Risk of progressive chronic kidney disease in individuals with early-onset type 2 diabetes: a prospective cohort study

Abstract

Background

The progression trajectory of renal filtration function has not been well characterized in patients with early-onset type 2 diabetes mellitus (T2DM) although albuminuria is often reported in this population. We aim to study the risk of progressive chronic kidney disease (CKD) in individuals with early-onset T2DM.

Methods

In total, 1189 T2DM participants were followed for 3.9 (interquartile range 3.2–4.7) years. Progressive CKD was defined as estimated glomerular filtration rate (eGFR) decline of ≥5 mL/min/1.73 m² per year. Early-onset T2DM was defined as age at T2DM diagnosis between 18 and 30 years.

Results

Compared with later-onset counterparts (N = 1032), participants with early-onset T2DM (N = 157) were more obese and had poorer glycaemic control at baseline. In the follow-up, 24.2% and 15.6% experienced progressive CKD in early-onset and later-onset participants, respectively (P = 0.007). Logistic regression suggested that participants with early-onset T2DM had 2.63-fold [95% confidence interval (CI) 1.46–4.75] higher risk of progressive CKD after accounting for multiple traditional risk factors. Furthermore, the excess risk of progressive CKD associated with early-onset T2DM mainly occurred in participants with preserved renal function [eGFR ≥60 mL/min/1.73 m², odds ratio (OR) 2.85, 95% CI 1.50–5.42] and was more pronounced in those with diabetes duration <10 years (OR 3.67, 95% CI 1.51–8.90).

Conclusions

Individuals with early-onset T2DM have a higher risk of progressive CKD. The excess risk mainly exhibits in early stage of CKD and cannot be solely attributed to traditional risk factors and a longer diabetes duration.

INTRODUCTION

The onset age of type 2 diabetes mellitus (T2DM) is falling and the prevalence of early-onset T2DM is increasing rapidly in both developed and developing countries presumably due to the concurrent obesity epidemic [1–3]. In some populations including Asians, the incidence of T2DM has outnumbered type 1 diabetes mellitus (T1DM) in young people [1, 4, 5].

The natural history of early-onset T2DM is still largely unknown. However, accumulating evidence suggests that patients with early-onset T2DM exhibit a more aggressive disease course and often develop vascular complications after a relatively shorter duration of diabetes than their later-onset counterparts and those with T1DM [3, 6–10]. Diabetic kidney disease (DKD) is one of the most frequently reported complications among individuals with T2DM onset at a young age, with albuminuria as the most often observed manifestation [7, 11–13]. Notably, DKD begins to appear as early as 3–5 years after diabetes diagnosis in this population, which is sooner than those with T1DM [7, 12, 14–16]. Moreover, patients with early-onset T2DM and microalbuminuria have a faster progression to persistent macroalbuminuria and a higher risk for end-stage renal disease (ESRD) [17, 18]. However, it remains to be elucidated whether this higher ESRD risk is due to a fast progression of chronic kidney disease (CKD) or the relatively lower competing risk of mortality conferred by young age.

Albuminuria and decline of renal filtration function are two manifestations of kidney impairment with overlapping but distinct pathophysiologic and genetic determinants [19, 20]. Although a high risk for albuminuria development and progression has been reported in patients with early-onset T2DM [3, 7, 12], to our knowledge, no study has characterized the progression trajectory of renal filtration function in this population. In the current work, we aim to study the risk of progressive CKD in individuals with T2DM diagnosed at their early adulthood.

MATERIALS AND METHODS

Participants

Participant recruitment for the SMART2D (Singapore Study of Macro-angiopathy and Micro-Vascular Reactivity in Type 2 Diabetes) cohort has been described elsewhere [21, 22]. In brief, 2057 T2DM outpatients aged between 21 and 90 years were recruited from a regional hospital and a primary care medical facility in the northern region of Singapore between August 2011 and March 2014. T2DM was defined as (i) fasting plasma glucose ≥7.0 mmol/L, (ii) random plasma glucose ≥11.1 mmol/L, (iii) haemoglobin A1c (HbA1c) ≥6.5% or (iv) on hypoglycaemic medications. Ascertainment of T2DM was mainly based on exclusion of T1DM and specific types of diabetes due to other causes. Because the prevalence of anti-glutamic acid decarboxylase autoantibody was only 30–40% in Asians with T1DM [23], we defined T1DM as requirement for continuous insulin treatment 1 year after diabetes onset without taking presence of autoantibodies as a prerequisite. For participants with onset age <40 years and body mass index (BMI) <25 kg/m², we sequenced three major maturity onset diabetes in the young (MODY) genes (HNF1A, HNF4A and GCK) by next-generation sequencing (Ion Torrent, Life Technologies, Carlsbad, CA, USA) and did not identify pathogenic mutations.

