Thank you to the authors for addressing a relevant and interesting area of research (Snorgaard et al., 2017).
The review was well planned, but the methodology lacks detail that enables the reader to understand the processes involved in the completion of the meta-analysis and some study limitations were not described.
Please could the authors clarify why the meta-analyses use both mean change from baseline and mean final value in the same meta-analysis (see Figure 2 and Figure 3 where the lower means indicate change from baseline and the higher means indicate unadjusted final values)? Also, could it be clarified why the selected arms from the three-arm trials were chosen over the arms that were omitted? Although the population, intervention and outcomes were defined in the methodology, the comparator was not.
Furthermore, when using the mean final HbA1c value in the meta-analyses, papers such as Krebs et al. (2012) and Guldbrand et al. (2012) have higher baseline HbA1cs in the lower-carbohydrate group, which was not mentioned in the paper, nor mentioned as a limitation to the meta-analysis. Guldbrand et al. (2012) demonstrate that HbA1c remained the same at two years (the time point the authors refer to) in the lower-carbohydrate arm but increased in the comparator arm by 0.2%. Therefore the low-carbohydrate arm was the superior intervention; however, the forest plot (Figure 3) suggests that the control intervention was slightly but not significant...
Thank you to the authors for addressing a relevant and interesting area of research (Snorgaard et al., 2017).
The review was well planned, but the methodology lacks detail that enables the reader to understand the processes involved in the completion of the meta-analysis and some study limitations were not described.
Please could the authors clarify why the meta-analyses use both mean change from baseline and mean final value in the same meta-analysis (see Figure 2 and Figure 3 where the lower means indicate change from baseline and the higher means indicate unadjusted final values)? Also, could it be clarified why the selected arms from the three-arm trials were chosen over the arms that were omitted? Although the population, intervention and outcomes were defined in the methodology, the comparator was not.
Furthermore, when using the mean final HbA1c value in the meta-analyses, papers such as Krebs et al. (2012) and Guldbrand et al. (2012) have higher baseline HbA1cs in the lower-carbohydrate group, which was not mentioned in the paper, nor mentioned as a limitation to the meta-analysis. Guldbrand et al. (2012) demonstrate that HbA1c remained the same at two years (the time point the authors refer to) in the lower-carbohydrate arm but increased in the comparator arm by 0.2%. Therefore the low-carbohydrate arm was the superior intervention; however, the forest plot (Figure 3) suggests that the control intervention was slightly but not significantly superior. The paper by Krebs et al. (2012) concluded that both groups had an increase in HbA1c of 0.1% at trial end suggesting no superiority of either intervention; however, again the forest plot (Figure 3) indicates that the effect size was slightly but not significantly in favour of the control group. It appears that the inclusion of these data in their current form may limit the reliability of the meta-analysis seen in Figure 3. Could the authors or editor please comment on this.
We thank colleagues for their critical comments that help to clarify relationships we have studied. We are not concerned with the frequency of what specific genes, high-risk or not, has increased recently. We are just making a general statement that with relaxed natural selection detrimental mutations may accumulate. The paper by Witas et al. cited by our critics uses the same rationale as we do when suggesting changes in type 1 d...
We thank colleagues for their critical comments that help to clarify relationships we have studied. We are not concerned with the frequency of what specific genes, high-risk or not, has increased recently. We are just making a general statement that with relaxed natural selection detrimental mutations may accumulate. The paper by Witas et al. cited by our critics uses the same rationale as we do when suggesting changes in type 1 diabetes (T1D) susceptibility gene frequencies under the operation of natural selection over the last 700 years [1]. For most of that time, excluding the last 100 years or so, natural selection was strong so it eliminated susceptibility genes. Selection only became relaxed in the recent decades. Besides the Human Leukocyte Antigen (HLA) gene group studied in the papers quoted by our critics, we know now that there are some 40-50 alleles located on various chromosomes that may increase susceptibility to type 1 diabetes [2, 3] . Our statements pertain to the genetic susceptibility to T1D that may be caused by all the T1D related genes, not just specific alleles. We argue that T1D related genes may have been accumulated in the whole population rather than in some particular samples. Having studied papers quoted by our critics to support their letter we found a fatal error in their logic. The studies [4-8] cited in the letter used only rather small samples of T1D patients of a limited age range and considered subsets of disease related genes, not all of them. Within those samples of patients, frequencies of alleles considered to confer greatest susceptibility decreased. Meanwhile frequencies of other alleles conferring some T1D susceptibility increased. Without consideration of the frequency of greatest susceptibility alleles, and others, in the total gene pool of the entire population from which limited samples were selected, it is impossible to tell whether the total frequency of susceptibility genes in a population has increased or decreased. The only conclusion that can be drawn from the studies cited by our critics is that frequencies of a limited number of alleles providing different levels of susceptibility to T1D have changed through time among people already suffering from the disease. This tells us nothing about total frequencies of these alleles in a population.
