Introduction
The thrifty phenotype hypothesis proposes that impaired foetal growth and low birthweight reflects adaptations to insufficient foetal nutrition, prioritising development of essential organs over muscle, liver, adipose tissue, and pancreatic beta cells. Reference Vaag, Grunnet, Arora and Brøns1–Reference Ross and Beall3 While these adaptations may increase short-term survival in scarce environments, they may predispose individuals with low birthweight to type 2 diabetes and associated comorbidities in conditions of affluence and obesity. Reference Vaag, Grunnet, Arora and Brøns1–Reference Ross and Beall3
The major risk factors for type 2 diabetes can be categorised into three groups: genetics, the intrauterine environment, and the postnatal environmental. Reference Hansen, Thomsen and Brøns4–Reference Zheng, Ley and Hu6 Proxies for these include polygenic risk scores (PRS) or family history of type 2 diabetes, birthweight, and BMI in adulthood, respectively. Reference Khera, Chaffin and Aragam7 Each factor appears to additively account for a majority of the lifetime risk of type 2 diabetes. Reference Wibaek, Andersen and Linneberg5 Among individuals with recent onset type 2 diabetes, low birthweight has been associated with a younger age at onset with less obesity, Reference Hansen, Thomsen and Brøns4 supporting the notion that low birthweight individuals are more sensitive towards the deleterious effects of obesity. Reference Wibaek, Andersen and Linneberg5
This study aimed to examine how the associations between birthweight and key clinical characteristics of diabetes, including cardiovascular comorbidities, differ between individuals with and without type 2 diabetes. By comparing these associations across both groups, we aim to provide examples of how differences between populations with and without type 2 diabetes may reflect important inferences relevant to the complicated interplay between early and late life exposures, as well as genetic predisposition, in the development and clinical course of type 2 diabetes.
Methods
Study participants
The UK Biobank is a prospective, population-based cohort of approximately 500,000 participants aged 40–69 years, recruited from 2006–2010 across the United Kingdom. Reference Elliott and Peakman8 At their initial visit to an assessment centre, baseline data were collected through interviews, questionnaires, anthropometric measurements, and blood and urine samples. Medication usage was self-reported, and comorbidities were identified using ICD-10 codes from the National Electronic Healthcare records.
The study cohort included participants with self-reported birthweight data excluding values over 6.5 kg as likely incorrect. Birthweight were categorised into three groups based on the first and third quartiles. Diabetes status was determined through physician diagnosis or ICD-10 codes (Supplementary Table 1). To limit misclassification of type 2 diabetes with type 1 diabetes, participants with self-reported insulin use within one year of diagnosis, those diagnosed <35 years of age, Reference Uglebjerg, Ahmadizar and Aly9 and those with a history of gestational diabetes were excluded. Additionally, participants who were adopted or part of multiple birth were excluded. Continuous variables were assessed using histograms and scatter plots, with values >5 standard deviations from the mean being excluded.
Covariates
Smoking, alcohol consumption, and physical activity were self-reported; participants who declined to answer were excluded (see Supplementary Table 1 for variable details). Anthropometric measurements and blood pressure were obtained at baseline. Hypertension was determined by a physician’s diagnosis prior to the baseline visit. Myocardial infarction (MI) and stroke were identified via physician diagnosis or ICD-10 codes, while deep venous thrombosis (DVT) was based solely on physician diagnosis. Biomarkers selected by the UK Biobank Design Phase Expert Group Reference Bycroft, Freeman and Petkova10 and used in this study included high-sensitivity C-reactive protein (CRP), glycated haemoglobin A1c (HbA1c), lipids (total cholesterol, LDL cholesterol, HDL cholesterol, triglycerides), and plasma creatinine. Genetic predisposition was assessed using both self-reported family history of diabetes and a polygenic risk score (PRS) for type 2 diabetes. The PRS, representing the sum of an individual’s risk alleles weighted by effect sizes from existing genome-wide association studies, was constructed using a Bayesian approach. Reference Thompson, Wells and Selzam11 The PRS was normalised to have a mean of 0 and a standard deviation of 1. Reference Thompson, Wells and Selzam11
Outcome measures
The study outcomes included age at type 2 diabetes diagnosis, anthropometric factors (BMI, waist and hip circumference, waist-hip ratio, waist-height ratio), systolic and diastolic blood pressure (SBP and DBP), hypertension, MI, stroke, DVT, and selected biomarkers at baseline.
