Addendum: hypercaloric diets with high meal frequency, but not increased meal size, increase intrahepatic triglycerides: A randomized controlled trial
Potential conflict of interest: Nothing to report.
Financial support: A.W.B. was supported in part by NIH grant P30DK056336. The content is solely the responsibility of the authors and does not necessarily represent the official views of the NIH or any other organization.
One of us (M.J.S., on behalf of the authors) published an article which compared a “high-sugar” (HS) versus a “high-fat, high-sugar” (HFHS) diet and “meal frequency” (F) versus “meal size” (S) in a two-factor, randomized, controlled design.1 Another of us (A.W.B. and colleagues, hereafter “the readers”) read the article with interest but raised concerns about the analysis. This letter identifies clarifications that need to be made regarding those concerns.
First, there are differences in baseline values in tables 1 and 3 in the article. These describe the same participants, and baseline values should have been identical. This is explained by the fact that table 1 reflected data from participants with complete (meaning all study measurements) data sets only (corrected in Supporting Table S1).
| HFHS-S | HFHS-F | HS-S | HS-F | P | P | P | P | |||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Delta | SD of Delta | Delta | SD of Delta | Delta | SD of Delta | Delta | SD of Delta | Overall | S vs. F | HFHS vs. HS | Interaction | |
| BMI (kg/m2) | 0.6 | 0.6 | 0.8 | 0.5 | 0.8 | 0.3 | 0.5 | 0.6 | 0.502 | 0.655 | 0.705 | 0.165 |
| REE (kcal/day) | 47 | 152 | 37 | 88 | 185 | 99 | −21 | 198 | 0.063 | 0.048 | 0.449 | 0.071 |
| Glucose (mmol/L) | 0.0 | 0.5 | 0.0 | 0.2 | −0.1 | 0.3 | 0.1 | 0.3 | 0.790 | 0.453 | 0.902 | 0.540 |
| Insulin (pmol/L) | 7 | 21 | 5 | 12 | 12 | 11 | −1 | 16 | 0.453 | 0.207 | 0.930 | 0.304 |
| Cortisol (nmol/L) | −66 | 235 | 35 | 105 | 10 | 83 | −74 | 167 | 0.454 | 0.881 | 0.783 | 0.122 |
| Glucagon (ng/L) | −10 | 23 | 4 | 19 | 9 | 13 | −8 | 25 | 0.284 | 0.849 | 0.645 | 0.059 |
| Leptin (ng/mL) | 0.8 | 1.0 | 1.7 | 1.7 | 0.9 | 0.8 | 1.4 | 2.0 | 0.596 | 0.196 | 0.834 | 0.751 |
| Triglycerides (mmol/L) | 0.1 | 0.5 | 0.3 | 0.2 | 0.2 | 0.3 | 0.2 | 0.4 | 0.793 | 0.488 | 0.948 | 0.485 |
| FFA (mmol/L) | 0.0 | 0.1 | −0.1 | 0.2 | −0.2 | 0.2 | −0.2 | 0.3 | 0.280 | 0.278 | 0.148 | 0.485 |
| EGP (μmol/kg/min) | −0.1 | 1.4 | 0.2 | 1.2 | 0.0 | 1.1 | 0.4 | 1.0 | 0.886 | 0.461 | 0.789 | 0.955 |
| EGP suppression (%) | 6.2 | 10.4 | −9.4b | 13.2 | −4.9 | 10.1 | 2.3 | 9.7 | 0.046 | 0.305 | 0.939 | 0.009 |
| Rd (μmol/kg/min) | −0.4 | 10.4 | −0.2 | 7.8 | −2.7 | 8.6 | −2.3 | 5.3 | 0.906 | 0.930 | 0.467 | 0.976 |
| FFA suppression (%) | 0.0 | 9.5 | −4.6 | 4.5 | −5.4 | 9.0 | −1.2 | 5.7 | 0.431 | 0.945 | 0.706 | 0.110 |
| IAAT (L) | 0.00 | 0.06 | 0.06 | 0.04 | 0.05 | 0.09 | 0.05 | 0.05 | 0.278 | 0.235 | 0.494 | 0.194 |
| SQAT (L) | −0.02 | 0.05 | 0.04c | 0.03 | 0.03 | 0.04 | 0.03 | 0.02 | 0.027 | 0.066 | 0.257 | 0.035 |
| VAT (L) | 0.02 | 0.03 | 0.02 | 0.03 | 0.02 | 0.06 | 0.02 | 0.03 | 1.000 | 0.985 | 0.912 | 0.985 |
| IHTG content (%) | 0.2 | 0.4 | 0.4 | 0.5 | 0.1 | 0.9 | 1.6a | 1.7 | 0.028 | 0.028 | 0.132 | 0.087 |
- a P = 0.06 versus HS-S with Bonferroni correction for multiple testing.
- b P = 0.07 versus HFHS-S with Bonferroni correction for multiple testing.
- c P < 0.05 versus HFHS-S with Bonferroni correction for multiple testing.
- Abbreviations: BMI, body mass index; FFA, free fatty acid; IAAT, intra-abdominal adipose tissue; Rd, rate of disappearance; REE, resting energy expenditure; SD, standard deviation; VAT, visceral adipose tissue.
Second, we would like to clarify the between-group outcome comparisons. In tables 3 and 4, outcomes were expressed as changes from baseline within treatments. The readers were concerned that some results were potentially consistent with the “differences in nominal significance” error. This can occur when within-group changes are compared instead of between-group differences, resulting in type I error rates up to 87.5% for four groups.2 Conclusions such as “hypercaloric diets with increased meal frequency … increase IHTG and abdominal fat in lean men whereas similar diets with increased meal size do not” describe within-group (rather than between-group) comparisons. The authors reanalyzed the data by two-way analysis of variance of change scores (“post” minus “pre” values; Table 1). The readers did not have access to the raw data, but they agree that the results presented below add detail to the originally reported results and show a significant overall between-group effect for suppression of endogenous glucose production (EGP suppression), subcutaneous adipose tissue (SQAT), and intrahepatic triglyceride (IHTG) content. For EGP suppression and SQAT, the interaction effect with post hoc between-group differences between HFHS-F and HFHS-S for EGP suppression shows a trend (Bonferroni corrected P = 0.07) and is significant for SQAT (Bonferroni corrected P < 0.05). Finally, for IHTG there is a significant main effect of S versus F (P = 0.028), as reported.1
These analyses confirm the original conclusion that a hypercaloric diet with high meal frequency compared to high meal size increased IHTG. However, the conclusions that “high meal frequency increased … abdominal fat” and “snacking … contributes to obesity” should have been phrased more cautiously as the interaction effect suggests that the effect on SQAT is dependent both on the HFHS/HS factor and the F/S factor and there are no significant between-group differences for other obesity-related variables (body mass index, intra-abdominal or visceral adipose tissue).
-
Mireille J. Serlie, M.D., Ph.D.1
-
Kasper W. ter Horst, M.D.1
-
Andrew W. Brown, Ph.D.2
-
1Department of Endocrinology & Metabolism Academic Medical Centre Amsterdam,
-
Amsterdam, The Netherlands
-
2Office of Energetics and Nutrition Obesity Research Center
-
University of Alabama at Birmingham
-
Birmingham, AL