Patients with autoimmune diseases (e.g. systemic lupus erythromatosis) or cancer on active treatments, pregnant women and those who could not fulfil informed consent were excluded. Other exclusion criteria included intake of non-steroid anti-inflammatory drugs and oral steroids equivalent to ≥5 mg/day prednisolone on the day of enrolment. Participants were subsequently followed at their regular clinic visits in the same institutions.

The current study complies with principles laid by the Helsinki Declaration and has been approved by the Singapore National Healthcare Group Domain Specific Review Board. Written consent has been obtained from each participant.

Definition of early-onset T2DM

We define early-onset T2DM as age of diabetes onset (diagnosis) between 18 and 30 years because, (i) early studies have shown that T2DM patients with onset age <30 years have a worse cardio-renal risk profile and a higher risk for progression to persistent proteinuria [9, 13, 24] and (ii) we intend to focus on participants with T2DM onset in their early adulthood in the current study. The remaining participants with age of T2DM onset >30 years were taken as ‘later-onset’ T2DM controls. In sensitivity analyses, we define early-onset T2DM as diabetes diagnosis age ≤35 or ≤40 years, respectively.

Trajectory of renal filtration function and definition of progressive CKD

Readings of estimated glomerular filtration rate (eGFR) were extracted from electronic medical records to study progression trajectory of kidney function. For participants with inpatient records (N = 192) in the follow-up period, only the first eGFR at entry and the last eGFR at discharge were included to partly reduce the influence of potential reversible acute kidney injury (AKI) during hospital stay. We identified 1232 participants with eGFR readings ≥3 mL/min/1.73 m² from date of cohort enrolment to April 2017. Trajectory (slope) of annual eGFR change was calculated by linear regression by which eGFR was regressed on follow-up time (year). Coefficient of time was taken as slope of decline [25, 26]. After excluding participants with <1 year follow-up or those with baseline eGFR ≤15 mL/min/1.73 m², 1189 participants were included in final analyses (Supplementary data, Figure S1).

CKD progression was defined according to Kidney Disease: Improving Global Outcomes criterion, i.e. an eGFR decline of ≥5 mL/min/1.73 m² per year [27]. All other participants with eGFR trajectory decline of <5 mL/min/1.73 m² per year were taken as controls.

Clinical and biochemical measurements

Baseline demographic, clinical and biochemical data were collected in a standardized form and entered into an online central data management server. Ethnicity and smoking status were self-reported. Age at diabetes diagnosis was also self-reported but cross-validated by review of the medical records for those who reported age at T2DM diagnose younger than 40 years. Blood pressure was measured three times in a seated position with a 5-min interval in-between, and the average reading was used. Blood and spot urine specimens were collected in the morning after an overnight fasting.

Serum triacylglycerol and high-density lipoprotein (HDL) and low-density lipoprotein (LDL) cholesterol levels were quantified by enzymatic methods (Roche Cobas Integra 700, Roche Diagnostics, Switzerland). HbA1c was measured by a point-of-care immunoassay analyser (DCA Vantage Analyzer, Siemens, Germany). Creatinine was measured by an enzymatic method, which was traceable to isotope dilution mass spectrometry reference. Urinary albumin was quantified by a solid phase competitive chemiluminescent immunoassay (Immulite, DPC, Gwynedd, UK) and albuminuria level was presented as albumin-to-creatinine ratio (ACR, mg/g). Plasma C-reactive protein was measured by a high sensitive immunoassay kit (R&D Systems, Minneapolis, MN, USA). Glomerular filtration rate was estimated by the CKD-Epidemiology Collaboration (CKD-EPI) formula. Serum creatinine concentrations for baseline and follow-up eGFR calculations were measured in the same central lab.

Statistical analysis

Data were presented as mean ± SD, median (interquartile range, IQR) or proportion where appropriate. Between-group differences in clinical and biochemical variables were compared by Student's t-test for normally distributed variables, Mann–Whitney U test for variables with skewed distribution or chi-square test for categorical variables.

