Christmas season is the time to eat lots of chocolate. And as science teaches us, your confectionery is actually the superfood which will make you healthy, slim and clever. Good for you, good for the chocolate industry which often generously sponsors such scientists.
In May 2016, I brought a story about chocolate health research and how it is funded by food industry giants Mars and Nestle. The main protagonist was Thomas Lüscher, cardiology professor at the University of Zürich and head of the heart centre at the University Hospital Zurich in Switzerland. Lüscher postulated that eating dark chocolate daily is beneficial for heart insufficiency patients and may prevent heart attacks, he now offers some additional advice. Now his peer, the Swedish vascular surgeon Jonas Malmstedt, provides his analysis of Lüscher’s publications below. Another study which Malmstedt unpicks, is an opus from the Luxembourg Institute of Health (Alkerwi et al, 2016), which declared that eating chocolate makes you younger and healthier, and prevents diabetes on top.
Lüscher himself admitted to have been constantly receiving funding from Mars and Nestle for his research, but when I submitted a Freedom of Information inquiry to his university I was officially told that what their professor received from the two food giants in the last ten years was nought. Not a single Swiss Franc, simply because the money was thoroughly mixed with all possible industry (mostly pharma) sponsorships and then re-routed through his foundation Zurich Heart House and the University Hospital. This is how Swiss secrecy works, and it is indeed very efficient.
Elusive flavanol effects
One common problem of all chocolate health studies is that they fail to exactly correlate their astounding health effects to any magic ingredient of chocolate. Sure, the claim goes that the chemical compounds of the flavanol family, especially the epicatechines, are the ones which stabilise your blood pressure and reduce the oxidative damage to your blood vessels (as it was originally postulated by Norman Hollenberg of Harvard Medical School and Helmut Sies of University Clinic Düsseldorf, see my previous report). However, the European Food Safety Authority (EFSA) issued already in October 2010 a “Scientific Opinion on the substantiation of health claims related to cocoa flavanols and protection of lipids from oxidative damage (…), and maintenance of normal blood pressure (…)”, DOI: 10.2903/j.efsa.2010.1792. The EFSA screened “one systematic review and 11 intervention studies in humans” which reported effects of flavonols on markers of oxidative stress or antioxidant status:
“On the basis of the data presented, the Panel concludes that a cause and effect relationship has not been established between the consumption of cocoa flavanols and protection of lipids from oxidative damage”.
For the alleged effect on blood pressure, EFSA scrutinised ten randomised clinical trials, and found that the ones with the strongest effect were “small and un-blinded”, while the “adequately powered and well controlled” clinical trials showed no effect of flavanols:
“On the basis of the data presented, the Panel concludes that the evidence provided is insufficient to establish a cause and effect relationship between the consumption of cocoa flavanols and maintenance of normal blood pressure”.
The retired Düsseldorf professor and oxidative stress specialist Sies however told me:
“The vascular effects of flavanols ((-)-epicatechin in particular) are scientifically proven […] [These] are largely short-term effects, namely over hours and days, and their existence is beyond doubt. Regarding the long-term effects on the cardiovascular health, there one needs to perform long-term studies similar as with cancer diseases. This is difficult, expensive and in fact tedious: this is why there is no conclusive information yet“.
Whether flavanols from raw cocoa have any health effect or not, is still under debate. But how do we know that these are indeed still present in the industrial chocolate which Mars and Nestle sell us? Sies explained to me:
“Fermentation, heat treatment, alkalisation all lead to the loss of flavanols, which can exceed 90%. With gentler methods the losses can be significantly reduced. This is why you can produce from the same cocoa bean crop both low-flavanol as well as high-flavanol cocoa powder. […] Whether dark chocolate really has a higher flavanol content than a less dark one, this must be proven through measurement of flavanols. This didn’t always happen”.
Indeed, Sies’ Swiss colleague Lüscher admitted to me that he never independently measured that flavanol content in the chocolate bars he distributed to his patients, but relied fully on Nestle’s own analyses. The Swiss chocolate maker supplied those together with their product „Nestlé Noir Intense“, sugar content 30%, to be distributed to Lüscher’s heart insufficiency patients. The control was white chocolate.
