The Association Between Age at Diagnosis of Type 2 Diabetes and Albuminuria in Chinese Adults

Background: Type 2 diabetes is increasingly diagnosed at a younger age worldwide and in China. Limited data are available regarding the association between age at diabetes diagnosis and risks of albuminuria, which is a common microvascular complication in diabetes patients. In addition, few studies on age at diagnosis and outcomes have fully accounted for the complex interplay between age at diagnosis and the actual age or diabetes duration, all of which are independently associated with vascular risks. Methods: We used data from a nationwide multicenter study with 207,961 participants recruited from 25 communities across mainland China during 2010-2012. Age, sex, and study sites were matched for 31,366 screen-detected type 2 diabetes and 31,366 normal controls. Age, sex, study sites, and diabetes duration were matched for 7,490 self-reported type 2 diabetes and 7,490 normal controls. Risks of having albuminuria in matched type 2 diabetes vs. normal controls were examined using multivariable logistic regression analysis in strata of age at type 2 diabetes diagnosis (<50, 50-59, 60-69, or ≥ 70 years). Albuminuria was dened as urinary albumin-to-creatinine ratio ≥ 30 mg/g. Results: Percentages of albuminuria were signicantly higher among type 2 diabetes patients compared with normal controls in each stratum of age at diagnosis. Although the absolute rate of albuminuria is higher in older adults, the odds ratio of albuminuria in type 2 diabetes vs. matched controls decreased with increasing age at diagnosis. For participants with diabetes diagnosed at an age of <50, 50-59, 60-69, or ≥ 70 years, the multivariable adjusted risk of albuminuria increased by 81%, 60%, 45%, and 33% for screen-detected diabetes, and 135%, 121%, 90%, and 58% for self-reported diabetes compared with their normal controls, respectively. Conclusions: A younger age at diagnosis of type 2 is associated with more signicantly elevated


Background
Although type 2 diabetes mellitus (T2DM) has been regarded as a condition that affects mostly elderly people, the number of younger people with diabetes is increasing. According to the International Diabetes Federation (IDF), 46% of deaths associated with diabetes among the 20-79 age group are in patients under age 60 [1,2] . Given the shifting age distribution of diabetes patients globally, the impact of age, age at diagnosis, and diabetes duration on vascular complications warrants investigation. Previous studies have established close associations of age [2] and diabetes duration [3,4] with the risk of vascular complications in patients with diabetes. In addition, studies have suggested that a younger age at diagnosis was related to an increased risk of macrovascular and microvascular complications as well as death in diverse populations including Chinese patients [5][6][7][8][9][10] . However, few studies have fully accounted for the complex interplay between age at diagnosis and the actual age or diabetes duration, all of which are independently associated with vascular risks. Furthermore, most previous studies have compared vascular risks in early-onset versus late-onset T2DM, rather than the effects of age at diagnosis across a wider age range.
Albuminuria is a biomarker of endothelial dysfunction and has been considered as the initial pathway to the progression of decline in renal function in patients with T2DM [11] . In addition, albuminuria has been regarded as a risk factor for cardiovascular diseases (CVD) and all-cause mortality in adults with or without diabetes [12,13] . Currently, limited data are available investigating the association between age at diabetes diagnosis and urinary protein/albumin excretion. One study examined the risk of nephropathy de ned by protein-to-creatinine ratio ≥ 0.5 g/g in participants with youth-onset diabetes compared with adult-onset diabetes in Pima Indians and the risk was not different [14] . Another study found signi cantly higher incidence of microalbuminuria in diabetes diagnosed at age 18-44 years compared with diabetes diagnosed at age ≥ 45 years [15] . Therefore, we used baseline data from a large nationwide cohort of Chinese adults and matched T2DM patients with normal controls on age, sex, and study sites. For diabetes not newly diagnosed, duration of diabetes was also matched. We examined the risk of albuminuria in T2DM patients compared with matched controls in each stratum of age at diagnosis to see whether the increased risk is similar for different age groups at diagnosis, therefore to examine the impact of age at diagnosis of T2DM on the risk of albuminuria in Chinese adults.

