Cocoa is one of the few foods with a genuinely interesting cardiovascular literature behind it, and most of what is sold has had the relevant compounds removed on purpose.
The compounds are flavanols, a subgroup of flavan-3-ols including epicatechin and the procyanidins built from it. They are what the cocoa research is about. They are also bitter, astringent, and reddish, which puts them in direct conflict with what a manufacturer wants from cocoa powder.
Alkalization solves that problem. Also called Dutch processing, it treats the cocoa with an alkaline solution, raising the pH from its natural 5.3 to 5.8 up to 7.6 or higher. The powder darkens, the harshness softens, and it disperses better in liquid. It is a cosmetic and sensory process. Nobody performs it for a nutritional reason.
The cost was quantified in a survey of commercial cocoa powders, and the relationship turned out to be linear. Untreated natural cocoa averaged 34.6 mg of total flavanols per gram. Lightly alkalized powder, at pH 6.5 to 7.2, averaged 13.8. Medium treatment brought it to 7.8. Heavily alkalized powder, the very dark kind, averaged 3.9 mg per gram. That is a loss of roughly 89 percent from natural, and even the gentlest commercial treatment removed about 60 percent.
Two things make this strange (in my opinion) rather than merely interesting.
The first is that darkness reads as quality. A deep, almost black cocoa looks stronger and more premium, and it is the tier with the least of the compound the marketing is implicitly referencing. The visual signal runs opposite to the chemistry.
The second is disclosure. In the United States, alkalized cocoa must be labelled, and it usually appears as “processed with alkali” or “Dutch process” in the ingredient list. That phrase is accurate and almost entirely uninformative to a shopper. It does not say what was processed out, and no front-of-pack claim is obliged to mention it.
This survey measured composition, not health outcomes, so it establishes what is present rather than what any given powder does in a person. And the work came from a chocolate manufacturer’s own research centre, which is unusual in that the finding runs against the sponsor’s commercial interest. That direction of bias makes it more credible, not less, though it remains an industry publication.
If cocoa is being eaten as a food, none of this matters much. If it is being eaten for the flavanols, the ingredient list decides whether that is happening at all, and the natural, lighter, more bitter powder is the one that still contains them.
Miller et al., J Agric Food Chem 2008. PMID 18710243
