PubMed, Embase, and Cochrane were searched systematically for articles relating to NADKD in patients with type 2 diabetes. The search term was “(diabetic nephropathy OR diabetic kidney disease OR diabetic glomerulopathy OR chronic kidney disease in type 2 diabetes OR renal dysfunction in type 2 diabetes OR impaired renal function in type 2 diabetes OR renal impairment in type 2 diabetes OR renal insufficiency in type 2 diabetes OR decline in renal function in type 2 diabetes) AND (non-proteinuria OR non proteinuria OR silent OR without protein OR non-albuminuria OR non albuminuria OR without albuminuria OR normoalbuminuria OR normoproteinuria)”. Limitations were set to retrieve “clinical trial” or “clinical study” (detailed search strategies are shown in the appendix file). The cutoff time for retrieval was November 5, 2021. No time restriction was applied in our search. The results were limited to full-text articles published in English. All of the reference lists from the included studies and relevant reviews were screened manually for additional eligible studies.

Two authors independently screened the titles and abstracts for inclusion of all the potential studies identified before. The full text was retrieved if necessary. All clinical studies were included in this evaluation. The literature quality evaluation was conducted by means of the Newcastle–Ottawa Quality Assessment Scale (NOS) for cohort studies and by means of the American Agency for Healthcare Research and Quality (AHRQ) methodology checklist for cross-sectional studies (19). Two authors independently evaluated the results. The NOS included eight items, with a total score of nine, and focused on three areas, including participant selection, comparability of study groups, and ascertainment of exposure. The evaluation tool of the AHRQ included eleven items (19) (details shown in Appendix Tables 2 and 3 ).

Data extraction was conducted by the two authors using a standardized data collection form. The following information was obtained from the included studies: type of study, country, first author’s name, year of publication, target population, definition of nonalbuminuria, research features, and all data related to clinical characteristics of patients and follow-up outcomes.

We systematically analyzed all the parameters of prevalence, clinical characteristics, pathology, and prognosis of patients with NADKD. When parameters were reported by only a few studies, descriptive analysis was performed. However, if the data were sufficient, meta-analysis was performed using RevMan 5.3 and STATA version 16 (Stata Corporation, College Station, TX, USA). The standardized mean difference (SMD) and risk ratio (RR) were calculated using random-effects models for the presence of heterogeneity (I 2 > 50%, p < 0.05). A p-value < 0.05 was considered statistically significant.

In 1992, Lane et al. identified eight female normoalbuminuric patients with type 1 diabetes who had decreased renal function and diabetic renal structural lesions (40). In 1994, Tsalamandris reported that a subgroup of diabetic patients had renal function decline without significant albuminuria independent of diabetes type (41). This subtype among adults with type 2 diabetes was highlighted in 2003 by Kramer et al. (20). Over the past two decades, the prevalence of albuminuria has decreased; however, the prevalence of eGFR decline has increased, as reported by Kume et al. using 1996, 2001, 2006, and 2004 cohort data from Japanese serial studies conducted at Shiga University of Medical Science (42). NADKD in type 2 diabetes has been increasingly recognized (43–45). In our study, after excluding 5 studies with inconsistent DKD definitions (8, 25, 30, 32, 39) and 2 case–control studies (28, 29), using the available data (missing data were not included), the overall prevalence of NADKD among patients with type 2 diabetes and renal insufficiency was 45.6%; among patients with DKD (either with albuminuria or renal insufficiency), the prevalence of NADKD was 24.7%, and among all the patients with type 2 diabetes, it was 8.4%, which means that almost half of patients with DKD and decreased eGFR did not develop albuminuria.

