In healthy, pregnant women with normal BP and without risk of PE, acute and chronic consumption of flavanol-rich chocolate was not associated with significant changes in either FMD, SBP or DBP. Our clinical study provides important insights into the feasibility, acceptability and methodology of a larger clinical trial to evaluate long-term chocolate intake in pregnant women. Our results confirmed the feasibility and good compliance of our intervention and its effectiveness in increasing blood theobromine and flavanol metabolite concentrations after acute ingestion.
The association between chocolate consumption and the risk of PE was explored for the first time in a cohort study  of 2,291 pregnant North American women who gave birth to a single, living baby. “Exposure” to chocolate was measured in 2 ways: data on the provision of caffeine-rich drinks and chocolate consumption since the beginning of pregnancy were collected during a structured interview around the 14th week of gestation. Consumption was also evaluated during the 3rd trimester of pregnancy in a post-natal interview. On the other hand, the serum concentrations of theobromine were quantified in a blood sample collected from the umbilical cord during delivery. Serum theobromine concentrations were inversely correlated with the risk of PE before and after adjustment for the main confounding factors (odds ratio (OR) = 0.31; 95% CI: 0.11 - 0.87, for the highest compared to the lowest quartile). Women who consumed 5 or more portions of chocolate per week during the 3rd trimester of pregnancy manifested a 40% decrease in PE risk compared to women who consumed less than 1 portion of chocolate per week (OR = 0.60; 95% CI: 0.30 - 1.24). However, this diminution was not significant. A case–control study  nested in a cohort does not support the previous finding. According to the authors, unmeasured confounding or reverse causation may account for the positive association previously reported as discussed above. A non-placebo, controlled, non-blinded study by Di Renzo et al  suggested that modest daily intake of high-cocoa content chocolate contributes to BP reduction during pregnancy. The results of these epidemiologic studies reinforce the importance to assess the effect of chocolate consumption using a clinical trial design.
In our study, DBP increased significantly from baseline to 12 weeks in the high-flavanol and low-flavanol chocolate groups, but the difference between the 2 groups was not significant. It is probable that this elevation was related to the normal increases from the second to the third trimester of pregnancy.
Although no prospective studies in pregnant women are available, the effects of daily flavanol-rich chocolate intake on endothelial function and BP have already been well documented in other populations [24, 25]. A more recent meta-analysis by Hooper et al., that included 11 studies and 373 participants, suggested improvement of FMD 2 h after chocolate ingestion and its chronic intake . Stronger effects were apparent at higher doses of epicatechin that could be a key contributor to the outcomes observed. They also found a significant effect of chocolate on DBP and MAP but heterogeneity of the results of the 11 studies included in this meta-analysis was significant. This heterogeneity could be partially explained by BP differences in participants at the start of chocolate intake. Indeed, recent investigations have reported unchanged  or no significant change  in SBP and DBP after dark chocolate consumption in normotensive, healthy populations. Taubert et al., in their meta-analysis, observed trends towards a marked effect of chocolate intake in patients with high BP .
In our clinical trial, although acute ingestion of our experimental chocolate bars was accompanied by a very significant increase in plasma epicatechin concentrations, daily, chronic epicatechin intake by pregnant women could have been too low to improve BP. In fact, in the meta-analysis by Hooper et al , subgroup analysis by epicatechin dose suggested greater effects on BP at doses exceeding 50 mg, but not with doses less than 50 mg. In our study, pregnant women had to consume 20-g of chocolate daily, which provided less than 50 mg of epicatechin if all chocolate bars provided were eaten.
Other investigations also showed an increment in plasma epicatechin concentrations after acute consumption of flavanol-rich products [30–32]. The amount of experimental chocolate administered in our study had no effect on increasing plasma catechin concentrations. Similar observations have been made previously [30, 31]. Epicatechin is the predominant flavanol in dark chocolate. Moreover, among the flavanols, epicatechin has been reported to occur in a primary bioavailable form . Holt et al  suggested that complex mixtures of chocolate dimers and oligomers may be degraded in the gut into epicatechin monomers, which promotes absorption and would explain the low catechin concentrations in plasma.
At 12 weeks, epicatechin was below the limit of detection. Blood sample extraction after a 12-h fast could have been responsible for the low epicatechin concentrations, considering that epicatechin clearance from the plasma compartment is very fast  and most absorbed epicatechin is cleared from the blood by that time [30, 34]. Moreover, Taubert et al  excluded the storage of plasma phenols after dark chocolate intake because the short elimination half-lives of flavanols prevented accumulation of plasma levels and thus no flavanols could be detected after 12-h post-intake of flavonoid-rich dark chocolate. Nevertheless, after daily dietary chocolate incorporation for 12 weeks, even if epicatechin concentrations were very low, plasma theobromine increased 4-fold, indicating very good compliance, as discussed below. Plasma epicatechin levels should be quantified 2 h after the chocolate dose because of the short elimination half-life of flavanols.