Statistics
Baseline characteristics are presented as medians with interquartile ranges for continuous variables and as frequencies with percentages for categorical variables, according to birthweight categories and diabetes status.
Associations between birthweight (kg) and continuous outcomes – including age of diagnosis, anthropometric values, SBP, DBP, and biomarkers – were analysed using linear regression. Restricted cubic spline regression models were used to explore potential non-linear patterns. Model fit was evaluated by likelihood ratio tests (p-value <0.05) and the Akaike Information Criterion (AIC). Using family history as the predictor, we also estimated the impact of reporting a family history of diabetes on birthweight. Logistic regression models were used to derive odds ratios (OR) and 95% confidence intervals (CI) for the associated risk between categorical variables and birthweight groups. Linear and logistic models were adjusted for key confounders sex, age at enrolment, and family history of diabetes. Since birthweight is an exposure defined at birth, no additional adjustments were made in the main model to prevent overadjustment. Later-life socio-behavioural, metabolic, and lifestyle factors were considered as potential intermediates in the pathway between birthweight and type 2 diabetes. Additional exploratory analyses involved adjustments for behavioural lifestyle factors (physical activity, smoking status, and alcohol consumption), Index of Multiple Deprivation (IMD), BMI, insulin treatment, antihypertensive medication, and lipid-lowering medication. To account for genetic predisposition to type 2 diabetes, we further adjusted our models for the PRS of type 2 diabetes. For models with BMI as the outcome, adjustments were made with and without waist circumference. Correcting p-values for multiple testing was not applied since the models were considered independent with distinct hypotheses derived from previous studies. Reference Rothman12
Sensitivity analyses were performed with birthweight stratified by conventional clinical cut-offs Reference Cutland, Lackritz and Mallett-Moore13
Data management and statistical calculations were performed using R version 4.3.2 (R Foundation for Statistical Computing, Vienna, Austria).
Results
Study population
The UK Biobank consists of 502,357 participants, with 225,456 (45%) participants lacking recorded birthweight data. After applying exclusion criteria (detailed in Fig. 1), the final study population comprised 263,888 participants.
Figure 1. Flowchart of study population. Flowchart of UK Biobank study population.
Among these, 9,442 (3.6%) participants had type 2 diabetes at baseline (2006–2010), while 254,446 (96.4%) did not. Baseline characteristics by birthweight are presented in Table 1. Individuals with diabetes had a median age at enrolment of 61 years, an average of five years older than those without diabetes, with a median age of 55 at type 2 diabetes diagnosis. The proportion of males was higher in individuals with diabetes (57.2%, n = 5,398) compared to those without (38.3%, n = 97,463). BMI was higher among participants with diabetes (31.4 kg/m2) vs. without diabetes (26.4 kg/m2) and participants with high birthweight had a higher BMI regardless of diabetes status (Table 1). Lifestyle factors, including smoking and physical activity, were similar across birthweight groups regardless of diabetes status. However, individuals with a lower birthweight were less likely to consume alcohol frequently (3–4 times per week or almost daily) regardless of diabetes status (Table 1). A larger proportion of individuals with type 2 diabetes self-identified as being of an ethnicity other than White compared to those without diabetes (5.6% [n = 528] vs. 2.8% [n = 7,215]). This trend was also observed in low birthweight groups, regardless of diabetes status (6.9% [n = 169] vs. 3.9% [n = 2,517]).
