Discussion
In a certain proportion of patients with diabetes, an eGFR decline precedes albuminuria. A new concept called DKD is advocated that also included the albuminuria-free patients. The prevalence of DKD in a large-scale database that integrated cohorts in Japan was indicated, and we revealed the risk factors of three subgroups of DKD. Additionally, we performed a longitudinal eGFR assessment to detect early decliners and the risk identification of early decliner using a trajectory analysis. Since the analysis has the advantage of proper classification by patients’ eGFR variation consists of several different points, an exact assessment of eGFR decline rate of every group can be achieved. These findings must help the early detection of DKD and early decliners so that medical intervention can be provided at an early stage for high-risk patients.
In our multicenter study, approximately half of the patients with diabetes met the DKD criteria. Next, we divided patients with DKD into three groups to explore the risk factors by DKD category. Regarding atypical patients with low eGFR but normoalbuminuria, the prevalence was 23% in all patients with DKD. The status of normoalbuminuric patients with DKD was more strictly controlled with lifestyle-related disease than patients with albuminuria, and their eGFR decline rates were the slowest among all participant groups.
In a past study of the United States national observance of DKD,16 the prevalence of DKD was 26%. Although the prevalence of DKD, a lower eGFR (<60 mL/min/1.73 m²) (14%), and albuminuria (15%) were lower than our findings, they were based on the doorstep screener of the general population; meanwhile, our data were provided by foundation hospitals including inpatients. Compared with past studies of DKD,5 16 17 our participants showed better control of BMI and serum lipid status. On the other hand, glycemic control estimated by HbA1c and blood pressure (BP) were almost equivalent to those of previous studies. The patients were prescribed antidyslipidemic drugs (52%) or antihypertensive drugs (56%) at a frequency almost equal to those of patients with diabetes in a study from the USA.16 The suggestion that patients with diabetes in our facilities received therapeutic intervention for metabolic syndrome was almost equivalent to patients in the USA.
Although smaller percentage of patients with type 1 diabetes corresponded to DKD compared with type 2 diabetes, it cannot be generalized. A possible reason for the result is that patients with type 1 diabetes were followed up from an earlier stage of the disease compared with patients with type 2 diabetes.
For the decreased eGFR and normoalbuminuric category, clinical variables like HbA1c, BP, TG, and both HDL and LDL cholesterol levels were almost within the target range recommended by the Kidney Disease Improving Global Outcomes, American Diabetes Association, or the Kidney Disease Outcomes Quality Initiative guidelines14 18 19: HbA1c ≤7.0% (53 mmol/mol) (aiming to prevent complications), BP <140/90 mm Hg, TG <150 mg/dL, HDL ≥40 mg/dL, and LDL <100 mg/dL. Since those variables were not different from those of patients with no DKD, the patients with decreased eGFR were at lower risk of microangiopathy such as diabetic retinopathy. This is compatible with the fact that the prevalence of diabetes retinopathy in the decreased eGFR group was not different from that in the patients with no DKD. Their eGFR decline rates were slower than those of other patients with diabetes. It is deducible from the fact that they had a lower ratio of complications like diabetes retinopathy than the other DKD group, and they were prescribed antihypertensives at a significantly higher frequency than the patients with no DKD; 45% of the patients in that category were taking RAAS inhibitors, which might suppress the onset of albuminuria.20 They would benefit from the medication. Regarding UA, similar results were reported that the serum UA level predicted the decline of eGFR independently of other risk factors.21 Although the international criteria of metabolic syndrome22 does not include UA level, those patients with a higher UA level were much more likely to have metabolic syndrome.23 24 It remains unknown why an elevated UA predicts an eGFR decline.
