Introduction

Diabetes is one of the most common chronic diseases affecting children and adolescents. There are currently more than 190,000 people younger than 20 years of age with diabetes in the USA [1] and this number is projected to increase two-fold or more by 2050 [2]. Historically, type 1 diabetes has been the predominant form of disease in children and adolescents. However, over the past two decades, the rise in childhood obesity has led to an increasing incidence of type 2 diabetes among children and adolescents, which now parallels and at times exceeds that of type 1 diabetes among minority youth, particularly after the age of 15 [1].

It has long been known that diabetes is associated with a significant increase in mortality, largely as a result of its long-term complications. More recently, this excess mortality has been found to be concentrated in the subset of people with diabetes who develop kidney disease, in both type 1 [3, 4] and type 2 [5] diabetes. These observations highlight the importance of diabetic kidney disease (DKD) at least as a marker of a population at highest risk of mortality and perhaps as a risk factor directly contributing to excess mortality.

While more severe stages of DKD take decades to develop and are thus rarely observed in childhood, kidney biopsies as early as 1.5–5 years after diabetes onset show structural changes characteristic of DKD in both adults and children [6–8]. This suggests that the DKD course begins soon after diabetes onset and that this early interval may provide a critical time-frame for detection and intervention in the disease course, supporting intensive monitoring and modification of risk factors in children and adolescents. Perhaps even more so than in adults, our current tools for early diagnosis of DKD in children and adolescents are few and flawed. Nonetheless, the heavy impact of childhood diabetes on morbidity and mortality later in life mandates our full use of all available tools and resources with a nuanced understanding of their strengths and limitations as well as a renewed effort toward the development of new diagnostic tools and therapies.

This review will discuss the natural history and risk factors for the development of DKD, the structural changes observed in DKD, and the pathophysiological mechanisms causing those alterations. We will also discuss the differences in natural history and outcomes between type 1 and type 2 diabetes.

Natural history

The natural history of DKD was classically described as an initial and progressive rise in urine albumin excretion, followed by progressive glomerular filtration rate (GFR) loss and eventual development of end-stage renal disease (ESRD) over several decades (Fig. 1). Microalbuminuria, defined as a urine albumin excretion of 30–299 mg/day (or an albumin to creatinine ratio of 30–299 mg/g creatinine in random samples) in at least 2 of the 3 measurements, is the earliest sign of DKD and occurs in 26 % of children and adolescents after 10 years and in 51 % after 19 years of diabetes [9]. In the classic DKD presentation, once microalbuminuria develops, urine albumin excretion continues to rise, particularly in presence of uncontrolled risk factors. Macroalbuminuria, defined as a urine albumin excretion ≥300 mg/day (or albumin to creatinine ratio ≥300 mg/g creatinine), heralds the onset of overt DKD and is thought to inexorably lead to impaired GFR (defined as an estimated GFR <60 ml/min/1.73 m2) and eventually ESRD (Fig. 1).

Fig. 1
figure 1

The classic course of the natural history of diabetic kidney disease (DKD). Classic DKD is thought to begin with the rise in urine albumin excretion, initially to microalbuminuria (lower dashed red line), then proceeding to macroalbuminuria (upper dashed red line), with the glomerular filtration rate (GFR) dropping only after macroalbuminuria

Full size image

More recent work has refined our understanding of the DKD natural history under current standards of care. First, longitudinal studies of type 1 diabetes have shown that while microalbuminuria does raise the odds of future DKD development, more adults [10–12] and children [13, 14] with microalbuminuria regress to normoalbuminuria (∼30–40 %) than progress to more advanced stages of DKD. Perhaps as a result, microalbuminuria is not associated with an increased rate of GFR loss [15], suggesting that rather than being a committed marker of specific and irreversible DKD, microalbuminuria might represent reversible endothelial injury. Microalbuminuria regression to normoalbuminuria is more common with improved glycemic and blood pressure control, but is independent of the renin–angiotensin system (RAS) inhibition both in adults [12] and in children [13, 14]. As such, the current KDOQI guidelines recommend treatment with RAS inhibitors only if microalbuminuria co-exists with other DKD risk factors (e.g., hypertension, diabetic retinopathy, dyslipidemia, etc.) [16].

