Reader Comments
Post a new comment on this article
Post Your Discussion Comment
Please follow our guidelines for comments and review our competing interests policy. Comments that do not conform to our guidelines will be promptly removed and the user account disabled. The following must be avoided:
- Remarks that could be interpreted as allegations of misconduct
- Unsupported assertions or statements
- Inflammatory or insulting language
Thank You!
Thank you for taking the time to flag this posting; we review flagged postings on a regular basis.
closeConclusion not fully supported by the data at hand; public health implications should be presented with greater caution.
Posted by PLOS_ONE_Group on 22 Sep 2010 at 11:11 GMT
[Comment posted on behalf of Jan Steijns, FrieslandCampina Research, Wageningen]
General remark
The authors have provided an overview of published studies on the blood lipid modifying effects of dietary trans fatty acids derived from various sources providing useful and informative perspective on this topic. The primary conclusion of the authors is that all fatty acids with a double bond in the trans configuration raise the ratio of plasma LDL to HDL cholesterol concentration. The authors go on to discuss the potential significance of their findings for public health. One of the perplexing issues from this article is the suggestion that both animal and industrial trans fatty acids have similar adverse effects on cardiovascular disease risk. We would like to suggest that this conclusion is perhaps not fully supported by the data at hand and that the public health implications should be presented with greater caution.
Indeed, there is a substantive body of evidence from both animal models and clinical studies to indicate that trans fatty acids derived from animal and industrial sources can have discrete and differential bioactivity. Additional consensus for this topic has also been communicated in a recent WHO review1 where one can read the following quote:
“The current growing body of evidence from controlled trials and observational studies indicates that TFA consumption from partially hydrogenated oils adversely affects multiple cardiovascular risk factors and contributes significantly to increased risk of CHD events. Although ruminant TFAs cannot be removed entirely from the diet, their intake is low in most populations and to date there is no conclusive evidence supporting an association with CHD risks in the amounts usually consumed. In contrast, TFA produced by partial hydrogenation of fats and oils should be considered industrial food additives having no demonstrable health benefits and clear risks to human health.”
Given the timing of the current article it may have been useful for Brouwer et al to incorporate some of the issues from the WHO Review, which was published in May 2009 (1)
Specific comments
1. The authors have chosen to use LDL / HDL ratio as an indicator of CVD risk for their analysis despite the growing consensus that other biomarkers (indices) may offer a stronger association with CVD risk. For example, TC/HDL ratio and/or non-fasting TG have both recently been considered a strong and perhaps more appropriate marker(s) to estimate CVD (1,2) As the authors highlight 5 papers in their Introduction (references 1-5) and a meta-analysis (reference 26) referring to the TC/HDL ratio, they could discuss the potential limitations of their marker in context to their findings.
2. The authors have chosen to recalculate the intake to isocaloric substitution of MUFA to TFA. In table 1 the doses (delta En% TFA) used in the various studies are listed relative to cis-UFA. When trying to reproduce the listed values, it is not clear how the doses of industrial TFA or ruminant TFA have been calculated as no details are provided on TFA and MUFA for baseline values. For example, consulting Chardigny et al (2006) (3), in which the protocol has been described including run-in values for individual fatty acids, we conclude that the values listed in Table 1 must have been derived by summing up trans C18:1 total, trans C18:2 and CLA. It is interesting to note that Chardigny et al (2008) (4) exclude CLA from their total trans calculation, as previous literature had indicated that this ruminant fatty acid may not have a negative effect on biomarkers of CVD risk.
3. The calculations for CLA have been based primarily on supplements that were given in addition to uncontrolled ad-lib dietary based studies. Furthermore assumptions on caloric intake for men and women were made, 2500 and 2000 kcal respectively per day, which may bias the outcome on blood lipid profile.
4. The topic of the source of CLA is also unclear. CLA supplements are generally vegetable oil based and due to isomerization processes, form two main isomers, roughly in equal quantity. One of the CLA isomers is rumenic acid, the main CLA isomer (cis9, trans11) in meat and dairy sources. However, in figure 1, only rumenic acid is shown, whereas it has been described that the trans10, cis12 CLA isomer and the cis9, trans11 isomer may have differential effects on body composition and plasma lipids in humans (5 ). In the first paragraph of the Method section the authors refer to industrial trans fatty acids, or conjugated linoleic acid, or other ruminant trans fatty acids. In the Public Health Implications section they also focus on TFA sources of animal origin without an appropriate discussion of effects of habitual intake from food sources. Collectively the data and conclusions appear inconsistent.