A particular study [6] may have shown that a large proportion of T1D patients have no family history of the disease, but this runs counter to the well-known fact that people of different ethnic ancestry (and thus derived from different gene pools) have vastly different prevalence rates of T1D and to the strong correlation of the disease appearance among monozygotic twins. The genetic component of T1D susceptibility is substantial and, as we have already said, provided by scores of genes from various chromosomes.
Our critics are also wrong in pointing that our conclusions can be based on the fact that prevalence of T1D increases with increasing GDP. In our statistical analyses we have excluded possible influence of GDP by using partial correlation analysis that stabilizes effects of controlled variables statistically, as if they were of the same magnitude in all analysed countries. The correlation between T1D and Ibs we have observed is independent of the correlation between GDP and T1D prevalence.
We are not denying that environmental factors may play some role in increasing prevalence of T1D, we are just pointing, in an appropriately structured statistical analysis of worldwide data, that relaxation of natural selection has an effect on T1D prevalence. Study of relative magnitudes of genetic and environmental influences on T1D increase requires careful quantitative analyses not just rhetorical statements "play much larger role" based on small-scale selective studies.
1. Witas, H., K. J??drychowska???Da??ska, and P. Zawicki, Changes in frequency of IDDM???associated HLA DQB, CTLA4 and INS alleles. International journal of immunogenetics, 2010. 37(3): p. 155-158.
2. Barrett, J.C., et al., Genome-wide association study and meta-analysis find that over 40 loci affect risk of type 1 diabetes. Nature genetics, 2009. 41(6): p. 703-707.
3. Bradfield, J.P., et al., A genome-wide meta-analysis of six type 1 diabetes cohorts identifies multiple associated loci. PLoS Genet, 2011. 7(9): p. e1002293.
4. Spoletini, M., et al., Temporal Trends of HLA, CTLA-4 and PTPN22 Genotype Frequencies among Type 1 Diabetes in Continental Italy. PloS one, 2013. 8(4): p. e61331.
5. Fourlanos, S., et al., The rising incidence of type 1 diabetes is accounted for by cases with lower-risk human leukocyte antigen genotypes. Diabetes care, 2008. 31(8): p. 1546-1549.
6. Parkkola, A., et al., Extended family history of type 1 diabetes and phenotype and genotype of newly diagnosed children. Diabetes care, 2013. 36(2): p. 348-354.
7. Patterson, C., et al., Is childhood-onset type I diabetes a wealth-related disease? An ecological analysis of European incidence rates. Diabetologia, 2001. 44(3): p. B9-B16.
8. Vehik, K., et al., Trends in high-risk HLA susceptibility genes among Colorado youth with type 1 diabetes. Diabetes Care, 2008. 31(7): p. 1392-1396.
In the article, "Type 1 diabetes prevalence increasing globally and regionally: The role of natural selection and life expectancy at birth" (You and Henneberg 2016), the authors find a correlation between worldwide type 1 diabetes prevalence and both life expectancy at birth and the "Biological State Index" (Ibs), a measure of population reproductive success. Based on these findings, they argue that "the correlation of Ibs to the...