Second, while the classic presentation of macroalbuminuria preceding the impairment of GFR is observed in a majority of cases in type 1 diabetes, in a sizable minority GFR loss occurs either after microalbuminuria (16 %) or even in the absence of any albuminuria (24 %) [15]. Third, while macroalbuminuria is associated with a 50-fold greater risk of progression to impaired GFR [17], 60 % of people with macroalbuminuria maintained an estimated GFR >60 ml/min/1.73 m2, even after 15 years’ follow-up (Afkarian et al., submitted). These observations suggested that macroalbuminuria was a sign of substantial renal injury, but that it did not set an inexorable course to impaired GFR or ESRD in all cases. In addition, significant renal injury could occur in the absence of macro- or even microalbuminuria.

Interpretation of albuminuria in children and adolescents is contingent on two additional considerations. First, the vast majority (75–95 %) of proteinuria in children and adolescents is due to benign causes such as transient and orthostatic proteinuria.[18–22]. Transient proteinuria may be associated with exercise, fever, cold, stress, dehydration or seizure and is reported to occur in as many as 50–74 % of cases of proteinuria observed in children and adolescents [18, 20]. Orthostatic proteinuria, another etiology of benign childhood proteinuria, is reported in 6–20 % [23] of healthy children and adolescents and leads to no adverse outcomes after up to 50 years of follow-up [24, 25]. Luckily, transient and orthostatic proteinuria can be readily identified by repeat measurement of urine protein excretion in the absence of the conditions causing transient proteinuria and in a first-void urine sample to exclude orthostatic proteinuria [26].

The second challenge in the accurate interpretation of albuminuria in children is the likely misclassification of albuminuria status by the application of a single threshold of 30 mg/day (or 30 mg/g of creatinine) in children of different ages, size, gender, and race. In children and adolescents, excretion of creatinine [27, 28] and albumin [27, 29] increases with age. Creatinine excretion is higher in boys [27, 30] and albumin excretion is higher in African–Americans compared with Caucasians [31]. Using a single threshold of 30 mg/g in spot urine samples is likely to overestimate albumin excretion in younger, smaller girls whose daily creatinine excretion is more likely to be much less than 1 g.

Structural changes and their correlation with functional parameters

In biopsies performed in both adults and children with type 1 diabetes, the first observed alteration is the thickening of the glomerular [6] and tubular [8] basement membrane (GBM, TBM) 1.5–2.5 years after diabetes onset. Mesangial matrix expansion with an increase in fractional volume of the mesangial matrix typically follows 5–7 years after diabetes onset [7]. Interstitial expansion is also observed, initially due to an increase in the cellular component and later on due to accumulation of fibrillar collagen in advanced DKD [32]. These various lesions progress at different rates within and between different patients with type 1 diabetes. However, by the time advanced DKD with renal insufficiency sets in, all patients display marked mesangial expansion and GBM thickening as well as interstitial expansion and fibrosis, tubular atrophy [33] and glomerulosclerosis [34]. Similar lesions are observed in children and adults with type 1 diabetes and normoalbuminuric DKD [35]. Diabetic retinopathy (DR) is nearly universal at the time of macroalbuminuria [36] and its severity is proportional to the degree of structural abnormalities in early DKD in type 1 diabetes [37].

Renal structural lesions are much more heterogeneous in type 2 diabetes [38, 39]. Adults with type 2 diabetes and micro- or macroalbuminuria, but GFR >60 ml/min/1.73 m2 display typical DKD lesions in only 30 % and 50 % of cases respectively, while the remainder show either atypical pathological features (little or no diabetic glomerulopathy, with chronic tubulointerstitial fibrosis, arteriolar hyalinosis, and glomerular sclerosis) or no significant pathological features [39]. Interestingly, some degree of DR is present in all cases with typical DKD pathological features, while none of the cases with atypical pathology or non-significant injury have proliferative DR [39]. On the other hand, in type 2 diabetic patients with micro- or macroalbuminuria and impaired GFR (<60 ml/min/1.73 m2), typical glomerular lesions are reported in 83 % and 100 %, respectively. However, less than 40 % of those with impaired GFR and normoalbuminuria show typical DKD glomerular lesions, while a majority show either severe renovascular and/or tubulointerstitial changes or no identified cause of kidney disease [40].