5. Another issue we wish to raise is the interpretation of data from figures 3B/3C. It may be possible that the correlation drawn by the authors is heavily reliant on a single (and potentially extreme) datapoint (based on an intake which is far above habitual intake). For example, removing this data point (references 13 and 27 respectively) reveals a very different and opposing relationship. Further, it is not clear why reference 62 is not included in the correlation analysis as it compares trans10, cis12 CLA and cis9, trans11 isomer enriched samples and shows differential outcomes on blood lipids. The authors mention that the sequential design of this study justifies exclusion, yet the first period of the sequence in this cross-over design with 6 weeks wash-out could have been analyzed separately.
6. It is also not clear why reference 63 was excluded (the authors state it showed an extreme
discordant effect). According to table 1, the measured effect was 0.75, which is not so inconsistent with findings from references 32 and 41 (effect of 0.64 and 0.67 respectively). The arrow in figure 3A with ref. 63 seems to indicate it is considered an outlier?
7. Based on the comments above we also like to refer to the results of reference 13 (TRANSFACT study), in which the main finding was that ‘compared with TFAs from industrially produced sources, TFAs from natural sources significantly (p<0.012) increased HDL cholesterol in women but not in men. Significant (p<0.001) increases in LDL-cholesterol concentrations were observed in women, but not in men, after the consumption of TFAs from natural sources. Apolipoprotein (apo) B and apoA1 concentrations confirmed the changes observed in LDL and HDL cholesterol. Analysis of lipoprotein subclass showed that only large HDL and LDL concentrations were modified by TFAs from natural sources but not by those from industrially produced sources’. The TRANSFACT authors conclude that “TFAs from industrially produced and natural sources have different effects on CVD risk factors. The HDL cholesterol–lowering property of TFAs seems to be specific to that from industrial sources. The observed responses, however, were greater in women than in men, and the mechanism underlying these biological effects warrants further investigation.”
8. The authors of the study of Motard-Belanger et al (reference 12), conducted with men only,
concluded that “a high intake of ruminant TFA (rTFA) may lead to deleterious changes in lipid CVD risk factors, similar to those that have been attributed to TFA from industrial sources. However, data also indicated that an intake of rTFA that may be practically attained by the consumption of large quantities of dairy products had no effect on CVD risk factors. On the basis of these observations and of data from previous studies, we propose that the current intake of rTFA by the population, which corresponds to approximately one-third of that achieved in the moderate rTFA diet in the present study, is unlikely to lead to deleterious changes in CVD risk.” A follow-up of that study with women only is in progress.
9. Indeed we recognize, that there is inherit challenge to address the discussion about ruminant and industrial TFA and the relationship with CVD risk. However, we would submit that in order to have valid data to draw conclusions for public health, it would perhaps be more appropriate to consider whether intake from foods at habitual consumption has adverse effects. It may well be that the linear model used by the Brouwer et al to address this question is not necessarily the most appropriate model.
Yours sincerely,
Jean-Michel Chardigny, PhD, Affiliation Head of JRU 1019 Human Nutrition, Clermont-Ferrand, France, Jean-Michel.Chardigny@clermont.inra.fr
Spencer Proctor, Associate Professor, Director Metabolic and Cardiovascular Diseases Laboratory, Alberta Institute for Human Nutrition, University of Alberta, Edmonton, Canada,
spencer.proctor@ualberta.ca
Adam Lock, Assistant Professor, Dept. Animal Science, Michigan State University, USA,
allock@msu.edu
Jan Steijns, PhD, Corporate Nutritionist at FrieslandCampina Research, Deventer, The Netherlands, jan.steijns@frieslandcampina.com
References:
(1) Uauy et al. EJCN (2009) 63, S68–S75; doi:10.1038/ejcn.2009.15. WHO Scientific Update on trans fatty acids: summary and conclusions.