Table 1. Baseline characteristics
Values are n (%) unless otherwise indicated.
Anthropometric measurements and blood pressure
In individuals with type 2 diabetes, linear regression analyses showed that each kg decrease in birthweight was associated with a 0.74 kg/m2 (95% CI 0.58, 0.90) lower BMI, a 2.15 cm (95% CI 1.78, 2.52) smaller waist circumference, and a 1.65 cm (95% CI 1.33, 1.97) smaller hip circumference (Fig. 2). Among those without diabetes, each kg decrease in birthweight was associated with a 0.21 kg/m2 (95% CI 0.18, 0.24) lower BMI, a 0.83 cm (95% CI 0.76, 0.91) smaller waist circumference, a 1.04 cm (95% CI 0.98, 1.09) smaller hip circumference,a 1.78 mmHg (95% CI 1.66, 1.89) higher SBP, and a 0.64 mmHg (95% CI 0.58, 0.71) higher DBP (Fig. 2). These associations remained after further adjustment for alcohol, smoking, physical activity, BMI, IMD, PRS for type 2 diabetes, insulin treatment, lipid-lowering and antihypertensive medication for both groups (Supplementary Table 2). Additionally, a slight increase in DBP (0.30 mmHg [95% CI 0.02, 0.59]) was observed in individuals with diabetes, though this association was attenuated with further adjustment. Each kg decrease in birthweight was also associated with slight increases in waist-hip ratio, waist-height ratio, total cholesterol, triglycerides, LDL cholesterol, HbA1c, CRP, and a small decrease in HDL cholesterol among individuals with diabetes (Supplementary Table 3). In contrast, those without diabetes showed a lower waist-hip and waist-height ratios with decreasing birthweight. No pattern of association was found between birthweight and age at diagnosis or biomarkers. Cubic restricted splines with up to 5 knots showed comparable results as the linear regression estimates, and the AIC did not differ significantly between the models (Supplementary Table 4 and supplementary Fig. 1).
Figure 2. Linear regression analysis. Linear regression analysis of BMI (a), waist circumference (b), hip circumference (c), systolic blood pressure (d), and diastolic blood pressure (e) according to birthweight. Estimate shows change in outcome per kg change in birthweight with 95% CIs. Adjusted for sex, age at enrolment, and family history of diabetes. Except for family history of diabetes which are adjusted for sex and age at enrolment.
Genetic predisposition and cardiometabolic outcomes
Reporting a family history of diabetes was associated with a higher birthweight regardless of diabetes status with double the effect size for individuals with diabetes; 0.06 (95% CI 0.03, 0.09) kg increase in birthweight and 0.03 (95% CI 0.00, 0.05: p-value 0.049) for individuals without diabetes (Table 2). In individuals with diabetes, those with a birthweight <2,860 g were more likely to have experienced a MI (OR 1.33 [95% CI 1.11, 1.60]) compared to those with a birthweight of 2,860–3,630 g (Fig. 3, Table 3). These associations persisted after further adjusting for BMI, smoking status, alcohol consumption, physical activity, IMD, PRS for type 2 diabetes, insulin treatment, lipid-lowering and antihypertensive medication (Supplementary Table 5). No pattern of association was found between birthweight and stroke or DVT (Fig. 3, Table 3). Additionally, in those with diabetes, a birthweight >3,630 g was associated with MI (OR 1.25 [95% CI 1.05, 1.49]) compared to those with a birthweight 2,860–3,630 g (Table 3).
Figure 3. Forest plot of logistic regression. Forest plot of family history of diabetes and diabetes-associated complications according to birthweight. Type 2 diabetes is indicated by orange colour (1st and 3rd line in each outcome) and blue for non-diabetes. For persons with type 2 diabetes a birthweight <25% = <2,860 g (n = 2,401) and >75% = >3,630 g (n = 2,189). For persons without type 2 diabetes a birthweight <25% = <2,950 g (n = 129,366) and >75% = >3,690 (n = 60,999). Adjusted for sex, age at enrolment, and family history of diabetes.