Multivariable logistic regression models were fitted to study the association of early-onset T2DM with risk of progressive CKD. CKD progression (binary) was a dependent variable and early-onset T2DM (yes or no) was an independent variable. Duration of diabetes was entered as covariate because it is an important determinant of diabetic complications, especially in those with early-onset age [10, 28, 29]. Next, demographic variables (index age, sex and ethnicity) and smoking, cardiometabolic risk factors (BMI, systolic blood pressure and HbA1c), baseline eGFR and ACR, and insulin and renin–angiotensin system (RAS) blocker usage (yes or no) were sequentially entered into the models in block. Similar logistic regression models were applied to study risk of progressive CKD associated with early-onset T2DM after stratification of participants by duration of diabetes or baseline eGFR levels.

Statistical analysis was performed by SPSS (version 22). A two-sided P < 0.05 level was considered as statistically significant.

RESULTS

Participant characteristics at baseline

T2DM onset age in SMART2D cohort (N = 2057) was 46 ± 12 years. Those not included were older, and had a shorter duration of diabetes but older age at T2DM onset. They had a less severe cardiometabolic risk profile as manifested by a lower BMI and triacylglycerol level, a higher HDL cholesterol level and were less likely to be on RAS blocker and insulin treatments. There was no clinically meaningful between-group difference in baseline renal function (Supplementary data, Table S1).

As shown in Table 1, participants with early-onset T2DM had diabetes diagnosed at 25 ± 4 years. They were 15 years younger but had a longer duration of diabetes at cohort enrolment. Also, they had a higher BMI, poorer glycaemic control, a higher proportion of current smokers, and were more likely on insulin treatment but less likely on statin intake. Notably, a higher proportion of participants with early-onset T2DM had albuminuria (ACR ≥30 mg/g) at baseline (56.5% versus 47.9%, P = 0.047).

A higher risk of progressive CKD in participants with early-onset T2DM in follow-up

Participants were followed for 3.9 (IQR 3.2–4.7) years with 1.6 (IQR 1.2–2.2) eGFR readings per year. There were no significant differences in follow-up duration and frequency of eGFR measurements between early- and later-onset T2DM participants. eGFR declined at an average of 2.6 and 2.1 mL/min/1.73 m² per year in early- and later-onset T2DM participants, respectively (P = 0.08). Age and baseline eGFR-adjusted decline slope was faster in early-onset T2DM but the difference did not reach statistical significance level (adjusted difference was 0.73 mL/min/1.73 m² per year, P = 0.06).

A significantly higher proportion of participants with early-onset T2DM experienced progressive CKD as compared with later-onset counterparts (24.2% versus 15.6%, P = 0.007). Logistic regression models suggested that participants with early-onset T2DM had 1.72-fold [95% confidence interval (CI) 1.13–2.61] higher risk of progressive CKD after adjustment for diabetes duration (Model 1, Table 2). Further adjustment for potential confounders, traditional risk factors, baseline eGFR and ACR level slightly increased magnitude of the association (Models 2–5, Table 2). Further adjustment for C-reactive protein or lipids profile (HDL, LDL cholesterol and triacylglycerol) above Model 5 did not materially alter the outcome [odds ratio (OR) 2.66, 95% CI 1.47–4.8 and OR 2.60, 95% CI 1.44–4.71, respectively].

In sensitivity analyses, the association of early-onset T2DM with progressive CKD remained statistically significant when definition of early-onset T2DM was extended from 30 to 35 years of age at diabetes diagnosis, which was ∼1 SD lower than the average onset age in SMART2D cohort (adjusted OR 2.25, 95% CI 1.29–3.92). Including six participants with onset age <18 years or excluding 42 participants with baseline eGFR ≤30 mL/min/1.73 m² did not materially change the outcomes either. However, no significant association of early-onset T2DM with progressive CKD was observed in both unadjusted and adjusted models when definition of early-onset T2DM was extended to onset age younger than 40 years old (data not shown).

