular functioning, reduce blood pressure, and decrease the risk of metabolic syndrome. Coffee drinkers consume up to 1 gram of chlorogenic acids per day; by comparison, a typical adult aspirin tablet weighs 325 milligrams, or one-third the amount consumed by coffee drinkers.

This discussion of coffee brings us back full circle to earlier discussions of chlorogenic acids. These chemicals, the same phenolics associated with protecting our health when we drink coffee, are among the toxins we’ve seen in earlier chapters. Chlorogenic acids that end up in the fog drip of eucalyptus trees kill other plants growing near them. And these same chemicals, found in the salt grass that montane voles eat, cause the rodents to stop reproducing.

In the Norwegian study on unfiltered coffee, the higher mortality rates among unfiltered-coffee drinkers (compared with the filtered-coffee drinkers and nondrinkers) may be caused by the LDL-raising effects of the terpenoids cafestol and kahweol. On the other hand, the lower death rates among filtered-coffee drinkers compared with the nondrinkers may be caused by the polyphenols, including chlorogenic acids. So, coffee contains toxins that are associated with both positive and negative effects on human health.

A cautionary note is called for. In any observational study, whether the Norwegian one, the UK one, or the umbrella one, selection bias could give rise to the observed effects between coffee drinking and health risks. If selection bias came into play, then some other variable besides coffee drinking underlies the pattern because it correlates with people who consume different levels, and types, of coffee. Yet in light of animal studies, including some human research, the potential biological mechanisms for both the LDL-increasing aspects of the terpenoids and the cardiovascular and diabetes risk-reducing aspects of the phenolics in coffee are clear. Still, we need to tease out all the factors by conducting a long-term, double-blind, placebo-­controlled study, which is difficult to do.

Although I’m neither a nutritionist nor a physician and so do not give dietary or medical advice, I’m convinced enough of the totality of this information to change my own habits. Therefore, I will both filter it and continue to drink a lot of caffeinated coffee, although not too much. I drink three or four large cups of filtered coffee per day, with a cappuccino or flat white sneaked in there a few times a week as a treat.

Beyond the potential cardiovascular protection of coffee, coffee drink‑ ing is also associated with dramatically lower risk of Parkinson’s disease in observational studies. Similarly, a large observational study in the United Kingdom found that drinking three to six cups of coffee or tea daily reduced the risk of dementia by 28 percent and stroke-induced dementia by 48 percent. What is driving these associations is unclear. One candidate is a serotonin derivative called eicosanoyl‑5‑hydroxytryptamide from coffee beans. The chemical can slow Alzheimer’s disease progression in laboratory rat models of the disease.

These findings certainly warrant further study. With coffee’s terpenoids and phenolics out of the way, we can finally get down to the big reason we consume caffeinated beverages like coffee: the alkaloid caffeine. We will now explore its origins, its biological effects, and our relationship with this alkaloid.

Brazen Beetles and Killer Coffee

Caffeine and the human mind seem like a match made in heaven. But even though billions of people imbibe caffeinated beverages each day, caffeine first evolved in the absence of humans. The two main species cultivated for coffee beans, Coffea arabica (the source of arabica beans) and C. canephora (the source of robusta beans), are both native to the highlands of Ethiopia but are now grown all over the world in similar climates.

As a PhD student, I used to buy coffee beans from a St. Louis roaster called Kaldi’s Coffee. It was there that I learned about Kaldi, the apocryphal ancient goat herder mentioned in a 1671 treatise on coffee by the Maronite chronicler and professor Antoine Faustus Nairon. He wrote of a “certain camel herder or, as others say, of goats” from “Arabia Felix,” who had complained to the monks that he had been awoken at night by his goats, which seemed to be “jumping.”

Nairon explained that the prior of the monastery decided to find out why. When he investigated, he found the goats eating the berries of the coffee plant. A potion he made from the boiled beans gave him insomnia, and so he then ordered the rest of the monks to drink it to help them stay awake during the night watch and evening prayers.

We don’t know for sure when coffee was first cultivated by humans. However, it was widely used in the Arabian Peninsula in antiquity, coming to Yemen through the Yemenite Sufi community around the fourteenth century BCE. But coffee didn’t make it to Europe until the early 1600s. The Dutch began cultivating it in glasshouses in Amsterdam by 1616 and then in plantations in the East Indies. Thereafter came cultivation by the French, Spanish, and British in their own colonies.

The root of the word coffee is traced to the Arabic qahwah, which may have originally referred to a kind of wine but has the root qahiya, which means “to have no appetite.” So the dark, red-wine-like color of coffee coupled with its appetite-suppressing effect are embodied in its name.

We didn’t breed plants to produce caffeine, although we’ve helped create different modern cultivars from their wild relatives through artificial selection. These and all other caffeine-producing plants were making caffeine tens of millions of years before any humans were walking the earth.

On October 7, 1984, a breakthrough was announced in the New York Times: “Caffeine Is Natural Insecticide, Scientist Says.” As the headline suggests, biologist James Nathanson had just demonstrated that caffeine is indeed a potent natural insecticide. Nathanson discovered this by incorporating powdered tea leaves and coffee beans into artificial caterpillar food, which he then gave to newly hatched tobacco hornworm caterpillars, which don’t feed on plants containing caffeine in the wild. The hornworm adult is also called a hawk moth.

What Nathanson found shocked the world (although, by now, probably not you): “At concentrations from 0.3 to 10 percent (by weight) for coffee and from 0.1 to 3 percent for tea, there was a dose-dependent inhibition of feeding associated with hyperactivity, tremors, and stunted growth. At concentrations greater than 10 percent for coffee or 3 percent for tea larvae were killed within 24 hours.”