Table 2. Linear regression estimates for family history as predictor of birthweight
Results are given as increase in birthweight (kg) per reported family history of diabetes. Medication = lipid-lowering, antihypertensive, and insulin treatment. *IMD = Index of Multiple Deprivation.*.
*Subpopulation for only UK Biobank participants from England
Table 3. Logistic regression estimate
Covariates column indicate the adjustment variables. Estimates given odds ratios (OR) with 95% confidence intervals (CI). Medication = lipid-lowering, antihypertensive, and insulin treatment.
For individuals without diabetes, a birthweight <2,950 g, compared to a birthweight of 2,950–3,690 g, was more likely to have hypertension (OR 1.22 [95% CI 1.20, 1.25]), MI (OR 1.23 [95% CI 1.13, 1.33]), and stroke (OR 1.24 [95% CI 1.14, 1.34]) (Fig. 3, Table 3). Conversely, a birthweight>3,690 g, compared to a birthweight 2,950–3,690 g, were also less likely to have hypertension (OR 0.89 [95% CI 0.87, 0.91]) (Table 3). These associations remained after further adjustment for BMI, smoking, alcohol consumption, physical activity, IMD, and PRS for type 2 diabetes, but the association with MI and stroke was attenuated with additional adjustment for insulin treatment, lipid-lowering and antihypertensive medication (Supplementary Table 5).
Sensitivity analyses
Re-analysis of categorical variables using clinically defined low (<2,500 g) and high (>4,500 g) birthweight yielded results comparable to the reference group of 2,500g–4500 g, though with a tendency toward larger estimates and less statistical precision (Supplementary Table 6). Both low and high birthweight groups were more likely to have had a DVT (OR 1.20 [95% CI 1.09, 1.33] and OR 1.16 [95% CI 1.02, 1.31], respectively) (Supplementary Table 6). Including multiplicative interaction terms between birthweight and diabetes status in linear regression analyses gave p-values <0.05 for BMI, waist circumference, SBP, and DBP (Table 4).
Table 4. Linear regression estimates
Covariates column indicate the adjustment variables. Estimates given as change in the outcome per one kg increase in birthweight, with 95% confidence intervals (CI). Interaction column refers to multiplicative interaction terms between birthweight and type 2 diabetes status. Medication = lipid-lowering, antihypertensive, and insulin treatment.
When matching on age, sex, and BMI, we found similar effects for anthropometric measurements and blood pressure (Supplementary Table 7a), although differences between individuals with and without diabetes were reduced, Categorical outcomes showed comparable results (Supplementary Table 7b).
Limiting analyses to participants with type 2 diabetes within 5 years (n = 4,762 participants) produced comparable results with wider CIs (Supplementary Table 8a and 8b), Notably, birthweight was now associated with 0.07 years (95% CI 0.02, 0.13) younger age at diagnosis for each kg decrease in birthweight.
Sex-stratified analyses for participants with type 2 diabetes mirrored the non-stratified results (Supplementary Table 9a), except for females being associated with a 0.88 mmHg (95% CI 0.08, 1.68) higher SBP for each kg decrease in birthweight. The impact of lower birthweight was about half in females compared to males. Among participants without type 2 diabetes, each kg decrease in birthweight was associated with approximately double the increase in SBP and DBP in females compared to males (SBP: 2.17 mmHg [95% CI 2.02, 2.33] for females vs. 1.12 mmHg [95% CI 0.95, 1.29] for males; DBP: 0.78 mmHg [95% CI 0.69, 0.87] for females vs. 0.43 mmHg [95% CI 0.33, 0.53] for males) (Supplementary Table 9a, 9b, and 9b.