In type 1 diabetes, mesangial expansion correlates significantly and inversely with GFR, presumably because of consequent reduction in glomerular capillary filtration surface area [7], a feature that correlates strongly with GFR loss [41]. Mesangial expansion also correlates strongly with albuminuria [7, 42] and hypertension [43]. GBM thickening correlates closely with albuminuria, but less so with GFR or hypertension [44]. Interstitial expansion and global sclerosis correlate with albuminuria and hypertension and inversely correlated with GFR [7]. These associations are similar, though less precise, in type 2 diabetes, with albuminuria correlating closely with both GBM width and mesangial expansion, while risk of progressive GFR loss in type 2 patients with albuminuria correlates significantly with the extent of mesangial matrix expansion [45]. Notably, unlike the structural studies in type 1 diabetes, all of those in type 2 diabetes were conducted in adults. In the light of rising incidence and poor prognosis of type 2 diabetes in children and adolescents, as discussed below, similar structural studies in this population have the potential to add much needed depth to our understanding of the disease process.

Pathophysiology

Animal models have shown dysregulation of many pathways in DKD and these findings have largely been corroborated in human studies. The Diabetes Control and Complications Trial (DCCT) demonstrated that hyperglycemia is the primary trigger for dysregulation of all these pathways. Hyperglycemia leads to end-organ damage in several tissues, perhaps particularly those that are not able to down-regulate glucose entry in hyperglycemia [46]. In 2001, Brownlee presented a unifying hypothesis [47] that aimed to connect all the implicated pathogenic pathways to elevated intracellular glucose concentration (Fig. 2). Drawing on several lines of evidence, including a hyperglycemia-induced increase in intracellular superoxide, the Brownlee hypothesis posited that elevated intracellular superoxide inhibits a key glycolytic enzyme, glyceraldehyde 3-phosphate dehydrogenase (GAPDH), thus blocking metabolism and disposal of excess intracellular glucose. The resulting accumulation of intracellular glucose and its glycolytic intermediates, fructose-6-phosphate and glyceraldehyde-3-phosphate, then feed into and induce four primary pathways: the polyol, hexosamine, and protein kinase C pathways, and the non-enzymatic generation of advanced glycation end-products (AGE). These pathways in turn activate many other pathogenic signaling cascades, most notably the TGFβ and RAS pathways, leading to distal effects such as excessive production and accumulation of extracellular matrix components, which give rise to the DKD structural and functional alterations discussed in previous sections, e.g., expansion of the mesangial matrix and thickening of the GBM.

Fig. 2
figure 2

Schematic of the unifying theory connecting hyperglycemia to distal pathophysiological mechanisms via elevated intracellular superoxide. Hyperglycemia increases mitochondrial superoxide generation, which activates poly (ADP ribose) polymerase (PARP). PARP inhibits glyceraldehyde 3-phosphate dehydrogenase (GAPDH) by ADP-ribosylating it. Inhibition of GAPDH leads to a build-up of preceding glycolytic intermediates, which feed into and activate the four primary diabetic kidney disease (DKD) pathophysiological pathways, the polyol (blue), the hexosamine (green), the advanced glycation end product (AGE; pink), and protein kinase C (PKC; orange). These pathways in turn trigger several distal cascades, upregulating expression of effector proteins such as components of the renin–angiotensin system and transforming growth factor-beta (TGF-β). ADP adenosine diphosphate

Full size image

While hyperglycemia is widely believed to be the inciting pathogenic stimulus, recently controversy has surrounded the role of intracellular superoxide in triggering the subsequent pathophysiological mechanisms. The hyperglycemia-induced increase in intracellular superoxide is at the core of the Brownlee hypothesis. However, recent observations of reduced superoxide levels in kidneys of DKD animal models [48] have led to an alternative hypothesis. This hypothesis suggests that hyperglycemia reduces superoxide generation which in turn decreases activity of 5′ AMP-activated protein kinase (AMPK), a major energy-sensing enzyme, and PGC1α, the master regulator of mitochondrial biogenesis. The resulting reduction in mitochondrial number and function promotes DKD progression. Pharmaceutical activation of AMPK and PGC1α were shown to increase kidney superoxide content and diminish activity of the distal pathogenic mechanisms, manifested in the reduction of albuminuria and lower expression of fibronectin, collagen, and TGF-β in the kidneys in several animal models of DKD [48]. Resolution of the controversy surrounding the effect of hyperglycemia on intracellular superoxide and its role in DKD pathogenesis await further studies.