Study population
The Risk Evaluation of cAncers in Chinese diabeTic Individuals: a lONgitudinal (REACTION) study is a nationwide multicenter prospective cohort study [16,17] , of which baseline data were used for the current analysis. During 2011-2012, a total of 259,657 adults aged ≥40 years were recruited from 25 communities across mainland China. Participants were examined comprehensively for their cardiometabolic health using a structured questionnaire, anthropometric measurements, and biochemical tests. Participants with other types of diabetes, diabetes diagnosed at an age <18 years, diabetes with a duration ≥10 years, diseases which may also cause albuminuria such as lymphoma and other renal diseases, and participants with missing data on diabetes status, age at diagnosis of diabetes, and urinary albumin to creatinine ratio (ACR) were excluded. Therefore, a total of 207,961 participants were used for matching ( Figure 1).
The REACTION study was approved by the Ethical Review Committee of Ruijin Hospital. All study participants provided written informed consent.

Matching
Matching was conducted in 2 steps (Figure 1). In the 1 st step, participants with T2DM were categorized into 4 groups according to their age at diagnosis (years): <50, 50-59, 60-69, or ≥70. Self-reported (i.e. previously diagnosed) T2DM and screen-detected (i.e. newly-diagnosed) T2DM were separated because diabetes duration is an important confounding factor for self-reported T2DM but not for screen-detected T2DM. For patients with self-reported T2DM (n = 11,810), diabetes duration (±1 year) was matched between the 4 groups in a ratio of 2:4:3:1 based on the number of participants in each group [18] . In the 2 nd step, controls were randomly selected from the general population without diabetes (n = 164,725) and were 1:1 pair matched with all T2DM patients in the 1 st step for age (±1 year), sex, and study sites. After matching, 31,366 screen-detected T2DM and 31,366 matched controls, 7,490 self-reported T2DM and 7,490 matched controls were nally included in the current analysis.

Data collection
At each study site, trained staff collected data according to a standard protocol at local health stations or community clinics. Using a standard questionnaire, information on sociodemographic characteristics, lifestyle factors, and medical histories was collected through personal interviews. Participants were asked whether they were previously diagnosed as having diabetes by a physician. If the answer was 'yes', types of diabetes and time of diagnosis were reported.
Height and weight were measured to the nearest 0.1 cm and 0.1 kg, respectively with participants wearing light-weight clothes and no shoes. Body mass index (BMI) was calculated as the weight in kilograms divided by the height in meters squared (kg/m 2 ). Blood pressure (BP) was measured 3 times with 1-min intervals after at least 5-min sitting rest using an automated electronic device (Omron Model HEM-725 FUZZY; Omron Co, Dalian, China). Participants were advised to avoid alcohol, coffee, tea, smoking, and exercise at least 30 minutes before BP measurement. The average of three readings was used for analysis.
Blood samples were collected after an overnight fast for at least 10 hours. Participants without a known history of diabetes underwent the oral glucose tolerance test (OGTT) and 2h blood samples after the glucose load were also collected. Fasting and OGTT-2h plasma glucose levels were measured at local hospitals using the glucose oxidase or hexokinase method within 2 hours of blood sample collection. The rst void urine samples were also collected from each participant in early morning. Serum and urine samples were aliquoted into 0.5-mL Eppendorf tubes within 2 hours after collection and were then frozen at -80 degrees at local hospitals. All samples were shipped in dry ice to the central laboratory at Shanghai Institute of Endocrine and Metabolic Diseases, accredited by the College of American Pathologists (CAP), where levels of serum creatinine (Cr), total cholesterol (TC), low-density lipoprotein cholesterol (LDL-c), high-density lipoprotein cholesterol (HDL-c), and triglycerides (TG) were measured on an auto-analyzer (c16000 system, ARCHITECT ci16200 analyzer, Abbott Laboratories, Illinois, USA), and levels of glycosylated hemoglobin A 1c (HbA 1c ) were measured using a high performance liquid chromatography method (VARIANT TM II System, BIO-RAD, Hercules, CA, USA). The estimated glomerular ltration rate (eGFR) was calculated using the Chronic Kidney Disease-Epidemiology Collaboration (CKD-EPI) equation [19] .
Urinary albumin concentrations were measured at the central laboratory by immunonephelometry using Siemens BNII nephelometers (Siemens Healthcare Diagnostics, Marburg, Germany). The lower limit of detection is 2.13 mg/l. The intra-assay and inter-assay coe cients of variation for urinary albumin were 2.1% and 2.3%, respectively. Urinary creatinine concentrations were measured at the central laboratory by an enzymatic method (ADVIA Chemistry XPT System; Siemens Healthcare, Erlangen, Germany). The intraassay and inter-assay coe cients of variation for urinary creatinine were 1.1% and 1.3%, respectively.