Meta-analysis was used to analyze the clinical characteristics of the patients. Nonalbuminuria was defined as ACR < 17–30 mg/g (albuminuria–creatinine ratio), AER < 20 µg/min (albumin excretion rate), or UAE < 30 mg/24 h (urinary albumin excretion for 24 h). Decreased eGFR was defined as eGFR < 60 ml/min/m2. There were 5 studies with inconsistent definitions (2011 Rajiv defined albuminuria as ACR < 220 mg/g; 2019 Masayuki and 2021 Tsutomu defined albuminuria as ACR < 300 mg/g; 2013 Mauro only included patients with GFR < 30 ml/min/m2, and 2016 Ivo defined decreased eGFR as <75 ml/min/m2 (as shown in Table 2 ) (8, 9, 30, 32, 39). After these 5 studies were excluded, 18 studies reported the characteristics of NADKD, and they were included in the meta-analysis (5, 7, 10, 12, 15–18, 22–24, 26, 31, 34–38) (as shown in Additional Table 2 ). Missing data were not included in the analysis. Compared with patients with albuminuria DKD, patients with NADKD were older (MD = 1.04 years old, 95% CI [0.52, 1.57], p < 0.05); were more often women (Male RR = 0.74, 95% CI [0.68, 0.81], p < 0.05); had shorter diabetes duration (MD = −2.9 years, 95% CI [−3.63, −2.18], p < 0.05), lower HbA1c levels (MD = −0.034%, 95% CI [−0.42, −0.25], p < 0.05), and lower blood pressure (systolic blood pressure MD = −6.21 mmHg, 95% CI [−9.41, −3.0], p < 0.05; diastolic blood pressure MD = −1.27 mmHg, 95% CI [−2.15, 4.0], p < 0.05); less frequently experienced diabetic retinopathy (RR = 0.58, 95% CI [0.51, 0.67], p < 0.05); and less frequently used renin–angiotensin–aldosterone system (RAAS) inhibitors (RR = 0.76, 95% CI [0.65, 0.89], p < 0.05). However, lipidemia, body mass index (BMI), smoking frequency, and history of CVD were not different between these groups (as shown in Figure 2 ).

The comparison between patients with albuminuric DKD without renal insufficiency and patients with NADKD revealed similar results in most respects, but BMI and low-density lipoprotein (LDL) seemed to be lower in the latter, and the frequency of retinopathy and the use of RAAS inhibitors were not different. Finally, when compared with diabetic patients without albuminuria or renal insufficiency, in patients with NADKD, the duration of diabetes was much longer, SBP and triglyceride levels were significantly higher, and the presence of retinopathy, CVD history, and the use of RAAS inhibitors were more frequent (as shown in Figure 2 ; details are shown in Appendix Table 1 and appendix forest plots).

Heterogeneity test results showed that heterogeneity existed in all the meta-analyses, with I 2 fluctuating between 73% and 100% and a p-value of less than 0.05. However, further subgroup analyses according to study quality assessment (whether assessed as high quality), study type (cross-sectional or cohort study), study sample (whether the sample of the NADKD group included fewer than 50 cases), and patient sources (whether from hospital sources) showed that the results were robust. No source of heterogeneity was found. Moreover, in sensitivity analysis, when each study was excluded one by one, the results did not change (as shown in Appendix Directory 2).

Since renal biopsy is not always applicable, available data on histological changes in NADKD are limited. Specific renal pathological lesions with diabetes were mostly studied in patients with type 1 diabetes, including thickening of the glomerular basement membrane, mesangial matrix expansion, nodular lesions, and glomerular sclerosis and arteriolar hyalinosis (1). In contrast, this symptom has been less well investigated in patients with type 2 diabetes (DM2). Classic nephropathy lesions in patients with type 2 diabetes are always less severe because the picture of DKD in type 2 diabetes is confounded by other contributors, such as aging, obesity, hypertension, insulin resistance, and vascular disease (1, 46). Few studies have investigated the renal pathological changes in NADKD among patients with DM2; to our knowledge, only four are available. In 2013, Budhiraja et al. collected renal tissue specimens from 10 patients with DM2 and normoproteinuria who underwent nephrectomies because of kidney cancer and first demonstrated that classical diabetic renal pathological changes could occur in the absence of proteinuria (28). Subsequently, the structural lesions of 8 patients with NADKD, 8 with microalbuminuria, and 17 with macroalbuminuria were compared by Ekinci et al., and the results showed that typical histopathological changes in diabetic nephropathy (DN) occurred in patients with DM2 but less frequently than in those with elevated albuminuria (29). In 2016, one study from an autopsy of individuals identified 20 patients with NADKD and demonstrated again that typical DN lesions may develop before the onset of albuminuria and other clinical findings (33). In 2019, Yamanouchi et al. retrospectively assessed 526 patients with type 2 diabetes and DKD [eGFR < 60 ml/min/1.73 m (2)] who had undergone renal biopsy from Japan’s nationwide multicenter renal biopsy registry and showed that normal or near-normal renal structure was the most common (62%) in the patients with NADKD, whereas typical DKD was the most prevalent (66%) in the proteinuric DKD group (8). Overall, NADKD was reported to tend to be associated with more advanced tubulointerstitial and vascular lesions but mild typical glomerular lesions compared with those in patients with albuminuria DKD (7, 29, 47).