Another aspect worthy of further study relates to the scarcity of novel diagnostic and therapeutic tools in the face of acknowledged advances in our understanding of the underlying DKD pathophysiology based on experimental animal models. Since the discovery of the beneficial effect of intensive diabetes treatment and RAS inhibition on DKD, no new therapies have proven successful. Most recently, dual RAS inhibition and activation of the Nrf antioxidant pathway failed to have an impact on DKD progression, joining a cadre of prior failed trials using inhibitors of advanced glycation end-products, protein kinase C, and aldose reductase. Given the strong association of DKD with mortality, understanding, and remedying, the causes of this delay in translation from pathophysiology to novel diagnostic and therapeutic options would have a significant impact on public health.

Risk factors

The seminal DCCT/EDIC study included 195 participants aged 13–17 years old at enrollment, among whom intensive diabetes treatment was associated with a 54 % reduction in microalbuminuria and this benefit persisted during the EDIC study [49]. This established hyperglycemia, measured by hemoglobin A1c, as the predominant risk factor for DKD. In addition, female gender, diabetes duration, hypertension [50, 51], high normal urine albumin excretion [52] elevated LDL cholesterol, triglycerides [53], high body mass index [54], and smoking [55] have been reported to be risk factors for the development of microalbuminuria in children and adolescents. The role of pre-pubertal diabetes duration in the development of microalbuminuria has been the subject of some debate, but overall the total disease duration, and not the duration before or after puberty, appears to determine cumulative microalbuminuria risk [50, 56] and DKD structural alterations in the kidney [57]. Hemoglobin A1c [9], male sex [58], smoking [59], baseline albumin excretion [9], and serum uric acid [60] have been reported to be risk factors for progression to macroalbuminuria in children and adolescents. Given the significant heritability observed for DKD [61], a family history of kidney disease is also a significant risk factor for DKD development.

Insulin resistance and type 2 diabetes in children and adolescents

In 2009, type 2 diabetes contributed to 23 % of cases of diabetes among US children and adolescents, but was disproportionately more common among ethnic minorities. For example, among 15- to 19-year-olds, type 2 diabetes caused 5.5 % of diabetes among the non-Hispanic Whites, but 34–38 % of diabetes in African–Americans, Hispanics, and Asian Pacific islanders, and 80 % of diabetes among American Indians [1]. Among youths diagnosed with type 1 diabetes, an additional 23 % were insulin-resistant with a higher prevalence of obesity, again more commonly among ethnic minorities [62].

Despite its lower overall prevalence compared with type 1 diabetes, the rise of type 2 diabetes in youth presents a grave public health problem for four reasons: youths with type 2 diabetes have a greater prevalence of risk factors for diabetic complications and mortality, show evidence of poor control of these risk factors, and display more rapid progression of complications, higher rates of mortality and, most alarmingly, unabated progression of complications, despite attempts to control risk factors.

First, compared with type 1 diabetes, young-onset type 2 diabetes is more frequently associated with several cardiovascular and mortality risk factors, even at diabetes onset. Hypertension is present in 24 % of youths with type 2 vs 6 % of those with type 1 diabetes [63]. Dyslipidemia is also more common among youths with type 2 diabetes: an LDL >130 is present in 24 % of children and adolescents with type 2 vs 19 % of type 1 diabetes; an HDL < 40 is present in 44 % vs 12 % of youths with type 2 and type 1 diabetes respectively [64]. Children and adolescents with type 2 diabetes were also much more commonly obese (79 %) than those with type 1 diabetes (13 %) [65]. Even more ominous is the considerable prevalence of early markers of cardiovascular disease at, or soon after, diabetes onset in youths with type 2 diabetes: microalbuminuria was observed in 22 % vs 9 % of children and adolescents with type 2 vs type 1 diabetes [66]. Left ventricular hypertrophy was found in 22 % of youths with type 2 diabetes up to 3 years after diagnosis [67].