Excess risk of progressive CKD associated with early-onset T2DM mainly occurred in participants with duration of diabetes <10 years or those with preserved renal filtration function

Given that excess risk of cardiovascular and kidney complications associated with early-onset T2DM has been mainly attributed to a longer diabetes duration in early studies [8, 28, 30], we performed further analyses after stratification of participants by diabetes duration. As shown in Table 3, participants with early-onset T2DM and a <10 years diabetes duration had 4.91-fold higher adjusted risk of progressive CKD as compared with later-onset counterparts. The adjusted risk was attenuated with prolongation of diabetes duration. Notably, the unfavourable cardio-renal risk factors in those with early-onset T2DM were even worse among the subgroup with a diabetes duration <10 years (Supplementary data, Table S2).

Cognisant that participants had heterogeneous baseline kidney function ranging from CKD Stage 1 to 4 [27], we further studied risk of progressive CKD after stratifying participants by baseline eGFR levels as an exploratory analysis. In subgroup with preserved renal function (eGFR ≥60 mL/min/1.73 m², N = 964), 26.7% versus 16.6% participants experienced progressive CKD in those with early- and later-onset T2DM, respectively, P = 0.005). However, no significant difference in risk was observed in subgroup with eGFR <60 mL/min/1.73 m² (N = 225, 9.1% versus 11.3% in early- and later-onset participants, P = 0.75). Multivariable logistic regression suggested that participants with early-onset T2DM and preserved kidney function had 2.85-fold (95% CI 1.50–5.42) higher adjusted risk for progressive CKD (Supplementary data, Tables S3 and S4).

DISCUSSION

In this longitudinal study, we observed that participants with T2DM onset in their early adulthood had a greater risk of progressive CKD as compared with those with later-onset T2DM. The excess risk mainly occurred in early stage of CKD and could not be solely attributed to a longer diabetes duration and traditional risk factors. To our knowledge, this may be the first study to characterize the trajectory of kidney filtration function in patients with early-onset T2DM although high prevalence of albuminuria and greater risk for albuminuria progression have been well-appreciated in this population [3, 7, 12, 13]. Our data added evidence to support that early-onset T2DM may have a more aggressive disease course in terms of renal function decline and potentially distinct pathophysiology as compared with the majority of T2DM patients with diabetes onset in their later adulthood.

The clinical phenotype of participants with early-onset T2DM in our current study was similar to that reported in other studies [8, 9]. They tended to be more obese, had worse glycaemic control and atherogenic lipids profile, and were more likely on insulin treatment but less likely on statin treatment despite having a similar or even higher LDL cholesterol level as compared with later-onset counterparts [7, 9, 10, 31, 32]. These unfavourable cardio-renal risk factors may at least partly contribute to the high risk of macro- and microvascular complications associated with early-onset T2DM [7, 12–16, 24].

The mechanisms underpinning the excess risk of progressive CKD in patients with early-onset T2DM remain to be elucidated. As mentioned above, the unfavourable cardio-renal risk factors may partly explain the high prevalence of albuminuria and high risk for progressive CKD in this group of patients [33]. However, adjustment for these traditional risk factors did not materially change the direction or magnitude of association between early T2DM onset and risk of progressive CKD, suggesting the existence of residual risk factors or novel pathophysiological pathways. These unobserved risk factors may include insulin resistance, which is associated with abnormal tubular-glomerular feedback and an increase in glomerular hydrostatic pressure [34]. Socio-economic and behaviour factors may also play a role in the excess risk of progressive CKD as indicated by a higher proportion of current smokers in participants with early-onset T2DM in the current study. Emerging data suggest that the pathogenesis of early-onset T2DM may involve a stronger genetic predisposition as compared with later-onset counterparts [3]. Therefore, it is possible that genetic predisposition may also partly contribute to CKD development and progression in concert with environmental and behavioural factors in the setting of early T2DM onset. On the other hand, it is reasonable to postulate that hyperfiltration due to uncontrolled hyperglycaemia and its subsequent resolution after treatment may underlie fast eGFR decline in early-onset T2DM. However, this is unlikely the case since neither the absolute rate of eGFR decline nor proportion of progressive DKD in participants with baseline eGFR ≥120 mL/min/1.73 m² (upper bound of 95% CI of eGFR distribution in SMART2D) differ significantly from those with baseline eGFR <120 mL/min/1.73 m² (data not shown).