(2) Mensink et al (2003) AJCN 77, 1146-1155. Effects of dietary fatty acids and carbohydrates on the ratio of serum total to HDL cholesterol and on serum lipids and apolipoproteins: a meta-analysis of 60 controlled trials.
(3) Chardigny et al (2006) Contemporary Clinical Trials 27, 364–373. Rationale and design of the TRANSFACT project phase I: A study to assess the effect of the two different dietary sources of trans fatty acids on cardiovascular risk factors in humans.
(4) Chardigny et al (2008) AJCN 87, 558-566. Do trans fatty acids from industrially produced sources and from natural sources have the same effect on cardiovascular disease risk factors in healthy subjects? Results of the trans Fatty Acids Collaboration (TRANSFACT) study.
(5) Terpstra AHM (2004), AJCN 79, 352-361. Effect of conjugated linoleic acid on body composition and plasma lipids in humans: an overview of the literature.
Dr Proctor has received research funding by Dairy Farmers of Canada and has accepted invitations to speak at the World Dairy Congress.
Dr Lock has received research funding from the United States Department of Agriculture National Research Initiative, Dairy Australia and Dairy Management Inc, and is a member of the Action Team on Fats and Oils of the International Dairy Federation Standing Committee on Nutrition and Health.
Dr Steijns is an employee of FrieslandCampina, a dairy company that develops and markets dairy products.
RE: Conclusion not fully supported by the data at hand; public health implications should be presented with greater caution.
brouweria replied to PLOS_ONE_Group on 21 Dec 2010 at 10:41 GMT
Reply to Chardigny et al.
We thank Chardigny and colleagues for their comments. We have recalculated various analyses according to their suggestions. These recalculations lead either to the same conclusions or to more unfavorable estimates of the effects of animal trans fatty acids on blood lipids and lipoproteins. This confirms that all fatty acids with a double bond in the trans configuration raise the ratio of total or LDL cholesterol to HDL cholesterol. It also provides another argument besides the high content of saturated fatty acids to lower the intake of ruminant animal fats.
We discuss the comments of Chardigny et al. one by one below.
1. “[total cholesterol]/HDL ratio and/or non-fasting [triglycerides] [are] perhaps more appropriate marker(s)”
The total cholesterol/HDL ratio and the LDL/HDL ratio are both good predictors of coronary heart disease risk, and they are numerically similar. The difference between them is due to VLDL cholesterol, a minor cholesterol fraction in blood.
We have not done a new quantitative review of the total cholesterol/HDL ratio as that would be a major undertaking. Instead, we have estimated the effect of the various types of trans fatty acids on this ratio for two extreme scenarios. In scenario 1, we assumed that ruminant trans fatty acids have the same effect on fasting triglycerides as industrial trans fatty acids, namely 0.019 mmol/L per percent of energy (1) In scenario 2, we assumed that ruminant trans fatty acids have no effect on fasting triglycerides.
We calculated VLDL cholesterol from the fasting triglycerides concentration using the Friedewald formula (2):
VLDL cholesterol (mmol/L) = 0.45 * triglycerides (mmol/L)
We calculated total cholesterol per study as:
Total cholesterol =HDL cholesterol +LDL cholesterol +VLDL cholesterol.
We assumed that the mean fasting triglyceride concentration was 1 mmol/L on the control diet in each study. If we used a value of 2 mmol/L instead, all total /HDL cholesterol ratios became 0.002 higher, but the differences between the various types of trans fatty acids remained identical.
Under scenario 1 the estimated lines for the total /HDL cholesterol ratio are [Change in ratio] = 0.065 * [change in intake] for industrial trans fatty acids, 0.045 for ruminant trans fatty acids and 0.053 for CLA. Under scenario 2, these values are 0.040 for ruminant trans fatty acids and 0.046 for CLA. Thus, the effects on the total/HDL cholesterol ratio are similar to those on the LDL/HDL ratio which were 0.055 for industrial trans fatty acids, 0.038 for ruminant trans fatty acids, and 0.043 for CLA under both scenarios. The real effect may lie somewhere in between these extremes.