First, the phenomenon may be modified by the development of renoprotective treatments, such as renin–angiotensin system inhibitors (RASi), lipid-lowering drugs, and sodium glucose cotransporter 2 inhibitors (SGLT2i), which reduce urinary albuminuria. However, RASi was not an independent risk factor for NADKD in multivariate analysis (26), and when excluding users of RASi or lipid-lowering drugs, the prevalence and general characteristics of NADKD resembled those of all patients (12, 23, 39, 48). Second, NADKD may be an early stage of classic DKD. As shown in the present systematic review, NADKD was associated with a shorter duration of diabetes and a lower prevalence of retinopathy. Moreover, in An et al.´s study, the prevalence of normal albuminuria decreased progressively with an increase in the diabetic duration, and among the 30 diabetic patients with renal insufficiency who had a diabetic duration longer than 21 years, none had normal albuminuria (24). Thus, it is possible that at least a proportion of patients with NADKD may be at an early stage of DKD. Third, the prevailing underlying pathology of patients with NADKD might be macroangiopathy instead of microangiopathy. In the present study, patients with NADKD were older than other diabetic patients, less frequently had concurrent diabetic retinopathy, and tended to have lower HbA1c levels and LDL and lower blood pressure. On the other hand, the annual eGFR decline tended to be slower in patients with NADKD and was even slower than in patients with nonalbuminuria and non-DKD in Yokoyama et al.’s study (16). However, the risk of CVD and all-cause mortality was higher than that in patients with nonalbuminuria and non-DKD and tended to be comparable to that in patients with albuminuria and normal renal function. These findings suggest that renal impairment in NADKD was not mainly caused by hyperglycemia or microangiopathy, and it is possible that aging, arteriosclerosis, and macroangiopathy might contribute more to NADKD. In accordance, Kume et al. reported that eGFR decline without albuminuria was associated with age-dependent arterial stiffness (42). Kim et al. demonstrated that arterial stiffness (measured by baPWV) was strongly related to albuminuria in patients with type 2 diabetes (34). However, MacIsaac et al. reported that the resistance index of the intrarenal arteries measured by duplex Doppler ultrasound, which reflects intrarenal arteriosclerosis elevation, was not different between patients with NADKD and those with albuminuric DKD (22). Fourth, genetic susceptibility might also play a role in the development of NADKD. Bhalla et al. reported that non-Hispanic white individuals are more likely to have nonproteinuric DKD than individuals of other races (27). A Japanese study reported that DKD in the absence of albuminuria was associated with polymorphisms of the protein kinase C-β gene (49). Yagil et al. identified an experimental model reminiscent of human NADKD and reported that the absence of proteinuria in Cohen diabetic sensitive rats with declined renal function may be genetically determined (50). Fifth, the divergence of pathologic changes may have played a role. The different clinical characteristics of NADKD mean that the underlying renal pathology may be different. The pathological characteristics of NADKD were different from those of albuminuria DKD. Klessens et al. reported that the presence of albuminuria was correlated with interstitial fibrosis tubular atrophy (IFTA) rather than DN, and it is possible that the onset of albuminuria may only occur after interstitial and tubular damages are severe enough to lead to insufficient reabsorption of the leaked protein from the glomerular filtrate (33). As a result, renal insufficiency due to glomerular damage may occur preceding the onset of albuminuria. Finally, the preponderance of the disease in female patients indicated that the pathogenesis of NADKD might be affected by estrogen. NADKD may be caused by multipathogenic factors. Superimposed factors such as cholesterol emboli, hypertensive nephrosclerosis, episodes of acute kidney injury, and concomitant additional renal disease have all been suggested to explain this situation.