In the article, "Type 1 diabetes prevalence increasing globally and regionally: The role of natural selection and life expectancy at birth" (You and Henneberg 2016), the authors find a correlation between worldwide type 1 diabetes prevalence and both life expectancy at birth and the "Biological State Index" (Ibs), a measure of population reproductive success. Based on these findings, they argue that "the correlation of Ibs to the type 1 diabetes prevalence rate has been observed, compatible with suggestion (sic) that lower opportunity for selection allows accumulation of unfavorable genes," and that, "reduced natural selection (Ibs) may be the major contributor to the increasing prevalence of type 1 diabetes worldwide with special regard to European countries." We do not agree with the conclusion that natural selection is playing a role in these correlations, particularly in light of an existing body of evidence that has already examined the question of whether or not the high-risk genes associated with type 1 diabetes have increased over the past decades--evidence that the authors failed to cite.
The existing body of evidence shows that the high-risk genes associated with type 1 diabetes have actually decreased over time, while more people with low-risk genes are now developing the disease. For example, Hermann et al. (2003) found that the high risk genotypes were lower, and protective genotypes were higher, in Finnish type 1 diabetes patients diagnosed more recently (between 1990 and 2001) as compared to those diagnosed between 1939 and 1965. Resic-Lindehammer et al. (2008) found that the high-risk type 1 diabetes genotypes were decreasing over time, and the lower-risk genotypes were increasing in Swedes diagnosed in three time periods between 1986 and 2005. Gillespie et al. (2004) compared the frequency of high- risk genes in UK patients diagnosed with type 1 diabetes more than 50 years ago compared to those more recently diagnosed (1985-2002). They found that, "The proportion of high- risk susceptibility genotypes was increased in the earlier cohort, especially in those diagnosed at age 5 years or younger, which is consistent with the hypothesis that the rise of type 1 diabetes is due to a major environmental effect." The strongest correlations found by You and Henneberg (2016) were in Europe, where these studies were conducted.
Outside of Europe, the patterns are the same. Steck et al. (2011) found that in two large cohorts representing multiple populations, the high-risk genotypes associated with type 1 diabetes decreased between 1965 and 2004-6. Fourlanos et al. (2008) analyzed the genotypes of Australians diagnosed with type 1 diabetes between 1950 and 2005 and found that, "The rising incidence and decreasing age at diagnosis of type 1 diabetes is accounted for by the impact of environment on children with lower-risk ... genes, who previously would not have developed type 1 diabetes in childhood." Vehik et al. (2008) genotyped Colorado youth with type 1 diabetes (diagnosed from 1978-1988, as compared to 2002-4), and found that "high-risk... genotypes are becoming less frequent over time in youth with type 1 diabetes of non-Hispanic white and Hispanic origin. This temporal trend may suggest that increasing environmental exposure is now able to trigger type 1 diabetes in subjects who are less genetically susceptible."
Witas et al. (2010) compared the DNA from bones from the 11th-14th centuries to those of today. They found that, "Contrary to the initial assumptions, genetic predisposition towards type 1 diabetes... is much lower contemporarily than it was approximately 700 years before present."
Moreover, the vast majority of people who develop type 1 diabetes have no family history of the disease. Even when taking into account both first and second degree relatives, the number of people diagnosed with type 1 diabetes without a family history approaches 80% (Parkkola et al. 2013).
Aside from natural selection, there are other reasons that could explain why reproductive success and life expectancy are associated with worldwide type 1 diabetes prevalence, such as the availability of insulin (which the authors do acknowledge) and other medications, technologies, and health care opportunities that are more available in countries with higher life expectancies and reproductive success. Numerous additional environmental factors that are positively associated with GDP, which the authors found was even more strongly associated with type 1 diabetes prevalence than the Biological State Index, could also explain these patterns (Patterson et al. 2001).
The authors' conclusion that "reduced natural selection (Ibs) may be the major contributor to the increasing prevalence of type 1 diabetes worldwide" is not supported by the existing scientific evidence. The studies that have examined this hypothesis via genetic analyses show the genetic risk of type 1 diabetes actually declining over time, and that people with lower genetic risk profiles are now developing type 1 diabetes more often than they did in the past. These patterns suggest that environmental factors play a much larger role than natural selection in the increasing rates of type 1 diabetes.