Statistical analysis
Variables were checked for the distributional assumption of normality using normal plots. Subject characteristics which conformed to a normal distribution were presented as means ± standard deviations (SDs), otherwise as medians (interquartile ranges). All categorical variables were presented as the numbers (proportions). The distributions of diabetes duration, TG, and urinary ACR were skewed. Analyses were done separately for self-reported and screen-detected T2DM. General characteristics of individuals with T2DM according to their age at diagnosis and matched controls were described. Covariate balances were evaluated by standardized differences, which contrasted the group means of covariates in units of the pooled SDs of the comparison groups. A standardized difference of less than 0.1 was considered well balanced [22,23] . We used logistic regression models to estimate odds ratios (ORs) and 95% con dence intervals (CIs) of albuminuria in T2DM patients vs. matched controls in each groups of age at diagnosis, either unadjusted or adjusted for covariates including education (high school education or above), current smoking (yes/no), current drinking (yes/no), history of CVD (yes/no), use of an angiotensin converting enzyme inhibitor (ACEI) or an angiotensin II receptor blocker (ARB) (yes/no), BMI, systolic BP, LDL-c, TG, and eGFR. All the tests were two-tailed, with a p value <0.05 considered to indicate statistical signi cance. Statistical analyses were performed using SAS version 9.4 (SAS Institute Inc, Cary, NC) and R version 3.6.3 (R Project for Statistical Computing, http://www.r-project.org).

Results
General characteristics of the study population are shown in Table 1 for screen-detected T2DM and matched controls and in Table 2 for self-reported T2DM and matched controls. Patients with T2DM were well matched with normal controls for age, sex, and study sites (all standardized differences < 0.001). In addition, median duration of diabetes was 2.7-2.8 years and was well-balanced between groups of age at diagnosis for self-reported T2DM. Compared with matched controls, participants with screen-detected T2DM had higher levels of BMI, systolic BP, diastolic BP, FPG, HbA 1c , TG, TC, LDL-c and lower levels of HDL-c, while participants with self-reported T2DM had increased proportions of current drinkers, history of CVD and use of an ACEI or ARB, and higher levels of BMI, systolic BP, FPG, HbA 1c , TG and lower levels of HDL-c (all standardized differences ≥ 0.1). Furthermore, current smoking, current drinking, FPG, HbA 1c , and TG decreased signi cantly across the groups of age at diagnosis for both screen-detected T2DM and self-reported T2DM (all p values for trend < 0.001).  The distribution of age at T2DM diagnosis after matching is presented in Fig. 2. The mean age at diagnosis was 60.0 years in patients with screen-detected T2DM and 60.9 years in patients with selfreported T2DM. The range of age at diagnosis was 40.0-94.2 years in patients with screen-detected T2DM and 31.3-88.0 years in patients with self-reported T2DM.
Levels of urinary ACR in participants with screen-detected T2DM, self-reported type 2 T2DM, and matched controls in strata of age at T2DM diagnosis are depicted in Fig. 3. ACR levels signi cantly increased across groups of diagnosis age and were signi cantly higher in patients with self-reported T2DM or screen-detected T2DM compared with controls in each stratum of age at diagnosis (all p values for trend < 0.001).
Percentages of participants with albuminuria in T2DM and matched controls, and ORs (95% CIs) of albuminuria in participants with T2DM vs. control in strata of age at diagnosis are shown in Fig. 4.
Percentages of albuminuria were signi cantly higher among T2DM patients compared with normal controls in each stratum of age at diagnosis. Logistic regression analysis revealed signi cantly increased risks of albuminuria in both screen-detected and self-reported T2DM patients compared with their normal controls, respectively in each stratum of age at diagnosis before and after adjustment. Furthermore, ORs of albuminuria in T2DM vs. normal controls decreased with increasing age at T2DM diagnosis. For participants with T2DM diagnosis at an age of < 50, 50-59, 60-69, or ≥ 70 years, the multivariable adjusted risk of albuminuria increased by 81%, 60%, 45%, and 33% for screen-detected T2DM, and 135%, 121%, 90%, and 58% for self-reported T2DM compared with their normal controls, respectively.