Second, risk factors appear to be less well controlled in children and adolescents with type 2 diabetes, compared with children with type 1 diabetes or even adults. Poor glycemic control, defined as an HbA1c ≥9.5 %, was reported in 27 % of youths with type 2 diabetes, compared with 17 % of those with type 1 diabetes and more commonly observed among ethnic minorities [68]. Only 32 % of youths with type 2 diabetes who had hypertension were aware of their diagnosis and only 6.6 % received treatment [63]. Among youths with dyslipidemia, an even smaller fraction (1 %) received treatment [64].

Third, youths with type 2 diabetes see earlier and more rapid progression of DKD, not only compared with those with type 1 diabetes [69–71], but also adults with type 2 diabetes [72]. In countries where the incidence of DKD among youths with type 1 diabetes has declined over the past two decades; DKD due to type 2 diabetes has not [69]. This more rapid progression of DKD in type 2 diabetes is at least partly due to the coexistence of other risk factors, including insulin resistance [73]. Consistent with a higher burden of comorbidities and complications, young-onset type 2 diabetes was found to be associated with greater risk of mortality at a younger age and with significantly shorter disease duration than type 1 diabetes, despite comparable glycemic control [74]. Consequently, American youths with type 2 diabetes have a life expectancy 15 years shorter and a markedly reduced quality of life owing to the development of severe and morbid diabetic complications by their 40s [75].

Perhaps most concerning were the recent reports from the Treatment Options for type 2 Diabetes in Adolescents and Youth (TODAY) study, suggesting that young-onset type 2 diabetes and its complications may be much more aggressive and treatment-resistant than its adult counterpart. TODAY was a multicenter trial of 699 adolescents 10–17 years old, with type 2 diabetes of less than 2 years’ duration, BMI ≥ 85th percentile, HbA1c ≤8 % on metformin, controlled blood pressure ,and creatinine clearance >70 ml/min, who were randomized to metformin alone, metformin and rosiglitazone, or metformin and intensive lifestyle management [76]. Despite good compliance with medications, after a median of 1 year, 46 % of participants had failed glycemic control and had to start insulin. This was associated by a much more rapid loss of β-cell function in this cohort than seen in adults [77]. After 4 years’ follow-up, the prevalence of microalbuminuria and hypertension rose by 3-fold. Hypertension was associated with male gender and an increase in BMI, and appeared to be resistant to treatment, requiring multiple medications in more than one-third of participants initially treated with one agent [78]. Dyslipidemia, defined as an LDL ≥130 mg/dL or use of LDL-lowering therapy, increased from 4.5 to 10.7 % over 3 years’ follow-up, while elevated inflammatory markers (hsCRP, PAI-1, and homocysteine) continued to increase over time and were not affected by lipid-lowering therapy [79].

Key summary points

  • In 2009, diabetes affected more than 190,000 children and adolescents in the US, where its prevalence is projected to increase two-fold or more by 2050.

  • Most (75–95 %) proteinuria in children and adolescents has benign causes (orthostatic and transient proteinuria), which have to be ruled out before further evaluation for micro- or macroalbuminuria.

  • Using random albumin or protein to creatinine ratio may overestimate albuminuria in younger and smaller children, particularly females in whom timed urine albumin excretion may be more appropriate.

  • Microalbuminuria, defined as a urine albumin excretion of 30–299 mg/day (or albumin to creatinine ratio 30–299 mg/g), is the earliest available sign of predisposition to DKD. Microalbuminuria can regress, remain unchanged or progress to macroalbuminuria.

  • Risk factors for microalbuminuria are high A1c, blood pressure, LDL, triglycerides, high normal urine albumin excretion, obesity, and smoking. Microalbuminuria is more likely to regress with improved control of risk factors (A1c, blood pressure, smoking cessation).

  • Macroalbuminuria, a urine albumin excretion ≥300 mg/day or albumin to creatinine ratio ≥300 mg/g, signals more severe kidney injury, but is also more likely to progress slowly with control of risk factors (A1c, blood pressure, smoking).

  • Significant DKD can be present, even in the absence of elevated urine albumin excretion, emphasizing the need for optimal risk factor control, even in the absence of albuminuria.

  • Earliest DKD structural changes in the kidneys appear as early as 1.5–5 years after type 1 diabetes onset.