Generally, DKD is considered to initiate with an increase in urinary albumin excretion, then progress to macroalbuminuria, and finally develop to renal insufficiency (16). However, this process has been challenged since trajectories different from this classical phenotype have been increasingly reported. In recent decades, the phenotype of NADKD has been increasingly recognized (14). Patients with NADKD tend to have a lower eGFR decline rate, which is even slower than that in diabetic patients without DKD (16), and they have a lower risk of adverse end-stage outcomes than patients with albuminuric DKD. The situation of this phenotype was relatively stable compared with those of other phenotypes. However, these patients still have a higher risk of ESRD, CVD, or death than diabetic patients without DKD, which indicates that a subset of patients might progress gradually to the end stage without albuminuria. In addition, it is possible that some patients might experience a rapid decline in eGFR (defined as ≥5 ml/min/1.73 m²/year), and the reported risk factors include high GFR, high systolic blood pressure, prior CVD, and albuminuria ². Krolewski et al. observed the development of 286 patients with type 1 diabetes and normoalbuminuria for 4–10 years and showed that 10% of them experienced a rapid eGFR decline (at least 3.3%/year) preceding the onset of albuminuria. Of note, most patients in his study had normal renal function (51). Thus, although a rapid eGFR decline was common in diabetic patients with albuminuria, the development of albuminuria was not indispensable for it. Alternatively, some patients with NADKD may develop albuminuric DKD, as we discussed above.

Over the past two decades, among diabetic patients, although the prevalence of albuminuria has decreased, the prevalence of eGFR <60 ml/min/1.73 m² is increasing (16), which means that the natural history of DKD may have changed worldwide. NADKD has already become the predominant phenotype of DKD with renal insufficiency, and its prevalence may increase in the future.

Although studies associated with NADKD have increased in recent decades, the understanding of NADKD is still far from sufficient. Microalbuminuria is always considered the first clinical sign of NADKD. However, the discordance between changes in albuminuria and GFR indicated that albuminuria cannot meet the diagnostic demand for early DKD in clinical practice. Available studies have reported the use of some new markers, ultrasound technology (such as the renal-resistive index), or other methods in the early diagnosis of DKD among patients with nonalbuminuria and normal renal function (52, 53). Cytokines and growth factors (IL-17A and macrophage inflammatory protein 1α) related to mild inflammation and fibrosis were reported to be associated with NADKD (54). However, no other studies have specifically examined the diagnosis of NADKD. Moreover, since NADKD is different from albuminuria DKD, the therapeutic strategies should be different. During the past two decades, renoprotective drugs have mostly aimed at reducing albuminuria. The prevalence of albuminuria in patients with DKD decreased over the past decade, which indicated that the treatment was effective (42). However, patients with NADKD are usually excluded from relevant clinical trials; thus, whether these strategies are sufficient to combat NADKD is unknown. At present, there is no specific treatment for NADKD. Therefore, further studies are needed to provide an additional understanding of NADKD.

To our knowledge, our study is the first systematic review associated with NADKD. Thirty-one available studies from different countries were included for analysis, among which 20 studies were evaluated as high quality and 9 were evaluated as medium quality. However, there are some limitations. First, most studies are not based on biopsy for the diagnosis of DKD and thus may include some other kidney diseases. However, studies based on biopsy showed that the incidence of nondiabetic nephropathy was relatively low (29, 33). Second, most studies used eGFR, which might not accurately reflect the real GFR. However, it has been reported that the difference between the isotopic determination of GFR and the MDRD equation estimation of GFR is slight and not significant (17). Third, patients in most studies had only one assessment of albuminuria without confirmation. Fourth, patient characteristics, laboratory methods, and study designs were different among the included studies. Fifth, there might be divergences among different races and countries, but we cannot determine them. Thus, our results should be interpreted with caution due to the above limitations and the heterogeneity in the meta-analysis.