The group with albuminuria but a normal eGFR had a significantly higher ratio of a previous history of diabetic retinopathy and significantly higher HbA1c and SBP than patients with no DKD. This finding is compatible with disease phenotypes of microangiopathy as an early stage of classical diabetic nephropathy. Diabetic retinopathy is considered a risk factor of macroangiopathy according to the changes in vascular conductance and calcification.25 In particular, this group showed a higher prevalence of retinopathy than the other groups, suggesting a longer history of diabetes.5
These data implicated that a better past glycemic control, BP, and lipid condition might suppress albuminuria or microangiopathies like diabetic retinopathy. Strict long-term glycemic control reportedly suppressed the pathogenesis and progress of albuminuria in cases of type 1 diabetes.26 27 Similarly, in patients with type 2 diabetes, strict glycemic control reduced the risk of microangiopathy and albuminuria.4 From recent studies, multidisciplinary therapy including glycemic, BP, and lipid control for type 2 diabetes significantly suppressed the pathogenesis and progress of albuminuria and decreased renal events like creatinine doubling28 or ESRD.29 Our results similarly indicated that good control of glycemic level, BP, and lipid condition might contribute to reducing renal damage and decreasing complications like microangiopathy. Our findings will be of great help in the early detection of patients with high-risk DKD and aid with early therapeutic intervention to avoid renal disorder.
Concerning the longitudinal analysis, we performed a trajectory analysis that could grasp yearly eGFR variations and classified the eGFR into three different patterns. An eGFR rapid lowering subgroup with eGFR normal baseline patients (14%) was identified as early decliners. Their eGFR would decrease to −29 mL/min/1.73 m² within 4 years, and soon they may face the crisis of renal insufficiency. Their baseline factors such as older age, higher SBP, and ACR contributed to the eGFR decline. The same trend was seen when adding LDL cholesterol as a variable instead of total cholesterol (online supplementary table S9).
A higher ACR level reportedly predicts an eGFR decline and ESRD.30 31
Regarding hypertension, despite several trials showing that strict BP control in the early stage of diabetes suppressed albuminuria onset or progression, there was no significant difference in eGFR decline within the follow-up term (4–8 years).32 33
However, an observational study of patients with hypertension but normal renal function showed that patients with poorly managed BP (≥160/95 mm Hg) experienced a rapid decline of eGFR compared with those with better controlled BP (<140/90 mm Hg).34 This finding is consistent with previous observations that poor BP control accelerates renal dysfunction.
Although a strong correlation existed between albuminuria and eGFR decline in our study, the eGFR reduction occurred despite baseline normoalbuminuria in a certain proportion of the patients with diabetes as seen in a previous report.35
These results will help us recognize early decliners by observing eGFR decline rate for approximately 3 years and risk factors such as a high SBP, high ACR, and older age to allow for early intervention in high-risk patients.
Although we excluded patients whose eGFR levels were considered too high, the definition of hyperfiltration remains obscure. Patients who underwent hyperfiltration reportedly had a high probability of a rapid eGFR decline.36 37 In this study, we set our threshold of hyperfiltration as 120 mL/min/1.73 m2 37 ; however, regarding type 2 diabetes, it was difficult to completely exclude patients with hyperfiltration only by eGFR value because hyperfiltration would be hidden by age-related changes in eGFR in patients >40 years of age.36
This study has several limitations. First, due to its cross-sectional observation design based on multicenter data, we could not avoid sampling bias. Because the baseline variables were not unified in this study, we could not elucidate the relevance of participant diabetes duration or complications other than renal damage, such as cardiovascular disorders, neuropathy, and malignancy. Second, we could not exclude various causes of renal damage except for diabetes since not all participants underwent a renal biopsy. It is possible that not all normoalbuminuric DKD can be actually attributed to DKD itself but can be a chronic kidney disease in patient with diabetes, which could be of multifactorial etiology like hypertension and atherosclerosis.
Moreover, the risk factors of DKD were calculated at only a moment in time; hence long-term follow-up in the future would reveal the characteristics of patients with DKD. In addition, we did not have enough data of prescription medicine including RAAS inhibitors to analyze the risk factors of DKD or early decliners although the data may affect the progression of renal damage.
However, our research included a large number and various types of patients with diabetes and analyzed the prevalence and risk factors of both DKD and early decliners using the same database.
In conclusion, here we revealed the epidemiology of DKD and identified past poor glycemic, SBP, and lipid control as risk factors of progression of renal damage and microangiopathy in patients with DKD. It is best to identify patients with DKD earlier and intervene in cases of hyperglycemia and other lifestyle problems in the early phase of diabetes.