Discussion
Using baseline data from a nationwide multicenter community-based cohort, we showed that risks of albuminuria increased signi cantly in T2DM compared with normal controls matched by age, sex, and study sites. The relative risk of albuminuria in type 2 diabetes decreased with increasing age at diagnosis. It was the highest among patients with age at diagnosis < 50 years and the lowest among patients with age at diagnosis ≥ 70 years. Compared with matched controls, the risk of albuminuria was even more markedly elevated in those with T2DM diagnosed at a younger age. This pattern of association was observed for both screen-detected and self-reported T2DM, indicating that age at diagnosis of T2DM was independently associated with the risk of albuminuria.
Previous studies have reported a signi cant association between age at diagnosis of diabetes and risk of cardiovascular complications. Although the absolute rate of CVD is higher in older adults, young adults with early-onset T2DM have a much higher risk of CVD relative to age-matched control subjects in health maintenance organization and nationwide register studies [5,15] . Recent studies also demonstrated a higher risk of microvascular complications such as progressive chronic kidney disease [6] and retinopathy [24] in individuals with early-onset T2DM. However, studies have suggested that patients with early-onset T2DM were at increased risk of cardio-renal complications mainly driven by longer diabetes duration [4,7,10,25] . A recent study from Australia found that diabetes duration was the predominant determinant of end-stage kidney disease (ESKD) and the incidence of ESKD was higher at a given duration in those diagnosed later, which may be partly related to the increased risk at an advancing age [10] . Because the actual age, age at diagnosis, and duration of diabetes are all important risk factors for diabetes complications such as chronic kidney disease and ESKD [20,26] , an appropriately designed study which takes into full consideration the confounding or impact from the actual age and diabetes duration is needed. The current study is one of the few studies which minimize the effect of these factors by matching on age and sex. Furthermore, for diabetes not newly diagnosed, duration of diabetes was also matched. Therefore, our data provide robust evidence regarding the association between age at diabetes diagnosis and albuminuria.
The potential mechanism for the increased risk at younger ages of T2DM diagnosis warrants investigation. In our study, young adults diagnosed with T2DM were more likely to have an unfavorable metabolic pro le compared with normal controls, which has also been reported in previous studies [8,27] .
Compared with older patients, younger patients with T2DM were less likely to have a healthy lifestyle. A younger age of T2DM diagnosis was associated with poor self-management and medication adherence, increased stress, and depressive symptoms [28,29] . In addition, genetic factors might play an important role in young-onset T2DM by affecting beta-cell development [30,31] . Management Algorithm [32] recommend less aggressive risk management in individuals with diabetes aged < 40 years. However, because the risk of microvascular complications such as albuminuria was more signi cantly elevated in individuals with diabetes diagnosed at an earlier age, more intensive monitoring and control of glucose and other risk factors might be needed to prevent microvascular and macrovascular complications in younger people with T2DM [33] .
Strengths of the current study included the large sample from a nationwide multicenter community-based cohort with a wide age range at T2DM diagnosis and the minimized confounding by matching on age, sex, study site, and diabetes duration. Limitations of the current study should also be acknowledged.
First, for T2DM not newly diagnosed, age at diagnosis was self-reported. Recall bias in self-reported T2DM cannot be excluded and might be more prevalent in those with longer diabetes durations.
Therefore, participants with a diabetes duration ≥ 10 years were excluded from the current analysis to minimize recall bias. Second, glucose measures and urinary ACR were tested only once, although centralized measurements were conducted in a CAP accredited laboratory with a standard methodology.
Third, antibodies such as the glutamic acid decarboxylase antibody were not measured to de ne types of diabetes. Nevertheless, T2DM is the predominant type of diabetes in community adults and participants with diabetes diagnosed before age 18 were excluded. Forth, this is a cross-sectional study and prospective studies are needed to demonstrate the interrelationship between age at diagnosis of T2DM and albuminuria. Finally, generalizability of the current ndings to populations with other ethnicities is limited.

Conclusions
A younger age at diagnosis of T2DM is associated with a more signi cantly elevated risk of albuminuria than an older age at diagnosis in Chinese adults. Therefore, age at diagnosis of T2DM could serve as an important risk stratifying factor in the screening, prevention, and management of this chronic condition [34] . Evidence from intervention studies in T2DM patients diagnosed at different ages is warranted to minimize risks of microvascular and macrovascular complications.