  • Historically, type 1 diabetes was the predominant form of diabetes in children and adolescents. With the rise of childhood obesity, the prevalence of type 2 diabetes has increased to 23 % of cases of diabetes in US children and adolescents, and 34–80 % of diabetes among US ethnic minorities.

  • Young onset type 2 diabetes is associated with more co-morbidities and risk factors, faster progression of DKD and other complications and higher mortality than type 1 diabetes and may be more aggressive and treatment-resistant than type 2 diabetes in adults.

Multiple choice questions (answers are provided following the reference list)

  1. 1.

    A 15-year-old boy suffering from type 1 diabetes of 10 years’ duration has a urine albumin excretion of 280 mg/day based on a timed mid-day urine sample. What is the next best step?

    1. a)

      Start an angiotensin converting enzyme inhibiting agent

    2. b)

      Obtain a repeat urine sample to confirm microalbuminuria (random albumin/creatinine ratio is adequate)

    3. c)

      Obtain a repeat first-void urine sample to exclude orthostatic proteinuria

    4. d)

      Determine if the patient had fever, dehydration, seizure, intense exercise or any other causes of transient proteinuria prior to collection of the urine sample

    5. e)

      c and d

  1. 2.

    A 7-year old girl suffering from type 1 diabetes and normal BP presents with a urine albumin/creatinine ratio of 150 mg/g in a first-void urine sample and you are assured by the parents that she has had no recent fever, seizure, dehydration or any potential causes of transient proteinuria. What is the next best step?

    1. a)

      Start an angiotensin converting enzyme inhibiting agent

    2. b)

      Repeat her urine albumin excretion in a timed urine collection to estimate her daily creatinine excretion.

    3. c)

      Optimize risk factors (A1c, blood pressure); she is too young to receive angiotensin converting enzyme inhibiting agents

    4. d)

      Repeat her urine albumin excretion in another random sample

    5. e)

      There is no need to follow her albumin excretion. DKD does not develop this soon after diabetes onset.

  1. 3.

    A 17-year-old Samoan girl suffering from type 2 diabetes for 7 years presents with BP 150/95, A1c 10, BMI 32, LDL 190, and urine albumin excretion of 200 mg/day. What is the next best management step to reduce her risk of progression to advanced DKD and other diabetic complications?

    1. a)

      Controlling A1c

    2. b)

      Controlling blood pressure

    3. c)

      Encouraging weight loss

    4. d)

      Treating macroalbuminuria (RAS inhibition)

    5. e)

      Treating dyslipidemia

    6. f)

      Work-up of secondary hypertension

    7. g).

      All of the above

  1. 4.

    A 11-year-old boy presents with type 1 diabetes (for 6 years), poor glycemic control (A1c 11), poorly controlled hypertension (BP 165/94), 2 g/day proteinuria, hematuria, and GFR ∼50. What is the best management?

    1. a)

      Control risk factors (A1c, BP). This is most likely rapidly progressing DKD.

    2. b)

      Initiate work-up for other causes of kidney disease

    3. c)

      Initiate work-up for secondary hypertension

    4. d)

      Obtain a repeat first-void urine sample to exclude orthostatic proteinuria

    5. e)

      b and c

  1. 5.

    A 16-year-old Hispanic girl with recently diagnosed type 2 diabetes presents with poorly controlled diabetes (A1c 9.8), hypertension (BP 163/98), dyslipidemia (LDL 189), obesity (BMI 38), and an elevated urine albumin excretion (130 mg/g). Which of the following statements is incorrect?

    1. a)

      Her albuminuria must be benign or have other causes of kidney disease because it takes at least 1 year to even see the first structural changes of diabetes and several more years to develop albuminuria.

    2. b)

      Given her young age, her hypertension requires work-up to exclude causes of secondary hypertension

    3. c)

      She is at high risk of developing and rapidly progressing overt DKD despite her short diabetes duration because of many coexisting risk factors, including insulin resistance, hypertension, poor glycemic control, and dyslipidemia.

    4. d)

      She is at high risk of glycemic failure on oral agents and requires transition to insulin.

    5. e)

      Her presentation is consistent with early DKD as young onset type 2 diabetes is often associated with microalbuminuria and other risk factors at diagnosis.

    6. f)

      In this young person with a high risk of early mortality, weight is one of the most significant risk factors to modify.