And as for measuring flavanol derivates in the patient blood samples after the sweet medicine was administered, Lüscher admitted that the exact values were rather irrelevant to him and pointed out that the measurement was done by his collaborator Mauro Serafini, “in a special laboratory”. I could not establish where and if this laboratory exists, Serafini did not reply to my email, while one his institutional account at Italian agricultural council CREA proved deactivated. This Italian professor once published a paper that eating dark chocolate (but not milk-containing chocolate!) boosted anti-oxidant capacity of blood, not somewhere, but in the journal Nature (Serafini et al, 2003). Serafini currently runs a Frontiers thematic series “Chocolate and Health: Friend or Foe?” which after several months failed to recruit any manuscripts (but 10 authors, whose “names will be published here as soon as their papers are accepted”).
Lüscher’s health advice
Several weeks after my previous report, I spoke with Lüscher on the phone, and these are the final statements he supplied to me by email afterwards:
“Sugar is, in my opinion, not really desirable and unfavorable inside the body, so one should never eat too much sweet chocolate (e.g., not one whole bar a day) and definitely prefer bitter black chocolate which contains more flavonols and less sugar.
This year we are going to make a randomized controlled study with Mars, where we will clinically test a sugar-free epicatechin drink [Lüscher again receives personal funding from Mars since at least 2013, -LS]. The epicatechin measurements are again completely blinded by Mars (that is, according to the best scientific standards), while all other measurements are made by our laboratories.
The epicatechin measurements serve only to demonstrate that, after taking the active substance, epicatechin is indeed detectable in the body – the primary endpoint in all studies was the vasomotion of the arteria brachialis and the coronary vessels, respectively.
Of course, as is customary in our lab, no employees of any company were directly involved in the investigations and were never present in our laboratory, nor will they ever be. The fact that supporting companies are presented with the final manuscript for review and comment prior to submission is commonly accepted practice (and a matter of decency) and in no way implies that the company was involved in the analysis of the data or preparation of the manuscript or the interpretation of the manuscript data – that was presented wrong [in my article, -LS]”.
Lüscher is not only a professional in such sponsored industry-academia interactions, he is also a champion of research integrity. In June 2014, he placed a German-language article in the journal Herz, titled “Quality and integrity in the production and publication of scientific results. Data trimming, manipulation and (self) plagiarism”. This is peculiar, because just a few months earlier, Lüscher received a written reprimand from his own university, for the almost unchanged copy-pasting of several pages of text for the textbook “Siegenthaler’s Differential Diagnosis”. This chapter was written by Lüscher’s former colleague, Christoph Scharf, although only his boss was then named as its sole author. Despite this incident, the admonished plagiarist Lüscher did not declare any conflicts of interest in his anti-plagiarism Lüscher, Herz 2014 article (he probably did not see any, just like he did not see them with his chocolate research). In that paper, Lüscher concluded:
“We should not believe everything which we are presented in print- critical reading remains important for everyone, for readers, peers, reviewers and publishers”.
This is now exactly what Lüscher’s peer Malmstedt now does below.
Post-publication peer review of chocolate health studies, by Jonas Malmstedt
When evaluating the hype of chocolate as a possible nutrient good for your cardiovascular health, one must put the proposed effects in perspective. First, it is not chocolate per se that has the proclaimed beneficial effects; it is the content of flavonoids which in turn is dependent on cocoa content and the processing of cocoa. Much of the industrial processing involved in chocolate production is known to alter or degrade flavonoid content (Hurst et al, 2011).
One must also remember that the main source of flavonoids is not from cocoa, but from berries, grapes and fruits (Vogiatzoglou et al, 2015). In fact, the contribution from cocoa products to total flavanol intake is only 5 % in Europe, corresponding to an average intake of 7-19 mg/day (Knaze et al 2012).
On Alkerwi’s chocolate-prevents-diabetes claims
The article by Alkerwi et al, 2016 is a cross-sectional survey where they sought to establish a relationship between chocolate consumption and metabolic phenotypes, with the main finding of a negative association between chocolate and insulin levels. For an increase in daily chocolate consumption by 100g the serum insulin levels were 0.16 µg/L lower.
The study is highly problematic in several ways. Firstly, there is a major selection bias as the participation rate was only 32% and subjects from working class were underrepresented (Alkerwi et al 2010).
Furthermore, the exposure was poorly described as the Food Frequency Questionnaire (FFQ) used was not validated for flavonoid intake. The validity of exposure is not determined by FFQ for chocolate and is only loosely correlated to fruit and vegetables: r = 0,10 – 0,19 and cross-classification was poor with Kappa 0,08 – 0,1 (Sauvageot et al, 2013). In fact, the specific questionnaire is not published (Alkerwi et al, 2010). The results are also scarce regarding the distribution of the parameters used in linear regression and one can suspect chocolate consumption being skewed (ie. non-normal distribution), which in turn casts doubt about the whole regression model. In fact, the chocolate consumption is not even measured as a continuous variable, as the FFQ used a frequency question with 6 levels. The levels were then transformed to approximate chocolate intake in grams per day, and apparently used in analysis as a continuous variable. It is also striking that BMI was omitted from regression analysis (for the dubious claim of being in the casual pathway). It is more likely that BMI was omitted due to the fact that inclusion rendered chocolate consumption non-significant.