The marginal but statistically-significant increase in theobromine levels in the experimental compared to placebo group at 180 minutes and 12 weeks post-randomisation could be attributed to different rates of intestinal metabolism, which is influenced by diet. Indeed, orally-administered, theobromine is rapidly and almost completely absorbed by the gastrointestinal tract . At baseline, theobromine concentrations were also slightly lower in the placebo group, but its sources in diet measured by FFQ in the last month preceding randomisation and each follow-up visit, were not significantly different between the 2 groups.
Some limitations  of the technique for FMD assessment could also explain our lack of significant effect. Measurement of endothelial function is challenging. Although a number of non-invasive techniques are now available, the most frequently-used method involving FMD of the brachial artery with ultrasound imaging was incorporated in our trial. This gold standard technique is not easy and requires extensive sonographer training as well as labour-intensive image analysis. Also, lack of standardization could influence methodological reproducibility in follow-up studies, particularly among pregnant women. Moreover, in this trial, we adopted M mode during image acquisition of the brachial artery. However, both M and A modes are applied to continuously measure brachial diameter, and these techniques may be subject to error due to tracking drift . In addition, M mode is more variable than B mode; thus, it is difficult to maintain the positional stability of measurements which must be performed at set time points (50 s), limiting reproducibility and showing lower FMD as a result . Furthermore, 40 to 60 patients are typically needed in a parallel-group study design to find significant improvement in FMD . Sample size in our study was likely too small to detect a significant difference between the 2 groups with this method. More novel fingertip-based methodologies, such as pulse amplitude tonometry, which has the potential advantage of an automated, computerized analysis system that minimizes operator dependency and inter-observer variability, would be more appropriate in prospective studies.
It is relevant to point out and discuss the consumption of low-flavanol chocolate bars as placebo to high-flavanol chocolate bars. In fact, our placebo chocolate probably did not have a neutral effect compared to experimental chocolate because it provided similar theobromine concentrations . Although the beneficial outcome of chocolate on FMD and BP has been largely associated with flavanols, a recent clinical trial indicated that theobromine could be partly responsible for the BP-lowering action of chocolate . Our results could have been masked by its effects, which would confirm the importance of carefully selecting an adequate placebo.
Interestingly, after checking the medical records of participants when the study ended, the pregnant women high-flavanol chocolate group had not incurred any kind of disease related to hypertension or PE, whereas a participant in the low-flavanol chocolate group had PE during late pregnancy (result not shown).
Important and significant theobromine concentrations, a good marker of chocolate intake, and the count of unused chocolate bars confirmed good compliance with the intervention during 12-week follow-up, which did not differ across arms. This finding suggests that 20-g chocolate bars could be incorporated in the usual diet of pregnant women for as long as 12 weeks. Based on the compliance difference reported by pregnant women (97.6% vs. 93.6%, experimental and placebo group, respectively) the slight difference at week 12 (p = 0.03) between the concentrations of theobromine in the two groups could reflect better compliance in the experimental group. Moreover, at baseline, the high-flavanol chocolate group had theobromine concentrations that tended to be higher compared with the low-flavanol chocolate group. Therefore, the experimental group could have had a higher consumption of dietary sources of theobromine from their regular diet at baseline and throughout the study, even though all women were instructed to avoid these foods.
Overall, 42 of 44 randomized pregnant women were retained in our study, and only 1 participant was not included in intention-to-treat analysis.
Weight increases linearly during the 2nd and 3rd trimesters of pregnancy, but is associated with escalating physiological weight in pregnancy. Mean gestational weight gain by our pregnant study participants was in accordance with new recommendations of the committee re-examining pregnancy weight guidelines . Caloric intake and nutritional dietary composition did not change significantly during the trial. Although chocolate has high energy density and is perhaps recognized as a fattening food, several studies [35, 39–41] have not discerned any weight increase after daily chocolate intake. In our pilot RCT, participants in the 2 arms were instructed not to add chocolate bars to their usual diet but to use them by replacing foods of similar energy and macronutrient composition. Moreover, to ensure support and motivation for adequate use of chocolate, women were met individually by a dietitian who offered global counseling. These strategies could have a favorable impact on weight gain during pregnancy. Therefore, we can speculate that weight gain was not attributable to chocolate intake.
The participants were in good health at baseline and probably represented mainly females with an interest in health and diet. Therefore we can speculate that it may have contributed to reduce the possibility of a significant chocolate intake effect on FMD and BP. Therefore, our results may not be applicable to high-risk pregnant women.