Early studies suggested that the high risks of cardiovascular and renal events in patients with early-onset T2DM were largely attributable to a long duration of diabetes and consequent longer exposure to cardiometabolic risk factors [8, 28, 30]. However, our data showed that the excess risk of progressive CKD could not be solely explained by a longer duration of diabetes. Instead, the excess risk mainly occurred in the first 10 years after diabetes onset. The discrepancy between our current study and early reports may be partly reconciled by the difference in study populations and definition of early-onset T2DM. The prior studies defined early-onset T2DM by diabetes diagnoses before 40 years of age [8, 30]. In our current study, we chose diagnosis ages of 30 and 35 years [13, 24]. We conceive that a 35-years old cut-off corresponds to 1 SD from average onset age in our cohort, which may be more likely to identify a subpopulation with ‘outlying’ clinical and pathophysiological feature. It has been known that Asians develop T2DM at an earlier age than Caucasians [35]. Hence, a younger cut-off age may be appropriate for study of early-onset T2DM in Asian populations.

Our finding of a high risk for progressive CKD in participants with early T2DM may have clinical implications. CKD progression as defined by eGFR trajectory is associated with a high risk of ESRD, cardiovascular disease and mortality [25, 36, 37]. Therefore, the rising incidence of early-onset T2DM worldwide may affect the epidemiology of DKD [5]. Our work lends evidence to support that the higher ESRD risk in early-onset T2DM as observed in early studies may be attributable to both a fast CKD progression and a lower competing risk of mortality conferred by young age [17, 18]. Additionally, given that the excess risk of progressive CKD mainly occurred in early-onset T2DM with preserved renal function and a shorter duration diabetes, an early and intensive surveillance on renal function is warranted in this T2DM subpopulation.

Our current study has the following strengths. Progressive CKD as defined by trajectory of eGFR is less influenced by short-term eGFR fluctuations [26, 27, 36, 37]. We have taken major traditional cardio-renal risk factors into consideration and performed several sensitivity analyses. It seems that the independent association of early-onset T2DM with a high risk of progressive CKD is robust. However, our data should be interpreted in the context of several weaknesses. First, it is difficult to ascertain the exact onset age of T2DM because many patients with T2DM remain asymptomatic and undiagnosed. Therefore, we have to take T2DM diagnosis age as onset age. In addition, participants with a long duration of diabetes may have recall bias and inaccurate report of diagnosis age. However, given that the excess risk of progressive CKD mainly occurred in those with a shorter diabetes duration, recall bias may be less likely to be major concern for the current work [38]. Secondly, although eGFR slope derived from long-term multiple eGFR readings may theoretically be less likely influenced by short-term eGFR decline due to AKI, the latter has been considered as a risk factor for CKD progression [39]. We could not address how AKI may alter CKD progression trajectory. Thirdly, the proportion of participants with early-onset T2DM in the overall diabetic population is relatively small. The low statistical power prevented us from testing multiplicative interactions before stratification of participants by CKD stage and diabetes duration. Therefore, our subgroup analyses are exploratory for hypothesis generation. Finally, early study has shown that eGFR decline may proceed albuminuria progression in patients with T1DM [40]. We could not address the temporal relationship between CKD progression and albuminuria progression in the current work.

CONCLUSION

Our current study suggests that patients with early-onset T2DM may have a distinct clinical phenotype and a high risk of progressive CKD. The excess risk of progressive CKD associated with early-onset T2DM mainly exhibits in the early disease stages and cannot be solely attributed to traditional cardio-renal risk factors and a longer duration of diabetes.

ACKNOWLEDGEMENTS

We thank Mr Darren Yeo, Ms Babitha Jeevith, Mr Dave Lee and Ms Jessie Fong from Khoo Teck Puat Hospital, Singapore for data set management, calculation of eGFR trajectory and assistance with medical record review.

FUNDING

The work was supported by Singapore National Medical Research Council [Grant CSA-INV/0020/2017 and CIRG/1395/2014]. The funder has no role in study design, data analysis, manuscript writing and decision to submit for publication.

AUTHORS’ CONTRIBUTIONS

J.-J.L. and S.C.L. designed the study, researched and interpreted the data and drafted the manuscript; S.L., R.L.G., K.A., W.E.T., C.F.S. and S.T. contributed important intellectual knowledge. All co-authors revised the manuscript critically for important intellectual contents and approved publication of the manuscript. S.C.L. is the guarantor of this work and, as such, had full access to all data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

CONFLICT OF INTEREST STATEMENT

None declared.

REFERENCES