Discussion
In this study using UK Biobank data, we observed several differential associations between birthweight and cardiometabolic traits in individuals with and without type 2 diabetes. Sex-stratified analyses revealed generally similar trends. Lower birthweight was consistently associated with lower BMI, waist, and hip circumference at enrolment, with the magnitude of these associations being considerably larger in individuals with type 2 diabetes, as also supported by multiplicative interaction terms. Additionally, low birthweight was associated with MI irrespective of diabetes status. Interestingly, family history of diabetes was associated with a higher birthweight regardless of diabetes status, although with a larger impact in individuals with diabetes. Moreover, lower birthweight was associated with higher SBP, DBP, and risk of stroke only in individuals without type 2 diabetes.
We propose that these extensive differences in associations between birthweight and clinically relevant variables in people with and without type 2 diabetes arise from comparing populations with different predispositions to cardiometabolic traits. These discrepancies likely stem from the increased inherent susceptibility of individuals with low birthweight to developing type 2 diabetes. Assuming that birthweight, genetic predisposition, and obesity are approximately additive risk factors for type 2 diabetes, it seems plausible that individuals with a predominant contribution of low birthweight (reflecting an adverse intrauterine environment) require a lower burden of other major risk factors, such as genetic predispositions and obesity, to develop overt type 2 diabetes. In populations with established type 2 diabetes, it appears that when one risk factor is dominant, other contributory factors may be less prominent.
Contrary to previous findings, Reference Hansen, Thomsen and Brøns4,Reference Paulina, Donnelly and Pearson14 we did not find an association between lower birthweight and age at type 2 diabetes diagnosis. However, when restricting the analysis to participants diagnosed with type 2 diabetes within the past 5 years, a small association emerged between lower birthweight and a younger age at diagnosis. This discrepancy between our and previous findings likely stems from differences in data collection methods. In the UK Biobank, both birthweight and age at diagnosis are self-reported, whereas the Danish Centre for Strategic Research in Type 2 Diabetes (DD2) cohort Reference Hansen, Thomsen and Brøns4 utilises data where diabetes diagnosis is recorded at the point of clinical diagnosis, and birthweight is obtained through original midwife records.
The relationship between birthweight and adult body composition, notably BMI, is complex and likely influenced by numerous factors. Our findings, which show that lower birthweight is associated with lower BMI and waist circumference regardless of diabetes status, are consistent with previous studies. Reference Hansen, Thomsen and Brøns4,Reference Zhao, Wang, Mu and Sheng15,Reference Stansfield, Fain, Bhatia, Gutin, Nguyen and Pollock16 To our knowledge, we provide the first direct comparison between individuals with and without type 2 diabetes, showing more pronounced associations between birthweight and BMI in those diagnosed with type 2 diabetes compared to those without. This reflects the notion that individuals with low birthweight may be more sensitive to the deleterious effects of obesity, potentially accelerating the onset of type 2 diabetes. Further corroborated by Wibaek et al., indicating that birthweight and BMI contribute to the development of type 2 diabetes in an additive manner, Reference Wibaek, Andersen and Linneberg5 and Hansen et al., showing that a low birthweight is associated with a markedly younger age with less obesity at type 2 diabetes diagnosis. Reference Hansen, Thomsen and Brøns4 The finding of a small inverse relationship between birthweight and abdominal obesity, reflected by waist-hip and waist-height ratios in individuals without type 2 diabetes, likely indicates a true relationship of importance for the increased risk of developing type 2 diabetes, as supported by several previous studies Reference Stansfield, Fain, Bhatia, Gutin, Nguyen and Pollock16–Reference Alves, Cavalcante and Melo19 Despite the limitations of self-reported birthweight and family history data, the large sample size of the UK Biobank allowed us to validate previous findings that reporting a family history of diabetes is associated with higher birthweight among individuals with of type 2 diabetes. Reference Hansen, Thomsen and Brøns4 Notably, the magnitude of this association was nearly twice as large in individuals with type 2 diabetes compared to those without This pronounced difference could be explained by an enrichment of mothers with gestational diabetes within this group, as gestational diabetes is associated with having larger offspring. Reference Kc, Shakya and Zhang20 This aligns well with the emerging understanding of the additive contributions from genetic predispositions (proxy: family history of diabetes and PRS of type 2 diabetes), the intrauterine environment (proxy: low birthweight), and the postnatal environment (proxy: BMI) in the aetiology of type 2 diabetes. This suggests that among people with type 2 diabetes, there is a dilution of individuals with a family history of diabetes among those with low birthweight and, conversely, an enrichment of individuals with a family history of diabetes among those with high birthweight.