References Cited
Fourlanos S, Varney MD, Tait BD, Morahan G, Honeyman MC, Colman PG et al. The rising incidence of type 1 diabetes is accounted for by cases with lower-risk human leukocyte antigen genotypes. Diabetes Care 2008; 31(8):1546-1549.
Gillespie KM, Bain SC, Barnett AH, Bingley PJ, Christie MR, Gill GV et al. The rising incidence of childhood type 1 diabetes and reduced contribution of high-risk HLA haplotypes. Lancet 2004; 364(9446):1699-1700.
Hermann R, Knip M, Veijola R, Simell O, Laine AP, Akerblom HK et al. Temporal changes in the frequencies of HLA genotypes in patients with Type 1 diabetes-- indication of an increased environmental pressure? Diabetologia 2003; 46(3):420-425.
Parkkola A, Harkonen T, Ryhanen SJ, Ilonen J, Knip M. Extended family history of type 1 diabetes and phenotype and genotype of newly diagnosed children. Diabetes Care 2013; 36(2):348-354.
Patterson CC, Dahlquist G, Soltesz G, Green A. Is childhood-onset type I diabetes a wealth-related disease? An ecological analysis of European incidence rates. Diabetologia 2001; 44 Suppl 3:B9-16.
Resic-Lindehammer S, Larsson K, Ortqvist E, Carlsson A, Cederwall E, Cilio CM et al. Temporal trends of HLA genotype frequencies of type 1 diabetes patients in Sweden from 1986 to 2005 suggest altered risk. Acta Diabetol 2008; 45(4):231-235.
Steck AK, Armstrong TK, Babu SR, Eisenbarth GS. Stepwise or linear decrease in penetrance of type 1 diabetes with lower-risk HLA genotypes over the past 40 years. Diabetes 2011; 60(3):1045-1049.
Vehik K, Hamman RF, Lezotte D, Norris JM, Klingensmith GJ, Rewers M et al. Trends in high-risk HLA susceptibility genes among Colorado youth with type 1 diabetes. Diabetes Care 2008; 31(7):1392-1396.
Witas HW, Jedrychowska-Danska K, Zawicki P. Changes in frequency of IDDM-associated HLA DQB, CTLA4 and INS alleles. Int J Immunogenet 2010; 37(3):155-158.
You WP, Henneberg M. Type 1 diabetes prevalence increasing globally and regionally: the role of natural selection and life expectancy at birth. BMJ Open Diabetes Res Care 2016; 4(1):e000161.
We will start by emphasizing that we cannot control
what has been written in the media about this study. We
will therefore limit our response to the reviewer's
comments that directly pertain to our manuscript.
The results of this study are clearly enumerated in the
paper, with the supporting data shown in every
instance[1]. We distinguish explicitly be...
We will start by emphasizing that we cannot control
what has been written in the media about this study. We
will therefore limit our response to the reviewer's
comments that directly pertain to our manuscript.
The results of this study are clearly enumerated in the
paper, with the supporting data shown in every
instance[1]. We distinguish explicitly between between-
group and within-group effects, noting that the
between-group effects are primary. In studies that
cannot be blinded, however, there are well-known
Hawthorne effects, which can attenuate statistical
power in unpredictable ways. Thus, isolated within-
group effects, though secondary, are nonetheless
relevant and of interest, and may speak to issues that
warrant reinvestigation in trials with greater
statistical power.
While we considered Hawthorne effects to the greatest
extent possible in our study design, it is very
difficult to anticipate their magnitude in any given
context. We were thus left to conjecture about such
effects in the interpretation of our data.
Interpreting data from different perspectives, rather
than relying on a single approach, helps to overcome
Hawthorne and other potential effects that may obscure
meaningful associations. This, of course, is uniquely
important in lifestyle interventions that cannot be
fully blinded. Presenting only between-group effects
would provide a stunted view, and perhaps a
misrepresentation of our findings. This is
particularly true given the relatively small sample
size. We do not mistake our secondary outcomes for
primary, and nor should our readers. We do, however,
think that these findings warrant readers' attention.