Other important confounders omitted from the analysis are poverty, fat and sugar intake and dyslipidemia.
A recent review found the evidence for anti-inflammatory effects of cocoa consumption to be low (Ellinger & Stehle, 2016). They also pointed out that the majority of studies with chocolate as source of flavanol did not find any anti-inflammatory effects and in those which found positive effects of chocolate, only single measure changed.
On Lüscher’s chocolate-prevents-heart-attacks claims
The small RCT (n=20) by Flammer et al 2011 showed an increase in endothelium dependent vasodilatation in patients with congestive heart failure (CHF) who consumed 80g Nestlé Noir Intense (94 mg catechin and epicatechin) during 4 weeks. All other long-term endpoints failed to show any difference, e.g. no change in antioxidant capacity, platelet function or baroreceptor function. Interestingly, the intervention group had a smaller baseline brachial artery diameter, indicating a possibly more vasoconstricted state, which may indicate a greater potential for change in vasodilatation.
A recent review of RCTs regarding cocoa consumption and inflammation concluded that little evidence exists that cocoa may reduce inflammation (Ellinger & Stehle, 2016). Furthermore, one large cohort study (n=24 885) with 11 years follow up concluded that there were no reduction of CVD risk by flavan-3-ol intake (Vogiatzoglou et al, 2015).
The big question is if change in surrogate markers such as vasodilatation, insulin levels and platelet reactivity really translates into meaningful clinical improvements such as reduced incidence of cardiovascular events and lower mortality? An important concern regarding chocolate as the flavonoid provider is the high content of sugar and fat and the link to cardiovascular disease (Bastien et al, 2014). The serving in the Flammer et al 2011 study translates to 13 kg fat and 7,5 kg sugar extra per year, or as the concluding remark in an editorial in the journal Heart (Donaldson et al, 2016):
“Thus, whether the advice on potential benefit of chocolate consumption will be a moment on the lips of scientists but a lifetime on the hips of the population remains to be seen”.
Given the relatively minor role of specific nutrients on overall cardiovascular risk, one can ask if research on chocolate can be justified in a population heath context. The research should be directed towards questions of how to implement the already well known dietary policies for cardiovascular disease, which will give true benefit in large populations (see for example the landmark study on Mediterranean diet in NEJM which showed an remarkable reduction in major cardiovascular events by 30 % (Estruch et al 2013)). For an excellent review of the state of art dietary recommendations one can start with the article by Mozaffarian in Circulation this year, “Dietary and Policy Priorities for Cardiovascular Disease, Diabetes, and Obesity: A Comprehensive Review.” (Mozaffarian, 2016), which stated that we should be
“focusing on foods and overall diet patterns, rather than single isolated nutrients; recognizing the complex influences of different foods on long-term weight regulation […]. […]. Evidence-informed dietary priorities include increased fruits, nonstarchy vegetables, nuts, legumes, fish, vegetable oils, yogurt, and minimally processed whole grains; and fewer red meats, processed (eg, sodium-preserved) meats, and foods rich in refined grains, starch, added sugars, salt, and trans fat.”
but are there lots of clinical trials that show chocolate to be harmful?
If it is the general rule for the chocolate producers to pay the researchers, and to have a right of review / veto of the final manuscript, I would be surprised if there were many negative results.
Isn’t this again the classic case of confusing correlation with causation? Even more so here, since there is not even a whole lot of correlation going on? Being it positive or negative correlation.
Who cares as long as it tastes good! I love Mars and Nestle
Pingback: Friday Links | Ruff Observer
Pingback: Successione (guerra di) - Ocasapiens - Blog - Repubblica.it
Pingback: Cacao sempre più meravigliao – OggiScienza
What statistical theory requires a predictor variable in linear regression to be Normally distributed? “one can suspect chocolate consumption being skewed (ie. non-normal distribution),” Can you provide a source? Call me old-fashioned but I always though and taught that it was the stochastic disturbance terms that had to be (approximately) IID Normal.
Pingback: Michael HotTiger of Zurich, patron of biomedical ethics – For Better Science
Pingback: Steven Houser and the Temple of Fraud – For Better Science