The foetal insulin hypothesis proposes that the link between low birthweight and type 2 diabetes may be partly explained by a genetically determined reduction in insulin secretion and/or action, which is causally related to impaired foetal growth and development. Reference Hughes, De Franco, Freathy, Flanagan and Hattersley21 While this hypothesis may not fully explain the relationship between low birthweight and type 2 diabetes, there are a few known type 2 diabetes susceptibility genes that also are associated with low birthweight. Reference Hughes, Hattersley, Flanagan and Freathy22 However, further adjusting all our models for a type 2 diabetes PRS did not change our associations, suggesting that the here reported associations have a predominant non-genetic origin. Importantly, the association between birthweight and family history of diabetes in both individuals with and without type 2 diabetes supports the notion that the relationship between low birthweight and type 2 diabetes risk is unlikely to be substantially confounded by genes that are simultaneously causally implicated in impaired foetal growth and diabetes susceptibility.
Low birthweight has consistently been associated with hypertension and CVD, even at relatively young ages, irrespective of diabetes status. Reference Hansen, Thomsen and Brøns4,Reference Knop, Geng and Gorny23–Reference Curhan, Willett, Rimm, Spiegelman, Ascherio and Stampfer26 Interestingly, the relationship between birthweight and CVD-related outcomes, such as hypertension, stroke, and MI, in this study appeared to be more pronounced in individuals without type 2 diabetes compared to those with type 2 diabetes. The fact that some of these associations are less pronounced, may be explained by competing and differential cardiometabolic risk factors including elevated lipids and glucose levels, and/or the higher use of medications among people with type 2 diabetes. This, especially, may explain the lack of associations between low birthweight, on one side, and hypertension as well as stroke on the other, within the type 2 diabetes subgroup.
Noteworthy, however, a recent study from our group reported that in a population of people with newly diagnosed type 2 diabetes, the increased risk of CVD associated with low birthweight becomes more apparent over time. Reference Hansen, Brøns and Engelhard27 This indicates that the associations between low birthweight and CVD are not solely mediated through the pathways of type 2 diabetes or hyperglycaemia, highlighting a more complex interplay between early-life growth and later cardiovascular health. In sex-stratified analysis, females with lower birthweight showed an approximately double increase in systolic and diastolic blood pressure compared to males for each kg decrease in birthweight. Previous studies have found a similar pattern of a more pronounced risk of type 2 diabetes and risk of CVD in males than females among individuals with low birthweight, although all a general increased risk. Reference Wibaek, Andersen and Linneberg5,Reference Hansen, Brøns and Engelhard27 This effect modification may be explained by differences in genes, vascular reactivity, endothelial function, or hormonal influences such as oestrogen, which are known to play distinct roles in cardiovascular regulation between sexes. Reference Regitz-Zagrosek and Gebhard28
Additionally, we found small effect size associations between birthweight and several biochemical biomarkers, including hsCRP, blood lipids, creatinine, liver enzymes, plasma glucose, and HbA1c, exclusively in individuals without type 2 diabetes. The absence of similar associations in those with type 2 diabetes may be due to the dysmetabolic state inherent to the condition, where the specific contributions of adverse foetal environment may be masked by the other metabolic disturbances. Furthermore, individuals in the type 2 diabetes population were likely already in relevant treatments aimed at managing their dysmetabolic state, which may further obscure these associations.