Chardigny et al also suggest that non-fasting triglycerides may be a more appropriate risk marker. Neither the papers cited by them nor authoritative reviews [e.g. (4)] consider non-fasting triglycerides a more appropriate marker of cardiovascular disease risk than LDL and HDL cholesterol
2. “it is not clear how the doses of industrial TFA or ruminant TFA have been calculated”
We calculated doses of industrial and ruminant trans fatty acids by adding up all fatty acids with a double bond in the trans configuration. We included CLA because it is a trans fatty acid. We did not take into account opinions of individual authors as to whether a particular trans fatty acid affected cardiovascular disease risk.
3a. “The calculations for CLA have been based primarily on supplements that were given in addition to uncontrolled ad-lib dietary based studies”
These uncontrolled diets increase variability and thus the chance of a null effect, but random variability does not bias the direction of the outcome. We combined outcomes of a large number of studies, which reduced variability and allowed the signal to rise above the noise.
3b. “assumptions on caloric intake for men and women were made, 2500 and 2000 kcal respectively per day, which may bias the outcome on blood lipid profile”
Changing the estimated intake of calories has little effect on the outcomes. If we assume intake was 2750 kcal per day the estimated effect of CLA on LDL becomes 0.039 instead of 0.038 and that on the ratio of LDL to HDL 0.044 instead of 0.043. The effect on HDL is unchanged.
4a. “in figure 1, only rumenic acid is shown”
Figure 1 shows examples of the major classes of trans fatty acids. There are many variants within each class. We depicted rumenic acid (9 cis, 11 trans C18:2) and mentioned the closely related 10 trans, 12 cis C18:2 in the legend.
4b “the trans 10, cis 12 CLA isomer and the cis 9, trans 11 isomer may have differential effects”
That is possible. However, in our own study (5) ( reference 27 in (3)) we gave a CLA preparation that contained 80% of the natural cis 9, trans 11 isomer, and it produced a similarly unfavourable effect on LDL and HDL as industrial trans fatty acids. Claims that particular isomers have substantially different effects are not supported by strong evidence.
5a. “the correlation drawn by the authors is heavily reliant on a single (and potentially extreme) datapoint” [in figures 3B/3C]
When we leave out the extreme datapoint No. 13 from figure 3B, the effect of ruminant trans fatty acids on the LDL/HDL ratio increases from 0.038 to 0.046 (CI -0.008 to 0.100). When we leave out the extreme datapoint No. 27 from figure 3C, the effect of CLA increases from 0.043 to 0.064 (CI 0.010 to 0.118). Thus, leaving out these points increases the unfavorable effects of animal trans fatty acids.
5b. “it is not clear why reference 62 is not included in the correlation analysis”
Reference 62 did not meet our pre-specified inclusion criteria as it had a sequential design without a true control group. We showed in the Results section (3) that inclusion of this study produced only a small change in the slope of the regression line, i.e. from 0.043 to 0.041.
6. “It is also not clear why reference 63 was excluded”
There was an error in Table 1 which has now been corrected, the value for reference 63 should have been 1.06 not 0.75. The arrow in figure 3A indicates that this value lies outside the frame of the figure. This was a small study which showed an aberrant, extremely strong adverse effect of industrial trans fatty acids.
7a “Significant (p<0.001) increases in LDL-cholesterol concentrations were observed in women, but not in men” [in reference 13 of (3)]
In reference 13 women showed a larger response than men did, but in reference 27 they showed a smaller response. Both may be due to chance.
7b “The TRANSFACT authors conclude..”
Our quantitative review combined the relevant data in a transparent, objective way. We did not do a formal review of the opinions given in the papers that we analyzed.
8a “an intake of rTFA that may be practically attained by the consumption of large quantities of dairy products had no effect on CVD risk factors” [in reference 12]
Reference 12 showed that high levels of ruminant trans fatty acids had deleterious effects on blood cholesterol levels. The study did not find a significant effect of a lower dose of 1.5 energy% ruminant trans fatty acids. This was probably due to a lack of statistical power. Our equations (3) predict that 1.5 en% should raise LDL by 0.067 mmol/L. The 95% confidence interval of the effect seen in reference 12 includes 0 but also includes this predicted effect. Thus, the study was underpowered to detect the effect of such a dose.