We clearly note the basis for our data interpretation
and the relevant limitations in our manuscript.
References
1 Njike VY, Ayettey R, Petraro P, Treu JA, Katz
DL. Walnut ingestion in adults at risk for diabetes:
effects on body composition, diet quality, and cardiac
risk measures. BMJ Open Diabetes Res Care. 2015 Oct
19;3(1):e000115. doi: 10.1136/bmjdrc-2015-000115.
eCollection 2015.
This study purports to show that eating walnuts lowers cholesterol level, a claim enthusiastically and uncritically repeated in the media (e.g."A handful of walnuts a day may help lower cholesterol:study", CTV News, November 25, 2015). Unfortunately, the claim is spurious, as the published data show no such effect.
The study is well-controlled in including a randomized control group...
This study purports to show that eating walnuts lowers cholesterol level, a claim enthusiastically and uncritically repeated in the media (e.g."A handful of walnuts a day may help lower cholesterol:study", CTV News, November 25, 2015). Unfortunately, the claim is spurious, as the published data show no such effect.
The study is well-controlled in including a randomized control group of subjects not eating walnuts. However, the results fail to show a significant difference in cholesterol level between this control group and the group which did eat the nuts. Instead of focusing on this central comparison of the study, the authors chose instead to emphasize a significant within-group decline in cholesterol level over the experiment in the group eating nuts. At the same time they ignored the fact that cholesterol also significantly declined in the group not eating nuts.
In their own words, they stated "When compared with the walnut-excluded phase, the walnut-included diet showed no significant improvement [in cholesterol]. Yet in their highlighted "Key message", they claim the opposite: "The inclusion of walnuts in habitual diet...improved cholesterol". A similar claim appears in their press release.
Given the decline in both groups, logic requires that if they are going to claim that eating walnuts lowers cholesterol, they must also claim that not eating walnuts lowers cholesterol. To avoid this nonsensical pair of conclusions, the authors should have refrained from a
selective, misleading, and contradictory interpretation of their data and told us instead only that they failed to show that eating walnuts lowers cholesterol.
Thank you to the authors for addressing a relevant and interesting area of research (Snorgaard et al., 2017).
The review was well planned, but the methodology lacks detail that enables the reader to understand the processes involved in the completion of the meta-analysis and some study limitations were not described.
Please could the authors clarify why the meta-analyses use both mean change from baseline and mean final value in the same meta-analysis (see Figure 2 and Figure 3 where the lower means indicate change from baseline and the higher means indicate unadjusted final values)? Also, could it be clarified why the selected arms from the three-arm trials were chosen over the arms that were omitted? Although the population, intervention and outcomes were defined in the methodology, the comparator was not.
Furthermore, when using the mean final HbA1c value in the meta-analyses, papers such as Krebs et al. (2012) and Guldbrand et al. (2012) have higher baseline HbA1cs in the lower-carbohydrate group, which was not mentioned in the paper, nor mentioned as a limitation to the meta-analysis. Guldbrand et al. (2012) demonstrate that HbA1c remained the same at two years (the time point the authors refer to) in the lower-carbohydrate arm but increased in the comparator arm by 0.2%. Therefore the low-carbohydrate arm was the superior intervention; however, the forest plot (Figure 3) suggests that the control intervention was slightly but not significant...
Show MoreWe appreciate the attention to our manuscript[1].
We will start by emphasizing that we cannot control what has been written in the media about this study. We will therefore limit our response to the reviewer's comments that directly pertain to our manuscript.
The results of this study are clearly enumerated in the paper, with the supporting data shown in every instance[1]. We distinguish explicitly be...
Nutty conclusion in walnut study
This study purports to show that eating walnuts lowers cholesterol level, a claim enthusiastically and uncritically repeated in the media (e.g."A handful of walnuts a day may help lower cholesterol:study", CTV News, November 25, 2015). Unfortunately, the claim is spurious, as the published data show no such effect.
The study is well-controlled in including a randomized control group...
Pages