Limitations
Discrepancies in cohort characteristics and definitions between our study and prior research could explain some of our divergent findings. The UK Biobank’s prospective design introduces potential participant bias, including healthy volunteer bias, with only a 5.5% participant acceptance rate. This cohort exhibits lower mortality and morbidity, higher education levels, and healthier lifestyles than the general population. Reference Davis, Coleman and Adams29,Reference Stamatakis, Owen, Shepherd, Drayton, Hamer and Bauman30 Given that low birthweight is associated with lower IQ, Reference Flensborg-Madsen and Mortensen31 socioeconomic status, Reference Barker32,Reference Barker33 and increased multimorbidity, Reference Vaag, Grunnet, Arora and Brøns1–Reference Ross and Beall3 it is plausible that the UK Biobank’s participants with low birthweight represent the healthiest subset of those born with low birthweight, likely underestimating comorbidities and complications and biasing results toward the null. Nonetheless, the UK Biobank’s internal validity and size ensure reliable exposure and outcomes assessments. Reference Davis, Coleman and Adams29,Reference Stamatakis, Owen, Shepherd, Drayton, Hamer and Bauman30 A further limitation is the reliance on self-reported data, such as birthweight, which could introduce recall bias. However, the consistency of birthweight findings in the UK Biobank with those from previous meta-analyses suggests that self-reported birthweight is a reliable data source. Reference Shenkin, Zhang, Der, Mathur, Mina and Reynolds34 Additionally, we lacked information on gestational age, preterm birth, and non-singleton status. The availability of gestational age or preterm birth status would possibly allow for a more precise estimation of the degree of adverse intrauterine environment. The main strength of our study lies in its large sample size, allowing for increased statistical power and precision, thereby improving the generalisability of our findings. We were able to adjust for several confounding variables and comprehensively assess the specific impact of birthweight on our outcomes. Despite these adjustments, we could not account for all potential lifestyle factors, leaving the possibility of residual confounding, particularly given the lack of data on maternal and paternal environments. Additionally, our observational study design limits our ability to establish causality. While we identified associations, this study cannot definitively determine whether low birthweight directly causes the observed health outcomes. Nonetheless, our analysis allowed for the exploration of multiple associations between birthweight and various important clinical cardiometabolic outcomes, highlighting differences between people with and without type 2 diabetes.
Conclusion
Our comparison demonstrates how stratification by type 2 diabetes reveals differential associations that facilitate a deeper understanding of the complex interplay between birthweight, the foetal environment, and the development of type 2 diabetes and its associated cardiometabolic traits and comorbidities. Our approach not only reinforce the link between birthweight and type 2 diabetes but also sheds light on the heterogeneity of type 2 diabetes presentation and its various endotypes, which may be explained by multiple factors including birthweight.
Supplementary material
The supplementary material for this article can be found at https://doi.org/10.1017/S2040174425000066.
Acknowledgements
The current research was performed as part of the UK Biobank Application ID 71,699 and acknowledges the same. All authors have read and approved the manuscript.
Financial support
TSA was supported by the Novo Nordisk Foundation Grant NNF18OC0052457 and internal funding from Steno Diabetes Center Copenhagen, Herlev, Denmark. CB owns stock in Novo Nordisk. This work was partly funded by the Swedish Research Council (EXODIAB, 2009-1039; 2018-02837). The study funders were not involved in the design of the study, the collection, analysis, and interpretation of data or writing the report. They did not impose any restrictions regarding publication of the report. The authors declare that there are no other relationships or activities that might bias, or be perceived to bias, their work.
Competing interests
None.
Ethical standard
Ethical standardEthical approval for the UK Biobank study was obtained from the Northwest Centre for Research Ethics Committee (REC) (REC reference: 11/NW/0382). This study was conducted under UK Biobank application number 71,699.
Open access