Further experimentation revealed that the level of caffeine naturally found in undried tea leaves (0.68 to 2.1 percent) or undried coffee beans (0.8 to 1.8 percent) was sufficient to kill all of the caterpillars. Nathanson found the same insecticidal effects of caffeine on mosquitoes, beetles, butterflies, and true bugs, including at concentrations found in nature.

His most telling experiments involved spraying a mixture of caffeine on tomato leaves, the typical host plant for hornworms. Tomatoes don’t produce caffeine, so these experiments were designed to mimic what the sudden evolution of caffeine might do to an herbivore that found itself eating a caffeine-producing plant. As the concentration of caffeine went up, there was a concomitant reduction in the amount of leaf chewed by the caterpillars. In other words, the caffeine protected the plant from attack by the hornworms.

A similar effect was found in 2002, when scientists in Hawaii acciden‑ tally discovered that a caffeine solution being tested as a toxicant to control the coqui, an invasive frog introduced from the Caribbean, also killed most of the large slugs found in their field plots. The researchers followed up by spraying or dipping vegetables in solutions containing caffeine concentrations of 1 to 2 percent, the same levels found in coffee beans, and offering them to the mollusks. Most of the snails and slugs died. And at far lower concentrations (0.01 percent), caffeine deterred them from feeding.

Although caffeine mimics the insecticidal effects of coffee or tea when sprayed artificially on plants that don’t make it, this surface application is quite artificial. After all, the caffeine is made naturally inside coffee and tea plants’ cells. Another way to sort it out would be to endow a plant species that does not normally synthesize caffeine in its cells with the ability to make it and see how resistant its leaves become.

Biologists did just that by genetically engineering caffeine production in tobacco plants, which don’t normally make caffeine. The researchers spliced three caffeine-producing genes from the coffee plant genome into the tobacco plant’s genome in the laboratory.

These transgenic tobacco plants produced levels of caffeine similar to those found in coffee plants. Leaves from tobacco plants carrying coffee plant caffeine genes and control leaves without these genes were fed to tobacco cutworm caterpillars. The leaves producing caffeine were 99.98 percent less susceptible to herbivory than were the control leaves.

In naturally occurring caffeine-bearing plants like citrus, coffee, and tea, genes encoding the enzymes used to make caffeine evolved from exist‑ ing genes that had performed a different function. Although we cannot board a time machine to determine why caffeine first evolved in any plant, the only known function of caffeine in plants is as direct or indirect defense against natural enemies. This role seems obvious, thanks to Nathanson’s experiments. But caffeine might have first evolved as a molecule the plant used to signal the presence of stressors, just as salicylic acid serves as a hormone that signals the presence of attackers, rather than as a defensive strategy. Under this model, only later did caffeine become co‑opted by plants as a toxin, just as willows took the ubiquitous plant-signaling hormone salicylic acid and turned it into a toxin by making much more of it.

Not surprisingly, some specialist herbivores of coffee plants have evolved the capacity to resist the toxic effects of caffeine. The most troublesome is the coffee berry borer, a small beetle that tunnels into the fruit and lays eggs in the beans. The larvae of this beetle consume the bean from within, rendering it unusable for coffee bean production. The beetle is native to the same African regions that gave rise to the two Coffea species now cultivated in the worldwide tropics.

The coffee berry borer has found coffee plants wherever they are, even in Hawaii, costing producers at least $500 million per year in damaged plants globally. Coffee is second only to petroleum in value as a global commodity, valued at $83 billion per year, and the borer is a major threat. The beetle itself cannot tolerate the insecticidal caffeine in its food. To get around it, the insect relies on enzymes produced by the bacteria living in its gut—its microbiome—which detoxify the caffeine.

When these beetles were treated with antibiotics that killed the bacteria and given their normal diet of coffee beans, they perished just as any other insect fed this diet would. The amount of caffeine consumed by one of these beetles in a meal is equivalent to the amount you’d find in five hun‑ dred cups of coffee, ten times the level that killed twenty-one-year-old Lachlan Foote in the early morning hours on New Year’s Day in 2018.

Foote’s tragic death in Australia was caused by an accidental caffeine overdose. He’d added one teaspoon of pure caffeine powder to a protein shake. After saying good night to his parents, he was found by his father in the bathroom, where he had died. It is not clear where he obtained the caffeine, but there was no warning label found on the bag he had used to store it.

Foote consumed at least five thousand milligrams of caffeine, equivalent to fifty cups of coffee. The FDA’s recommended daily allowance for adults is four hundred milligrams. In the wake of his death, the family successfully pushed the Australian government to ban food additives with caffeine concentrations above 5 percent and liquids with levels above 1 percent. A ban went into effect less than a year later.

In October 2014, I was out of town when I learned that my dog was in critical condition at an emergency veterinary hospital in Tucson. She had broken into the dog sitter’s backpack and torn into a bottle of gelatin-coated caffeine pills. The dog ate more than the eight to thirteen pills needed to achieve caffeine toxicosis, given her size. The dog sitter desperately tried to save her and, with the help of my neighbor Erika, brought her to the vet. My dog fell into a coma despite their heroic efforts.

I raced home. The vet suggested that we prepare ourselves that the dog might not survive. But under their steady care, and with a lot of luck, she began to awaken, weak and confused but with her personality intact.

These dark lessons teach us that plants don’t make caffeine for our benefit. They make it as a defense against being eaten. However, we’ve learned to use caffeine to improve our own lives because, unlike the tiny bodies of insects, our own big bodies can handle fairly large doses of this alkaloid and because our brains seem prewired to interact with it.

Excerpted from the MOST DELICIOUS POISON by Noah Whiteman. Copyright © 2023 by Noah Whiteman. Reprinted with permission of Little, Brown and Company. All rights reserved.