8b. “we propose that the current intake of rTFA by the population […] is unlikely to lead to deleterious changes in CVD risk”
We agree that intake of ruminant trans fatty acids is modest. We estimated that elimination of ruminant trans fatty acids would lower the total trans fatty acid intake in the United States and Europe by about 0.5% of energy and would reduce coronary heart disease risk by 1.5 to 6%. However, ruminant trans fatty acids now provide more than half of the remaining trans fat in the diet in Europe, and the USA is moving in the same direction. Attempts to further reduce trans fat intake cannot ignore ruminant trans. Also, milk and beef fat not only contain trans fatty acids but they are also rich in saturated fatty acids. Our results provide an additional argument to lower the intake of such fats. Our data also argue against current attempts to increase the trans fatty acids content of milk fat.
9. the linear model used […] is not necessarily the most appropriate model”
The closest fit to the data is indeed not a straight line but a logarithmic model. This model actually predicts a more unfavourable effect of animal trans fatty acids on blood lipoproteins at low intakes than the linear model, which we used. However, as described in the Discussion Section of Brouwer et al (3), the logarithmic model does not make sense because it predicts that the LDL to HDL ratio equals -∞ (minus infinity) at zero intake of ruminant trans fatty acids .
Yours sincerely,
Ingeborg A Brouwer 1, Anne J Wanders 1,2, Martijn B Katan 1
1 Department of Health Sciences and the EMGO Institute for Health Care Research, Faculty of Earth and Life Sciences, VU University, Amsterdam, The Netherlands.
2 Division of Human Nutrition, Wageningen University, Wageningen, The Netherlands.
References
1. Mensink RP, Zock PL, Kester ADM, Katan MB. Effects of dietary fatty acids and carbohydrates on the ratio of serum total to HDL cholesterol and on serum lipids and apolipoproteins: a meta-analysis of 60 controlled trials. Am J Clin Nutr 2003;77:1146-1155.
2. Friedewald WT, Levy RI, Fredrickson DS. Estimation of the concentration of low-density lipoprotein in plasma, without use of the preparative ultracentrifuge. Clin Chem 1972;18:499.
3. Brouwer IA, Wanders A, Katan MB. Effect of animal and industrial trans fatty acids on HDL and LDL-cholesterol levels in humans – a quantitative review. Plos One 2010;5:e9434.
4. Greenland P, Alpert JS, Beller GA, Benjamin EJ, Budoff MJ, Fayad ZA, Foster E, Hlatky MA, Hodgson JM, Kushner FG, Lauer MS, Shaw LJ, Smith SC, Taylor AJ, Weintraub WS, Wenger NK. 2010 ACCF/AHA Guideline for Assessment of Cardiovascular Risk in Asymptomatic Adults: A Report of the American College of Cardiology Foundation/American Heart Association Task Force on Practice Guidelines. Circulation 2010 Nov;:CIR.0b013e3182051b4c.
5. Wanders AJ, Brouwer IA, Siebelink E, Katan MB. Effect of a high intake of conjugated linoleic acid on lipoprotein levels in healthy human subjects. PLoS One 2010;5:e9000.
Dr. Katan has acted as Supervisor for PhD students whose research was funded by the Wageningen Centre for Food Sciences and its successor, TI Food and Nutrition. This partnership receives funding from the Netherlands Ministry of Economic Affairs, five research organizations (University of Groningen, Wageningen University and Research Centre, Maastricht University, NIZO Food Research, and TNO Quality of Life) and six Dutch food industries. Between 2004 and 2007, Dr. Katan attended scientific symposia organized by Nestlé Research, and expenses related to attendance were paid by Nestlé. Dr. Brouwer was employed by Wageningen University and posted to Wageningen Centre for Food Sciences (see above) for 100% of her time from 1999 to 2005. She then moved to VU University but continued with Wageningen Centre for Food Sciences and its successor TI Food and Nutrition for 40% of her time in 2006 and for 10% in 2007 and the first half of 2008. Her research involved B vitamins and omega-3 fatty acids, and was unconnected with the topic of our present review. In 2004 and 2005, Dr. Brouwer assisted in a study on blood pressure that was supported by a grant from Unilever.